One of the most important contributions of language and sexuality research over the past 20 years has been to counter reductive beliefs that there exists a binary distinction between homogenous homosexual versus heterosexual (or ‘lesbian/gay’ versus ‘straight’) subjectivities. Instead, work in the field has provided compelling evidence that the ways in which sexuality is experienced, constructed, and perceived are inextricably linked to other categories of social life (e.g. gender, race, nation, religion, social class) and to the norms, expectations, and constraints of local interactional contexts (e.g., audience, setting). In this chapter, I review research on how these intersections between sexuality and other dimensions of lived experiences are manifested in variable linguistic practice. I focus in this review on two broad areas of scholarship in variationist sociolinguistic research: work that has investigated the diversity of practices subsumed under the umbrella label ‘LGB’ or ‘queer’ and work that examined instead the ways in which sexuality and other relevant categories mutually constitute one another. I argue that research in both of these traditions has contributed to the destabilising of binary understandings of sexuality, and has forced a fundamental rethinking of what we mean when we speak of sexuality-linked linguistic practice. I illustrate these arguments with a brief discussion of the interrelationships between language, sexuality, and gender among a group of lesbians in London.
The first question I am often asked by non-linguists about my work is whether it is true that gays and lesbians speak differently than heterosexuals. My answer to this question is inevitably that ‘it’s complicated’. It is complicated because sexuality is not a homogenous category, such that we can speak about what all ‘gays and lesbians’ do (in the same way that we cannot speak about what all of those who are not gay or lesbian do). It is also complicated because there is never a one-to-one correspondence between language and social identity. Just because someone identifies as a lesbian does not mean that she will necessarily speak in a particular way any, some, or all of the time. And just because someone speaks in a particular way any, some, or all of the time does not mean that she is a lesbian. This is not to say that there is no relationship between sexuality and language. There clearly exist ways of speaking that are stereotypically associated with different sexual identity categories, and research over the years has been able to demonstrate that, in some communities at least, certain linguistic features appear more (or less) frequently in the speech of gays or lesbians than among their heterosexual counterparts. But identifying surface-level correlations like these between linguistic features and identity categories can only ever tell us a partial, and often inaccurate, story. The reality of the connection between language and sexuality is a more complicated one.
In this chapter, I outline some of the ways in which we can approach this complexity in our research. Focusing on research within variationist sociolinguistics, I review certain key studies in the area of language and sexuality and describe the different methodological tools that have been developed for investigating how variable patterns in language use participate in the construction of sexuality. In order to do so, I also review some important theoretical concepts that variationist sociolinguists draw on to frame their analyses, including indexicality (e.g. Ochs 1992), intersectionality (e.g. Crenshaw 1989), and performativity (Butler 1990, 1993). Finally, I close with a brief illustration of some of these theories and methods ‘in action’ by discussing pitch variation within a community of lesbians in London (based on data collected and analysed by Lawrence 2014). By the end of the chapter, the reader will understand why it is problematic to treat sexuality as a binary characteristic (i.e. ‘gay’ versus ‘not gay’) in linguistic research. I will also introduce some of the theoretical tools that can be used to move away from this type of binary thinking, and will provide an illustration of variationist research on sexuality that places these tools at the centre of its analysis. Before getting to that, however, I first begin in the next section with a brief overview of the study of sexuality from a variationist perspective. This is important because it helps to contextualise the methodological discussion that follows.
We can divide variationist research on sexuality into three basic types based on the kind of theoretical approach the research takes (see Levon and Mendes 2016 for a more detailed discussion of this taxonomy). The first type is research using what we can describe as a correlational approach to the subject (Eckert 2012: 94). Correlational research assumes that the language practices we observe are directly determined by some element of the underlying social structure. In the case of sexuality, the assumption would be that there exists such a thing as the ‘gay and lesbian community’, and that membership in this community gives rise to a set of distinctive social and linguistic practices. This is the perspective that was adopted by the earliest variationist research on language and sexuality, where studies focused on identifying the specific phonological, lexical, or discursive features that were believed to define the unique experiences of lesbians and gays (see Jacobs 1996; Kulick 2000; Queen 2007 for full reviews). Moonwomon (1985), for example, examined pitch differences in the speech of lesbians and heterosexual women in the US, and found that lesbian speakers had lower mean pitch levels and lower overall pitch ranges than the heterosexual speakers she studied. Similarly, Leap (1993, 1996) identified certain conversational features that he argued were specific to interactions among American gay men. In both cases, the authors suggested that it was the speakers’ identities as gays and lesbians that caused them to speak in distinctive ways.
