GIF You’re Happy and You Know It: Reaction GIFs and Images in a Gay Male Twitter Community of Practice

Abstract

Reaction GIFs and reaction images appear as common multimodal linguistic objects in digitally mediated communication. While past research has tended to focus on the paralinguistic functions of these communicative devices, less attention has been paid to how these digital tools enable their users to strategically enact and embody particular social identities on social media. This article offers a stance-based computer-mediated discourse analysis of a small, gay UK- and Ireland-based Twitter community of practice. Through qualitative analyses of the eight members’ tweets containing reaction GIFs and images (n = 991), as well as their responses to an online survey, this article demonstrates how these self-identified gay men construct four distinct feminine-coded personae: the Sassy Queen, the Hun, the Battle-Axe and the Flamboyant Queer. Each persona exhibits linguistic (e.g., features from British English or African American Language) or stance-based collocations. This analysis identifies common qualities or traits that all four personae possess that these Twitter users may identify with or value, potentially motivating their recurrent constructions. The ability of these non-traditional linguistic resources to conduct identity work is discussed. More broadly, this study foregrounds the significance of social media as a series of digital platforms where online identities are continually developed, (co-)constructed and negotiated.

1. Introduction

For over one hundred years, authors from various disciplines (including psychology, sociology and, importantly, linguistics) have explored the idea that gay men exhibit linguistic behaviours that are distinct from their heterosexual counterparts (Cameron & Kulick, 2003; Pavia, 1910). A particular recurring focus has long been placed on how some self-identified gay men use linguistic features that index femininity, such as pronouns (she, her), gendered slurs (bitch, slut), sex terms (pussy) and so forth (Baker, 2002; Cory, 1951; Legman, 1941/2006). These feminine-marked linguistic items serve numerous communicative functions. Past linguistic research has primarily focused on how the indexing of femininity in social contexts enables gay men to construct and perform a diverse array of gay identities, both offline and online (Barrett, 2017; Ilbury, 2020). These identities are forged via the prevailing sociocultural association between effeminate/non-masculine ways of behaving or speaking and ‘gayness’ (Halperin, 2012). In probing how linguistic tools are strategically marshalled to enact digital gay identities or personae, existing sociolinguistic studies have largely focused on traditional linguistic resources (e.g., lexical items, phonological features, etc.) (Ilbury, 2020, 2022). This is despite the fact that the late twentieth and early twenty-first centuries have afforded internet users a range of relatively new communicative devices which, as of yet, remain comparatively understudied in terms of their potential to construct and convey online queer identities. To add to this sociolinguistic literature, the present study will focus on two multimodal linguistic objects: reaction GIFs and reaction images.1 Specifically, this article will show how these objects are employed in diverse and meaningful ways by the members of a gay, male Twitter (now X)2 community of practice (Eckert & McConnell-Ginet, 1992)—comprised of eight White British and Irish, gay, male Twitter users in their twenties and thirties—to digitally do ‘being gay’.

GIFs are looping, audio-free animations of the .gif (Graphic Interchange Format) image file type, created in 1987 (Fan, 2022). A specific form of GIF is reaction GIFs, which provide “stand-ins for users’ own verbal and nonverbal (re)actions by animating GIF characters’ speech, facial expressions, gestures, and body movements” (Lindholm, 2024, p. 104). Reaction GIFs are often used to capture and circulate extracts from popular media (television series, films, etc.) and apply them to new conversational scenarios. The introduction of in-built GIF keyboards on smartphones and social media interfaces has enabled users to quickly locate and send GIFs that fulfil their interactional goals (Fan, 2022). As such, users have the choice to employ pre-made GIFs that match their search criteria or even create and send their own if they so wish. Similar to reaction GIFs are reaction images, a term this paper uses in specific reference to static image files (.png, .jpeg, etc.) that convey a reaction to a (hypothetical) situation or conversational turn. The major difference between reaction GIFs and images is animatedness. It has been argued that the animated quality of GIFs increases their communicativeness (Miltner & Highfield, 2017; Church et al., 2023), as their portrayal of movement can express complex emotions (e.g., a sudden change from happiness to sadness) and exaggerate emotional responses. A benefit of reaction images, however, is their ease of creation. While creating reaction GIFs can require specific editing software or the use of online GIF makers, reaction images are typically more straightforwardly produced, whether they be screenshots from online videos, photographs taken by the users themselves, or otherwise.

Some linguistics-based/-informed research has been conducted on reaction GIFs and reaction images (Diedrichsen, 2020, 2022; Lindholm, 2024; McCulloch, 2019). Such research has demonstrated how these items can convey paralinguistic information by standing in for nonverbal cues (e.g., shrugs, eyebrow raises) (McCulloch, 2019), construct and reinforce in-group status (Miltner & Highfield, 2017), convey affective stances (Tolins & Samermit, 2016), function similarly to conversational routines and situation-bound utterances (Diedrichsen, 2020), and so forth. Yet, less research has examined these linguistic objects with specific reference to their potential to enact and embody digital personae; an under-explored area that this paper duly addresses. To do so, I will qualitatively analyse a manually collected corpus of tweets (n = 991) from eight self-identified gay, male Twitter participants through a stance-based computer-mediated discourse analysis (Du Bois, 2007; Herring, 2004). As a result, this study will highlight the immense potential that these multimodal items possess in terms of accomplishing identity work on social media. Specifically, this article will show that four unique archetypes are digitally embodied by this study’s participants—the Sassy Queen, the Hun, the Battle-Axe and the Flamboyant Queer—each with specific linguistic or stance-based collocations. While these personae examinations are largely Twitter data-based, this investigation will be supplemented and analytically consolidated by brief considerations of the participants’ answers to a short online questionnaire, as detailed in Section 4.1. This work thus foregrounds that reaction GIFs and images merit further consideration from sociolinguists interested in the complex relationship between (online) language and identity.

This article is organised as follows. Section 2 situates this paper within ongoing sociolinguistic discussions that consider how language use enables the performance of gay identities. This section will achieve this by providing a short overview of studies that have explored how some gay men perform identities by exploiting the indexical link between (linguistic) femininity and gayness. Section 3 then summarises existing works that discuss reaction GIFs or images in relation to online identity performance. Section 4 details the methodological and analytical approaches taken in this study. Next, Section 5 introduces the Twitter community of practice (CofP) under analysis. Finally, Section 8 concludes the study by summarising its significance and outlining potential future research directions.

2. Gender-Marking as an Index of Sexual Identity

Following Ochs’ (1992) work on in/direct indexicality and Eckert’s (2008) concept of the indexical field—in which linguistic forms can be connected to “a constellation of ideologically linked meanings, any region of which can be invoked in context” (Eckert, 2012, p. 94)—sociolinguistic research has frequently considered how feminine-coded linguistic resources can be used to perform gayness. Podesva (2007, 2008, 2011), for instance, has produced numerous studies on the language of gay men. To illustrate, in his examination of the language of Heath, a gay medical student, Podesva (2007) analyses the subject’s use of falsetto across various settings: a barbecue with friends, a phone call with his father and a meeting with a patient. Most notably, Podesva states that in the barbecue setting—where falsetto is used most—Heath demonstrates a wider fundamental frequency (F0) range, a longer mean falsetto duration and a higher mean F0 maximum in falsetto utterances than in the other settings. Podesva argues that this falsetto use is strongly associated with ‘expressiveness’, which Heath takes advantage of to construct a flamboyant diva persona. This construction is made possible “due to the dominant ideology that performing expressiveness is a non-normative behavior for men” (Podesva, 2007, p. 495), meaning that expressiveness is socioculturally coded as feminine. In a later examination of the same speaker, Podesva (2008) expands on his earlier work by highlighting how a variable like falsetto can contribute to the construction of multiple personae for a single speaker in different (linguistic/conversational) contexts. In a similar vein, Podesva (2011) examines the intonational patterns of three gay professionals’ speech. In this study, extreme F0 excursions in the speech of his participants Heath and Regan are seen to convey ‘animatedness’. Podesva relates this to the iconisation of pitch changes as an index of emotion, which consequently connects to the stereotype of women being ‘overly emotional’. However, Regan and Heath both use these drastic F0 excursions (which are thus associated with femininity) to produce two quite different personae: the gay partier and the diva, respectively. These distinct personae emerge because of the linguistic contexts in which the extreme falls occur (e.g., in discussions with friends about tequila bars and partying versus expressing indignation during a debate about shampoo).

