‘Bisexual oysters’: A diachronic corpus-based critical discourse analysis of bisexual representation in The Times between 1957 and 2017

Corpus-based critical discourse analysis

This study is framed within an analytical tradition that understands discourse as ‘practices which systematically form the objects of which they speak’ (Foucault, 1972: 49). In other words, our knowledge of the social world is largely a product of culturally and historically contingent discourses which are, in part, constituted by language. This is exemplified in our contemporary notions of sexuality. While same-sex desire has always existed, the notion of a ‘homosexual’ or ‘bisexual’ identity is a product of psychological, medical and criminal discourses that emerged in Europe during the late 19th century (Foucault, 1978). Over time, such ideas were reproduced and mediated through language, eventually becoming accepted as common-sense categories that help people to identify themselves and others. In this way, we can see how bisexuality is not a universal identity across time and space, but rather a discursive construction that emerged in a particular time and location. How and when such a process occurred can be mapped by using diachronic corpus-based critical discourse analysis.

Discourse analysis looks to examples of naturally occurring language for ‘traces’ of discourse (Fairclough, 1989; Sunderland, 2004; Van Dijk, 1993). Corpus linguistic methods can enhance this analysis by approaching a representative sample of language with computational methods, thereby revealing elements that would not have been visible in a smaller quantity of data. In this study, I will be focusing on collocation and the close reading of concordance lines as an entry point to how bisexual people have been discursively constructed.

Collocation refers to the frequent co-occurrence of words in a particular corpus (Stubbs, 2001). While the frequency of a single word may provide insight into certain aspects of the corpus, the analysis of how words interact with one another can also ‘be useful in revealing how meaning is acquired through repeated uses of language, as certain concepts become inextricably linked together over time’ (Baker, 2014: 18). In other words, if certain lexical items frequently occur together then the occurrence of a word in isolation may still ‘trigger’ the connotations and encoded cultural concepts that are produced through its collocate (Stubbs, 1996, 2001). For instance, if bisexual frequently occurs near the word promiscuous, then a ‘strong’ theory of collocation would suggest that bisexual might come to act as a psychological trigger for the idea of promiscuity. This point is of particular significance in the context of how marginal identities are represented in the media. According to Stubbs (1996), ‘if particular lexical and grammatical choices are regularly made, and if people and things are repeatedly talked about in certain ways, then it is plausible that this will affect how they are thought about’ (p. 92). Frequent collocates may, therefore, affect how bisexual people are represented and how they are perceived.

In addition to collocation, Baker (2010) suggests carrying out ‘concordancing work in order to identify exactly how the relationship is manifested in language’ (p. 24). For instance, even if bisexual frequently collocates with promiscuous (as in the example above), an analysis of concordance lines may reveal that in many cases the two words appear together because authors are challenging the idea that bisexual people are being represented as promiscuous. Concordance lines may reveal grammatical features that indicate something relevant to representation, for example, past tense verbs may indicate that bisexual promiscuity is being discussed as something that occurred in the past. Qualitative analysis of discourse, therefore, enhances the data that have been generated through corpus linguistic techniques, enabling a more robust and nuanced analysis of how bisexual people have been represented through language.

This study combines two approaches to corpus-based critical discourse analysis. On the one hand, I am taking a ‘naïve’ view when approaching the question of how bisexuality has been represented, allowing the data to guide the analysis (see Partington et al., 2013). On the other hand, I also want to interrogate the corpus for ‘traces’ of bisexual erasure as well as how bisexual people have been misrepresented in the discourse of the British Press. This latter approach is more congruent with critical discourse analysis in which the researcher begins by focusing on a social wrong in its semiotic aspect and then selects for texts in which the problem is ‘semiotically realised’ (Fairclough, 2010: 245). My political position is that media representations of marginal identities and minority communities impact on how such groups are perceived. Uncovering and drawing attention to such inaccurate or negative representations is a critical step in mitigating such discourses and improving the lives of those affected. The following discussion will, therefore, further explore discourses of bisexual erasure and misrepresentation in order to frame how the corpus data will be interpreted.

The erasure and misrepresentation of bisexual people

Contemporary theories of bisexuality have pointed to the manifold ways in which bisexuality is both erased and misrepresented (Angelides, 2001; Du Plessis, 1996; MacDowall, 2009; Yoshino, 2000). For instance, while female bisexuality has been particularly salient in popular culture (Johnson, 2016), its authenticity has often been questioned, especially in cases where it is clear that such depictions have been construed for the gaze of heterosexual men (Johnson, 2016). Questions of authenticity, however, have also been extended to the very idea of a bisexual identity, characterizing bisexuality as not having the same legitimacy as homosexuality and heterosexuality (Angelides, 2001; Du Plessis, 1996; MacDowall, 2009). Angelides (2006) argues that this is because bisexuality has not been permitted the status of a legitimate sexual identity, as it would destabilize the ‘epistemological integrity of the very categories of man, woman, heterosexual and homosexual’ (p. 132). This argument is congruent with one posited by Yoshino (2001), who explains that homosexual and heterosexual people are mutually invested in denying the visibility and existence of bisexuality and have conceded to an ‘epistemic contract of bisexual erasure’ in which bisexual people are not permitted the same ontological status as monosexual identities, that is, homosexuality and heterosexuality.

