Sydney Dawson

​Department of Anthropology, University of British Columbia, Canada. 

sdawso74@student.ubc.ca

​​

Abstract    Stigmatizing language can be hugely influential to the social perception of marginalized communities, and social media has amplified these effects. Exploring over 3000 Instagram comments, this research features a critical discourse analysis of how referential language is used differentially and potentially harmful on social media posts made by LGBTQ+ couples. This research establishes a theoretical framework to evidence stereotypes that label LGBTQ+ identities as liminal and unserious, and demonstrates how infantilizing language can further heteronormativity as a covert and benevolent form of hierarchy establishment. Furthermore, speech accommodation theory supports an understanding of the motivations behind infantilizing and asexualizing language aimed toward LGBTQ+ couples. This study measures reported and actual language use and addresses the motivations speakers may have to adjust their language when speaking to and about LGBTQ+ subjects. This research aims to help speakers understand the complex relationships between motivation and effect in allyship-related language.

 

Résumé     La stigmatisation d'un certain langage peut avoir une énorme influence sur la perception sociale de communautés marginalisées, et les réseaux sociaux ont amplifié ces effets. En explorant plus de 3 000 commentaires sur Instagram, cette recherche présente une analyse critique des discours sur la façon dont le langage référentiel est utilisé de manière différentielle et potentiellement nuisible dans les publications sur les réseaux sociaux rédigées par les couples LGBTQ+. Cette recherche établit un cadre théorique pour mettre en évidence les stéréotypes qui caractérisent les identités LGBTQ+ comme liminales et frivoles, et démontre comment l'infantilisation du langage infantilisant peut renforcer l'hétéronormativité dans une forme plus discrète et paternaliste d'affirmation hiérarchique. De plus, la théorie de l’ajustement de la parole appuie notre compréhension des mobile derrière l'usage d'un langage infantilisant et asexualisant destiné aux couples LGBTQ+. Cette étude mesure l'utilisation de la langue telle que déclarée et telle qu'effective et aborde les motivations que les locuteurs peuvent avoir à ajuster leur langage lorsqu'ils parlent de sujets LGBTQ+. Cette recherche vise à aider les locuteurs à comprendre les relations complexes entre motivations et effets dans le langage lié aux allié.es LGBTQ+.

 

Keywords                     language; allyship; LGBTQ+; heteronormativity; social media

​​

​

​​Introduction

 

Language is the architect of thought; it shapes our socialization, how we conceptualize ideas, and even the window of discourse that is considered acceptable to speak about at all. Needless to say, language is a powerful tool that holds often underestimated influence. Referential language, or the way we use language to refer to others, is one of the ways in which language shapes our worldview, including our metacognition, how we think about others, and the social issues we share. While this fundamental feature of language is universal, its effects are especially pronounced in socially marginalized communities whose very position in society can be significantly influenced by their narration. 

 

In this research, I take particular interest in the notion that members of the LGBTQ+ community are uniquely vulnerable to stigmatizing referential language that both the users and recipients can internalize. One significant effect of this involves perceptions of identity and relationship validity, which can be negatively influenced by linguistic and social tools such as infantilization and covert stereotype reliance. Dean’s (2016) research on the social value of being perceived as non-homophobic carries notable implications for furtive shifts in homophobic language and stereotypes, suggesting that as LGBTQ+ identities become more widely accepted, homophobia may become more discrete, change its form, or otherwise shift to avoid negative attention. Heteronormative bias, as Kiesling (2019) notes, is possibly the most prominent feature of normativity in language. LGBTQ+ identities continue to gain mainstream social acceptance and visibility, and so, in many ways, become less stigmatized. However, in the face of contentious discourses in North America about the appropriateness and validity of Queerness, we must approach this assumption with a lens that attends to how even covert language can shape social perceptions and persistent and emerging stereotypes. 

 

Language use both reflects and shapes our ideologies, and while it is easy to understand our language choices as circumstantial and singular, in reality, they are small units of larger aggregate trends that compound into recognizable discourses. Our language choices can be understood as both products of and informants of worldview with profound and material impacts on everything from personal action and opinion to public policy and safety for LGBTQ+ individuals. 

 

As we become increasingly reliant on technology for cultural communication, social media’s vast potential for linguistic and cultural influence provides unique insight into language’s role in worldview formation; it is, in many ways, a digital archive of language use, discourse, and culture. Social media also plays a notable role in surveillance and identity construction, and how these forces may lead to self-conscious performances of allyship, or the act of supportively aligning oneself with a marginalized group, particularly when social value is often placed on this moral attestation. As Thai and Nylund (2023) note, allyship can come with social and material rewards, and the desire to appear unprejudiced can, at times, outweigh genuine concern with and understanding of marginalized groups; self-interested allyship, which is more interested in virtue signaling and social alignment than sincere solidarity and advocacy, is a particular concern on social media (Voogt 2022). 

 

So, how does allyship interact with language and Queerness on social media? With particular attention to the cooperative role of feminizing and infantilizing language in hierarchy reinforcement, it becomes evident that the linguistic treatment of LGBTQ+ individuals and relationships on social media is steeped in salient but covert stereotypes (Huot 2013; Ponterotto 2014). Stereotypes such as “the LGBTQ+ phase” and imposed platonism (the framing of relationships as platonic rather than romantic or sexual) employ liminality and unseriousness as archetypes through which to paternalistically shape language use and contribute to the prevalence of anti-LGBTQ+ stigma (McBean 2016; Monaghan 2016). Moreover, applying speech accommodation theory to this notion allows us to understand that speakers may adapt their treatment of others based on perceived stereotypical qualities, exposing a direct connection between social perception and language use (Brown and Draper 2003). The canon of research on these forms of referential language use advises that subjects who internalize linguistic infantilization may experience negative self-image and sometimes even adopt these stigmatizing language trends themselves (Berg 2015; Frost and Meyer 2009; Martin 2016). 

