GENDER IN LANGUAGE PROJECT

For a language to be identified as “having gender” linguistically, it must fulfill three criteria: 1) it must have system-wide nominal classes and all of the language’s nouns must be assigned to one of these classes, 2) the gender value of the noun must trigger patterns of morphosyntactic agreement on dependent elements, and 3) there must be a basis to the gender assignments, whether formal, semantic, or both in combination (Dixon, 1982; Corbett, 1991; Kramer, 2015). Languages that violate any of these three criteria (including some that have gender morphology) are not considered to have the feature of gender, and because the systems of nominal classification this theory describes are more often called “gender,” non-qualifying languages are subsequently identified as “genderless.” However, this canonical labelling of languages as linguistically “gendered” or “genderless” crucially cannot account for the ways in which linguistic gender at any level is correlated with social gender, even in so-called “genderless” languages. Queer, trans, nonbinary, and other gender-nonconforming speakers of different languages have identified features that they believe to mark normative masculine and feminine (social) gender and proposed innovations meant to be inclusive of people of other genders. In doing so, they point out the main reason our definition of “linguistic gender” should change: it has material consequences.

These consequences manifest chiefly in structural discrimination, which produces psychological and bodily harm. At the institutional level, language academies, such as the Real Academia Española have taken reactionary stances against gender-inclusive language that discourage the widespread use of gender-confirming language in society. While rejecting gender-inclusive language, these bodies indirectly cite linguistic theories of gender which state that the feature is arbitrary (e.g. Ibrahim, 1973). Most saliently, the Académie Française’s characterization of inclusive French as putting the French language in ‘péril mortel’ (AF, 2017) has emboldened homophobic and transphobic rhetoric in the French press, whereby ‘inclusive writing’ and related terms (e.g. ‘novlangue’) become far-right dog-whistles against progressive causes in general. On the individual level, where linguistic marking of speakers as normatively masculine or feminine is obligatory, trans, non-binary, and gender-nonconfirming individuals are repeatedly misgendered, which not only has profoundly negative psychological impacts (Langer, 2011) but can also prevent individuals from accessing healthcare (Namaste, 2000), and can cause other bodily harm. Thus, canonical definitions of linguistic gender do not correlate with the lived experiences of speakers of those languages, which (socially) gender referents in other salient ways.

By contrast, the Gender in Language Project proposes an alternative model for defining linguistic gender which begins from the perspective of social gender. We argue that a new definition focused on the multivariate ways that social gender categories may become encoded in language not only resolves differential understandings of the concepts of social and grammatical gender, but it also empirically strengthens the connection between them, and allows for a theory which corroborates the lived experiences of queer, trans, nonbinary, and other gender-nonconforming people globally with language.

The realization of social gender distinctions in the grammar and lexicon of different languages is described in the theory of morphological gender (Kramer, 2015), which defines languages whose grammars are structured in a particular way as “having gender,” and all others as being “genderless.” Because they do not fulfill each of the three criteria outlined in this theory, many languages like English, Vietnamese, and Tagalog are considered genderless linguistically even though they mark masculine and feminine gender in different ways, including morphologically. This paper presents a typological analysis of the aforementioned languages and details the multivariate features of gender they exhibit in order to problematize the notion that they are “genderless.”

While English is a famously “genderless” language, it features pronominal, lexical, and even traces of morphological gender. Queer speakers have innovated neutral and/or specifically nonbinary personal pronouns; the extant gender-neutral pronoun they has gained widespread popularity and numerous series of neopronouns (third-person singular pronouns other than he, she, or they), including ze/hir and xe/xem (Bertulfo, 2021) have been attested to provide more expansive ways to self-identify. Many English words exist as gender-paired lexical items by way of semantics, and many of these pairs lack neutral alternatives (e.g. nephew, niece). Lastly, English features feminizing suffixes (-ette, -ess, -ix) that gender-mark words in a manifestly morphological way (e.g. bachelorette, actress; Baron, 1986).