Beginning in the 1990s, a number of developments challenged the assumptions underlying this correlational model of language and sexuality. From outside linguistics, Butler (1990, 1993) popularised the notion of performativity, or the belief that identity is not the cause of observed behaviour, but rather its result. In other words, we do not act in a certain way because we are lesbian; we are socially constituted as lesbians because of how we act. Within linguistics, this concept of performativity was refined by Ochs’ (1992) definition of indexicality. Arguing that the link between a linguistic feature and a social category is rarely a direct one, Ochs claimed that features in language index particular stances, acts, and activities that are then ideologically linked to salient social categories. According to this account, tag-questions, for example, do not directly index the category ‘woman’. Instead, they are taken to signal a stance of ‘uncertainty’, which is itself linked to stereotypes of womanhood. Together, performativity and indexicality gave rise to a new type of constructionist research on language and sexuality. Rather than attempting to catalogue a characteristic lesbian or gay way of speaking, research in this paradigm sought to identify how people use language to actively construct sexual personae. Barrett (1995, 1997), for example, described how African American drag queens in Texas juxtapose features that are stereotypically linked to both white women in the US South and African American men in order to variably construct themselves as gay men, as African Americans, and as drag queens. The crucial point is that research in this paradigm did not assume that individuals would speak in a particular way because they were lesbian or gay, but rather that speakers draw on the indexical power of language to construct their sexualities through linguistic practice (see Cameron and Kulick 2003; Livia and Hall 1997 for more detail).
In the same way that the constructionist approach challenged certain underlying tenets of the correlational perspective, theoretical developments in the mid-2000s began to critique the assumptions of constructionist work. The crux of this critique was the assertion that the meaning of variation in constructionist research was very often still reduced to the cultural formations it was used to construct. ‘Gay language’, for example, though not necessarily viewed as an inherent correlate of gay identity, was nevertheless understood as that set of linguistic features used to construct a gay ‘self’. Scholars like Eckert (2008, 2012) argued that this was problematic because it fails to recognise the multiple possible meanings that a particular variable can have (and the multiple functions a speaker can use a variable to perform). Instead, Eckert developed an emergentist framework for analysing sexuality-linked variation, in which the focus is on understanding the more local actions speakers use variation to perform. Podesva (2007), for example, discusses how a man he calls Heath draws on the ability of falsetto voice to index ‘expressiveness’ to construct distinct personae in different settings. When at a barbecue with his friends, Heath uses falsetto to help adopt an expressive stance (i.e. as a person who explicitly expresses his thoughts and emotions) that, in conjunction with other relevant features, results in the creation of a ‘diva’ style. At the medical clinic where he works, in contrast, Heath’s use of falsetto serves instead to index expressivity as part of the creation of a ‘caring doctor’ persona. Crucially, while the same linguistic feature is deployed in both contexts, the ultimate meaning of the feature, in terms of the persona it helps to construct, is context-dependent. Moreover, while he acknowledges that the perception of ‘gay identity’ may emerge from Heath’s use of falsetto, Podesva argues that this is in a sense a potential by-product of Heath’s use of the feature and that the primary motivation behind Heath’s observed practice is the construction of situationally relevant personae. Research on sexuality within an emergentist paradigm thus does not ignore the fact that identities may result from variation. But it does not assume that ‘doing identity’ is a speaker’s ultimate aim. Instead, emergentist research looks first at what immediate interactional goals speakers are trying to achieve and only then attempts to explain how the linguistic attainment of those goals may link to the emergence of salient social identities in interaction. In this respect, variationist research within the emergentist paradigm shares certain similarities with other sociolinguistic traditions for the study of gender (e.g. interactional sociolinguistics, discourse analytic perspectives), though as variationist research it remains committed to identifying systematic (i.e. quantitative) distributional patterns of language use.