 

Considering what is known about the social capital of allyship on social media and the motivations and effects of linguistic accommodation, it becomes necessary to examine how underlying sentiments about LGBTQ+ subjects present themselves in language use on social media. Through a critical discourse and thematic analysis of comments on Instagram—a social media platform that privileges idealized and aspirational identity performance (Bailey 2021; Dawson 2024; Duffy and Hund 2015; Shadijanova 2020)—and data from a quantitative survey I conducted in 2021 discussing LGBTQ+ identities, language, and social media, I will draw a comparison between actual and reported referential language use on Instagram.

 

Heteronormativity functions to exclude homosexuality from the baseline of cultural narratives, further marginalizing LGBTQ+ individuals and reaffirming heterosexism. In this way, it can become a tool to insidiously serve homophobic agendas with little attention from mainstream anti-discrimination movements. In a world that is slowly but surely accepting sexual diversity, benevolent homophobia is revealed in subtle ways, sometimes even as a well-intended overcompensation of allyship. Specifically, this research focuses on the notion that referential language use directed at LGBTQ+ subjects on Instagram displays both a hypercorrection of sexualized stereotypes (the visibility of which can be identified by allies) and invalidation of same-gender relationships through infantilization and feminization (which reflect less-visible stereotypes and social desires to course-correct LGBTQ+ individuals back onto a heterosexual path). By analyzing these observable linguistic patterns, through which heterosexuality is normalized and homosexuality is dismissed, trivialized, and policed, we can conclude that this language significantly impacts the societal coherence of LGBTQ+ romantic subjects and has consequential implications for LGBTQ+ self-image and social status.

 

 

Background and Literature Review

 

Feminization and Infantilization

 

The gendered nature of infantilization has been well-documented in previous multidisciplinary work and is described excellently by Sut Jhally in the film The Codes of Gender (2009): “If you’ve never heard of the ‘infantilization’ of women, allow me to introduce you; it is an incredible phenomenon by which our society systemically equates femininity with things like vulnerability, submission, uncertainty, and childhood. To be womanly today is to be, in many senses, infantile.” The notion of femininity as infantile has been explored by many other scholars, including Diane Ponterotto in her work on referential language and female athletes in sports journalism (2014). Ponterotto suggests that tactics such as infantilization are weaponized against women who defy gender roles to undermine their seriousness and professionalism. While her work focused on athletic commentary, Ponterotto’s points on metaphoric taxonomies that proclaim women are children, women are embryos (still evolving), and the specifically desexualized language used to describe non-hegemonic femininities thought to be “heterosexual failures (2014, 98) has compelling implications in discussions of the linguistic treatment of LGBTQ+ individuals. It is important to note that feminization and infantilization work cooperatively to trivialize subjects, but the process of feminizing an individual or group through language does not apply only to women; the patriarchal classification of masculinity as privileged and femininity as devalued has complex circumstantial applications when it comes to LGBTQ+ identities (Hoskin 2020). 

 

Through the mechanisms of hegemonic binary gender,[1] maleness and femaleness exist as two singular options on opposite and complementary poles, a notion that is foundational, indispensable, and deeply intertwined with heterosexuality and heteronormativity. Therefore, individuals who express or associate with qualities of the gender opposite to their sex assigned at birth may be marked deviant in both their sexuality and gender expression; they are disrupting not only the binarism of gender expression, but, by association, calling their (hetero)sexuality into question. This (mis)embodiment of masculinity or femininity can then mark subjects not only as gender non-conforming but as “heterosexual failures,” as noted by Ponterotto. Therefore, the simultaneous feminization and infantilization of LGBTQ+ individuals through language can be attributed— regardless of gender— to a social desire to trivialize and desexualize non-normative sexualities. 

 

While LGBTQ+ identities can be associated with sexualized stereotypes, which this article will later discuss, Ponterotto’s work allows us to see how language can be used to reframe cultural narratives of people who “do not respect the [patriarchal] canon of femininity and sexuality,” (2014, 98). Ponterotto’s notion that women are children and embryos (2014) positions femininity as childlike and teleologically incomplete. By this logic, feminized subjects are taken less seriously, are less intelligent, easily misled, naive, and are in a liminal state of development where they cannot be fully considered adults (Ponterotto 2014). Resultingly, infantilization and desexualization begin to interact with one another due to children’s social position as pre-sexual and the taboo surrounding juvenile sex(uality).[2] Ponterotto’s work can be applied as a framework to conceptualize the connection between feminization, infantilization, and desexualization of subjects who transgress normative gender and sexuality expectations.

 

Homosexuality as Deviant

 

While feminization is policed through stigmatizing language, so too are non-normative sexualities. This stigma comes from commonly reiterated stereotypes or archetypes of LGBTQ+ identities, including hypersexuality, “the LGBTQ+ phase,” and imposed platonism. Rubin’s (2006, 153) theory of the “charmed circle” of sexuality is an excellent theoretical tool for examining standards of normative and socially acceptable sexuality. The charmed circle, or the boundaries of “good” sexual subjecthood, includes heterosexual, monogamous, procreative sex, amongst other socially favored forms. Contrastingly, it marks “abnormal” or “bad” expressions of sexuality, such as homosexuality, promiscuity, or non-monogamous sex, on the periphery. The charmed circle constructs certain (heteronormative) relationships as socially safe and familiar, and others as deviant and unknown. Since this system of marginalization discards LGBTQ+ identities and casts them to the margins, their status as unknown and therefore frightening attaches to outward perceptions of their identity, leading to a host of stereotypes which feed off of fears of sexualities cast as hypersexual and deviant. 