Also interested in how men index gay identities via pitch is Levon (2009). Levon organised numerous Israeli gay and lesbian activist groups into three main categories: the ‘Mainstream group’, the ‘Radical group’ and the ‘Community Centre’ group. Interviews were then held with participants from each of these groups, covering two differing discourse types. ‘Narratives’ involved discussing one’s personal life (education, family, coming out as gay, etc.), while ‘opinions’ involved discussing wider social and political matters (the Israeli–Palestinian conflict, gay rights, etc.). Within both ‘narratives’ and ‘opinions’, a further distinction was made between ‘gay’ and ‘non-gay’ topics. Arguably, Levon’s most striking findings relate to the Mainstream men, who maintain an integrationist approach to their activism, and who Levon argues index two separate gay personae across the two conversational frames. When talking about ‘gay opinions’, the Mainstream men speak with a higher mean pitch (compared to when they talk about ‘non-gay opinions’), which Levon describes as being associated with a stereotypically feminine style of speech. Levon (2009, p. 48) argues that this depicts a “more public, out-group presentation of self that corresponds to (positive) Israeli stereotypes of gay men” as being members of a ‘cool’, identifiable social type. In contrast, within the ‘narratives’ frame, Levon (2009, pp. 48–49) states that gay men speak with a lower mean pitch when discussing gay-related matters (as opposed to non-gay narrative topics) to construct “a more private, in-group” self-presentation. This self-presentation is linked to masculinity ideals in the Israeli gay community. This importantly demonstrates how style shifting can be both responsive (i.e., adapting one’s self-presentation depending on the conversational frame) and initiative (i.e., constructing distinct gay personae within these frames).

Finally, Barrett (2017) analyses six diverse gay male subcultures, revealing how these different groups index masculinity and/or femininity via language and other signs. Importantly, Barrett stresses that these subcultures each hold different ideologies about the link between sexual identity and gender and that they conceptualise gender in very different ways. This suggests that it must be carefully considered what ‘masculinity’ and ‘femininity’ actually mean in the particular group(s) under analysis. To illustrate, while the African American drag queen subculture is shown to index their sexual identity through femininity (by using a ‘White-woman style’ of speech that includes features like hedging, hypercorrect grammar, etc.), this femininity is “highly constructed” and is oftentimes associated with high-class glamour (Barrett, 2017, p. 217). In contrast, members of the radical faerie subculture unreservedly mix feminine and masculine signs (like having facial hair while wearing stereotypically feminine articles of clothing), treating femininity as “the expression of a natural and essential aspect of gay identity” (Barrett, 2017, p. 217).

3. Identity Performance Through Reaction GIFs and Images

Previous scholarly works have accentuated how multimodal embodiments of identity can be accomplished online. Jackson’s (2019) work is significant because it spotlights the disproportionate depiction of (particularly female or feminine) Black individuals in the reaction GIFs and images used by non-Black people (McCulloch, 2019). Jackson relates this overrepresentation to the concept of ‘digital blackface’ and argues that Black people are regarded by non-Black individuals as “superfluous” except for when they feature in reaction images and GIFs and let “the likes of Meghan McCain and scores of white gay men find their inner diva” (2019, p. 98). This immediately highlights how Black femininity can be indexed and performed via images and GIFs to construct a particular sassy, bold or spirited gay persona. As Miltner (2018, p. 298) contends, Jackson’s work ultimately reveals the fact that “GIFs and GIF keyboards are in fact deeply political and operate at the crossroads of identity and power”.

Miltner’s (2018) own study primarily explores cases where heterosexual women index drag femininity through the RuPaul’s Drag Race Keyboard app (2015–2017). This app enabled users to send emojis, ‘stickers’ (which are similar to reaction images) and GIFs that depict the stars of the (originally American) reality television series RuPaul’s Drag Race, in which drag queens compete to earn the title of ‘America’s Next Drag Superstar’. Despite the show’s primarily male audience (according to this study), Miltner states that approximately fifty per cent of the keyboard’s users were female. Miltner (2018, p. 292) explains that heterosexual women indexing drag femininity through the resources provided by this app allow heterosexual women to reject the “[p]erpetual self-negation” that is “at the heart of hegemonic femininity” by drawing on the self-resilience and exaggerated femininity of the show’s drag queens. These GIFs and other resources also enable heterosexual women to perform a form of femininity that is bawdy, bitchy and brash. That is, a version of femininity that society has deemed unladylike and uncivil. Also interested in women’s uses of reaction GIFs is Kuo (2019), who specifically analyses how women use feminist reaction GIFs to express anger, dissatisfaction and frustration. Of particular relevance to this article is Kuo’s consideration of how these GIFs enable women to perform (and reclaim) the feminist killjoy persona, a persona that openly celebrates “the unhappiness of those who are iconizations of white supremacy and patriarchy” (2019, p. 190). Notably, Kuo argues that the feminist killjoy is both gendered and raced, with the associated GIFs often depicting White little girls (along with parodic ‘White male tears’ GIFs). This is important because the reaction GIFs referenced in this study aid in the manifestation of a persona that is quite distinct from the sassy, feminine, Black caricature that Jackson (2019) describes. As such, it is already apparent from these studies that different reaction images and GIFs allow their users to index different versions of femininity, which may well be relevant when considering how gay men use reaction GIFs and images in their online communications.

Writing more recently on the uses of GIFs and memes is Wald (2024), who is particularly interested in GIF files featuring the African American reality TV star Tiffany Pollard. Noting that appropriations of Black affectations doubtlessly do occur within White/non-Black (queer) networks, Wald contends that examining Tiffany Pollard GIF/meme circulations through the lens of queer ironic consumption enables a presumption of jocularity and admiration for Pollard, rather than contempt or ridicule.

4. Approach

4.1. Methodological Approach

This study’s participants were recruited via the Twitter search tool, with participation ultimately entailing the subjects (1) consenting to the collection and analysis of their relevant tweets and (2) providing responses to an online survey (more below). As the Twitter search tool does not permit searching via images and GIFs, the decision was made to search for participants using the search term gay in combination with the emoji ‘face with tears of joy’, which, as of January 2022, was Twitter’s most used emoji (Broni, 2022). This work is derived from a longer doctoral thesis chapter that also considers emoji use within the same CofP (O’Neill, 2025), hence why an emoji was used for this search. When scrolling through the search results, if a potential participant was identified as fitting the participation criteria (gay, cisgender, male) and a glance at their profile indicated that they had some regular interaction with other individuals who also fit these criteria, each member of this group was contacted via Twitter’s direct message (DM) function. These users were also asked to refer any of their Twitter acquaintances who they felt may be interested in participating (i.e., snowball sampling).

Though it was not a participation requirement, a slight preference was held during recruitment for CofP groups with members who possessed similar demographic backgrounds to the researcher (i.e., a White, Irish, English-speaking male in his late twenties who has lived in England for several years). The rationale for this was that close sociocultural links to the group under analysis could provide a closer understanding of the data (e.g., of British or Irish pop culture references from a particular era). As will be explained in Section 5’s overview of the CofP, eight White British and Irish, gay, male Twitter users, all aged in their twenties and thirties, were eventually recruited.

After indicating provisional interest and providing the researcher with their email addresses via DM, the participants were emailed a participant information sheet and an informed consent form, which explained how their pre-existing tweets would be collected and analysed. After returning their consent forms, the participants completed a short online survey, hosted on https://www.onlinesurveys.ac.uk/ (accessed on 23 March 2022). This survey gathered some basic sociodemographic information and queried their use of reaction GIFs and images (see Appendix A for a full list of survey items). For the purposes of the present article, where the main analytical focus is firmly placed on the Twitter data, only some of the questionnaire results will be discussed. Specifically, only those survey items that evoked responses deemed particularly analytically insightful and/or immediately relevant to the upcoming arguments will be drawn upon.

In total, 991 tweets were collected, spanning an almost ten-year period (15 August 2012–20 May 2022). In total, 829 of these tweets contained a GIF (including 3 with broken links) and 162 included a reaction image. Ahead of their detailed introduction in Section 5,

Table 1. GIF use breakdown.

User	Number of Tweets Featuring a GIF	Percentage of 829 Total
@determinedapple13	344	41.495778045838364
@proudchild22	220	26.537997587454765
@gentleinflation46	91	10.97708082026538
@widehat91	74	8.926417370325694
@normalratio59	60	7.237635705669482
@ourhistory05	33	3.9806996381182147
@famouswords86	6	0.7237635705669482
@consistenterror70	1	0.12062726176115801

and

Table 2. Reaction image use breakdown.

