What Women Want? A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters

Sexist media depictions of female politicians have been studied. However, studies regarding portrayals of female voters and their hetero/cisnormative narratives are lacking. Therefore, this study explores UK media constructions of female voters in the context of the 2016 EU Referendum and the 2017 General Election via corpus-assisted CDA. It mainly focuses on the exclusion of LGBTQ+ women as voters via hetero/cisnormative language use. Female voters tend to be depicted in relation to men, or children and a family in general, while mentions of queer women are rare. Moreover, LGBTQ+ rights chiefly feature in political discourse to further a nationalist agenda through homonationalist arguments.


Introduction
The UK has been in a state of 'permalection' as the previous 4 years have seen three general elections (i.e. 2015, 2017, 2019) in addition to the 2016 referendum regarding whether the UK will leave the EU, the so-called Brexit referendum. There are a fair few studies that examine the representation of female politicians during times of election, in both the UK and other countries (e.g. Cameron, Shaw 2016), but what about the representation of female voters? This question appears to be particularly pertinent since female voters have been put somewhat front and centre, in the UK but also in the US for example, with women's marches happening across the world, 'stunts' such as pink buses sporting 'woman to woman' slogans to attract female voters and the founding of a Women's Equality Party in the UK. Such studies are far and few between. Which made me curious as to who is in/excluded from the group of 'female voters' and more specifically 'LGBTQ+ female voters' and what their priorities are perceived to be. 1 Will the media discourse include non-cis/non-heterosexual women, or will they be excluded through the use of heteronormative and/or cisnormative language? In short, how are (LGBTQ+) female voters constructed linguistically by the UK media in the lead-up to both the 2016 EU Referendum and the 2017 General Election (henceforth 2017 GE)? And is there a difference between left-wing and right-wing newspaper representation?

Politics and the Media
The media is viewed as the real public space in which politics take place and through which people understand politics and its processes (Lewis 2013). Media institutions purport to be objective and neutral in the political debate. However, many scholars argue that the media and their implicit biases play a mediating and constructing role in the political process and in turn in society's views and power imbalances (Chouliaraki, Fairclough 1999;Wodak, Meyer 2009). Furthermore, many media institutions in fact explicitly state their political allegiances. The UK is particularly known for its partisan press which is the least trusted in Europe (Hardy 2017). British broadcasters are legally compelled to be impartial, yet British newspapers do not have to adhere to such tight regulations 2 (Starkey 2017) and

Media Representations of (LGBTQ+) Women
Media images are often people's predominant source of knowledge about issues relating to underrepresented groups (e.g. women). Therefore, the manner in which women are portrayed is significant to the formation of public opinions, as negative and biased representations lead to stigmatization. The news media are a prime site for the analysis of gender representations as they are ineradicably linked to gender and power (Williams 2002), and women are often excluded from and/or severely underrepresented in the news (Jaworska, Hunt 2017;Gibbons 2000). Furthermore, "because the news is made by men, it is thought to reflect the interests and values of men" (Van Zoonen 1998, 34). Consequently, the news often reflects and reinforces patriarchal discourses by perpetuating gendered stereotypes that sexualise and objectify women by focusing on their physical appearance rather than their achievements (Ross 1995(Ross , 2000. Additionally, representations of women tend to be more negative than representations of men and/or even misogynist (Ajzenstadt, Steinberg 1997). A vast literature also demonstrates that women tend to be viewed as more sensitive and emotional, in contrast to the assertive and dominant nature of men (e.g. Len-Rios et al. 2005). Lastly, it is also a commonly used legitimation strategy within media discourse to 'other' and demonize oppressed groups such as women (Thompson, Yates 2017) and specifically women whose identities intersect with other oppressed groups (e.g. LGBTQ+ women, Muslim women, women of colour) (Baker 2006(Baker , 2014Gabriel 2017). The aforementioned 'sexual object', and 'mother' stereotypes also tie in with the rather pervasive discourse of heteronormativity. Women are primarily assumed to be heterosexual, whilst LG-BTQ+ identities are stigmatised and 'othered', perpetuating the norm of compulsory heterosexuality (Coates 2013;Motschenbacher 2011). Furthermore, heteronormative discourses also favour spe-cific types of heterosexual relationships and gender roles, exemplified by the: nuclear family, involving a stable, monogamous (preferably marital) and reproducible […] sexual relationship between two adults […] whose social and sexual roles are differentiated along conventional lines. (Cameron, Kulick 2003, 9-10) Accordingly, women have also been found to be more strongly associated with marital status than men (Pearce 2008;Jaworska, Hunt 2017). Lastly, previous research has indicated that age is another intersecting identity which connects to the trivializing and infantilizing of women. Jaworska and Hunt (2017) found that woman and girl tend to be used synonymously, whereas this is not the case for man and boy. This echoes Bolinger's (2014) suggestion that women tend to be represented as never growing up, and always remaining subordinate to men.