 

Herek (1991) identifies hypersexuality as a stereotype of significant concern for the LGBTQ+ community, and notes that it can lead to homophobic attitudes that correlate with increased rates of anti-LGBTQ+ physical assaults, verbal abuse, and discrimination. Therefore, hypersexuality stereotypes tend to affect diverse LGBTQ+ identities in different ways. Gay men have long been villainized for perceptions of their sexuality as monstrous or predatory (Suffredini 2001), amplified during the AIDS crisis where gay male sexuality was pathologized and was perceived as a new medicalized threat to heterosexual populations, reframing public notions of homosexuality and contagion (Florêncio 2018; Herek 1991). Other negative stereotypes tend to rely upon the idea that homosexuality is predatory, fulfilled by fears by straight individuals of being aggressively pursued against their wishes by a gay person, often including age differences and playing into fears of sexual assault (Suffredini 2001).

 

This notion of predatory homosexuality extends into lesbian stereotypes as well, with the ‘predatory lesbian’ claiming a position as a prominent homophobic trope in film and television that embeds itself in social consciousness (Darren 2011; Lott-Lavigna 2015). Media plays a noteworthy role in constructing stereotypes around LGBTQ+ identities, particularly lesbians; hypersexualization of lesbians in media emphasizes the male gaze, portraying lesbians as straight women by all metrics but sexual activity (Hoogland 1995). Ciasullo (2001) deems this character the “luscious lesbian,” a stereotype which portrays (hetero)sexualized lesbians in media as “a kind of lesbian representation that is directed at and meant primarily for a straight male audience—one that typically appears in straight porn films.” (2001, 606). She explains persistent notions that “[straight] men… are aroused by the idea of two women having sex with one another... Male fascination with female coupling is so universal, in fact, that some researchers consider the erotic response to it a reliable indicator of heterosexuality” (Ciasullo 2001, 606). Indeed, lesbian identity has been pornographized to such a degree that, as Khazan (2016) notes, it is one of the most searched terms on pornography websites in the United States. Diamond (2005) relays that the relatedly popular adage of “I’m straight but I kissed a girl” is a leading stereotype for lesbian representation, positing that Queer female sexualities are fetishized through media and pornography, leading to misconceptions about the material realities of these relationships. In addition, she suggests that these representations may influence straight women to commodify presentations of their identities by invoking the idea of sexual fluidity to gain status in the (hetero)sexual marketplace (2005).

 

Homosexuality as Liminal

 

The notion of LGBTQ+ sexuality as a sexually commodified action rather than a felt and lived identity feeds into another common LGBTQ+ stereotype: “the LGBTQ+ phase,” or the idea that Queer identities are childlike fixations that will be forgotten with time. On this note, Shire (2015) proposes that: “…describing a person’s homosexuality as a “phase” is one of the most familiar and dismissive things many [LGBTQ+ individuals] hear when… talking about their sexuality.” This dismissal of same-gender attraction as a viable sexuality is often socialized early in childhood; LGBTQ+ experimentation in youth can be disregarded as  unserious “practice” for a future heterosexual pairing, and while it often remains unspoken and stigmatized, these relationships are constructed as a relatively “safe” form of sexuality and relationship play which children will outgrow (Lamb 2001). While sexuality can be fluid and change over time, Toft and colleagues (2019) affirm that research shows LGBTQ+ individuals typically wait an average of three and a half years after realizing they identify as Queer before ‘coming out’ socially, and that it is overwhelmingly uncommon for these identities to be temporary. Despite these facts, representations of LGBTQ+ identities (especially those influenced by media archetypes) often portray queerness as “a temporary departure on the journey towards adult heterosexuality” (Monoghan 2016, 3).

 

Similarly trivializing is the stereotype of imposed platonism, also known as the “gal pal epidemic” or the “bromance phenomenon” (McBean 2016), where same-gender relationships are ubiquitously reduced to friendships through strategic employment of euphemisms. Sometimes historically documented as “Boston marriages” (Rothblum and Brehony 1993), McBean states that this phenomenon is more common in representations of relationships between two women, noting that employing this framing “appears to indicate a resistance to reading the (confirmed or likely) sexual nature of these pairings, masking the lesbian possibility with euphemisms of friendship” (McBean 2016). She argues that this “… infantilizes sexual relationships between women, comparing them with adolescent intensities- temporary placeholders for future heterosexual couplings” (McBean 2016), a description that exemplifies the interwoven nature of stereotype construction. This relationship miscategorization, paired with the discursive refusal to recognize the pairing as romantically valid, presents the underlying message that the speaker views the relationship as less serious than a heterosexual partnership. 

 

Furthermore, the conflation of same-gender desire with friendship can have complicated consequences; as McBean (2018) explains, platonic homosocial friendships can be misrecognized as romantic or sexual relationships as well, blurring the lines between close friendships and romantic or sexual attraction (Factora 2021; Steele 2018). When homophobic notions like hypersexuality, the LGBTQ+ phase, and imposed platonism permeate, they have a high capacity for internalization by LGBTQ+ individuals, which can lead to mental health struggles, low self-esteem, low sexual satisfaction, and decreases in the functionality and communication of romantic relationships (Frost and Meyer 2009; Logie and Earnshaw 2015).