User	Number of Tweets Featuring a Reaction Image	Percentage of 162 Total
@widehat91	58	35.80246913580247
@proudchild22	54	33.33333333333333
@ourhistory05	23	14.19753086419753
@gentleinflation46	14	8.641975308641975
@determinedapple13	12	7.4074074074074066
@famouswords86	1	0.6172839506172839
@consistenterror70	0	0
@normalratio59	0	0

provide tabular information regarding each CofP member’s Twitter contributions:

Concerning the 829 GIFs, the data collection process yielded 148 examples of what I refer to as ‘demonstrative GIFs’ and 681 examples of true ‘reaction GIFs’. It must be stressed that this distinction was not always clear-cut. Qualitatively categorising GIFs as either ‘reactive’ or ‘demonstrative’ is a subjective task. Some GIFs teetered the line between reactive and demonstrative, and other researchers may well have classified these GIFs differently. In any case, demonstrative GIFs are those which do not embody a specific emotion or reaction on behalf of the responder and instead simply depict the subject of the addressee’s (or one’s own) tweet. These are in contrast with reaction GIFs which, as previously explained, index some sort of emotional or cognitive reaction to a stimulus. Other tweets that appeared in the same conversational threads as the tweets featuring reaction GIFs/images but that did not contain the communicative devices themselves were not included in the dataset. However, these will be reproduced/referenced/analysed in this article to contextualise the use of these resources, so long as these tweets were also produced by one (or more) of the participants of this study. Participants were made aware of this when providing their consent.

Each collected tweet was manually categorised and coded in Microsoft Excel. As well as noting authorship, communicative device type and the linguistic/interactional context, the primary coding variables were depictions of humanness versus non-humanness, illustrations versus live-action shots, gender and sexuality presentation (considering both overt displays of these identities and background knowledge, where applicable), race/ethnicity, and, later, emergent personae types. Considering this paper’s interest in gender indexing, the gender presentation of each salient individual person or character depicted in each GIF/image was noted where possible (and within reason). Background knowledge about the source material and the depicted subjects was also noted. There is the possibility that gaps in the author’s knowledge of popular culture led to some of these individuals being incorrectly classified, for instance, if a short animation loop of a GIF did not indicate any flamboyance or campness of a celebrity who is well known to others in the community for embodying these qualities. In fact, coding and categorising reaction GIFs and images in general are both subjective and often difficult due to the polysemous nature of these/the fact that any two viewers may interpret the same GIF or image in different ways. However, owing to my demographic similarity to the participants (as mentioned earlier in this section, I am an Irish, currently UK-based Twitter user in his late twenties who is familiar with similar Twitter communities and the media often discussed), I felt justified in utilising this coding method, as it is likely that I shared a significant amount of overlap in cultural references with the CofP members.

4.2. Analytical Frameworks

A qualitative, computer-mediated discourse analysis (CMDA) approach was used to examine the data (Herring, 2004). CMDA describes “the use of spoken and written discourse analysis tools in the context of computer-mediated communication” (Benson, 2017, p. 81). Rather than describing a single method of analysis, CMDA describes an approach to (linguistic) analysis that explicitly acknowledges the computer (or other electronic device) as the medium of communication (Herring, 2004). Given that contemporary individuals regularly access social media via devices not typically thought of as ‘computers’ (smartphones, tablets, etc.), some may argue that the ‘CMDA’ label itself is outdated. A term like ‘digitally mediated discourse analysis’ may better reflect the diverse ways that these platforms are presently reached and utilised. Still, this article continues to employ the term ‘CMDA’ as it situates the present research within a recognised paradigm while highlighting this study’s connection to a larger body of work in sociolinguistics that critically considers online discourse. In a CMDA, the analysis may be entirely qualitative, largely quantitative or involve a mixed-methods approach. According to Herring (2004), CMDA has been previously utilised by linguists to investigate a range of linguistic phenomena at both a micro-level (lexical choice, syntactic structures, code-switching, etc.) and a macro-level (identity and community formation, coherence, etc.).

This analysis was primarily interested in exploring how the participants assumed distinct identities or personae in interaction with each other. Accordingly, the analysis followed the example of researchers like Ge (2019), who utilise CMDA as a means of investigating stance. Specifically, this CMDA draws on Du Bois’ stance triangle (2007). Du Bois (2007, p. 163) writes that ‘stance’ is a “public act by a social actor, achieved dialogically through overt communicative means, of simultaneously evaluating objects, positioning subjects (self and others), and aligning with other subjects”. The stance triangle thus provides a framework for analysing these objective, subjective and intersubjective relations through its three main components. These are (1) evaluation, which is conveyed by interlocutors through, for example, stance predicates like dreadful or wonderful; (2) positioning, attained by using pronouns like I in respect to epistemic (know) and/or affective (glad) states, for instance; and (3) alignment, which is achieved by employing stance verbs such as agree, using stance markers like yes and no, etc. (Du Bois, 2007).

5. The Community of Practice

This analysis concerns the communicative practices of the members (n = 8) of a small Twitter CofP. The members of this group, whose pseudonymised usernames are provided in Figure 1, all shared important commonalities: they self-identified as White, they were all native English speakers, seven of the eight members labelled their nationality as British (with one Irish member), seven of the eight were in the age bracket of 30–39 (while one was in the 20–29 age bracket) and they all self-identified as cisgender, gay men. These men shared the common endeavour of utilising Twitter to discuss popular media with each other, including their opinions on television shows, movies and, most significantly, music. Indeed, as a whole, Twitter appeared to be commonly used by this CofP to discuss (primarily pop) music, singers and groups, concerts, and so on. While discussions that revolved around popular media did dominate the dataset, more general discussions surrounding these men’s personal relationships, occupations, observations from daily life, etc. also occurred frequently.

The participants were unknown to the researcher prior to the research. Some basic sociodemographic information was directly gathered, including the participants’ genders and sexualities, which were confirmed when signing up for this study. Additional information about race, age, etc., was collected during the survey phase. However, for the most part, much of the knowledge I acquired about the lives and relationships of the participants came from immersing myself in their Twitter profiles while reading and collecting their tweets (i.e., an ethnographic approach). It appears that the members of this CofP all met online. At least three of the eight members had connections to an internet forum that discusses pop music, while two of the men met in a gay chatroom. During data collection, it soon became apparent that some of these participants had also forged offline relationships. For instance, @gentleinflation46 and @determinedapple13 were married and lived together, and several of the participants had met in person to attend each other’s weddings, go to concerts together and so on. Figure 1 (above) illustrates the participants’ ‘following’ relationships with each other, showing that the members of this CofP could be broadly categorised according to their status as ‘core’ versus ‘peripheral’ members (Holmes & Meyerhoff, 1999). In determining core versus peripheral membership in this CofP, the number of followers that one had from within the group was an important factor, as was the number of group members that a user himself followed. For instance, @proudchild22 followed and was followed by each of his fellow participants, thus making him an exemplary core member. @ourhistory05, on the other hand, followed and was followed by only two of his fellow participants, which seemingly suggests that he was more of a peripheral member.

6. Analysis

A qualitative dive into the dataset immediately reveals a strong tendency for these tweeters to utilise reaction GIFs and images that feature female or feminine subjects (e.g., cartoon characters, puppets, paintings, etc.). Drag and male femininity or queerness (e.g., flamboyant or publicly out male subjects) were also commonly visually depicted via both reaction GIFs and images.3 Heterosexual maleness—and, more generally, a lack of femininity/effeminacy—was noticeably less well represented than (overt) femaleness, femininity and male queerness combined. In other words, reaction GIFs and images featuring heterosexual males as their primary subject(s) were comparatively marked compared to femininity and effeminacy. Given the survey results, this finding is somewhat surprising. When asked to indicate their level of agreement with the statement “When choosing a reaction GIF or image to use, the gender(s) of the person(s) or character(s) featured in the reaction GIF or image factor(s) into my choice”, only four of the eight participants responded with ‘Somewhat agree’, while the other half indicated disagreement (two for ‘Disagree’ and two for ‘Somewhat disagree’). Of course, there are numerous possible reasons why the participants may have orientated themselves towards or against the notion that gender is an important variable in their GIF or image selections. For example, the direct mention of gender itself might have influenced their responses, those who indicated disagreement might have done so out of a desire to challenge the stereotype that all gay men behave effeminately and so on. Despite this, what is certain is that reaction GIFs and images that feature women/girls or feminine/female characters did often recur throughout the dataset.