Media Representations of Female Voters
Gender bias within politics and the use of gender stereotypes in media representations of politicians are well documented, (e.g. Semetko, Boomgaarden 2007;Valenzuela, Correa 2009). Yet, the literature on media representations of and appeals to female voters is rather lacking in both breadth and depth, even though it has been established that the underrepresentation of women in politics and negative, sexist portrayals of female politicians may put women off standing for office and/or voting (Katwala, Ballinger, Mattinson 2016). There is some research on appeals to female voters, yet the focus tends to lie on female politicians, or the electorate as a monolithic whole, a "people" (Zappettini, Krzyzanowski 2019). For instance, Scullion (2015) found that the electorate as a whole was portrayed with reverence, or infantilization, but no word on whether women were more or less revered or infantilized, as women tend to be (also see Farrell 2016 on Brexit coverage). Furthermore, there are several studies on the voting behaviour of women and how it tends to reflect a preference for female candidates (Dolan 2012;McElroy, Marsh 2010). Recent studies have also looked into the political engagement of women on Twitter (Mitchell 2015;Parry 2015), or (heterosexual) women as members of male-focused political fandoms in the UK (e.g. the Milifandom: Ed Miliband fans) (Cameron, Shaw 2016;Hills 2017;Norris 2017).
Among the previous studies on the actual media representation of female voters in the UK, Adcock (2010) appears to be the most salient and exhaustive. Adcock found that 'ordinary women', or voters, featured more heavily in news sources than expected, but they were Lotte Verheijen A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters

A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 129 Language,Gender and Hate Speech, portrayed as "uninformed, irrational, confused or apathetic mothers, housewives, shoppers, workers, and patients" (2010,148), and their views were judged as inappropriate in the masculine arena of politics. Women were also framed as 'adoring' fans of male politicians who are easily charmed and cannot be reasoned with. However, like the Adcock (2010) study most previous research is rather narrowly focused on one election or referendum (e.g. Harmer 2016; Ross 2016 on the EU Referendum; Harmer, Southern 2017 on 2017 GE), or comparing 'ordinary women' whose voices tend to be drowned out or constrained (Savigny, Warner 2015;Shaw 2006) to female politicians (Harmer, Southern 2017). These studies suggest that female voters tend to be more often portrayed as citizens than as experts and addressed as 'women' in a rather general sense (Ross 2016). Women are viewed as a homogeneous group, but further research is necessary on this topic. Moreover, large-scale, intersectional and diachronic studies that compare several election/referendum campaigns also appear to be lacking from the literature. For instance, heterosexuality has featured in a minor way in previous research, but neither heteronormativity nor LGBTQ+ voter representations have explicitly been identified. Regarding the studies on political appeals to women, foci have comprised gendered appeals by politicians identifying as mothers (Quirk 2015), or gendered 'stunts' such as Harriet Harman's Pink Bus (Savigny, Warner 2015) instead of serious appeals, which begs the question whether appeals to women are more personal and perhaps less jargon-heavy?