 

For the purposes of this inquiry, it is crucial to contextualize the nature of these three archetypes. While the LGBTQ+ phase and imposed platonism represent the imposition of heteronormative relationship development standards onto LGBTQ+ subjects, hypersexuality and its marginality in the charmed circle represent a more threatening transgression of normative (hetero)sexuality. Harris’s (2003) idea of “can-do” and “at-risk” subjects is relevant to this analysis; can-do subjects are positioned as capable (and destined for) “successful” subjecthood, while at-risk subjects are positioned as “failures in the making,” (2003, 25) a notion heavily tied to their participation in deviant sexuality. So, in this way, LGBTQ+ subjects can be considered can-dos who become at-risk by embodying an LGBTQ+ identity. It then becomes the job of heteronormative society to “relentlessly manage… [them] back onto a success track” (Harris 2003, 32), that is, a heterosexual identity. Notably, Dumas (2002) posits that infantilization is a device which operates continually to affirm (and, with time, ensure) a “failure to achieve one’s potential” (2002, 1), an imperative connection between this linguistic trend and notions of can-do sexual subjecthood. 

 

The LGBTQ+ phase and imposed platonism target subjects whose (hetero)sexuality is still considered can-do and in need of guiding back onto the normative path, but not yet failing at successful subjecthood (Harris 2003); they will, after all, grow out of their homosexuality with time by the logic of these stereotypes. However, hypersexual stereotypes and performances of sexuality that exist outside the confines of the charmed circle already represent a failure at successful sexual subjecthood, something to be avoided and feared. Ultimately, when cataloging the damage done to the LGBTQ+ community by these three stereotypes, hypersexuality represents more of a material threat than the LGBTQ+ phase or imposed platonism, which fall under a more latent or covert categorization. Both have adverse effects on the societal status of the LGBTQ+ community and implications for internalized homophobia, but the same concern with correcting one’s path to success need not be expended for heterosexual failures, to borrow Ponterotto’s term. 

 

Performative Allyship

 

When examining LGBTQ+ allyship on social media, these stereotypes are of primary importance; objectifying and stereotyping a marginalized group can influence the formation of prejudices and increase stigma, something that allyship practices strive to avoid (Blair 2002). As Dean’s (2016) research on non-homophobic heterosexualities suggests, sentiments discouraging discrimination have led straight individuals to perform their identities in ways that markedly disassociate from homophobia; contemporarily, the undeniable social capital of allyship has created a culture that commodifies alignment with social justice and progressive movements. There is much evidence that this commodification can be partially attributed to and is perpetuated by social media, where portraying oneself as socially conscious may become part of self-branding while being hailed as courageous and socially profitable (Thai and Nylund 2023; Vis et al. 2020; Voogt 2022). 

 

With this in mind, Viki and colleagues’ (2003) theory of benevolent sexism and paternalistic chivalry can be used as a theoretical framework to understand how homophobia is expressed through more covert or unconscious means within allyship-based contexts. They describe benevolent sexism as “compris[ing] a set of attitudes that favor keeping women in restricted roles, but are subjectively positive in feeling tone. Such attitudes may result in male behavior that could be considered prosocial,” (Viki, Abrams, and Hutchison 2003, 533). They also note that paternalistic manifestations of these attitudes “may be marked by extreme politeness and considerate behavior toward women but also place restrictions on the roles women may play ...” (Viki, Abrams, and Hutchison 2003, 534). Using these theories as a model, we can understand that benevolent homophobia can be expressed through paternalistic allyship; the language and attitudes may come across as positive and supportive, but they may reflect deeper biases and cultural desires to police the roles of LGBTQ+ subjects. As Viki and colleagues note, these attitudes reflect and reinforce more explicit forms of marginalization, but they can outwardly disassociate themselves from socially unfavorable discriminatory attitudes through more disguised or condescending measures.  

 

Knowing that hypersexuality is an overtly homophobic stereotype, and the LGBTQ+ phase and imposed platonism are often more covertly harmful, people striving to practice or capitalize on allyship may address hypersexuality as a primary concern for their supportive actions (or may not even register more latent stereotypes like the LGBTQ+ phase or imposed platonism in the first place). Therefore, there is a potential for hypercorrection, where heterosexual allies may fall into patterns of linguistic infantilization in efforts to conceptually oppose objectification. 

 

Based again upon Dean’s (2016) work, to appear supportive of same-gender relationships, heterosexuals may feel a desire to comment on social media posts featuring same-gender partnerships as a way to publicly showcase their allyship and support, distancing themselves from homophobia in the process. But, these comments may inadvertently reaffirm broader notions of frivolity and the childlike nature of same-gender couples through their lack of awareness of more concealed stereotypes. If the primary interest is in distinguishing oneself as non-homophobic rather than sincerely unlearning and deconstructing homophobic bias, these taken-for-granted stereotypes are less likely to be examined and deconstructed than their more overt counterparts. This phenomenon has been deemed “performative allyship,” which Lachman (2018) describes as allyship that can be “commodified by members of the dominant group who reap the benefits of being seen as an ally [without] having to take risks or make sacrifices” (2018, 13). Thai and Nylund’s research further emphasizes the role of weighing positive benefits and negative attributions or risks in distinguishing between sincere and performative allyship (2023).

 

For these reasons, some commentary on LGBTQ+ social media posts can be read through a lens of performative allyship, when commenters are primarily concerned with cementing themselves as non-homophonic allies without any sincere advocacy or ideological deconstruction. While this is not to say that all allyship efforts on social media have ulterior motives— Instagram, as a text and image-sharing app, functions particularly well to spread and share information to mass audiences— the self-conscious visibility of the platform has documented consequences, including self-comparison (Jiang and Ngien 2020). We can then understand the visibility and self-branding as a motivator, however (un)conscious, in these discussions where it would not be in offline activism or efforts without a curated audience. 