The gendered economy of GIFs should also be considered here, that is, how (un)representative GIF websites and keyboards are of real-world gender demographics. Gender stereotypes dictate that women are more emotionally expressive, and men are routinely discouraged from outward displays of negative emotions, such as sadness or anxiety (‘boys don’t cry’). Therefore, it might be anticipated that emotive reaction GIFs featuring females are more commonly created, selected and shared than GIFs depicting males. Yet, this should be carefully balanced against the fact that men have historically been, and frequently still are, overrepresented in global media (Santoniccolo et al., 2023). Furthermore, though research in this area is scarce, some evidence points to an underrepresentation of females in popular GIFs. Through both automated and human coding, Álvarez et al. (2021) analysed 747 of the most popular reaction GIFs on three GIF databases (Tenor, Giphy and Gfycat), focusing specifically on GIFs with a sole primary character. Of these 747 GIFs, only 279 (37.3%) featured female primary characters, in contrast to the 468 (62.7%) featuring male main subjects. While further research is necessary to more firmly establish large-scale trends relating to gender and sex representation in GIF repositories, the finding that femaleness and femininity are routinely depicted via GIFs in this gay CofP is especially striking in light of Álvarez et al.’s work.

A potential issue that concerns the data is whether using resources featuring female or feminine individuals is a deliberate attempt by these gay men to directly index femininity or not. That is, one may assert that the femaleness or femininity of the depicted subject is actually secondary (or even inconsequential) to the intended indexes, which might instead be lower-level stances, qualities or traits (sassiness, extravagance, melodrama, etc.). Indeed, a possible argument is that subjects visually representing femaleness or femininity is just a by-product of their primary intention: to digitally do ‘being gay’ by assuming stereotypically feminine traits or positionings. As touched upon in Section 2, because a heteropatriarchal society disapproves of men embodying emotions ideologically associated with femininity and vulnerability, one might contend that the only way that gay men can perform effeminate queer masculinities and convey emotionality, sassiness, etc., is by taking stances (in this case, via GIFs and images) that are feminine-coded. This theory helps to account for why half the group rejected the notion that gender factored into their choices of reaction GIFs and images, despite femininity prominently featuring throughout the dataset.

There are a few difficulties encountered by such an argument, however. Firstly, though survey results can reveal self-reported goals behind the uses of linguistic and semiotic resources, it is difficult to ever ascertain the ‘true’ intentions behind participants’ uses of communicative resources. What is more immediately apparent, though, is the actual output: these GIFs repeatedly contained visual depictions of femaleness and femininity. This is why—hypothesised or self-reported intentions aside—I refer to these men employing overt, visible depictions of female or feminine subjects as directly indexing femaleness or femininity. A theory that attempts to explain these performances of gayness without noting the immediate, clear connection of the resources used to gender also fails to capture how closely entwined ideas of male, gay sexuality and femininity remain in contemporary society (Halperin, 2012), as well as, more broadly, the inextricable link between gender and sexuality performance in the first place (Cameron & Kulick, 2003). That is, suggesting these men are not ‘trying’ to index femininity runs the risk of suggesting that ‘doing gender’ and ‘doing sexual identity’ can be neatly set apart. Further, separating these indexes of sexual identity from gender fails to acknowledge that gay men have a long history of strategically using feminine (linguistic) resources, both directly and indirectly (Cameron & Kulick, 2003, 2006). In addition, the previous paragraph’s argument does not show that these resources are also regularly used for emotions, behaviours, reactions and gestures that are not as strongly feminine-gendered (e.g., laughter, smiling). This supports the notion that womanhood and femininity or effeminacy hold particular value when it comes to these CofP men selecting reaction GIFs and images. Considering the participants’ interests in popular culture and, particularly, pop music, it also does not seem coincidental that these men often avail of resources depicting female (pop) stars, including the female artists that they show particular enthusiasm for (which is also a well-established practice in gay male culture; see the next subsection on gay male diva worship (Kooijman, 2019)). On these grounds, I argue that it is reasonable to state that these men are indexing their gayness by directly indexing femininity while stressing that queer maleness and effeminate maleness are not one and the same.

Moreover, what also becomes clear when looking through the dataset is that by indexing femininity and assuming various stances, distinct (feminine) personae—which reflect identifiable social identities—emerge across these gay males’ tweets. In other words, it appears that by utilising reaction GIFs and images as a means of virtually “substitut[ing] the embodiment of the speaker” (Kuo, 2019, p. 178) and presenting one’s emotions through the means of an avatar of sorts, various personae are constructed. Importantly, the upcoming persona labels were not derived from CofP self-usage. All four personae emerged from the data and reflected well-known social tropes, and the names I have subjectively selected (as a non-CofP member who shares vital sociocultural overlaps with the participants) seek to foreground this. While the Sassy Queen and the Hun are established gay, male personae in the existing sociolinguistic literature (Ilbury, 2020, 2022)—as is, arguably, the Flamboyant Queer (see Section 6.4 on its relation to Podesva’s (2007, 2008, 2011) diva)—the Battle-Axe is, to the best of the author’s knowledge, not. These personae, as will soon be demonstrated, appear to be oftentimes associated with certain linguistic repertoires or stances/topics. Though the volume and richness of the data collected for this study mean that this article alone cannot exhaustively analyse all the emergent personae, the following subsections will discuss a selection of the personae that have been deemed particularly salient. This perceived salience was based on an interplay of multiple factors. These included (1) the frequency of each persona (i.e., their repeated emergence throughout the corpus), (2) their spread across the group (universal or near-group-wide personae were favoured), (3) the clarity of these personae (i.e., how clear-cut and distinguishable they were) and (4) the discernability of clear stance or linguistic correlates (given the sociolinguistic nature of this research).

6.1. The ‘Sassy Queen’

The Sassy Queen (SQ) is a persona identified by Ilbury (2020). In his study, Ilbury demonstrates how a group of White, gay men from the UK utilised various lexical, phonological and morphosyntactic features from African American language on Twitter to construct an SQ persona. The SQ heavily relies on the problematic stereotypes of the ‘sassy Black woman’ and ‘angry Black woman’, which commonly recur in internet memes and beyond (Ilbury, 2020). Similar performances of the SQ persona emerge in this dataset, occasionally alongside linguistic resources linked with African American language. Six of the eight participants (everyone bar @famouswords86 and @consistentterror70) constructed this persona at least once.

As shown in Figure 2, @proudchild22 performed the SQ persona in a tweet that informed his followers that he had a relatively quiet work schedule for a period. @proudchild22 shared this information despite his understanding that this may have been seen as bragging (‘I don’t want to be a total arsehole but…’). Here, the distinctly British lexical item arsehole forms an interesting juxtaposition with the following GIF. Indeed, @proudchild22’s knowledge that this post may be read as self-congratulatory or boastful appears to trigger his performance of the SQ persona, as evidenced by his GIF choice: American singer Mariah Carey is shown kissing her fingers and giving the onlooker a wave goodbye as she begins to walk off-frame. In this context, these gestures could be read as connoting diva-ness, as one interpretation of this GIF use is that Mariah Carey is an embodied representative of @proudchild22 as he leaves work. Blowing a kiss and walking away from one’s co-workers can be read as both self-important and over the top. The objects and styling apparent in this GIF also connote diva-ness, beauty and femininity: from the silver jewellery to the painted red fingernails and the luxurious curly hair that Carey sports. Though Carey is a mixed-race woman, McCarroll (2009, p. 219) lists Carey among the multiracial celebrities who are “continually read as black” by the public. Given that Carey is often involved in constructions of the SQ in this CofP, it appears that one need not be fully Black, just recognisably so, to be pigeonholed into this character trope.

@proudchild22’s performance of the SQ persona triggered @normalratio59 to perform the same archetype in his response. In @normalratio59’s GIF, Porsha Williams, star of The Real Housewives of Atlanta, is denoted as snapping her fingers and remarking ‘Two snaps for you’. Williams sports glamorous jewellery, makeup and well-styled hair, set against the backdrop of a luxurious home. The gesture of snapping is notable as an index of African American femininity. As Johnson (1995, p. 129) explains, African American women (as well as gay men) are well known to “use snapping to compliment someone’s looks, a hairdo, or even inanimate objects”. Therefore, not only does the subject’s physical appearance denote African American femininity, but so too do her physical gestures (which, in this context, embody @normalratio59’s own reaction to @proudchild22’s work schedule). The congratulatory nature of @normalratio59’s tweet may not represent what one might imagine a typical ‘sassy’ reaction to be, given that it is neither overly opinionated nor bold. However, what this GIF does express is a level of liveliness that can be read as distinctly sassy. The exaggerated snapping may be interpreted as an over-the-top reaction to rather mundane news, which highlights and elevates the previous stance.