UK Politics. LGBTQ+ Issues
Even though the political rhetoric and possible impact of Brexit have been explored from many angles (Jackson, Thorsen, Wring 2016), there has been limited research on how Brexit and the other 'permalection' elections relate to and/or affect LGBTQ+ people in the UK (Danisi, Dustin, Ferreira 2019;Wintemute 2016), despite the fact that hate crimes against LGBTQ+ minorities have risen starkly since the EU Referendum (Stonewall 2017). Some studies have looked at LG-BTQ+ issues in relation to major events of (inter)national and 'nationalist' importance and pride such as the 2012 London Olympics. Hubbard and Wilkinson (2015) found that Puar's (2007) original 3 no-tion of 'homonationalism' in which nations are seen as more 'progressive' because of their supposed 'acceptance' of gay people, reiterating self/other dichotomies that demonize 'foreign' others (Browne, Nash 2014). This acceptance and tolerance is then also seen as a source of national pride and used to position such 'progressive' nations against other nations that are viewed as ostensibly less tolerant. Not only during the 2012 Olympics did the UK boast its gay-friendliness, as the UK government generally promotes LGBTQ+ rights both at home and abroad, positioning itself as 'world-leading' in terms of gay rights (Hubbard, Wilkinson 2015). This despite the fact that, according to Danisi, Dustin and Ferreira (2019), the EU has been the catalyst for change in the UK, which begs the question if and how the notion of homonationalism ties in with the Brexit debate and the 2017 GE.

Rationale for This Study
As established by the previous paragraphs, underrepresented groups such as women in general, and female voters and LGBTQ+ female voters in particular are both underrepresented by the media as well as in studies of such media representations. This is the case even though it has been established that the underrepresentation of women in politics may put women off standing for office and/or voting. Previous studies have tended to be somewhat narrowly focused, lacking diachronic analyses across multiple political campaigns, and viewing 'women' as a homogenous or even monolithic group rather than a multitude of both intersecting and varying identities. For instance, heteronormativity and LGBTQ+ voter representations have not explicitly been identified in previous studies. Therefore, the present study aims to fill this research gap and contribute directly to reduce the underrepresentation discussed above. Furthermore, the British press is a prime site for such research as it is almost unique in large democracies in terms of its reach, ubiquity and one-sidedness. It dominates national conversations around politics in Britain and can therefore potentially affect election outcomes. Due to this wide reach and the notion that the news often reflects and reinforces heteronormative and cisnormative patriarchal discourses, the representation or non-representation of (LGBTQ+) female voters will have far-reaching effects and consequences, and are worth further exploration. Especially when one considers that hate crimes against LG-BTQ+ minorities have risen starkly since the EU Referendum and arguments concerning homonationalist ideas might have affected Brexit in a myriad of ways.

Lotte Verheijen A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 131 Language, Gender and Hate Speech, 125-146 7 The Present Study. Theoretical and Methodological Framework

Critical Discourse Analysis
In this (preliminary) study I employ methods of analysis associated with Critical Discourse Analysis, henceforth CDA. CDA is a set of approaches, theories and associated methodologies which can be viewed as an overarching framework of reference or a critical perspective (Van Dijk 2001), concerning the study of language use, or discourse, and its wider social context. CDA views language as a social practice (Fairclough, Wodak 1997). Language itself is not powerful, rather "it gains power by the use people make of it and by the people who have access to [it]" (Baker et al. 2008, 280). Moreover, CDA is problem-oriented and focuses on "social problems, and especially on the role of discourse in the production and reproduction of power abuse or domination" (Van Dijk 2001, 96). This ties in with Johnstone's (2008) explanation of 'discourse', employed in this study, as conventional ways of talking or writing which produce and are produced by conventional ways of thinking, which in turn constitute (power-based) ideologies in society. CDA then aims to critically investigate "opaque as well as transparent structural relationships of dominance, discrimination, power and control as manifested in language" (Wodak, Meyer 2009, 10) which may not be apparent in everyday life (Van Dijk 1993).

Triangulation. Corpus-Assisted CDA
A widespread criticism regarding 'cherry-picking' of results within CDA can be offset by the use of quantitative corpus techniques to complement as well as serve as a basis for the more qualitative CDA analyses. Thus, triangulating your results by means of corpus-assisted discourse analysis (Partington 2009). Frequency analyses of articles and/or certain terms and their linguistic environments by means of a corpus software can help elucidate the reach and possible impact of the linguistic phenomena under investigation by accounting for all of their instances and consequently facilitating quantitative comparisons between corpora (Baker et al. 2008). Such frequency analyses may also provide an initial focus for analysis in the higher frequency terms and/or phenomena lending credibility to generalizations, which might also counteract the danger of 'cherry-picking' interesting phenomena that might support pre-conceived hypotheses, but may not be frequent or salient in the data (Mautner 2009;Widdowson 2008).