 

Speech Accommodation Theory

 

With the above-mentioned stereotypes and allyship motivations in mind, speech accommodation theory is an excellent framework for examining adapted referential language use. Thakerer and Giles first proposed the term in 1981 to describe conscious or unconscious adaptations in language that speakers make depending on whom they are speaking to, mainly influenced by their relative perception of the other. Much sociolinguistic work has been done on this theory since its inception, primarily in the medical field, nursing, and the care of older adults. 

 

Brown and Draper (2003) describe linguistic changes through speech accommodation theory when “people modify aspects of their speech in response to their evaluation of another person... often based on stereotypes. Consequently, a person’s pattern of speech may give clues about their evaluation of another’s competence, and functional and social status. As such, speech can be taken as a marker of one person’s attitude toward another person” (2003). Their application of this theory shows that, when caring for older adult patients, healthcare workers may take an infantilizing tone or increase their use of patronizing terms of endearment. Brown and Draper describe this as over-accommodation, a form of speech adapted to stereotypical perceptions of older adults as ingenuous that “resembles that which an adult would use when speaking with a language-learning child” (2003, 16). They note that this over-accommodated speech has harmful psychological effects on the recipients, who understand that inappropriately infantilizing, paternalistic, and condescending language directly results from the speaker’s assessment of their competency and social status.

 

A strong comparison can be drawn between this documented application of speech accommodation theory and instances of infantilizing language used with LGBTQ+ individuals on social media. Speakers’ desires to appear non-homophobic (and therefore avoid acknowledging hypersexual stereotypes) and stereotypes which prime speakers to associate LGBTQ+ identities with childlike qualities, liminality, and frivolity work cooperatively. Speakers, therefore, make a covert-stereotype-based linguistic change when commenting on LGBTQ+ relationship posts on Instagram as a form of over-accommodation and a self-conscious performance of allyship. These actions may be well-meaning—Keisling (2019) and Kitzinger (2005) affirm that heteronormativity’s prevalence in language means that heterosexism can present itself through speech in ways that speakers may not even be aware of. Nevertheless, intentions aside, the prevalence of these stereotypes in language is imperious, repetitive, and over-compensatory.

​

Methodology

 

Data Collection Methodology

 

In addition to a literature review, I conducted this research with a threefold approach. The first stage involved a discourse analysis of approximately 3000 comments on Instagram posts and pages featuring both opposite-gender couples and LGBTQ+ couples. The second stage was a survey, followed by the final stage of follow-up interviews with participants who either completed the survey or chose only to complete an interview.  

 

Identifying qualifying pages for social media research must consider ethical guidelines for internet research as well as methodological suitability— intended audience is a key consideration. For this reason, the pages used in the dataset were all public profiles accessible from a browser without an Instagram account. Public pages were also chosen based on users’ willingness to showcase their romantic relationship, and were identified by SEO keyword/hashtag searching and algorithmic suggestion. There are social media markets for both heterosexual and LGBTQ+ couple pages, and many couples who run lifestyle pages heavily featuring their relationship do so specifically to reach a large audience for personal or financial gain. The identities of non-celebrities (including influencers) and all commenters were anonymized.

 

An important consideration during social media data collection is that algorithmic and general account popularity may be biased. For example, specific characteristics or content genres could prevent posters from gaining a large follower platform due to racism, homophobia, fatphobia, or ableism, among others. Algorithmic bias like this is prevalent on Instagram (Alrasheed and Lim 2021), and to combat bias and represent these groups more proportionately, keyword searches were amended when appropriate. While these measures cannot truly eliminate algorithmic and follower bias from the data set, deliberate efforts were made to counter these forces when possible while avoiding artificial inflation of the data pool. For future research on this topic, while a certain degree of researcher selectivity cannot be eliminated, a larger sample size or varied methodological approach would be an asset. 

 

While the public accessibility of a profile was the only true constraint for eligibility, selected posts needed to feature both members of the couple with the romantic nature of their relationship easily identifiable from body language, captioning, or page context, could not be milestone or event-themed, could not be an advertisement or paid promotion, and, when possible, should not feature children. Up to the first 100 comments from each post were analyzed to help mitigate selection bias and standardize analysis while allowing significant data to represent each post. In total, 1507 comments were analyzed from 17 posts featuring opposite-gender couples, and 1501 comments were analyzed from 24 posts featuring LGBTQ+ couples, for a total of 3008 comments; either due to visibility, popularity, or engagement level, LGBTQ+ posts in this dataset received fewer comments on average.

 

Discourse Analysis Methodology

 

Once the comments were exported, each dataset was thematically organized into eight categories: Childlike Indicators (CI), Quality of the Match (QM), Favouritism (FV), Explicit Life Steps (ELS), Speculatory Life Steps (SLS), Positive Emojis, Negative Emojis and Symbolically Separating Comments (SSC). 

 

CI was a category of particular interest, and included phrases[3] such as “so cute!!,” “AWE! She is so fucking ADORABLE!,” “Cuties! ❤️,” “Mygod so wholesome,” “This is soooo pure,” “Awww how sweet ❤️,” or “Precioussss 😍😍.” These words positioned the subject as infantilized or childlike, consciously or inadvertently. As indicated in Huot (2013), language like “cute,” causes subjects to be perceived as less respectable and mature but more childlike. This mobilization of infantilizing language can reinforce superiority dynamics between the speaker and referent (Martin, 2016). 