The SQ persona need not always be performed through GIFs that feature actual women/girls or African American people/characters. In Figure 3, for example, @determinedapple13 uses a GIF in which the character Shellsea (a pink fish from the Disney cartoon Freshwater High) is denoted as uttering ‘GURRRRRLLLLLLLLLL’. Though the character of Shellsea in the show is in fact closer to the valley girl persona (she is a fashion-loving cheerleader who speaks with an exaggerated creaky voice), girl is a culturally toned diminutive that has been associated with both African American women’s speech and the language of gay men (Barrett, 1997). Owing the fact that Ilbury (2020) identified the eye dialect form gurl as one of the linguistic features utilised by his participants to construct the SQ and that the GIF here was employed to assume a stance of sassy disalignment with the addressee—whose negative evaluation @determinedapple13 clearly disagrees with—this appears to indeed be an example of an SQ performance.

Taking a step back, it is vital to consider not only the potential motivations behind performances of the SQ by these self-identified White, gay men but also the deeper implications of these uses. As the traits of sassiness and fierceness are highly valued in the gay male community (Ilbury, 2020), performing the archetype of the SQ through digital embodiment allowed the gay men of this CofP to claim these traits for themselves, enabling self-presentation as spirited and lively. Considering how widespread the performance of this persona appeared to be throughout the group (and, impressionistically, how often it was performed relative to the other personae), it might also be argued that a group identity of sorts was formed around performances of the SQ. Thus, performing and (co-)constructing this persona allowed the participants to (re-)establish a specific collective identity while simultaneously signalling their own membership within the group. It makes sense, then, that @famouswords86 and @consistentterror70 did not perform this persona. It appears that these two participants were on the periphery of this group based on their following relationships and, therefore, they possibly did not feel the need to/did not feel entitled to perform this in-group persona.

Given that many of the reaction GIFs and images throughout the corpus portrayed Black/mixed race singers (Mariah Carey, Janet Jackson, etc.), the GIF and image usage in this CofP may be linked to the concept of gay male diva worship (Kooijman, 2019). This involves the celebration and idealisation of female stardom. This adoration of glamorous, successful and assertive (Black) divas and the very “art of mimicking a diva—of pretending, inside, to be divine” (Koestenbaum, 1993, p. 133) is clearly reflected in the digital embodiment and enactment achieved through these men’s reaction GIF and image usage. But, while the (Black) diva traditionally functioned as an icon of empowerment, Kooijman (2019, p. 7) points out that she is also frequently transformed into a “clichéd representation or caricature” in gay culture. The identities of the divas with African heritage are reduced to the traits of boldness, boastfulness and fierceness, which these gay men co-opt for their own social benefit.

This conversation links back to the concept of digital Blackface (see Section 3). As Hornback (2018, p. 274) writes, supposedly neutral reaction GIFs (and images) disseminate and reinforce “the age-old stereotypes and underlying beliefs of blackface traditions”, namely, of the animated, Othered, over-the-top Black individual. This concept is clearly applicable to the sassy persona formulation in this dataset, where the participants utilised reaction GIFs and images to express disalignment, convey bold affective stances, brag about their own achievements and so on. What is particularly interesting here is that when responding to the survey prompt “When choosing a reaction GIF or image to use, the race(s) of the person(s) or character(s) featured in the reaction GIF or image factor(s) into my choice”, every participant who chose to expand upon their answer (five of the eight participants) acknowledged that using reaction GIFs/images featuring Black people could be regarded as appropriation. Responses referenced how these ‘might [feed] into the minstrel trope’, how they might ‘perpetuate any ‘sassy Black woman’ stereotypes’ and one participant directly referenced the concept of digital Blackface by name. Hence, despite the mixed levels of agreement given in response to the survey prompt itself (one participant chose ‘Strongly agree’, two chose ‘Agree’, two chose ‘Somewhat agree’, one chose ‘Disagree’ and two chose ‘Strongly disagree’) and despite the participants’ self-proclaimed hesitancy around playing into racial stereotypes, it is undeniable that the SQ played an important role in both the group dynamic of this CofP and in the individual identities of the majority of its members.

6.2. The ‘Hun’

‘Hun culture’ describes a British cultural phenomenon that rose to prominence in the latter half of the 2010s and early 2020s (Ilbury, 2022; Levine, 2020; Minor, 2023; Staples, 2021). Staples (2021, para. 2) describes ‘Hun culture’ as follows:

Hun culture celebrates a specific British type of “low culture” and it normally revolves around stanning women, known as huns, who embody this spirit. Normally it’s because they’re relatable, ever so slightly out of touch (but not quite to the ‘live, laugh, love’ level) or funny in a way that’s completely unintentional.

To expand on this, Hun culture involves (often millennial) gay men and women celebrating a distinctive category of White British women, particularly those who have made (often niche or kitsch) British media appearances (Ilbury, 2022). Specifically, the ‘Hun’ is described by Ilbury (2022, p. 489) as a woman with working-class roots who is economically mobile, but who still enjoys stereotypically working-class activities. Importantly, not all Huns fit the narrow category of ‘White, blonde, working-class, etc.’. Levine (2020) classifies celebrity cook Nigella Lawson as a ‘posh Hun’, while television presenter/celebrity chef Ainsley Harriott is a Black male who has appeared in Hun memes and is, in many ways, a ‘camp Hun’. Still, the archetypical Hun does possess the four qualities of femaleness, Whiteness, Britishness and working-classness. Iconic Huns include reality TV star Gemma Collins, glamour model Katie Price, the stars of British soap operas like EastEnders and so on, all of whom are flamboyant (in the sense of being over the top/excessive) yet still relatable, with a nostalgic, unmistakably British charm. Hun culture operates primarily online through social media accounts that share memes featuring Huns alongside relatable and/or humorous captions. However, Hun culture is also celebrated in the offline sphere, particularly in gay spaces like nightclubs that host Hun-themed nights, as well as in offline interactions between women and gay men (Ilbury, 2022). The significance of this persona in gay culture is further underscored by the song “UK Hun?” (2021). Performed as part of a challenge during the second series of RuPaul’s Drag Race UK, this song interweaves British, drag and Hun culture—spotlighting the significance of the Hun in contemporary British gay culture.

Ilbury (2022) analyses posts from two Instagram accounts (both run by gay men) dedicated to sharing memes related to Hun culture. He demonstrates how the Hun persona is performed through a variety of stylistic features. These include discourse features such as x as a sign-off, non-standard spellings like <partehhh> for party (which is intended to index Northern British English) and characterological tropes like the employment of cheesy mottos, all of which contribute to the composition of a distinct digital commodity register. While Ilbury’s paper focuses on how the Hun persona is constructed in memes, with less attention paid to how this persona functions in interactions between interlocutors, my data show how the Hun persona is strategically employed in the digitally mediated conversations of this gay male CofP. Five users engaged in the construction of this persona: @proudchild22, @ourhistory05, @widehat91, @normalratio59 and @determinedapple13. While not as widespread throughout the group as the SQ, the Hun was still highly distinguishable where it did occur. Figure 4 and Figure 5 highlight this:

In Figure 4 and Figure 5, some of the features that Ilbury (2022) identified as being enregistered as part of the Hun persona are evident in @ourhistory05’s use of GIFs. Beginning with Figure 4, @ourhistory05 employed a GIF of Gemma Collins, former star of the British reality show The Only Way Is Essex. Throughout this interaction, @proudchild22 and @ourhistory05 riff on the idea that an event has been cancelled because of British pop singer Melanie C falsely claiming to be sick (‘She mustn’t feel well’, ‘As long as she’s got a note from a doctor she found on the Internet’). @ourhistory05 used this GIF to embody his own reaction to Melanie C cancelling her appearance,4 and this GIF enabled him to perform the Hun persona. Derived from the British reality show Celebrity Big Brother, the GIF shows Collins slamming down a phone and shaking while shouting ‘FACK OFF GILLIAN MCKEEEEF!!!’. In the original context, Collins disaligned herself from her addressee, fellow reality star Gillian McKeith. Yet, placed in its new context, this GIF—with its visual connotations of extreme impatience and anger, along with its imperative caption—disaligned the recontextualised addresser (@ourhistory05) from the recontextualised addressee (Melanie C). As Miltner and Highfield (2017, p. 3) write, GIFs not only embody one’s emotional state but also act as a “demonstration of cultural knowledge”. Here, the cultural knowledge that the Scottish television personality Gillian McKeith was a contestant on the show I’m a Celebrity…Get Me Out of Here! who was well-known for allegedly pretending to faint is required to fully understand that @ourhistory05 was accusing Melanie C of feigning illness. Simultaneously, using this GIF in response to the mention of an internet-sourced sick note plays off doubts surrounding McKeith’s credentials as a self-declared holistic nutritionist. To illustrate, after being approached by the UK’s Advertising Standards Authority, McKeith—who previously made media appearances as ‘Doctor Gillian McKeith’, having trained at the non-accredited American Holistic College of Nutrition—agreed to stop using the honorific Dr in her advertising (Clout, 2007). Indeed, this demonstrates the fact that successful participation in the Hun persona oftentimes requires niche knowledge of British popular culture. Moreover, the Th-fronting (where /θ/ is fronted to [f]) of Collins’ Romford accent is orthographically represented here in the respelling of McKeith as <MCKEEEEF>, as well as a likely representation of the Essex vowel fronting of /ʌ/ (to the TRAP vowel) in the rendering of fuck as <FACK>. This is consistent with Ilbury’s (2022) findings that Hun culture memes often exploit orthographic variants to portray the phonological features of British English (including, but not limited to, Northern English working-class varieties). What is particularly interesting here, though, is that Collins in fact pronounces ‘fuck’ as /fʌk/ in the original clip from Celebrity Big Brother. Yet, the GIF creator clearly (unconsciously or consciously) took Collins’ Romford/Essex background into account while transcribing her speech, perhaps knowingly to accentuate her Hun status. In turn, this strengthened @ourhistory05’s own performance of the Hun persona.

In Figure 5’s GIF featuring the character Linda La Hughes from the British comedy series Gimme Gimme Gimme, the conative interjection (an interjection “directed at someone else who may be expected to fulfil the wishes of the speaker” (Ameka, 1992, p. 257)) ‘shhhhh’ follows the item ‘ere’. ‘ere is an interjection that is used in British English, formed via the contraction of look here. Under Kiesling et al.’s (2018, p. 688) understanding of stance, where the object/focus of stance can be an act (such as a request), “affect has to do with whether the act itself is overtly positive or negative”. As ‘shhhhh’ functions as a bald imperative (i.e., it is not mitigated by the adverb please, for instance), this GIF can thus be understood as expressing negative affect. The implication of this tweet is that @ourhistory05 wants @proudchild22 to keep it quiet that he has pirated the song under discussion. The rendering of <‘ere> orthographically represents H-dropping (i.e., the deletion of the fricative [h]), which is common in many English dialects spoken in England. Notably, this H-dropping is mirrored by @proudchild22 in his reply (‘You dirty ‘omo’), which suggests that he is ‘playing along’ with his addressee’s performance of the Hun. While La Hughes may not commonly appear in online Hun memes, she does possess many notable features of the archetypical Hun, which led to her classification as such: the character is a working-class, White, British woman who is out of touch with reality (egotistical and delusional) and who appears in a nostalgic British sitcom.

In the two former examples, the visual (i.e., the women depicted) and the linguistic contents of the GIFs work in tandem to allow @ourhistory05 to index a persona that is not only coded for gender (womanhood) but also for race (White), class (working-class) and nationality (British). As such, this very British persona contrasts strongly against the more commonly occurring SQ persona, who is distinctly African American. What the Hun has in common with the SQ, though, is the context that tends to trigger its performance. Similar to how the SQ emerges in this group in conversations where a participant is able to present himself as sassy, bold and lively, the Hun in this group largely occurs in scenarios where one is able to portray themselves as being somehow flamboyant and over the top (whether in terms of their emotions or their actions).

Though other linguistic characteristics of the Hun were also apparent in the corpora (e.g., the non-standard spelling of babes as <bbz>), linguistic features from British English were often not present when indexing this Hun persona. For instance, @widehat91 responded to @gentleinflation46’s news of his trip to Japan in Figure 6 with a reaction image that depicts Welsh singer Lisa Scott-Lee. This image depicts Scott-Lee in a lilac jacket, open-mouthed and with flared nostrils. While the connotation of Scott-Lee’s clothing is flamboyance and vibrance, Scott-Lee’s facial expression connotes extreme distress/upset. This sharp juxtaposition between her clothing choices and emotional state arguably provides a humorous effect. The purpose of this reaction image was to convey the affective stance of dismay and, in doing so, @widehat91 participated in the embodied performance of the Hun persona.

In his analysis of the Hun, Ilbury (2022, p. 500) notes that, unlike some other feminine personae, the Hun persona is “(intended to be) a depiction of a relatable high-camp identity” which functions as a form of “shared cultural capital” for both British gay men and (heterosexual) women. Ilbury stresses that the parodic nature of Hun memes is meant to celebrate this memetic figure’s relatability, rather than to offend or disparage women who belong to the associated social category. Yet Ilbury also briefly highlights the potential for this persona to be interpreted as misogynistic, which Minor (2023) expands upon. Noting that while Hun culture humour is often regarded as inclusive of the celebrity Hun by its participants, Minor argues that this laughter is in fact polysemous. For Hun fans from British, working-class backgrounds—or even just those who share similar tastes or sensibilities with the Hun (e.g., enjoying cheap alcohol or budget holidays)—Hun humour is a celebration of relatability and ‘bad taste’. Yet Hun culture can be equally patronising, depending on whether the actual women who embody the Hun archetype are ‘in on the joke’ or not. If the comedy/appeal of Hun memes stems from the celebrity Hun’s unawareness that her self-presentation or middlebrow tastes are actually ‘bad’/tacky (e.g., if she doesn’t know that earnestly saying ‘holibobs’ is cringe), then there is an element of laughing at rather than laughing with (i.e., because the fan ‘knows better’ than the Hun). Concerning social class, participants who are not from the same class background as the typical Hun, particularly those from wealthy or privileged backgrounds, may rely on the comedy of “how huns are ‘like us’ while simultaneously ‘punching down’ to reassert (class) difference” (Minor, 2023, p. 857). This would be apparent in the imitation or (re)production of linguistic features associated with working-class British English (or, more generally, variations from standard English like misspellings/typos) for comedic purposes by speakers who do not usually employ these features in their daily lives, for example. While there is the concern that some celebrity Huns would not find their meme status to be flattering if they knew what other people find so amusing/appealing about it, at least some Huns have openly embraced their Hun title, including celebrities such as Gemma Collins (Johnston, 2023) and Natalie Cassidy (Staples, 2024).

The question of whether the Hun persona is problematic or patronising remains an open one and likely varies on a case-by-case basis. As with the SQ, the argument could be made that the participants’ performances of this persona patronise the often working-class women featured in these GIFs/images and are thus rooted in misogyny and classism. While the present analysis does not possess information on its participants’ class backgrounds, Hun constructions in this dataset do not appear to be maliciously motivated. Rather than truly mocking the Hun and her interests or tastes, members of this gay male CofP seemingly resonate with and appreciate them, at worst playfully poking fun at these. Performances of the Hun appear to not only be motivated by a nostalgia for British media but also seemingly reflect a pride (that is, at times, ironic) in British (popular) culture. This is coupled with a celebration of its relatable—albeit oftentimes eccentric—female media stars. This argument is supported by the owner of the Instagram page @loveofhuns, who likewise states that (performing) the Hun persona allows gay men and (straight) women to convey “a streamlined and concentrated idea of what makes [British] culture so magical and special” (Staples, 2021, para. 9).