Corpus-Building. Sources and Time Periods
In order to analyse the linguistic representation of (LGBTQ+) female voters by the UK news media, a set of (sub-)corpora were constructed. I used the online database NexisUK to search for and collect relevant articles from national UK newspapers. National newspapers were the most relevant option for this study, as such newspapers have the broadest readerships and generally the largest influence on public discourse. Liberal centrist n/a n/a The Independent n/a n/a n/a The Mirror Centre-left, populist Labour Remain The Observer Centre-left Labour Remain The People Centre-left, populist n/a n/a The Sun Right-wing Conservative Leave The Telegraph Centre-right Conservative Leave The Times Centre-right Conservative Leave In order to ensure the longitudinal nature of the corpus the articles from the above-mentioned newspapers were collected from two sep-

Lotte Verheijen A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 133 Language,Gender and Hate Speech, arate campaigning periods within the 'permalection' period. This longitudinal framework facilitates the drawing of cross-campaign comparisons with regard to the representation of (LGBTQ+) female voters. Both in terms of a comparison between two consecutive years and two different types of campaigns: a GE and a Referendum. Furthermore, such diachronic analyses have, to my knowledge, not been done before with regard to the representation of (LGBTQ+) female voters in the news media.
The starting point for each of these periods was set at the official start of each campaign and the endpoints coincide with the GE date in 2017 (18 April-08 June) and the EU Referendum date in 2016 (15 April-23 June). The specific dates used are based on previous studies on these electoral events (Deacon et al. 2015;Moore, Ramsay 2017).
In order to limit the size of the corpus and the breadth of the analyses, while still maintaining the longitudinal aspect of this study, these yearly sub-corpora do not span the immediate aftermath of the election and referendum. The periods of campaigning were deemed to be a feasible and comprehensive sample with a clear start and end. Moreover, his is an approach that has been employed by other election-based studies (see Jackson, Thorsen, Wring 2016).

Search Terms
The search terms were chosen on the basis of other gender-based/ focused media research and pilot studies involving my own corpora. The majority of media studies on female representation merely look at neutral terms such as women, female (Cameron, Shaw 2016;Jaworska, Hunt 2017). In addition to these terms, I also wished to include terms carrying stronger connotations, such as lady, girl, as I aimed for a broad and representative perspective on female voter presentation. Therefore, I included a host of other nouns (and their plurals) 4 that are used to describe women or that are often used as synonyms for a woman such as mother or daughter: Gendered terms: female, woman, lady, girl, lesbian, ((great-)grand)mother, (grand) mum(my)/mom(my), mamma, ((great-)grand)daughter, wife, grandma, matron, aunt, niece, girlfriend, sister, mumsnet, nan, gran, stepmother, stepmum/mom, gal, lass One can also direct Nexis to search for specific combinations of terms. The following list of terms related to elections and voting in combination with the search constituent 'w\p', ensures that the gendered terms appear 'within the same paragraph' as these election terms. This excludes articles with separate unrelated paragraphs (e.g. one paragraph on women and another paragraph on voting) yet includes the combination of the two sets of terms within a sentence as well as articles where the gendered term and the election term are separated by a few lines.

Voting-related terms: vot!, 5 election, Brexit, referendum
After developing the search terms, sources and time periods discussed above, I collated the overarching corpus and its relevant subcorpora. I also annotated each article with specific tags in order for the articles to be sorted into corresponding synchronic sub-corpora, partially based on the political orientation and referendum stance categories listed in Table 1 to facilitate synchronic as well as diachronic CL and CDA analyses.

Corpora Characteristics
The 2016 and 2017 corpora differ greatly in size and composition (see Table 2 below). The 2016 corpus has almost double the number of articles of the 2017 corpus (742 vs. 438). One partial explanation for this could be the longer campaign leading up to the Referendum, 70 days, compared to the 52 days of the 2017 GE campaign. However, despite the longer campaigning period the average amount of articles per day is significantly higher in 2016 (10.6 vs. 8.4). Possible explanations could be that the EU Referendum was viewed as a more important and salient vote with more long-term effects than a snap GE. The aftermath has certainly been more eventful and drawn-out, as Brexit has yet to happen at the time of writing. There are also noteworthy differences evident in the sizes of the referendum stance and political orientation sub-corpora. The Leave and right-wing sub-corpora are substantially larger than their Remain and left-wing counterparts both in terms of number of articles and tokens, while the average number of articles per newspaper is also higher for right-wing/Leave newspapers in 2017 (see Table 1 below). However, the left-wing/Remain newspapers display a higher average number of ar-5 The '!' sign yields all words starting with the letters preceding it (e.g. vote(rs), voting).