Vietnamese is similarly considered a genderless language, yet it forms gendered distinctions in novel ways, including via processes of compounding involving normatively gendered adjectives (e.g. trai ‘boy’, gái ‘girl’) and otherwise gender-neutral nouns (e.g. con ‘child/dear one’), which compound in ways that disallow neutral alternatives prescriptively (e.g. con trai ‘son’, con gái ‘daughter’; Ngo, 2020). These adjectives double gender-mark certain gender marked roots (e.g. ông nội ‘paternal grandfather’, bà nội ‘paternal grandmother’).

Finally, Tagalog presents the most egregious example. As a result of Spanish colonialism, hundreds of loanwords (nouns and adjectives) bear masculine-feminine morphological gender that parallels Spanish; the loanword’s gender aligns with the referent’s gender and is inflected using canonical Spanish gender morphology (e.g. abenturero/abenturera ‘adventurer’, santo/santa ‘holy’), even when paired with native nouns (OEI, 1972). Queer speakers have improved on some of these items using gender-inclusive Spanish strategies (e.g. pilipinx), further proving their similarities (FIERCE, 2018). The discovery of a morphological gender system in a subset of the grammar and lexicon challenges the ability of the prevailing theory of morphological gender to identify all gender morphology cross-linguistically.

The information we lose in enforcing the label “genderless” is many-fold: it trivializes the diverse features of gender-marking present in English, Vietnamese, Tagalog, and other languages, and disregards important historical changes to gender systems brought about by language contact as well as contemporaneous innovations in language occurring in queer communities. We argue the necessity of a more expansive definition of gender in language which unifies features that are disjointed in the current theory, in turn positioning the focus of the study of gender in language on social gender. 

When most people think of the term “grammatical gender,” they probably think about masculine-feminine gender languages, wherein the genders of most words referring to people align with the gender of the person being referenced, yet this is actually a narrow understanding in linguistic theory. Masculine-feminine (including masculine-feminine-neuter) morphological gender is one of the systems of nominal classification described by Corbett (1991). While biological sex and social gender are fundamental organizing principles of these languages, how these features interact with the grammar is not well-described. The fact that these systems often include no other animate genders besides masculine and feminine leads to nonbinary, trans, and other gender-nonconforming people engineering their own solutions for linguistic self-representation. In doing so, they identify many features (e.g. pronominal, lexicosemantic, morphophonological) that transcend definitions of linguistic gender as purely morphological and are found in languages considered genderless by the same theory.

In order to critically analyze the ways that even languages considered to “have gender” challenge the current theory of morphological gender, we present a typological analysis of four masculine-feminine gender languages: Spanish, Catalan, Portuguese, and Modern Irish. We assembled a corpus comprised of data in prescriptive grammars and in proposals from queer communities that document where social gender is distinguished in these languages, including all known data on the realization of gender-inclusive forms in each.

Though all four languages have morphological gender, they also form gendered distinctions in ways that move beyond inflectional (including affixal) gender morphology—for instance, there exist pairs of lexical items that are gender-marked by what seems to be semantics only (e.g. hombre ‘man’ and mujer ‘woman’ in Spanish). In Modern Irish, gender is more tenuously related to semantics (e.g. cailín ‘girl’ is masculine due to its form) and morphology (e.g. the -(e)ach ending is feminine for mass nouns, but masculine for countable nouns), and the gender of the referent is implicated in case-specific, word-initial morphophonological mutations following certain articles (e.g. a madra [ə madra] ‘her dog’, a mhadra [ə wadra] ‘his dog’; Stenson, 2020). Finally, all four languages have attested gender-inclusive innovations in their personal pronominal systems (e.g. elle in Spanish, elu in Portuguese, elli in Catalan, and siad in Modern Irish; Acosta Matos, 2016; Lobo & Gaigaia, 2014; Fajardo Martín, 2021; Ní Choistealbha, 2018). In the case of Spanish, Portuguese, and Catalan, these pronouns are tied to inclusive gender morphemes (e.g. -e/-x, -e/-i, and -i/-x, respectively) which together are the canonical elements of additional morphological genders designating neutrality and/or specifically nonbinary gender identities.