 

QM phrases included “Couple goals❤️,” “The definition perfection wow 😍😍,” “Such a beautiful couple! ❤️🔥,” or “soulmates is an understatement.” These comments assessed the value of the couple as a whole and the appropriateness of the pairing. FV comments included phrases such as “God LOVE this couple!!!!!,” “my favs 😭😭💞💞,” and “I love these two,” and served to remark on the commenter’s subjective evaluation and appreciation of the couple comparatively.

 

ELS comments directly reference life steps or relationship milestones, such as “please invite my wife and I to your wedding” or “You guys are married everyday.” Similarly, SLS muse about the possibility of life steps or relationship milestones; for example, “They need a real baby,” or “you’ll be the prettiest parents ever omg.” Positive emojis were comments that mainly or entirely consisted of emojis with pleasant connotations, such as hearts, rings, or flowers. Contrastingly, Negative Emojis were comments that consisted mostly or entirely of emojis with negative connotations, like angry/sad/sick faces, weapons, certain animals (snakes, cows), or emojis accompanied by hateful speech.

 

SSC included comments such as “Mom & Dad 😍💛,” “ADOPT ME PLEEEEASE,” “wow my parents in the sun 🤧,” or “Mother and daughter?! 😮.” They function within a theoretical framework of sexual subjecthood; that is, who is considered (or granted the status of) a sexual subject. The term “symbolic separation” works in conversation with concepts such as symbolic annihilation (when a social group is represented only negatively or is completely absent from media) and other representative violences (Caluya 2006; Mulvey 1973; Owen 2018; Stonewall 2015). To symbolically separate a subject is to remove the possibility of sexual agency from their outward narrative by positioning them as incapable, unwilling, or socially unable to participate in sexuality and sexual agency. For example, children or relatives are considered symbolically separate from one’s own understanding of their sexuality, which can be attributed to many social factors ranging from (incest) taboos or explicit laws that prohibit these subjects from engaging in relational sexuality wholly or within specific contexts. Family members or parents would fall under this category due to the socialized taboo or “cultural inhibition” of acknowledging a relative’s sexual agency (Amann-Gainotti 1986), and their sexuality is stigmatized within these contexts. So, SSC remove the possibility of sexual subjecthood by framing the poster(s) as the parents of the commentator, related to each other, strictly platonic, or as children. In this way, CI works in conversation with SSC, and they were frequently found together.

​

Survey Methodology

 

To complement the actual-use dataset, a survey was released to better understand subjectively reported language use. The survey provided a qualitative measure to collect background information and subjective interpretations of language use and cultural understanding, supported by quantitative and limited statistical data that could fill any gaps in analysis caused by reporting bias or social desirability bias (Rosenman et al. 2011). The survey received over 180 responses and focused on three categories of questions: word association (aimed at gaining contextual insight into the semiotics and pragmatics of words and phrases), LGBTQ+ identity and perception (aimed at understanding the prevalence of LGBTQ+ stereotypes and the cultural understanding of these identities and relationships), and social media (aimed at understanding the self-reported usage, culture, and pressures of social media). 

 

Survey Demographic Data

 

Survey demographics help better understand the applicability of findings at large. 83.53% of survey respondents identified themselves as between the ages of 18-24, 9.41% as 25-34 years, 2.35% as 35-44 years, 4.12% as 45-54 years, and 0.59% as 55-64 years. Self-reported ethnicity was more than half White/Caucasian (56%), but also included East Asian (13%), South Asian (7%), Southeast Asian (4%), Hispanic (4%), Middle Eastern (4%), Black (3%), Latinx (3%), Indigenous (First Nation, Metis, Inuit; 2%), and Other (5%). While the survey was accessible regardless of geography, approximately 90% of respondents listed Canada as their country of residence, with the remaining 10% indicating the United States of America, Australia, China, Guatemala, Jamaica, Qatar, Hong Kong, India, Germany, and Russia. This element of the demographic information was to establish any underlying cultural or identity-based contexts that may inform survey responses.

 

The demographic data also asked identity-based questions that would help inform the context of the data in terms of gender identity and sexual orientation. 65% of respondents identified as Cisgender (55% female, 10% male), followed by Non-Binary (8%), Gender Non-Conforming (4%), Transgender Male (4%), Gender Fluid (3%), Agender (2%), Feminine of Centre (2%), Two-Spirit (2%), Transgender Female (1%), Masculine of Centre (1%), and Other (6%). For sexual/romantic orientation, 34% of respondents identified themselves as heterosexual, followed by Bisexual (14%), Queer (12%), Lesbian (8%), Asexual (7%), Questioning (7%), Homosexual (5%), Pansexual (5%), Gay (4%), Aromantic (4%), and Two-Spirit (1%). No respondents selected Other. The majority of respondents were cisgender 18-34-year-old Canadian residents of White/Caucasian or Asian descent, many of whom fell under the LGBTQ+ umbrella.  

         

Results and Discussion

            

The discourse analysis, which included 3008 Instagram comments on relationship-showcasing posts, showed framings of LGBTQ+ relationships as more childlike, less permanent and serious, and having less potential for sexual agency than opposite-gender couples’ posts. In both LGBTQ+ and opposite-gender posts, positive emojis were the most common type of comment, accounting for 19.27% and 22.34% of the total comments, respectively. Michelle, a heterosexual interviewee, suggested that positive emojis can be a substitutive feature for interactions that require only positive feedback or acknowledgement, particularly in acquaintance relationships where personal rapport may not yet be established. She described emojis as an invaluable tonal indicator that can assist in portraying meaning in writing-based contexts and stand alone as an interpretive object. Negative emojis were also typically co-speech or illustrative (McCulloch 2019), such as vomiting and angry faces, but occasionally became more emblematic when referencing more complicated cultural messaging. Notably, negative emojis were 454% (5.54x) more likely to appear in the comments of LGBTQ+ posts than opposite-gender posts in this dataset.

