Democratization and Gender-neutrality in English(es)

2.1. Language and Gender Research: An Overview

The concept of “gender” was first used in linguistic studies as a social variable accounting for observed variation (see below, the “survey era”). From that sociolinguistic perspective, gender has nearly always been framed as binary, i.e., including two variants, men and women, and only those two. This operationalization is simplistic, glossing over much of the complexity associated with the notion of gender and the fluidity of gender identity categories, and in the twenty-first century new gender theories (e.g., queer linguistics, see Coates 2013:218-220) provide a more inclusive, and, one could argue, more democratic representation of reality, including people who do not identify as either men or women (e.g., Zimman, Davis & Raclaw 2014; Zimman 2017; Jones & Mullany 2019; Paterson, this issue). In what follows we provide an overview of the main tendencies in the study of language and gender and, unavoidably, gender will most often be used as a binary term, because that is its status in the literature reviewed.

The study of how social meaning emerges combining demographic variables, such as gender, age, or ethnicity, is said to fall into three waves (Eckert 2012). ¹ The first wave (the “survey era”), beginning with Labov’s (1966) New York study, comprised studies which took as their starting-point predetermined social categories (male, female, working class, middle class, etc.) and did not embrace issues related to power and social order (Meyerhoff 2014:89). The second wave turned to ethnographic methods to achieve better descriptions of the local dynamics of variation. Key works include Labov’s (1963) study of Martha’s Vineyard and Trudgill (1972), ² which brought the vernacular to the fore and established the basis for the distinction between overt and covert prestige (Meyerhoff 2014:90). Also crucial to this second wave were concepts such as network and identity (e.g., Eckert 1989). The third wave of variation studies sought to emphasize stylistic practices, and “places speakers not as passive and stable carriers of dialect, but as stylistic agents, tailoring linguistic styles in ongoing and lifelong projects of self-construction and differentiation” (Eckert 2012:97-98).

While gender is considered a relevant variable in these three waves, gender-neutrality is not at the center of those sociolinguistic studies, unlike in feminist linguistics, which has addressed issues relating to gender-neutrality and the presence of sexism in language since Robin Lakoff’s (1975) pioneering work. Lakoff (1975) identifies patterns in the speech regularly used by men and by women, arguing that these differences contribute to male dominance, and that men tend to use language to dominate women. This view was complemented by Tannen (1990), whose study of gender differences concluded that some miscommunications can be explained by the diverging speech styles of males and females (“genderlects”). Like early sociolinguistic work in general, these studies, too, adopt a binary view of gender, which is increasingly recognized as limited and exclusive. Another development in feminist linguistics emerged under the influence of post-feminism (Butler 1990), which argues that speakers use language to perform a given identity.

Since feminist linguistics cannot be interpreted without reference to (social) feminism, Mills (2008) uses the labels “Second Wave” and “Third Wave” feminist linguistics to differentiate these two clear approaches to the study of sexism and language. While the Second Wave looks at the inherent meaning of words, the Third Wave focuses on how meaning is constructed (Mills 2008:25-26).

Thus, Second Wave feminist linguistics identifies androcentric language, which can be seen in the common linguistic tendency to use the masculine pronoun he to refer to high-status occupational terms (lawyer, scientist, etc.) as antecedents, while low-status occupational terms (e.g., secretary, nurse) are often followed by anaphoric she (Hellinger 2001:108). More evidence of linguistic androcentrism is, paradoxically, seen in feminized terms such as authoress, which make women visible but carry derogatory connotations (Holmes 2001:127). In fact, semantic derogation lies behind several derivational affixes with feminine meanings (such as -ster in spinster and -ette in usherette), which “tend to be associated with connotations of smallness, triviality or imitation” (Hellinger 2001:108-109).

Along with identifying androcentric language, Second Wave feminist linguistics proposes reforming language to eliminate sexism. The underlying assumption is of a Whorfian nature: just as language determines thought, sexist language perpetuates sexism. What exactly is meant by “non-sexist language” may vary depending on the language and its morphological characteristics. According to Romaine (2001:156), the tendency in English has been in favor of gender neutralization (or “degendering”), while languages with a richer inflectional morphology, such as German and French, have opted for visibilizing women though feminization (“engendering” or “regendering”).