The review in the preceding section is by no means an exhaustive overview of variationist research on sexuality (for that, see Queen 2013, 2014). It does, however, give a sense of the kinds of theoretical developments that have taken place, going from seeing language as the result of identity (correlational), to seeing language as a tool with which to ‘do’ identity (constructionist), to finally seeing language as an instrument for accomplishing local interactional goals through which embodied identities also emerge (emergentist). These developments have allowed us to provide much more nuanced analyses of the relationship between language and sexuality than would otherwise have been possible. Yet despite these advances, and as I have argued elsewhere (Levon 2015), our analyses of sexuality have tended to remain framed in terms of unitary categories of experience (like ‘gay’ or ‘lesbian’). Thus, while we have developed sophisticated accounts of how particular linguistic forms take on sexualised meanings and of how those meanings are then recruited by speakers in interaction, we have been somewhat less attentive to the fact that those sexualised meanings are also simultaneously gendered, classed, raced, and age-, culture-, and region-specific. Intuitively, we know that individuals do not experience life through the prism of a single identity category. Each of us maintains multiple affiliations and identifications, and these different components all influence our own experiences of self. It is therefore both theoretically and empirically inaccurate to conceive of sexuality in terms of simple binary contrasts (e.g. homosexual versus heterosexual) since each half of that binary itself encompasses a huge range of diverse stances, orientations, and behaviours.
A useful way for dealing with the complexity of sexuality as a lived experience is through the prism of intersectionality theory. A term originally coined by legal theorist Crenshaw (1989), and itself drawing on a long history of early Black feminist theorising (e.g. Anthias and Yuval-Davis 1983; Davis 1981; hooks 1981; Hull, Scott and Smith 1982), intersectionality refers to the idea that lived experience cannot be defined in terms of membership in a single identity category (e.g. ‘woman’, ‘black’). Rather, both our own, inner understandings of self and the kinds of access, opportunity, and treatment we receive are the product of multiple and intersecting systems of social classification. Because of this, an intersectionality perspective argues that no one analytical category is sufficient if we are to provide a rigorous analysis of the social practices we observe. Instead, we must investigate how a multiplicity of categories come together in the formation of individual subjectivity.
Since its popularisation in the early 1990s, intersectionality has become the dominant construct for theorising identity across the humanities and social sciences, including in fields as diverse as gender studies, sociology, philosophy, and politics (Collins and Bilge 2016; Davis 2008; Lutz, Vivar and Supik 2011; McCall 2005). However, this does not mean that intersectionality is a unified social theory. There are numerous debates in the relevant literature about the framework’s key concepts and about how to methodologically implement an intersectional perspective (cf. for example, Cho, Crenshaw, and McCall 2013; Choo and Ferree 2010; Weldon 2008). For the purposes of the current chapter, we can nevertheless identify two main assertions that all forms of intersectional analysis maintain. The first assertion is that if we assume that lived experience is ultimately intersectional, then we must place this intersectional complexity at the heart of our analyses. In practical terms, we can accomplish this by engaging in what Matsuda (1991: 1189) describes as ‘asking the other question’: When I see something that looks racist, I ask, ‘Where is the patriarchy in this?’ When I see something that looks sexist, I ask ‘Where is the heterosexism in this?’ When I see something that looks homophobic, I ask ‘Where are the class interests in this’.
When I see something that looks racist, I ask, ‘Where is the patriarchy in this?’ When I see something that looks sexist, I ask ‘Where is the heterosexism in this?’ When I see something that looks homophobic, I ask ‘Where are the class interests in this’.
A deceptively simple method on the surface (Davis 2008), ‘asking the other question’ forces us to go beyond analyses in terms of categories in isolation to consider how these categories intersect with equally important others.