 

Comments on LGBTQ+ posts led in only two of the thematic categories identified: they received 90% more CI and 50% more SSC overall. However, opposite-gender posts led in ELS (receiving 1363% more comments), SLS (receiving 416% more comments), FV (receiving 44% more comments) and QM (receiving 66% more comments) in this dataset. These results have interesting implications for the differential linguistic treatment of LGBTQ+ identities and relationships through referential language on social media. They demonstrate that LGBTQ+ relationships are being referred to in infantilizing terms nearly twice as often as opposite-gender couples, and are being symbolically separated with similar consistency. These couples are also receiving substantially fewer comments that position them as having a serious future together or acknowledge the seriousness of the relationship, with approximately 1900% fewer comments than their opposite-gender counterparts in these combined categories. Additionally, these numbers suggest that comments acknowledging the quality of the partnership and admiration amongst peers for LGBTQ+ couples are significantly outweighed by those that connote their childlike, impermanent, unserious, and desexualized status. 

 

These findings correlated with survey responses, which show that “cute” and “adorable” were the leading descriptors for play relationships between children among respondents. “Cute” was overall the most frequent description for young children regardless of gender, accompanied by “sweet” and “adorable” as adjectives for female children, affirming the gendered nature of infantilism (after “cute,” adjectives for boy children of the same age tended to emphasize curiosity, athletics, and energy levels rather than physical characteristics). With these crowd-sourced adjectives in mind, it is especially notable that 87% of participants found that words like “cute” and “sweet” carried a dismissive connotation as opposed to words like “committed.” Notably, “committed” was the most frequently suggested descriptor for long-term heterosexual dating relationships. This trend continued, as participants were slightly more inclined to perceive a long-term heterosexual pairing as viable for commitment-based life steps such as marriage. However, interviews with participants suggested that this discrepancy could be related to perceptions that LGBTQ+ individuals may be less likely to participate in the institution of marriage, may prefer anti-establishment or alternative commitment styles such as cohabitation, or may face governmental or religious barriers in doing so rather than perceptions that LGBTQ+ relationships are incapable of serious commitment.

 

Reflecting on the stereotypes which may have influenced these language accommodation trends, LGBTQ+ respondents substantially reported (nearly 50%) that they had been told their identity was a phase “frequently,” most commonly by parents, peers, and friends. Contrastingly, heterosexual participants overwhelmingly reported that they had never been told their identity was a phase (nearly 70% noted that they had never been told this), despite some participants indicating that they felt others may have thought this but had never been told directly. Heterosexual respondents reported that they “never” told others their relationship was a phase (8.6%), an inconsistency between groups’ reports that may reflect the specific respondents’ experiences, survey phrasing, or reporting bias. The notion of the LGBTQ+ phase was reported to be most heavily internalized by those who identified as lesbian or pansexual. When asked why she felt beliefs of the LGBTQ+ phase persist, Michelle said:

 

Because… I think that there are a lot of people that have been conditioned to believe that it isn’t normal. So [they] address it as ‘this person is going through a phase, they’ll come back over to where they need to be… at some point.’ Bottom line, I think if we can call it a phase, then we don’t have to learn. We don’t have to know, we don’t have to address our own thoughts, our own feelings about it, our own sexuality… We don’t have to learn… that everybody isn’t like us, right? If we just call it a phase then it’s gonna end.

 

Michelle’s description calls back to Harris’s theory of can-do citizenship and the desire to encourage subjects to “come back over where they need to be,” or, as Ahmed states, “the homosexual subject... gets read as having got lost on the way ‘toward’ the ‘other sex’” (2006, 79). Nick, a heterosexual male, added that he felt the expectations of straight parents that their children embody heteronormative sexuality may be another reason that this stereotype persists, a sort of sexuality policing that again returns Rubin’s notion of the charmed circle (2006) and the desire to keep subjects on the “straight line” towards successful heteronormative citizenship (Ahmed 2006; Rubin 2006). 

 

Additionally, several LGBTQ+ participants noted that family members had used platonic terms to refer to their romantic relationships euphemistically, such as “special friend” when describing the pairing. LGBTQ+ respondents also reported pet names that situated them in parental roles, such as “my lesbian moms” or “mom,” by friends on social media, which they recalled as notably gendered terms of endearment.

 

Instagram, a site well-known for its performative activist presence (Bailey 2021; Shadijanova 2020), was the host platform for the discourses analyzed in this research. Survey respondents of all sexualities noted to varying degrees that they felt posting on social media platforms like Instagram informed people about their stances on politics, that they had witnessed others using social media to share posts about allyship and social justice, and that they used social media to absorb information about their friends’ social values and interests. 

 

However, puzzlingly, respondents widely noted that they did not feel their social media pages reflected their own social values and interests; this understanding shows a dissonance between how respondents perceived peers’ intentions when using social media (a sincere reflection of their political alignment and interests) and how they perceived their own (more accurately a gesture towards their political alignment and interests, but not wholly reflective). Essentially, these answers suggest an individual in-group bias, which makes generalizations about others which do not apply to oneself.