Gender-neutrality in English, it is said, can be achieved through various means (examples from Hellinger 2001:109-110): the avoidance of stereotyping (e.g., avoiding the expectation that delegates are men implicit in utterances such as transport will be provided for delegates and their wives) and marked forms (e.g., female doctor) in accordance with the principle of symmetry, and the elimination of linguistic items that carry a masculine bias, such as replacing pseudo-generics like chairman with gender-neutral chair, or replacing the pronoun he used with general antecedents with combined he or she and singular they. Such recommendations are included in many style guidelines (e.g., APA, n.d.) and have largely been accepted by publishers, trade unions, and universities (Mills 2008:17). Nonetheless, reforms introducing these proposals after Second Wave feminism have provoked reactions among speakers and linguists alike. A much quoted response is that of Lass (1997:339), who claims that “avoiding ‘sexist pronouns’ is not a paradigm example of the kind of change that historical linguist-persons are typically concerned with, but an ideological excrescence,” which he subsequently expands on:

My use of generic he here (as throughout this book) illustrates the difference between a tendentious or ideological “act” [. . .] and a piece of structure. In my variety of English (and my wife’s as well!) he is the only pronoun usable for unselfconscious generic reference. Using s/he (which of course can’t be pronounced: does anybody say ‘ess-stroke-he’?) or he or she or they or whatever would count as an “act” (a deliberate flouting of grammatical convention in this case); but use of generic he is not, since it’s simply historically given, and I can’t not use it (without a conscious decision of a type not at all characteristic of ‘normal’ change) and still be speaking ‘my own language.’ Like all normal speakers, I am bound by the historically given. (Lass 1997:368)

A more recent and far more widely publicized example is the debate surrounding the use of gender-neutral pronouns sparked by Jordan Peterson’s objection to the statutory use of alternative pronouns preferred by transgender people and those who identify outside the binary, including but not limited to they, in Canada (e.g., Airton 2018). Despite this opposition, Mills (2008:11) suggests that in the twenty-first century overt sexism is generally seen as anachronistic “and signalling very conservative views of women, which are at odds with current views of gender relations.” However, as Graddol and Swann (1989:110) note, language reform does not remove sexism from language, because “[t]he existence of unmarked expressions ‘in the language’ does not mean that these will be used and interpreted in a neutral way.” Sexism and gender discrimination may take more subtle forms, for example, through the simple failure to mention women (Romaine 2001:154; Baker 2010), which constitutes a form of indirect sexism (Mills 2008:11 et passim). In addition, other non-sexist innovations such as chairperson and singular they “have been appropriated by a sexist speech community” (Ehrlich & King 1994:59) and have thus contributed to the persistence of sexism. Ehrlich and King (1994) cite, for example, a study that shows how individuals tend to identify the referent of an unmarked noun as male, even if the pronoun used to refer to it was singular they or combined he or she, something also discussed in Paterson (this issue). The only individuals in which the effects of language reform tend to be observed are women (Khosroshahi 1989, as cited by Ehrlich & King 1994:63; see also Romaine 2001:168).

Thus, while Second Wave feminism aimed at eliminating sexism from language by fostering gender-neutrality, the results do not seem to have contributed to reaching gender-equality. In fact, if anything, promoting gender-neutral language has made sexism more sophisticated and ambiguous, thus more difficult to identify (Baker 2014:5-6). This is why Third Wave feminist linguistics is concerned with the idea that there is diversity in both men and women, and hence focuses on how meaning is co-constructed (Mills 2008:26). The degree of complexity of what sexual discrimination means is indeed much higher than what can be appraised by language reforms, as the binary approach to gender traditionally adopted in studies on language and gender is clearly not inclusive, as it excludes people who do not identify as men or women (see, e.g., Zimman, Davis & Raclaw 2014; Jones & Mullany 2019). Following this line of thinking, what has been defined as sexist needs to be re-examined, along the lines suggested by queer theory (e.g., Cameron & Kulick 2003:74-106), in which speakers are not expected to possess identities that are reflected in their use of the language, but to construct their own identify through language use. As Motschenbacher and Stegu (2013:522) put it:

Queer Linguists question whether one needs to pre-assume and contrast two binary macro-categories, female and male, whose average behaviour is treated as a normative yardstick. Such a procedure further polarises the two gender categories and leads to a stigmatisation of people who deviate from the average pattern.