6.3. The ‘Battle-Axe’

The term ‘battle-axe’ is defined as “[a] formidable or domineering woman” (Oxford University Press, 2024), with Stollznow (2020) clarifying that battle-axe is most frequently an insult that is older woman-specific. In Mortimer’s (2019, p. 42) terms, the stereotypical battle-axe is a “monstrous figure on account of her failure to comply with the feminine ideal” due to her assertive, confrontational and disagreeable nature (i.e., she embodies the masculine). Though the term battle-axe originated in U.S. slang (Oxford University Press, 2024), its use has expanded well outside of the U.S. It is an archetype that has been applied to female politicians by the press (Baxter, 2018; Shaw, 2020), women in the workplace more broadly, women regarded as outspoken (Talbot, 2020) and more. This trope also emerges repeatedly in children’s literature and media. Prime examples include the schoolteacher Miss Battle-Axe from Francesca Simon’s Horrid Henry book series, as well the fearsome Miss Trunchbull from the novel Matilda (1988) by Road Dahl, an author who regularly draws on this imposing archetype. The intimidating nature of the stereotypical battle-axe is often tied to her (occupational) position of power or authority, though her stubbornness, temper and overbearing attitude may also be more generally linked to her advanced age. Throughout the Twitter CofP dataset, a recurring usage of GIFs featuring middle-aged and (much more frequently) elderly women was evident, particularly when assuming stances of blunt disalignment and/or negative evaluation. This Battle-Axe persona was performed by six of the eight participants (again, everyone bar @famouswords86 and @consistentterror70) at least once. Figure 7 and Figure 8 demonstrate two constructions of this persona:

Figure 7 and Figure 8 are similar in that the GIFs used convey a negative evaluation of their stance objects (the preceding reply in both cases) solely through the facial expressions of their depicted subjects. In Figure 7, @determinedapple13’s GIF denotes Dorothy Zbornak (a character from the American sitcom The Golden Girls) glancing upwards, with an unmistakably stern look in her eyes, while she slowly stops pouring a hot beverage. As such, the posing of the subject clearly connotes disapproval of @gentleinflation46’s non-nonsense advice that @determinedapple13 should ‘Step away from the internet’. Likewise, @normalratio59’s GIF in Figure 8 shows the character Professor Minerva McGonagall (from the Harry Potter film series), pursed-lipped and wide-eyed, while she cocks her head. Again, this facial expression communicates disapproval quite clearly, which is reflected in the alt text of the GIF (‘Maggie Smith Disapproval GIF’). Notably, this GIF was used in response to a conversation that was ladened with sexual innuendoes (‘I might give you a Jazzle if you’re nice’, ‘You could stick your tongue in my Fizz Wiz?’), playing into the reproachful, asexual nature of the stereotypical battle-axe (McAllister & Brien, 2020).

The Battle-Axe persona had an extremely clear distribution throughout this dataset, as it appeared to be performed almost exclusively in situations where disalignment or negative evaluation were expressed. While this persona was coded for gender (femaleness, though she is not feminine) like the two previously discussed personae and appeared to be also coded for race (White), it was unlike the previous personae in that it was more strongly coded for age (middle-aged plus). The SQ and Hun, by contrast, appeared less age-linked, though they tended to be younger than the stereotypical Battle-Axe. In addition, while the SQ and the Hun were both coded for distinct nationalities (American and British, respectively), the GIFs featuring the Battle-Axe in this dataset depict subjects from North America and Britain, suggesting that this persona was not so coded for nationality. While the SQ and the Hun both had linguistic correlates that co-occurred with at least some of their performances (features from African American language and British English, respectively), performances of the Battle-Axe did not appear to correspond to any particular linguistic features aside from, quite broadly, captions that conveyed negative affect or disalignment.

It is important to consider what may have inspired the constructions of this Battle-Axe character type. The projection of disapproving or stern stances onto older women is easily understood when considered in terms of their general cultural association with (often quasi-maternal) bossiness. Women are most intelligible as authority figures who inspire fear and indignation when positioned as mothers and institutional mother-like figures (the headmistress, the Big Nurse, the nanny and so on) (Cameron & Shaw, 2016), which is a pervasive mainstream sexist/ageist trope. However, it can be argued that the gay male use of this persona suggests a degree of identification with—and perhaps even admiration for—this oft-resented figure. Indeed, the case can be made that the gay men of this CofP respect the Battle-Axe’s open expressions of her dislikes. This is possibly because her sexual undesirability does not carry as much weight for them as it does for heterosexual men and thus they utilise this persona to represent their own inclinations towards portraying sternness and expressing disapproval.

Another possible explanation relates to the ages of the participants themselves, with the majority of them (seven out of eight) being in the age bracket of 30–39. While collecting the Twitter data, I witnessed tweets from a few of the participants acknowledging the fact that they are ageing/no longer feel young. Performances of the Battle-Axe thus playfully exaggerate their apparent loss of youth; drawing attention to this potential insecurity in a light-hearted, jovial manner.

6.4. The ‘Flamboyant Queer’

This final section demonstrates that a gay identity can be indexed via reaction GIFs and images that depict a certain kind of flamboyant, effeminate queer man/assigned-male-at-birth individual (excluding transgender women). Specifically, a distinct persona that emerged in this dataset was the ‘Flamboyant Queer’ (FQ). The nomenclature for this persona links to the work of Savin-Williams (2016). Through interviewing several young gay men, Savin-Williams found the FQ (whom he labels more rigidly as the ‘Flamboyant Gay’, described as a “super femmy gay”) to be a highly negatively perceived stereotype among his participants, associated with terms like bitchy, cattiness, bickering and drama (2016, p. 269). The FQ might be deemed a version or offshoot of the gay diva trope previously discussed by Podesva (2007, 2008, 2011), a persona who is equally unafraid to voice his opinions (and who is appearance-focused, prissy and attached to particular sociophonetic resources). Still, I favour the ‘FQ’ label for the social type that arises in this specific CofP context. This title appropriately highlights the reliance of this persona on direct, explicit visualisations of non-cishet maleness via digital objects. Like the SQ and the Hun, the FQ is flamboyant in the sense of being over the top and dramatic. Yet, the FQ is also flamboyant in the queerer sense of the word, where flamboyance refers to non-masculine behaviours or styles stereotypically exhibited by gay men. This is relevant to this chapter’s dataset because a recurring theme was the use of reaction GIFs and images that depicted queer men/assigned-male-at-birth people who (through various means) indexed queer femininity and whose embodied reactions were somehow critical or catty. Indeed, seven of the eight participants performed this persona at least once, with the one non-performer once again being on the periphery of the CofP (@consistentterror70).5 The following two excerpts demonstrate how @normalratio59 performed this persona.

As shown in Figure 9, @normalratio59 disaligned himself from the actor Roger Moore’s opinion of the singer/model/actress Grace Jones by employing a GIF featuring Tayce, a contestant from RuPaul’s Drag Race UK. The linguistic content of this GIF bluntly evaluates Moore’s opinion through a list of negatively evaluative nouns (‘The cheek, the nerve, the audacity, the gall and the gumption’). Tayce’s negatively evaluative utterance is uprooted from its original reality television context and strategically reapplied in response to Moore’s supposed impudence, thus offering an example of ‘recontextualisation’ (Bauman & Briggs, 1990). This GIF indirectly indexes queer male femininity in several ways. For instance, this GIF denotes Tayce turning his head slightly, rolling his eyes and lifting a limp-wristed hand into the frame. In his work on feminine stereotypes in characterisations of gay men, Rodríguez González (2008, p. 233) writes that limp wrist is a slang term that is sometimes applied to (effeminate) gay men: “dysphemistic in intention”, this label belongs to a “series of synonymic expressions which refer more specifically to the manners of a queen” (e.g., lightfoot, broken wrist, etc.). Indeed, the limp-wristed gesture has a long sociocultural association with effeminacy and, as such, the connotations of Tayce’s pose are campness, flamboyance and femininity. Tayce is also shown wearing an oversized colourful coat, chain necklace, hoop earring and pink buzzcut, all of which index gay male femininity. These indexicalisations rely on widely held gender stereotypes about clothing and self-presentation. While a buzzcut may be interpreted as masculine, the colour pink is highly linked to stereotypical femininity. Similarly, some forms of jewellery are normalised for men (gold chains for rap artists, wedding rings, etc.). In this case, though, it is the pairing of the necklace with the earrings—in combination with the bold, multicoloured print of the coat—that cements this outfit as stereotypically non-masculine. Contemporary (heterosexual) male fashion is typically thought of as less extravagant than female fashion. Thus, the showy nature of Tayce’s styling indexes gay effeminacy, owing to its departure from conventional masculinity (Halperin, 2012). It is clear, then, that by using this GIF, @normalratio59 indexed a particular form of queer male femininity, which, in turn, allowed him to construct his own gay identity. A similar negative evaluation is shown in Figure 10, where @normalratio59 negatively evaluates @determinedapple13’s claim that the British sweets Parma Violets are not tasty. @normalratio59 achieved this negative evaluation via a GIF that features the adjective ‘INCORRECT’. This indexed an epistemic stance (i.e., by implying that @normalratio59 ‘knows better’), consequently positioning himself and @determinedapple13 at opposite ends of the epistemic scale when it comes to ‘understanding’ the taste of Parma Violets. This GIF features the flamboyant, canonically pansexual David Rose, a character from the sitcom Schitt’s Creek. Effeminacy is indexed here primarily through the subject’s pose: Rose is denoted with his thumb and index finger touching, as he dramatically moves his hand in a circular motion and tilts his head slightly. This hyperarticulated gesture mirrors Rose’s hyperarticulation of the word ‘INCORRECT’, which the GIF represents as having strong emphasis placed on both morphemes via a changing text colour. Given the cultural associations between male-attracted-males and effeminate, expressive and dramatic hand gestures (Rodríguez González, 2008), the pose in this GIF connotes queer male femininity which, again, allowed @normalratio59 to construct his own gay identity in interaction with his interlocutor.