 

While participants did think social media was an effective facilitator for allyship, they reported that they felt moderate pressure to participate in activist movements online or present themselves in particular ways. Survey results showed that LGBTQ+ individuals may be more engaged in sharing posts about social issues, perhaps in part because they more strongly indicated that they used social media to reflect their interests, but were less inclined than heterosexual participants to think this constituted effective allyship (47% of the time compared to 60% of the time for heterosexuals). Heterosexual respondents reported they were most likely to comment on a post featuring a friend and their dating partner to show they support the relationship and because they felt it would make the poster happy. However, LGBTQ+ individuals more frequently suggested they would do so to make the poster happy and compliment them on their partnership. Comparisons between the group responses showed that heterosexuals felt more strongly than LGBTQ+ individuals that positive comments were good demonstrations of allyship, made the poster feel supported and validated, and that comments like “good match” or “perfect” were equally validating as “cute” and “sweet.” Altogether, these responses suggest that heterosexual respondents saw positive social media comments on LGBTQ+ relationship posts as stronger displays of allyship and saw less connotative differentiation between infantilizing and quality-based comments than their LGBTQ+ counterparts, who felt comments were ineffective forms of allyship.

 

Overall, the discourse analysis and survey data paint a convincing picture of the position of LGBTQ+ identities in linguistic treatment. As scholars like Huot (2013) affirm, stereotype threat — or, the idea that stereotypes can be internalized and incorporated into one’s self-image — is a particular concern regarding infantilizing language, as it makes a compelling case for “language as a social reality,” as Huot argues (2013). Berg and colleagues (2016) also reinforce the idea that linguistic microaggressions and negative attitudes towards LGBTQ+ identities have vast potential for internalization, which are reflected in diminished perceptions of one’s own relationship and social validity, further affirmed by Frost and Meyer (2019). Moreover, as these connotations are internalized and reproduced by LGBTQ+ individuals, they can reinforce the stereotypes this language use emerged from within the LGBTQ+ community, further entrenching social marginalization (Martin 2016). The language we use, particularly on influential forums such as social media, can therefore have profound implications not only for the persistence of LGBTQ+ stereotypes in wider heteronormative society but also on self-image, language use, and community perception within the LGBTQ+ community.

 

Conclusions

 

The way one uses language reveals much about the speaker, their attitudes towards the people they address, and unconscious societal ideologies ingrained in language. As Kiesling (2019) and Kitzinger (2005) proposed in their respective work on salient themes in language, heteronormativity plays a significant role in the way we code the world automatically. This normativity can negatively impact individuals who do not conform to the norms of a heterosexual world; in fact, it can even shape the narratives used to conceptualize these marginalized groups. 

 

Applying analyses of metaphoric taxonomies of feminization and infantilization, LGBTQ+ stereotypes such as hypersexuality, the LGBTQ+ phase, and imposed platonism to survey responses and Instagram comments, it becomes clear that infantilizing language can be mobilized selectively against LGBTQ+ relationships at higher rates than opposite-gender pairings. Using the notion of can-do citizenship (Harris 2003), non-homophobic heterosexualities (Dean 2016), the charmed circle of sexuality (Rubin 2006), and speech accommodation theory (Brown and Draper 2003), we can understand that these expressions of referential language on social media may reflect benevolent homophobia and paternalistic allyship, an impulse which heterosexual allies may feel performs social solidarity for LGBTQ+ individuals while presenting themselves in a way that increases social capital. However, these infantilizing and symbolically separating linguistic trends fall into patterns that can trivialize LGBTQ+ individuals and relationships by relying on covert stereotypes to over-accommodate these notions in speech. As a result, heterosexist social hierarchies are routinely affirmed through infantilizing referential language directed at LGBTQ+ individuals on social media, an assertion that LGBTQ+ individuals are actively internalizing. 

 

Social attitudes towards homosexuality have taken massive strides in recent years, a fact that should be celebrated and encouraged. I aim not to suggest that social media users eliminate words like “cute,” or “adorable,” from their referential vocabulary; these words do not carry an inherent moral value. I also do not aim to suggest that only heterosexual individuals are capable of or participating in the infantilization of LGBTQ+ individuals— research and data would show that, in many cases, these trends of infantilization end up being mobilized by the very communities at which they are aimed (Monaghan 2016). 

 

However, these findings suggest that we can all be more critical of how we use language referentially and how those choices may impact others; while our speech may feel like collections of singular interactions, they reflect and contribute to wider patterns and ideologies. Allyship is a practice that requires constant reflexivity and reflection on internal biases. Amending a quote by Viki and colleagues, it is possible for heterosexual allies to be supportive and considerate to LGBTQ+ individuals without simultaneously placing restrictions on how these individuals and their relationships can exist in digital spaces (2003, 534).

 

[1] Elements of this research design refer to gender in binary terms, such as “same-gender” or “opposite-gender,” for the purposes of situating the data within the structure of larger social and power relations created by gender binarism and heteronormativity as mainstream/normative discursive frames.

[2] While sexist infantilization can also have the opposite effect, creating a form of sexualization and objectification that relies on the notion that childlikeness is sexually desirable, that form is one which this particular project does not seek to analyze. 

[3] Direct quotes edited only for spelling when necessary for legibility.

​

Acknowledgements     

I would like to acknowledge the Department of Anthropology at Western University for their support of this research through an Undergraduate Summer Research Internship (USRI), under the supervision of Dr. Tania Granadillo. Thank you to Dr. Granadillo for her support, advice, and assistance in this project. Thank you to my survey respondents and interviewees for sharing your perspectives with me. This research received NMREB approval in July, 2021 (REB # 119238).