Lotte Verheijen A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 135 Language, Gender and Hate Speech, 125-146 ticles in 2016 and this difference as well as the aforementioned difference in 2017 is much smaller than the difference in total number of articles. This indicates that the large difference in number of articles is primarily due to the fact that the British newspaper landscape is overwhelmingly right-wing [Tab. 1] and in fact left-wing/Remain-supporting newspapers tended to dedicate more articles to the plight of (LGBTQ+) female voters than their right-wing/Leave counterparts. Lastly, the difference between the sub-corpora is greater in 2017, which is mainly due to the fact that The Times, a prolific Conservative-supporting newspaper (23 articles included in both 2016 and 2017) which supported Remain in 2016 is included in the right-wing sub-corpus of 2017.  Figure 1 shows the number of articles related to female voters published per 5 days of the 2016 EU Referendum campaign, a period of 70 days between 15 April 2016 and 23 June 2016. Intervals of 5 days were chosen to provide a more detailed view of the period as a whole and allow for salient sub-periods of either increased or decreased coverage to emerge from the data.

Newspaper Article Frequency
Overall, 742 articles from this period were included in the corpus, which is an average of 10.6 articles per day, and 53 articles per 5 days. The highest article rate per 5 days of 172 was observed during the last week of the campaign between 19 and 23 June and the lowest rate per 5 days comprising 17 articles was observed between 5 and 9 May. The number of articles included appears to have undergone five distinct phases: a) steady flow between the first day of the campaign and 19 May; b) sudden, steady increase between 19 May and 3 June; c) renewed steady flow, or plateau, of coverage between 4 and 13 June; d) small decline in coverage between 14 and 18 June; e) stark increase up to 172 articles in the last week of campaigning between 19 and 23 June. These periods, the peaks and troughs in particular, are not random and could be explained by more general political coverage trends as well as events specific to this campaign. In general, one would expect a steady increase in coverage as an election, or in this case ref-erendum draws near (Jackson, Thorsen, Wring 2016). This is why the dip in coverage towards the end stands out as unexpected. Possible explanations might lie in a general oversaturation of 'Brexit' coverage. However, it could also be possible that topics not specifically tied to voter gender might have been laid by the wayside. The stark increase in articles from 19 June onward could be interpreted as contrary to the suggestion of women becoming 'less' important to the overall coverage and debate closer to the Referendum date. However, one could also argue that this increase is merely in line with the expected increased coverage and therefore does not indicate a possible increase in the attention to issues related to female voters.  From this period, 438 articles were included in the corpus, which is an average of 8.4 articles per day, and 33.7 articles per 4 days of the referendum campaign. The highest article rate per 4 days of 70 was observed during the last 4 days of the campaign between 5 and 8 June and the lowest rate of 17 articles per 4 days was observed between 24 and 27 May. The number of articles included appears to be rather unstable and much more 'event-based' than 2016's Referendum coverage which displayed a steadier increase over the course of the campaign. In fact, five distinct peaks can be identified which all relate to salient occurrences in the campaign.
First, there is the peak spanning two bouts of 36 articles between 22 and 29 April. This peak is linked to the snap GE being called by then PM Theresa May and a woman dubbed 'Brenda from Bristol' expressing her dismay during a BBC interview on the morning of May's announcement (Wheeler 2017). The second peak of 34 articles occurs between 4 and 7 May and can be linked to increased coverage due to

Lotte Verheijen A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 137 Language,Gender and Hate Speech, local elections in the UK and what they might mean for Theresa May and female voters in the GE. Peak 3 of 33 articles occurs between 16 and 19 May and coincides with the release of the political parties' manifestos outlining their plans (Thorsen, Jackson, Lilleker 2017). The fourth peak of 48 articles per 4 days transpires between 28 and 31 May. This peak can be linked to not one, but two salient events occurring at once: the publication of new poll results predicting significant Labour gains and Labour leader Jeremy Corbyn's "car-crash" interview on the Woman's Hour radio show (Thorsen, Jackson, Lilleker 2017) in which he is stumped on the costs of childcare, a cause closely associated with the plight of female voters. Lastly, the fifth and highest peak of 70 articles is situated during the last 4 days of the campaign between 5 and 8 May which ties into the more general trend of a steady increase in coverage as an election draws near. Overall, 2017's article frequency displays more peaks and troughs compared to the steady rise of 2016 and thus 2017's frequency is more event-based.