The other main assertion of intersectional research is that categories do not only intersect but mutually constitute one another (Choo and Ferree 2010). This is a somewhat more contentious aspect of the theory (Crenshaw 2011) since it argues that intersections are not to be viewed as simple points of contact between two (or more) already existing categories (Shields 2008). Instead, the argument is that intersections are themselves formative of the categories in question. In other words, the idea of mutual constitution suggests that constructs such as class, race, and gender do not exist as entities unto themselves. Rather, they crucially depend for their meaning on their relationship to the other categories with which they intersect. Thus, there is no ‘gender effect’ to be discovered and analysed in a dataset; there is only the effect of gender in relation to class, race, etc. This is a strong claim, and there is debate in the literature as to whether such a strong position is necessary. Without getting in to the details of this discussion, it is nevertheless important for us to note how the concept of mutual constitution pushes the envelope of intersectional analysis further, encouraging us to move beyond seeing things in terms of compartmentalised categories to focus instead on the relationship between categories as formative of lived experience.
These two assertions – that lived experience is intersectional in nature and that categories mutually constitute one another – correspond to two main avenues for intersectional research in sociolinguistics and related fields. In this section, I very briefly review the two approaches and cite some key examples of work in these areas. For fuller examples of this type of work, including both non-variationist studies and studies on topics other than sexuality, see Levon (2015) and Levon and Mendes (2016).
In her well-known discussion of methods for doing intersectionality research, McCall (2005) defined what she terms ‘intra-categorical’ intersectionality, or intersectionality research that focuses on the diversity of more specific articulations of identity that exist within a given category. For example, an intra-categorical approach would critique a category such as ‘lesbian’, arguing that we need to examine the different ways in which the label ‘lesbian’ can be experienced and lived (e.g. ‘black lesbian’, ‘butch lesbian’, ‘middle-class lesbian’, etc). This type of intra-categorical approach is central to what intersectionality is about, and it serves a crucial theoretical, empirical, and political role in bringing to light a variety of lived experiences that would otherwise be obscured (Crenshaw 1989, 1991; Morgan 2004). The intra-categorical approach is also the most common form of intersectionality research within the variationist paradigm (Levon 2015).
I explore intra-categorical intersectionality in some of my own prior research in Israel (e.g. Levon 2009, 2010, 2011). In that work, which was based on a year-long sociolinguistic ethnography of members of 12 gay and lesbian activist groups from across the Israeli political spectrum, I demonstrate how groups of Israeli gays and lesbians use pitch differently as a way of signalling their distinct political affiliations. Those speakers who align with more centrist political discourses about Israeli society and Israeli nationalism (and whom I refer to as members of the ‘mainstream’ group) adopt mean pitch levels that align with dominant Israeli gender norms. In contrast, those who reject normative Israeli conceptualisations of the nation (who I describe as members of the ‘radical’ group) engage in linguistic practices that flout Israeli sociolinguistic gender norms. In this way, I illustrate how language varies among lesbians and gays in Israel as a function of their broader political beliefs and alignments. Podesva and Van Hofwegen (2016) make a similar argument in their exploration of /s/ variation among lesbian and gay speakers in rural California. There, they demonstrate that rural gay male speakers, for example, produce backer articulations of /s/ than gay men in nearby San Francisco do. Podesva and Van Hofwegen argue that this difference is due to the social context in which these men live, and the strong pressure to conform to a more normative masculine style (which includes the use of backer articulations of /s/). The language used by gay men in rural California is thus not only related to their sexuality (their articulations of /s/ are significantly fronter than their heterosexual rural male counterparts). Crucially, it is also affected by their overall orientations to ‘country’ versus ‘town’ culture. In both of these studies (and others like them), the main goal of the analysis is to understand how other aspects of the social context affect the lived experience of (homo)sexuality, and thus demonstrate the impossibility of thinking in terms of a simple homosexual versus heterosexual binary.