While these features (pronominal, lexicosemantic, morphophonological) are unified in languages considered to “have gender,” all of them are also found in languages considered “genderless,” signaling that the extant theory of morphological gender fails to explain the realization of all gender in “gendered” languages. Analyzing language from the perspective of social gender allows us to construct a new definition of gender in language in which nonbinary, trans, and gender-nonconforming speakers are at the center of this understanding, not morphology alone.

Contemporary theories of gender in language define the feature as morphological (and to a lesser extent, pronominal; Corbett, 1991). Isolating languages reveal a major weakness in this definition in that they encode normative gender despite their lack of inflectional morphology. Social gender is indeed marked in canonically “genderless” languages, including those that lack clear morphemic boundaries. In this paper, we focus on features of gender in three languages of East Asia: Korean, Mandarin Chinese, Cantonese. The former is alphabetic while the latter two are character-based, and all demonstrate features of gender that transcend the current definition of linguistic gender as purely morphological. These features (lexicosemantic, pronominal, radical, character) are not described by any predominant theory of gender in language, yet they hold the potential to improve upon this literature to explain the reality of gendered language for queer and trans people globally.

In alphabetic languages like Korean, letters indicate the phonology of the word they compose, and it is rarely the case that individual letters have an inherent meaning. There exist many pairs of words in Korean that have normatively masculine and feminine forms (어머니 ‘mother’, 아버지 ‘father’); these may be analyzed as lexicosemantic features of gender which can serve as the base forms for related words ((외)할아버지 ‘grandfather’, (외)할머니 ‘grandmother’; Yeon & Brown, 2019). As Korean has been heavily influenced by Chinese, the languages share related features: the Korean words 남자 ‘male’ and 여자 ‘female’ are derived from the Sino-Korean 男 ⼦ and ⼥⼦, respectively, both of which use the Chinese radical ⼦ ‘son’. This male bias is reflected in the Korean word 사람 ‘person’, which is commonly understood to mean ‘man’, invoking the use of 여자 ‘female’ as a more specific alternative.

Distinct from alphabetic languages, character-based languages often bear meaning in their orthographic systems. In Mandarin Chinese and Cantonese, both of which use Chinese orthography, radicals and non-radical subparts comprise each character, though not all are realized phonetically. In both languages, there exist two distinct semantic radicals, the ‘male’ or ‘human’ radical (亻) and the ‘female’ radical (⼥) which distinguish normatively male (他) and female-specific (她) third-person singular personal pronouns in written Chinese. To collapse this distinction, Mandarin speakers have proposed the use of the innovative X radical (X也) and the radical ⽆ ‘none/not any’ (⽆也; Lai, 2020; Zhu, 2021). In Cantonese, ‘male’ and ‘female’ radicals can double gender-mark certain kinship terms (伯⽗ ‘father’s older brother’, 姑媽 ‘father’s older sister’) and produce pejorative definitions in the feminine (伎 ‘skill’, 妓 ‘prostitute’), as in Mandarin (Chin & Burridge, 1993).

Together, these features represent crucial omissions from contemporary definitions of linguistic gender. These definitions particularly fail Chinese orthography, which features radical- and character-based (and often phonetically unrealized) methods of encoding gender. In reanalyzing linguistic gender from the perspective of social gender, languages of East Asia contribute greatly to an understanding which does not privilege linguistic gender as purely morphological, in turn allowing us to imagine pathways to language change for queer, trans, and nonbinary speakers of these languages. 