Search Term Frequency
The frequency of search terms, calculated by means of the corpus tool AntConc (Anthony 2018), reveals similar trends as well as differences within the 2016 and 2017 corpora (see Table 3 below). In both years the most frequent term is the general woman/women, while in 2017 the other general term female is relatively more frequent than in 2016 (rank 2 vs. 4). Furthermore, both corpora show a propensity for a heteronormative and cisnormative use of traditional, heterosexual, reproductive 'nuclear family' terms (Cameron, Kulick 2003). In fact, female voters are often talked about or interviewed in relation to the men in their family ("and his wife", "his mother", "mother of [MALE NAME]"), or a family in general ("mother-of-two"), while mentions of LGBTQ+ women are rather infrequent.
The 2016 corpus appears to be slightly more hetero-and-cisnorma-  tive, as wife is ranked higher (rank 2 vs. 4) than in 2017 while mother is ranked similarly (rank 3). Another difference between the two corpora relates to the inclusion of lady/ies in 2016. Upon closer inspection of the use of this term, the 2016 articles appear to be more geared toward older women, or 'ladies', as it is used to describe female politicians, (older) female voters of nobility and older/middle-aged female voters in general. On the contrary, the phrases 'young lady/ies' barely show up. In 2017 despite lady/ies being relatively less frequent, 'young lady/ies' is more frequent and the 2017 campaign appears to be geared more toward younger female voters. Perhaps because the effects of Brexit might be stronger for them. The lack of LGBTQ+ terms among the top search terms and the heteronormative/cisnormative implications of the terms that were included indicate the lack of importance accredited to LGBTQ+ issues during the campaigns covered in this study. In order to explore if and how LGBTQ+ women were represented and appealed to in the two campaigns I scrutinized the concordance lines of the following LGBTQ+ terms: lesbian, gay, bisexual, trans(gender). These analyses exposed that queer issues appear to be more prevalent in 2017, as articles detailing Theresa May's voting record on LGBTQ+ issues appeared in The Independent. In addition, there were certain particular themes present within the representation of LGBTQ+ women. Firstly, in general the LGBTQ+ community is addressed as a whole, instead of addressing queer women specifically, which mirrors how female voters are seen as a homogeneous group (Adcock 2010). References to gay men were also more frequent than references to gay women. Example 1 below describing a trans woman's voting preference (i.e. the Green Party) and Example 2's mention of "the lesbian vote" landing on the Green Party, which also includes a reference to "older gays", are two of the few exceptions to this rule.