Variationist research has also examined intra-categorical complexity from the perspective of perception research. Work on this topic attempts to identify the ways in which listeners’ perceptions of sexuality may be affected by other socially salient aspects of a speaker’s voice. Pharao, Maegaard, Møller, and Kristiansen (2014), for example, discuss how the identification of a voice as sounding ‘gay’ in Danish depends on its perceived ethnicity, such that non–white-sounding voices are never perceived as ‘gay’ even when they contain the same sexuality-linked linguistic features as white-sounding voices (see also Maegaard and Pharao 2016; Pharao and Maegaard 2017). Similarly, research by Mendes (2016) considers how the use of non-standard plural noun-phrase agreement in Brazilian Portuguese acts as a marker of femininity in certain voices, but not in others. Like the work by Pharao and colleagues, Mendes’ results demonstrate that we cannot make broad claims about what ‘sounds gay’ (or ‘lesbian’) in speech. Instead, we need to look at the specific linguistic context in which relevant features occur. Finally, a growing body of work has made the same type of argument in relation not only to linguistic contexts, but also to social ones. Studies by Drager (2011) in Hawaii, Mack (2010) in Puerto Rico, and Rácz and Papp (2016) in Hungary, among others, have all shown that there are specific cultural differences in what features listeners pick up on when making judgements about a speaker’s sexuality. As in the production studies described above, research on intra-categorical complexity in perception serves to highlight the diversity of ways in which one can be (or sound) lesbian or gay. In doing so, it pushes us to embrace – rather than ignore – this complexity in our own work.
The intra-categorical method outlined above succeeds in responding to the first claim of intersectionality theory, namely that lived experience is itself intersectional and so we must examine how multiple categories come together to influence observed social practice. It does not, however, force us to consider how these different categories constitute one another. For example, in my own work on Israel I was able to describe how the ways in which people embody identities like ‘gay’ or ‘lesbian’ are influenced by their other social identifications and affiliations (such as their political beliefs, and particularly how they saw the relationship between sexual politics and national politics more broadly). In doing so, I was able to identify a point of mutual influence between the categories ‘sexuality’ and ‘political beliefs’, but I stopped short of describing how these two categories may in fact define one another (such that sexual identity categories are politicised, and political beliefs are sexualised). To do this, we need to open up our analytical gaze to look across categories, and allow linguistic features that normally ‘mean’ one thing to be recruited in the service of another. In other words, one way to embed the idea of mutual constitution in our analyses is to examine how a feature that we normally think of as related to one identity category (gender, for example) is used by speakers to help construct a different identity category (e.g. region). I illustrate below what I mean with two brief examples.
Podesva (2011) describes how a speaker named Regan, a 31-year-old gay man in Northern California, varies his use of certain vocalic features that are all part of the California Vowel Shift (CVS). The CVS is a coordinated change in a number of different vowels that is currently underway in California. As its name implies, the CVS is most saliently associated with region, such that speakers with CVS-shifted vowels are heard as ‘Californian’. In his study, Podesva examines Regan’s use of CVS vowels in three contexts: out with friends at a gay bar, in a meeting with his supervisor at work, and at a casual dinner with a friend. Podesva demonstrates that Regan uses the most advanced realisations of CVS features when he is out at the bar, the most conservative with his supervisor at work, and an intermediate level with his friend at dinner. Podesva does not argue, however, that this finding means that the CVS features index a gay identity. Rather, drawing on the principles of mutual constitution, Podesva suggests that elements of the CVS system have become enregistered (Agha 2007) as markers of the ‘fun’, ‘laidback’, and ‘carefree’ lifestyle that is stereotypical of Californians. Podesva argues that when he is out at the bar, Regan draws on this regional meaning of the features to help him construct a fun, carefree, and laidback ‘gay partier’ persona. In other words, Regan recruits a regional meaning to help him construct a sexual one, thus demonstrating how, in Podesva’s research at least, the categories ‘Californian’ and ‘gay’ mutually constitute one another. A similar example can be found in Ilbury’s (2019) discussion of stylised uses of African American Vernacular English (AAVE) among young British gay men on Twitter. In that study, Ilbury describes how men who would normally not use AAVE in their everyday speech employ tokenistic elements of the variety in their tweets as a way of portraying a ‘sassy queen’ persona. Like Regan, the men in Ilbury’s study recruit a feature that has come to be stereotypically associated with one social category (in this case, African Americans) and use it to help construct another (‘sassy queen’). That they do so relies on various reductive and racist assumptions about the speech of African Americans and ignores the power imbalances inherent in the appropriation of AAVE by white speakers (cf., e.g. Hill 2008). Nevertheless, it also illustrates how for these individuals racialised conceptualisations of what it means to be ‘sassy’ partially constitute what it means to be ‘gay’.