Within linguistic theory, Global Portuguese is identified as a “masculine-feminine gender language,” a subtype of morphological gender in which all parts of speech must be masculine or feminine linguistically, except in the verbal system. In Portuguese, these genders are explicitly based on “biological sex” and/or “social gender” in words referring to people (Hutchinson & Lloyd, 2003), challenging the notion that gender is an inherent property of nouns. That the property of gender instead comes from the person referenced is made apparent by the insertion of a person into a Portuguese sentence lacking any other noun (e.g. Fernando é simpático ‘Fernando is kind’). Further proving the interconnection of social gender and morphological gender, many names in Portuguese pattern with its morphological gender system (e.g. Antônio, Antônia). As queer, trans, and nonbinary people have pointed out, the biggest limitation of the language is that gender is maximally binary, begging the question of how Portuguese can become inclusive.

This paper presents a novel corpus compiling where gender is marked in Portuguese along with inclusive forms proposed by queer, trans, and nonbinary Portuguese speakers. We surveyed prescriptive language documents, identifying all sites where gender is encoded in the language. This analysis revealed that gender is realized in Portuguese in ways that are not purely morphological. For instance, certain nouns have inherent lexical gender (e.g. pai 'father', mãe 'mother'); gender is also realized in pronouns (e.g. ele 'he', ela 'she'). Queer, trans, and nonbinary speakers of Portuguese have established not only gender-inclusive personal pronouns (e.g. elu, ilu, ile, ili) but also inflections which can be applied to the entire grammatical system of gender in the language. In combination, these innovations can be assigned to what we've identified as two new linguistic genders: the e gender (Lobo & Gaigaia, 2018; Ribeiro de Moura, 2021) and the i gender (Gaigaia, 2014). The e gender presents two allomorphs, -e and -u. If the masculine form ends in -o, -e surfaces as -e (menino ‘boy’ → menine ‘child’); if it ends in -e, -u surfaces instead (dele ‘of/by him’ → delu ‘of/by them [SG.]’). The i gender presents only one allomorph, -i (lindo ‘handsome’ → lindi ‘good-looking’).

Similarly to Spanish (Papadopoulos, 2022), our research shows that queer, trans, and nonbinary speakers of Portuguese have innovated two new linguistic genders which can be applied extensively to the whole grammatical structure of their language (e.g. Elu é ume alune aplicade ‘They are a hard-working student’). These innovations modify the existing function of gender in Portuguese, increasing its capacity to represent nonbinary and gender non-conforming speakers in the language. Additionally, these developments corroborate an important pathway to language change in modern Romance languages, exacerbating the understanding that in words for people, gender is a property of the person before it is a property of the noun. These robust findings necessitate a re-analysis of the concept of linguistic gender that is not based solely on morphology and that constructs a new typology of gender in which nonbinary people are centered. 

Modern Irish is considered a masculine-feminine, morphologically gendered language, like many Romance languages (Corbett, 1991). Gender assignments in these languages must be assigned based on formal and/or semantic criteria; in the case of Romance, "biological sex" and social gender play a fundamental role in the gender assignments of words for humans. While Irish is identified as "having gender" as Romance languages are, morphological gender in Irish is more weakly tied to both form and semantics, with its own distinct gender features: it largely lacks inflectional gender morphology, instead displaying features besides the morphological (lexicosemantic, morphophonological, pronominal). In fact, Irish is much more similar to English (a "genderless" language), especially due to the fact that most Irish speakers are also English speakers. The influence of English has caused an erosion of the formal gender system in Irish, and gender inclusive innovations in Irish more closely parallel those in English than those in other gendered languages. Queer and nonbinary speakers of Irish have targeted semantics to the exclusion of the morphological gender system, which prompts us to investigate how social gender is meaningfully realized in Irish.