Because of space constraints, the papers in this issue cannot approach the study of gender-neutrality and democratization from all its possible perspectives, and, notwithstanding the highly democratizing role of the language used to refer to people who do not identify within the binary, this approach falls outside our scope. In this paper and this issue we explore the intersection between gender-neutrality and democratization by focusing on the dichotomy of man and woman and on how the language used to refer to them varies across territories, registers, and media.

2.2. Corpus Linguistics in Language and Gender Research

Corpora and corpus linguistic methods have become mainstays in many areas of linguistic research, yet, as noted by Kreyer (2014), they are not found so frequently in studies of gender. Such a claim is supported, for example, by the selection of topics in the recent The handbook of language, gender, and sexuality (Ehrlich, Meyerhoff & Holmes 2014), where separate chapters are devoted to major methods used in the field, including variationist studies, ethnography, conversation analysis, and Critical Discourse Analysis (CDA), but not to corpus linguistics, the treatment of which is limited to a number of individual contributions. On the other hand, corpus linguistics is indeed singled out as a major methodological framework in other overviews (e.g., Harrington, Litosseliti, Sauntson & Sunderland 2008), and the term appears frequently in Motschenbacher (2012). However, in his introduction to a virtual special issue in Gender and Language, Baker (2013) notes that it is actually common for studies in this area to mention corpora, but that the term usually refers simply to the data set used in the study, not necessarily to any specific application of a corpus-linguistic methodology.

The reason why corpus methods have remained only moderately popular in language and gender studies is due to current thinking in the area, where, in accordance with Third Wave feminist linguistics, gender is seen as discursively constructed rather than residing in individuals and existing pre-discursively (Ehrlich, Meyerhoff & Holmes 2014:4). Accordingly, current work on language, gender, and sexuality attempts to move away from essentialist ideas of how men and women use language towards describing multiple ways of “doing gender” in multiple local contexts, typically using qualitative methods. Such thinking is seemingly at odds with corpus linguistic methods, which are seen to rely on quantification and aim to describe general patterns of language use, much in the spirit of first-wave sociolinguistics, with the help of fixed independent variables like age and gender. These categories appear particularly problematic in studies on language and sexuality, as sexual identifications are even more elusive to corpus compilers than gender labels (Motschenbacher 2018:147-148).

At the same time, an increasing number of very large corpora have been made available in recent decades (see e.g., Hiltunen, McVeigh & Säily 2017 for an overview), enabling a variety of analyses which were previously unfeasible due to a lack of data. For Kortmann (2018), the development of corpora has been a major driving force behind what he calls “the quantitative turn” in linguistics. This “turn” entails not only the use of increasingly sophisticated statistical techniques in quantitative research, but also the expectation that even primarily qualitatively oriented studies will contain some degree of quantification of the phenomenon in question (see also Janda 2013). As a result, recent years have seen an increasing number of data-driven studies on language, gender, and sexuality, mainly focusing on aspects of the representation and construction of identities (e.g., Caldas-Coulthard & Moon 2010; Moon 2014; Baker & Levon 2015a, 2015b; Potts 2015; Norberg 2016; Taylor 2016); Konnelly (this issue), on the varying patterns of representation of men and women in US broadcast talk, contributes to this growing body of research. The advantage of large corpora in such studies lies in affording researchers the ability to uncover latent patterns of representation and to avoid “cherry-picking” only those results that fit their expectations, although, as noted by Baker (2014), corpus linguistic studies always involve an element of subjectivity, and therefore researcher reflexivity and the use of triangulation remain extremely important.