Both of these examples illustrate one way of exploring the mutual constitution of categories by using variationist methods. In both cases, a linguistic feature that we would normally think of as meaning one thing (‘Californian’, ‘African American’) is strategically deployed by speakers to mean something else (‘diva’, ‘sassy’). The reason the features can be used in this way is because it is possible to ideologically elaborate a first-order indexical meaning to create a new one, such that a feature meaning Californian (first-order) comes to mean things that we associate with California (e.g. laid-back, carefree) (second-order). Once that ideological elaboration has taken place, the feature is then available to be used to express this new meaning (so that if I want to sound ‘laid-back’, I can use that California feature). What is important for the present discussion is that tracing how these ideological developments happen – what new meanings are created and how features are put to use in creative ways – provides us with a window into the internal composition of an identity: it allows us to see the gendered, regional, classed, racialised, and other dimensions that comprise it. This is why this type of method is helpful for intersectional analysis.
In the previous sections, I outlined some of the basic principles of intersectionality theory as it has been applied to variationist research and summarised a number of studies that have engaged with the principles of intersectionality in their analyses. To better enable the reader to anchor an intersectional perspective in their own work, in this section I provide a more detailed discussion of an examination of pitch variation in the speech of a group of young lesbians in London (Lawrence 2014). In my presentation of this work, I highlight the methodological steps the author took to investigate the relevance of intersectionality to her findings. This is not intended as a prescriptive rulebook for how to conduct a study of this kind. Instead, my aim is to provide a guide for thinking through questions of intersectionality in relation to a body of data and an illustration of how variationist techniques can be brought to bear on these issues.
Lawrence (2014) examined various aspects of language use within a lesbian friendship group in London. Her goal in doing so was to examine the extent to which a salient stereotypical divide between ‘butch’ versus ‘lipstick’ articulations of lesbian identity influenced the ways in which the women understood their sexualities and the kinds of social practices in which they engaged. Lawrence’s study is thus a clear example of an intra-categorical approach to intersectionality. What she is interested in is the diversity of lesbian experiences across individuals, even among members of the same friendship group. In this sense, Lawrence’s work helps us to move beyond simplistic binary assumptions about ‘lesbian’ versus ‘non-lesbian’ ways of speaking, and instead allows us to explore the various other factors that together with sexuality help constitute individual subjectivity.
The so-called butch–lipstick dichotomy (sometime also termed butch–femme) is a stereotypically very salient one within lesbian communities (and, arguably, within society at large). The labels themselves refer to two imagined archetypes of lesbian identity that occupy opposite positions on a spectrum of gendered and sexual presentation (Eves 2004; Munt and Smyth 1998). While popularly viewed as simple embodiments of more ‘masculine’ versus more ‘feminine’ styles, respectively, scholars have argued that butch and lipstick identities serve as powerful forms of resistance to heterosexual (and heterosexist) norms, allowing women to create a lesbian aesthetic within which to reimagine gender (e.g. Case 1988). Abstracting away from their political potential, the concepts ‘butch’ and ‘lipstick’ denote opposing positions within a complex economy of embodied practices (such as dress and demeanour), emotional styles, and sexual and romantic roles. In investigating the butch–lipstick dichotomy, Lawrence (2014) is not aiming to describe distinct and homogenous ‘butch’ versus ‘lipstick’ styles of speech. Rather, she attempts to understand how the women she observed orient to these concepts variably in their daily lives, and how they draw on language as a resource for materialising these orientations in interaction (see Jones 2011, 2012, 2014 for another discussion of butch–lipstick dynamics among lesbians in the UK).
To achieve these goals, Lawrence (2014) focuses on pitch variability among the women. She considers a number of different acoustic properties of spoken pitch, including mean pitch (the average pitch level in an utterance) and pitch range (the overall span of pitch across an intonation phrase). These are both features that have been extensively studied in the literature on language and sexuality, though the majority of that work has focused on (gay) men (e.g. Gaudio 1994; Levon 2006, 2007; Smyth, Jacobs and Rogers 2003). In the studies that have examined lesbian speech (e.g. Moonwomon 1985; Pierrehumbert, Bent, Munson, et al. 2004), results with respect to pitch have been either contradictory or inconclusive (see also Munson and Babel 2007). Nevertheless, there is a strong stereotype in the UK (and elsewhere) that butch women will have lower mean pitch levels and narrower pitch ranges, while lipstick women will have higher mean pitch levels and wider pitch ranges (Queen 1997; Van Borsel, Vandaele and Corthais 2013; Waksler 2001). These stereotypes in relation to the butch–lipstick dichotomy are analogous to the stereotypes that circulate with regard to the speech of men versus women (as well as the speech of more ‘masculine’ versus more ‘feminine’ men) (e.g. Henton 1989, 1995). For this reason, variation in pitch is a salient feature to examine among the women in question.