Nouns in Irish are either masculine or feminine, but this gender is more often not marked morphologically. Many Irish language learning texts note that a word's gender isn't made obvious by its form, with the only sure way to find out a word's gender being a dictionary (Stenson, 2020). The gender assignments of Irish nouns are also more weakly tied to semantics, as in the word cailín 'girl' which is masculine because of the suffix -in, though this same suffix can also be a diminutive ending that retains the gender of the original word. Instead of being expressed morphologically, most masculine-feminine gender distinctions in Irish are expressed in terms of lexical semantics (e.g. fear 'man' and bean 'woman'). On the other hand, social gender plays a much stronger role in the morphophonology of certain grammatical categories: some possessive noun phrases trigger a case-specific, word-initial mutation that lenites the noun; in this case, the mutation is defined by the (social) gender of the referent (e.g. máthair 'mother', a mháthair 'his mother', cf. a máthair 'her mother'). Finally, social gender is marked in masculine (sé/é) and feminine (sí/í) third-person singular pronouns, which queer speakers have improved upon by borrowing from the plural paradigm (siad/iad).

The disconnect between social gender and morphological gender, on top of a dense gender morphophonology, makes it incredibly difficult for speakers of Irish to upkeep the formal gender system. Language contact between English and Irish has influenced a weakening of the formal gender system, such that speakers are starting to generalize the masculine gender (Frenda, 2011). Irish also shares certain commonalities with English: siad is now used as a gender-inclusive singular personal pronoun, much in the same way that the English pronoun they functions. Thus, gender-inclusive innovations in Irish focus primarily on semantics—a common trait of these innovations crosslinguistically—prompting a reanalysis of how we define gender in language. 

It is well-studied that listeners can distinguish between normative male and female voices with almost perfect accuracy due to changes caused by physiological differences between speakers (e.g. vocal fold length) based on “biological sex” (Jacobs et al., 2000). A unique phenomenon takes place, however, when trying to distinguish between the voices of queer and non-queer speakers perceived to be of the same gender; listeners can accurately distinguish between these voices (Mack, 2010), leading to the theory that acoustic differences influence queer speech perception, even when physiological differences are not presumed to be the determining factor. Most stereotypes often cite pitch as the main factor in queer speech perception (Kachel et al., 2018), though the production of sibilants (Zimman, 2017), vowels (Smyth & Rogers, 2008), VOT (Pahis, 2017), and length of word segments (Esposito, 2020) have also been correlated with this phenomenon. However, while the production and perception of queer speech has been studied at length, most extant data comes from English, limiting our knowledge of queer speech production and perception in other languages.

The present study aims to investigate if the acoustic patterns described above play a role in identifying the speech of queer Spanish speakers. To do so, recordings were collected from four groups of interest: self-identified queer men, queer women, heterosexual men, and heterosexual women. The recordings were then analyzed in PRAAT to extrapolate patterns linking acoustic correlates of their speech to their self-reported gender and sexual self-identifications. A main focus was placed on the production of vowels (F1-F4), the production and articulation of sibilants (through analysis of COG), as well as any changes in pitch (F0) or word duration that occurred. A second experiment was conducted in which these recordings were modified and presented to new participants in a matched-guise task. The aural stimuli presented were divided into two different groups: one accompanied by visual stimuli introducing a bias in participants’ responses, while the others contained only the audio. A total of four voices were chosen and presented three times, resulting in twelve critical stimuli meant to determine whether implicit and explicit stereotypes influence queer speech perception.

Participants exhibited sociolinguistic variation in their results yielding two conclusions. In the cases that participants’ accuracy at matching the voice to the correct sexuality was highly dependent on the visual stimuli being presented (i.e. matching queer voices to “stereotypical queer-looking people”), it is presumed that visual cues played a larger role in perception. However, when the opposite effect was observed, we can conclude that there were acoustic cues (e.g. sibilant production, vowel formants, pitch, and segment duration) that allowed for the perception of sexuality through speech, further corroborating the results of prior research. While queer speaking practices are often studied in terms of morphology and semantics, this sociophonetic analysis bridges the many gendered features found in language with audiovisual identity, and queers acoustic theory in a way that makes it more representative of non-English-speaking populations, and queer voices around the world as a whole.