Despite concerns about essentialism and the reification of gender differences, approaches combining variationist analysis and corpus linguistics (see, e.g., Romaine 2008; Szmrecsanyi 2017) make an important contribution to language and gender research. The method allows for a systematic investigation of what factors are involved when language users choose between elements that are differentially marked for gender, like generic pronouns and nouns, and how those choices may be linked to sexism, as masculine gender has often been considered the norm, while feminine gender is considered to be marked. Previous studies on both synchronic and diachronic pronoun usage are numerous (e.g., Laitinen 2007; Balhorn 2009; Paterson 2014), and Paterson (this issue) and Loureiro-Porto (this issue) shed new light on pronoun variation, focusing on combined pronouns in British English (BrE) and American English (AmE), and on epicene pronouns in World Englishes, respectively. Similarly, studies on generic nouns (e.g., Holmes, Sigley & Terraschke 2009) have drawn attention to how terms supposedly used for generic reference (e.g., occupational terms) are in fact “false generics” (Eckert & McConnell-Ginet 2003:246), in that they advance male referents as the default interpretation and thus underline the distinctiveness of female referents.

The three papers in this issue showcase different ways in which research on gender neutrality can benefit from corpus linguistic approaches. The corpora represent both genre-balanced corpora like the ICE corpora and the Brown/LOB families, which have been widely used in previous decades, and more recent big-data corpora (GloWbE and COCA), which aim to capitalize on the advantages of the huge amounts of data now available on the internet, while maintaining a corpus design that enables meaningful linguistic analyses (Davies 2019).

3.1. Definition and Scope

Many recent studies on the history of English have convincingly argued that large-scale sociocultural processes related to societies in the process of becoming more democratic have had an important impact on language use and language change, and that such developments have been particularly evident in English during the latter half of the twentieth century (and also earlier, as noted by Myhill 1995, who suggests that around the American Civil War, changes in the modal domain may have been motivated by social and psychological factors, such as a change in social hierarchy). Perhaps the most widely used term to address this process is “democratization,” which goes back to Norman Fairclough’s work on CDA. In Discourse and social change, he uses the term to refer to “the removal of inequalities and asymmetries in the discursive and linguistic rights, obligations and prestige of groups of people” (Fairclough 1992:201). He argues that this process has profoundly shaped discourse, for example, by an increasing acceptance of non-standard varieties in institutional discourse, the reduction of overt markers of power, and traditionally formal discourse types becoming more informal. Since Fairclough’s work, the term has been used in various studies in different areas of linguistics without necessarily subscribing to the principles of CDA (e.g., Leech, Hundt, Mair & Smith 2009; Flowerdew 2012; Spirling 2016; Seoane & Hundt 2017). Examples of specific linguistic changes attributed to democratization include the decrease of deontic modals and the rise of semi-modals, and the elimination of overtly sexist features in language, such as generic he and occupational terms ending in –man (Farrelly & Seoane 2012:394).

Despite the attention that democratization has received in the literature, several terminological overlaps have been identified. Thus, for example, Farrelly and Seoane (2012) distinguish three related senses in which the term democratization is used in historical linguistic studies: “democratization proper” (i.e., removing overt markers of power asymmetry to promote equality), “colloquialization” (i.e., the use of speech-like features in writing), and “informalization.” ³ They further suggest that distinguishing these three senses is difficult due to the interrelatedness and overlap between them. In addition, alongside these terms, previous studies have also employed other labels for describing the relationship between sociocultural processes and language, and once again these have partially overlapping meanings: “personalization,” “tabloidization,” “conversationalization,” “mediatization” (e.g., Culpeper & Nevala 2012; Baker 2017). How these terms relate to each other, and to the issue of gender-neutrality in particular, is complex, and clearly requires both theoretical reflection and an appraisal of empirical evidence, and hence would benefit from interdisciplinary collaboration and dialogue (Farrelly & Seoane 2012:399).