Data are drawn from the speech of 5 women between the ages of 20 and 27 (mean age: 23) living in the London area who were all members of a London university’s LGBT society. Lawrence observed and recorded the women both in their casual interactions as a group (i.e. when out together at a local pub) and in individual sociolinguistic interviews she conducted with each of them. The analysis of pitch below is based on speech from these interviews. The interviews themselves were semi-structured, an approach which encourages natural speech and allows interviewees to have some control over the direction and content of their talk. This helps to encourage more relaxed, informal conversation, and enables interviewees to speak in more detail about topics they consider important while simultaneously ensuring comparability across the dataset (e.g. Schilling 2013). All interviews included talk about the interviewees’ social backgrounds, schooling, current work/studies, leisure activities, media consumption habits, and opinions about lesbian life. Once recorded, interviews were segmented into intonational phrases (IPs) and the mean pitch level and pitch range of each IP was measured. In addition to these linguistic measurements, IPs were also coded for whether they were taken from talk on ‘gay’ or ‘non-gay’ topics. ‘Gay’ topics include the women’s personal histories of ‘coming out’, their participation in a lesbian community or ‘scene’, and their opinions about current sexual politics. The division between ‘gay’ and ‘non-gay’ topics allows Lawrence to investigate whether there are patterns of topic-linked intraspeaker variation in the dataset (see, e.g. Levon 2009). Finally, each of the five speakers was also assigned a value on a butch–lipstick index, based on Lawrence’s observations and ethnographic knowledge of the participants (for further details of this index, see Lawrence 2014).
Initial results with respect to mean pitch levels are presented in Figure 3.1. Participants are ordered in terms of their scores on the butch–lipstick continuum, with those rated as more ‘butch’ on the left side of the plot and those rated more ‘lipstick’ on the right side. We see that the five speakers divide into roughly three groups: Bow with the lowest mean pitch level (average: 156.4 Hz), followed by Jane and R (average: 186.9 Hz and 184.2 Hz, respectively), and finally Meredith and Lizzie (average: 202.8 Hz and 200.1 Hz, respectively). What this pattern shows is that, in general, the more ‘lipstick’ a participant’s embodied style (i.e. the more it conforms to traditional notions of femininity), the higher her average mean pitch levels. This is confirmed by quantitative regression analysis, which demonstrates that there is a positive correlation between placement on the butch–lipstick continuum and mean pitch (p = 0.02). This finding is important because it illustrates that even though all the women self-identify as ‘lesbians’, their different positionings in relation to the constructs of ‘butch’ versus ‘lipstick’ have an impact on their linguistic practices. This is thus a clear example of intra-categorical complexity among the women under investigation.
Figure 3.1 Average mean pitch level in Hz for speakers ordered from left to right by position on the ‘butch’–‘lipstick’ continuum.