Lotte Verheijen A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 139 Language,Gender and Hate Speech, (1) "I want to vote Green. I feel the party is much more citizencentred in its policies and sticks up for minority groups and working people. That resonates with me because I'm a trans person who was assigned male at birth but now identifies as female". (The Observer, 28 May 2017) (2) There are a lot of rich, older gays in Brighton. They probably tend towards Tory [Conservatives]. The lesbian vote tends towards Green or Labour. (The Times, 30 May 2017) Other themes prevalent regarding LGBTQ+ issues relate to such issues being described as non-issues, as "we have gay marriage" and alleged equality with regard to sexuality, as exemplified by Example 3 below.
(3) we have gay marriage and sexuality is rarely a barrier to any form of personal or public achievement. (The Guardian, 2 May 2017) On the other hand, some articles describe LGBTQ+ issues, or "gay rights", as a "threatening" topic to a "lot of people" which appears to contradict the argument that full LGBTQ+ equality has been achieved and LGBTQ+ issues are no longer an urgent matter (see Example 4 below).
(4) The fact of the matter is that gay rights and feminism is very threatening to a lot of people. (The Guardian, 5 May 2017) Lastly, notions of (common) homonationalism, which as I will discuss below were the most salient theme for 2016, also appear in 2017 in order to express Islamophobic and xenophobic viewpoints. In this case LGBTQ+ issues are only seen as important as a stance against a party's or a journalist's biased notion of Islam as a religion that views "women or gay people" as "second-class citizens" (see Example 5 below).
(5) To introduce a "social attitudes" test as part of a points-based immigration system which would stop people who believe women or gay people are "second-class citizens" from entering the country. (telegraph.co.uk, 25 May 2017) In the 2016 corpus homonationalism in both Puar's (2007) original definition and the new and more common definition (Puar 2013) appear to be the most salient theme(s) regarding LGBTQ+ issues in the Brexit campaign. In the unfounded 'facts' of Example 6, from a Leavesupporting newspaper, one once again sees how gay rights are on-ly viewed as important as a stance against an Islamophobic view of Muslims as homophobic and therefore un-British and against specifically Muslim immigration, which according to Vote Leave could be stopped when the UK leaves the European Union, as it would then supposedly be free to tighten its immigration policies.
(6) THE fact more than half of UK Muslims want gay sex outlawed and almost a quarter want Sharia law shows how little they are willing to integrate. (The Sun, 15 April 2016) On the other hand, Remain-supporting newspapers display the same pride in gay rights (and women's rights), but also pride in "racial equality" (see Example 7) and view such rights as not just UK, but EU values. Consequently, leaving the EU would cast the UK back to times without "gay marriage", but with women being confined to "the kitchen and bedroom". This EU-wide homonationalism, so to speak, then views the EU as a progressive 'nation'. However, further research is needed to explore the homonationalist notions present in Remain and Leave rhetoric in more detail.
(7) Voting to leave would mean voting against racial equality and gay marriage and in favour of confining women to the kitchen and bedroom. (The Observer, 31 May 2016)

Conclusions
This preliminary study shows that although the 2017 corpus is smaller there appears to be an increase in the representation of female voters in general and LGBTQ+ female voters in particular from 2016 to 2017 (e.g. women has become a keyword). 6 However, heteronormative and cisnormative terminology prevails, albeit to a lesser extent in 2017. Because of this, LGBTQ+ female voters remain excluded through the perpetuation of hetero/cisnormative language use. Akin to a plethora of studies on the representation of women in the media showing 'the mother' to be a feminine archetype (e.g. Jaworska, Hunt 2017), as well as Adcock's (2010) findings regarding female voters being portrayed as 'mothers', and 'housewives', the high frequency of (heteronormative and cisnormative) nuclear family terms such as wife (as opposed to the non-marital girlfriend, for instance) and mother in this study show that female voters appear to be synonymous with mothers of men and wives of men. They are talked 6 Keywords are words that are more frequent in the studied corpus than in reference corpora and indicate the 'aboutness' of said corpus (Baker 2006).

Lotte Verheijen A Multimodal Critical Discourse Analysis of UK Media Constructions of (LGBTQ+) Female Voters
Quaderni del Comitato Unico di Garanzia dell'Università Ca' Foscari Venezia 1 141 Language,Gender and Hate Speech, about and interviewed with regard to the men in their lives and consequently it appears to be the case that their views might be judged as inappropriate in the masculine arena of politics. They are mere citizens, not experts (Harmer 2016), and addressed as 'women', the most frequent search term in both corpora, in a general sense (Ross 2016), and in turn as a homogeneous group. Furthermore, the presence of lady/ies as a frequent search term in 2016 and its links to older female voters bears further research, as older women tend to be rather underrepresented in the media. Instead the focus often lies on infantilized younger women (Bolinger 2014). Moreover, mentions of queer women remain infrequent in both 2016 and 2017 and almost solely confined to discourses surrounding the LGBTQ+ community as a monolithic whole, and/or notions of homonationalism linked to Islamophobia. Moreover, homonationalism, and its differing definitions, is particularly salient in the Brexit debate and further research is needed to explore the links between the Remain camp, which displayed a higher average article frequency per newspaper and therefore perhaps a heightened dedication to the plight of the (LGBTQ+) female voter, and the original definition of homonationalism, and the Leave camp and common definitions of homonationalism.