3.2. Democratization across Varieties of English

The close connection between social movements and linguistic change in the diffusion of gender-neutrality and democratization demands the analysis of varieties spoken in different territories and by different cultures. In conducting such an analysis, and for the purposes of this work, we will use the terms coined by Kachru (1985) in his Concentric Circles model, namely “inner-,” “outer-,” and “expanding-circle” varieties, although, as is well-known, this model has been found controversial based on its static nature and also on the fact that it considers language at the level of the nation state, thus not accounting for the heterogeneity of speech groups, identity construction, and language attitudes (Seoane 2016:4; Buschfeld & Kautzsch 2019:55). Nonetheless, Kachru’s (1985) labels remain useful in that they serve to provide a schematic representation of a sociolinguistically complex situation when the aim is to describe language variation and change, rather than to theorize about what model fits better the reality (as done, e.g., in D’Arcy 2013; Fox 2019).

The study of outer-circle varieties of English has flourished since the 1980s with the realization that “[t]he more English spreads globally, the more heterogeneous it becomes internally” (Mair 2013:255). These varieties, variously called World Englishes, New Englishes, and Postcolonial Englishes (see Schneider 2011:29-30 for a discussion), have been addressed from multiple viewpoints, such as contact linguistics, dialect typology, and second-language acquisition. Such studies have come not only to describe the different varieties, but also to help understand and explain general processes of language change, including contact grammaticalization (e.g., Ziegeler 2014), colloquialization (e.g., Collins 2013), linguistic globalization (Blommaert 2010), and the Americanization of English (e.g., Meyerhoff & Niedzielski 2003). Thus, any attempt to describe linguistic democratization must clearly include both inner-circle and outer-circle varieties.

Linguistic democratization is closely linked to social changes, and in its origins is a western-centric concept that might be claimed to be inaccurate for the description of varieties spoken in non-western territories. In fact, the label democratization itself is not often used in the World Englishes literature, and when it does appear it tends to refer to social changes such as increases in democratic rights (along the lines of Grugel 2002) or in the writing of a constitution (e.g., Schneider 2007), and hence is not used in Farrelly and Seoane’s (2012) sense. In the linguistic sphere, it may also refer to language planning and teaching, since attitudes to the colonial language, namely English, may lead to calls for different policies (e.g., Kembo Sure 2003, on Kenya) and for different approaches to teaching the language, as Tupas (2010) shows in arguing for the democratization of the school system in the outer circle, making a plea for teachers and students to be able to freely choose the variety to be taught and learned (i.e., native-like English or English as a Lingua Franca).

Because we are approaching the study of democratization from the perspective of gender-neutrality, and this concept has been related to women’s movements (see section 2.1), we cannot overlook that what we generally label feminism usually refers to a particular intersection between gender, race, and class (on the notion of intersectionality more broadly, see Crenshaw 1989, 2020); in other words, while feminism has traditionally referred to white, western women, gender (in-)equality in other territories requires specific analysis, as seen in Brooks (2007) and in Roces and Edwards (2010). The degree of variation in gender (in-)equality in territories within the same geographical region has been shown to go from places with a “high level of social development,” such as Singapore and Hong Kong, to places which are “more problematic on issues of gender equality giving rise to considerable disparities regionally,” such as the Philippines (Brooks 2007:3). Women in territories like India have been found to be “upheld as preservers of ethnic and religious authenticity” (Rydstrøm 2010:11), while the fate of Hong Kong women, for example, has followed the same path as those in western societies, in that they entered the labor force in very large numbers in the 1970s as a result of industrialization (Göransson 2010:199-200). Related to all these differences, the feminist movement in these territories has followed different paths of development, from mirroring the major phases of feminism in the West, as in the case of Hong Kong (Lim 2010:144), to developing later, as with India (Madhok 2010:225). All of these issues are, then, taken into account in the papers in this issue, since they constitute landmarks in the promotion and diffusion of gender-neutral language and thus may help to explain different degrees of democratization.