Further investigation of the women’s speech indicates that the pattern observed in Figure 3.1 is actually more complex than it appears. In Figure 3.2, mean pitch levels are again shown by speaker, but this time they are also divided by speech topic (‘gay’ versus ‘non-gay’). There, we see that the correlation between orientations to ‘butch’/’lipstick’ and mean pitch varies depending on the topic of conversation. For gay topics (the dark bars in Figure 3.2), we again see the relationship evident in Figure 3.1, with the dividing into three groups (Bow, Jane, and R, Meredith and Lizzie) and a general increase in mean pitch levels the more a speaker orients to the ‘lipstick’ end of the continuum. However, on non-gay topics (lighter bars in Figure 3.2), no such relationship exists. Instead, all of the women are shown to have roughly similar mean pitch levels. This is interesting for a number of reasons. First, it demonstrates that the women vary how they speak in relation to the topic of their talk. For this reason, we cannot speak of a characteristic ‘lesbian’ way of speaking, or even of a characteristic ‘butch’ or ‘lipstick’ way of speaking, since how an individual speaks clearly depends on other elements on the speech context (like topic). It is also potentially meaningful that the pattern of interest with respect to mean pitch obtains when the women are speaking about ‘gay’ topics. This could indicate that the variability in mean pitch that we find is in some way related to women’s constructions and presentations of particular sexual selves. Bow, for example, may strategically lower her pitch on gay topics to enact a more ‘butch’ persona, whereas Lizzie may strategically raise hers in order to enact a more ‘lipstick’ one. While we would want further evidence to support this kind of claim, the quantitative pattern in Figure 3.2 is consistent with an analysis in which mean pitch is a symbolic resource that the women can deploy to enact specific intersectional selves at particular interactional moments (for more on variation as a symbolic resource for enacting a particular interactional self, see, e.g. Bucholtz 2009; Eckert 2008; Levon 2011; Schilling-Estes 2004; Sharma and Rampton 2015).
Figure 3.2 Average mean pitch level in Hz for speakers ordered from left to right by position on the ‘butch’–‘lipstick’ continuum and divided by speech topic.
The foundational intersectional principle of moving beyond simple binaries was operationalised in Lawrence’s study by first looking at diversity of language use within a group of lesbians (rather than assuming that there is a singular ‘lesbian’ way of speaking that can be contrasted with the speech practices of others). Detailed quantitative examinations of mean pitch among women in the group revealed a correlation between higher average mean pitch and more ‘lipstick’ embodiments of lesbian identity. This correlation allowed Lawrence (2014) to detail the kind of intra-categorical complexity that exists within the group. Further analyses also revealed this complexity was itself dynamic in nature, only emerging in certain interactional contexts (i.e. when speaking on ‘gay’ topics in interviews). This more detailed result points to an understanding of mean pitch as a strategic indexical resource that is recruited by the women as a way of enacting the different types of lesbian genders with which they identify, and so enriching the intersectional understanding of the women under investigation.
It is worthwhile noting that Lawrence’s (2014) analysis does not venture into an exploration of how sexuality or ‘butch’/’lesbian’ personae are mutually constituted by their intersection with other categories. If we wanted to explore this avenue of enquiry, we could, for example, ask why it is that mean pitch is the tool that women use to enact differently gendered selves. In addition to research on pitch and gender/sexuality, prior research has also shown that mean pitch levels are also associated with perceived clarity, intelligence, sophistication, and even height. It could therefore be the case that part of what it means to be a ‘lipstick’ lesbian is to speak in a ‘clearer’ and more ‘sophisticated’ fashion and that one way of doing so is to use an elevated mean pitch. This type of exploration would be one way of considering how categories like ‘butch’ and ‘lipstick’ are mutually constituted by relevant others.
The preceding discussion of a selection of findings from Lawrence (2014) provides a brief, but nevertheless useful illustration of how we can approach one of the foundational principles of intersectionality – intra-categorical complexity – in variationist research on sexuality. Yet it is the other foundational principle of intersectionality – the mutual constitution of categories – that I believe is the next frontier in research on language, gender, and sexuality research. To date, most research within variationist sociolinguistics that has attempted to move beyond binary ways of thinking has tended to do so by adding new subdivisions to the categories of identity we consider (e.g. woman > lesbian woman > ‘butch’ lesbian woman) – a process that Eckert (2014: 530) describes as ‘nesting the terms of the [gender] binary within each side of the binary’. While this work is crucial in highlighting under-researched aspects of lived experience, I argue that its theoretical and methodological potential for overcoming binary modes of thinking is ultimately limited. What is needed is research that examines how binaries are themselves sustained, what social forces conspire to scaffold and reproduce binaries in other domains, and, most crucially, how individuals negotiate, resist, and transform these binaries in situated sociolinguistic practice (see Davis, Zimman, and Raclaw 2014). Only once we have done so do I believe that we will be able to provide a fuller analysis of the complex subjectivities of the individuals we analyse, and of the crucial role of linguistic variation in bringing that complexity to light.