Unlike the previously discussed personae, the FQ did not appear to be coded for race. There did seem to be a tendency for these performances to involve GIFs/images that depicted North American individuals; however, there were notable exceptions to this rule (e.g., Figure 9). In a way, this persona was quite similar to the Battle-Axe in that neither appeared to have any readily apparent linguistic correlates in this CofP and performances of both appeared to be largely restricted to scenarios where one wished to signal negative evaluation and/or disalignment. As McKinnon (2017, p. 91) notes, scholars have long recognised that some members of the LGBTQ+ community, including but not limited to Black speakers, employ jovial taunts, teasing and ritual insults (i.e., ‘reading’/‘throwing shade’) “to humorously build in-group solidarity” (Heisterkamp & Alberts, 2000; Jones, 2007; Murray, 1979). It is thus unsurprising that such a persona was performed by this group. Yet, given the prevalence of femmephobia in many gay male communities (Miller & Behm-Morawitz, 2016)—and, on the other hand, the enduring fetish of masculinity—why was such a flamboyantly queer, non-female persona performed by these participants? One possible explanation is that these men performed the FQ to reclaim and openly/publicly perform a type of subversive male femininity. Another explanation is that because the stereotypical FQ is so linked to over-the-top bitchiness and cattiness, using this persona immediately suggests that the negative evaluation or disalignment apparent in these tweets should be interpreted as being light-hearted rather than serious. As such, visually assuming this persona can operate as a particular framing device or contextualisation cue (Gumperz, 1982). In fact, while the participants often used reaction images and GIFs that clearly indexed femininity, they often did so to ‘take the piss’ out of each other or ‘throw shade’ in gay terms, which is stereotypically more characteristic of homosocial male interactions (Coates, 2003). In a way, these participants were doing ‘being masculine’ in a manner that was reminiscent of the fraternity-based CofP described by Kiesling (2002). In his study, the fraternity members utilised terms of address like honey and bitch boy to reproduce hegemonic heterosexuality and reinforce their homosocial bonds. However, in this case, the participants were being subversive with respect to masculinity by using feminine-associated language/resources. They strategically used cattiness for queer homosocial bonding: to show that they were friends with each other and build community.

7. Discussion

Returning to Eckert’s (2008) concept of the indexical field (see Section 2), it can be argued that just as a traditional sociolinguistic variant can be associated with a web of stances, qualities and characteristics that can enable the performance of socially meaningful personae, so too can reaction GIFs and images. Such an analysis is beneficial because it not only helps to pinpoint some of the qualities that are being identified with or presented in an admiring way by these participants but also helps to distinguish the overlapping qualities that recur across the constructions of these personae.

Figure 11 presents an indexical field analysis of the personae manifested via GIFs and images in this CofP. While each persona had its own unique qualities, two main characteristics were shared by all four personae: outspokenness and the refusal to ‘take any guff’. From the charmingly British Hun (‘FACK OFF GILLIAN MCKEEEEF!!!’) to the harsh Battle-Axe (‘DEAR GOD…’), it appears that all four of these personae are unafraid to be forthright with their opinions and negatively evaluate people/opinions/ideas that they find nonsensical or unpleasant. What might drive (some) performances of these personas, then, may be an identification with the outspokenness that the subjects of the GIFs/images display, or it may indeed be an admiration of these qualities (e.g., perhaps drawing on these personae demonstrates a veneration for subjects who do not contend with unpleasant people, events or circumstances). This again underscores the relevance of Du Bois’ stance triangle (2007) to the present analysis. Appealing to this framework clearly demonstrates how these social types coincide and differ in gearing towards certain types of situational positioning while accentuating how this repeated stance-taking effectuates the construction of discernible personae. As explained, the FQ and Battle-Axe appeared more stance-restricted than the SQ and Hun, with the former two being very closely linked with disalignments or negative evaluations. Yet, assuming recognisable, feminine-coded personae often signalled that the unfolding conversation was good-natured, non-serious or fun(ny), suggesting that even contributions that superficially indicated disalignment could affirm queer community and index solidarity within this CofP.

What also bound all four personae together was their relation to the aesthetic mode of camp (Sontag, 1964). If camp is classified as relying on exaggeration and the “subversion of gendered conventions” (Shugart & Waggoner, 2008, p. 22), it is clear to see how this notion applies to the feminine personae performed here. But, even beyond this, if camp is defined as “a performative practice rich with potential for disruption of heteronormativity” (Shugart & Waggoner, 2008, p. 22), that is, a practice that purposefully deconstructs hierarchies and default masculinities, it can be argued that all four personae discussed do this in different ways. This can be achieved via performing and/or celebrating Blackness; being working class and bold enough to exhibit low-brow tastes; constructing oldness, sexlessness and independence; or simply being queer. With this in mind, I assert that the constructions of the personae in this article crucially rely on either a strong identification with or admiration for the qualities and/or affect displayed by the depicted subject (the embodied avatar).

Importantly, Figure 11 does not intend to suggest that oppositely positioned personae are diametrically opposed, i.e., the Hun and the SQ are not ‘opposites’. The personae’s positionings are interchangeable, as the main purpose of the Venn Diagram is to highlight the traits and qualities shared by all four archetypes.

8. Conclusions

By analysing a small selection of modern communicative devices, this article demonstrated how gay men indexing femininity in their online interactions can aid in the (re-)construction of virtual gay identities. As Comer (2022, p. 191) notes, “[g]ifs [and reaction images] are a visualization of ‘embodied enactment’ in social interaction”. In online communication, people are aware of and seek to transcend this disembodiment, and the present work demonstrated how these resources aid in the performances of distinct personae. Furthermore, it is evident that looking at not only how these personae are performed but also who by is instructive. This approach reveals an interesting pattern of the same six participants almost every time, thereby suggesting that those who did not perform these main personae (@consistentterror70 and @famouswords86) were on the outskirts of this CofP. Though the participants’ survey answers revealed that half of the participants deemed the use of reaction GIFs/images that depict women as inoffensive (one vote for ‘Very inoffensive’, two for ‘Inoffensive’ and one for ‘Somewhat inoffensive’)—with the other half voting ‘It depends’—this research argued that using reaction GIFs and images that index femininity is not always a neutral process. To give but one example, this dataset showed how performances of the SQ persona may play into what Bailey (2021) describes as misogynoir: a specific form of racist misogyny aimed at Black women.

Furthermore, in discussing how the gay male participants of this CofP strategically and repeatedly embodied femininity via reaction GIFs and images, this article contributes to the discussion of how these multimodal linguistic objects may be utilised online to index sexual identity. Rather than being a new phenomenon, this article situated this communicative practice in the wider history of non-heterosexual individuals using language to perform gay identities. In other words, while reaction GIFs and images themselves are relatively new, the use of feminine-coded GIFs and images is a digital extension of the well-documented interactional custom of gay men performing gayness via feminine-marked linguistic items. Though this study has suggested that GIF/image selection may be guided by an admiration for or identification with the GIFs’/images’ subject(s), additional research may wish to substantiate this claim. Further research interested in online personae performance may also seek to unpick the question of whether cyberbodies, free from offline restrictions, can embody personae more completely than offline bodies (where language can be used but generic archetypes cannot be visually embodied in the same way). In all, this study not only highlights the relevance of reaction GIFs and images to sociolinguistic research but also emphasises the importance of analysing social media as a series of websites or applications in which complex identity work is accomplished via language and other communicative tools.