While bisexual identity may have been repressed, bisexual practice has often been pathologized, having been represented as being responsible for transmission, infection and – ultimately – scapegoated for the spread of HIV/AIDS to the heterosexual community (Kaestle and Ivory, 2012). The ramifications of such discourses strengthened the notion that bisexual people were more promiscuous than the general population while also misrepresenting the methods through which the heterosexual population saw a rise in infection. In an article debunking the myth of the bisexual infector, Worth (2003: 72) contended that not only were bisexual men ‘deemed promiscuous, unreformable, dishonest fence sitters’, but that their very identity had become conflated with HIV/AIDS meaning that they were now also perceived as ‘dangerous’.

While the effects of bisexual erasure and misrepresentation are likely to be highly varied according to, inter alia, gender identity, ‘race’, ethnicity, class, ability, age and region, Johnson (2016) demonstrates that bisexual people have worse health outcomes than the general population as well as when compared with the gay and lesbian population. Such outcomes are also manifest in higher rates of depression, mood anxiety disorders and higher rates of suicidality (Johnson, 2016). Such outcomes are also attributed to the negative media representations which depict bisexual people as promiscuous, mentally unstable, lying about their sexuality or performing their bisexuality for the consumption of others (Johnson, 2016). In sum, ‘poor media representation contributes to widespread biphobia, which contributes to poor mental health’ (Johnson, 2016).

The following section will describe how this study addresses the preceding issues by building and analysing a corpus of representative language concerning bisexuality.

The evolution of bisexual – from oysters to space stations

Between 1957 and 1967, iterations of the lexeme bisexual only occurred 16 times, with a closer reading of the concordance lines revealing that only 5 out of the 16 occurrences of bisexual referred to an individual who is sexually or romantically attracted to both male and female identified individuals. The remaining 11 occurrences suggested that the meaning of bisexual has changed significantly. For instance, the corpus revealed that The Times more frequently used bisexual to refer to someone or something, for example, oysters, that had both male and female reproductive organs and capabilities:

Concordance 1: Bisexual – organisms that are not sexually differentiated

OPEN IN VIEWER

his or her is correct, for the British oyster is	bisexual.	The American oysters, by the way, are males or females
250-150 million years ago, provide evidence of	bisexual	division. The Gonophyll theory has excited great interest
found all the species that he has studied to be	bisexual.	They have a closed circulation system containing large

Another unexpected use of bisexual was situations that involved both men and women. Two of the occurrences used the term to debate the possibility of a ‘bisexual priesthood’ in which women could become ordained priests in the Church of England, while one other occurrence was used in reference to sports that could be played by both men and women (Concordance 2).

Concordance 2: Bisexual – comprising members of both male and female identified people

OPEN IN VIEWER

and they will become much more tangled with a	bisexual	official priesthood." Women and Holy Orders, Church
attached to the Report seem to fear that a	bisexual	priesthood could exacerbate personal jealousies
It is common experience that more individual and	bisexual	things like fencing, canoeing, and dancing tend to attract

Mixed gender situations such as these were complemented by references to ‘bisexual fragrances’, a product that would now also likely be labelled ‘unisex’ as well as the ‘bisexual nature’ of certain instruments. Of the five occurrences that did refer to bisexuality as a sexual identity, two were ambiguous and could possibly have referred to someone who was androgynous in appearance and another whose ‘bisexual nature’ compelled them towards homosexuality, while yet another was used in a book review where the author describes how the letters sent by a male character to both a man and a woman were ‘bisexual’, but not the person who had written them (Concordance 3).

Concordance 3: Bisexual – other uses

OPEN IN VIEWER

torment of the poet’s life lay in his	bisexual	nature and the resultant homosexual
performance in the part, he makes Lussurioso a	bisexual	exquisite, more absurd than evil.
to dissuade him, and fill much of the novel with	bisexual	letters back home to boy friend and wife. But

While such uses of bisexual would appear unusual in an issue of The Times today, they comport with meanings that were in use at the time and date back to the 18th century. According to the Oxford English Dictionary (OED) (2018a), the term bisexual is first attested in 1793 in reference to a flower that contained both male and female reproductive organs. Indeed, a significant number of entries for bisexual in the OED allude to ‘hemaphroditism’ (OED, 2018). Of the eight entries for bisexual in the OED, only three denote sexual or romantic attraction to members of ‘both sexes’ – a usage that became significant in the early 20th century (OED, 2018). In addition to entries on biology and physical characteristics, the remaining entries support some of the uses in the corpus, referring to situations ‘involving or comprising individuals of both sexes’ (OED, 2018: n.p.), or items that are ‘suitable for or applicable to both men and women’ (OED, 2018: n.p.). Further support for this range of usage is explored by Macdowall (2009) who delineates three different uses for the term that have been in use since the 19th century. He claims that the first usage of bisexual refers to ‘forms of life that are sexually undifferentiated or thought to exhibit characteristics of both sexes’ (Macdowall, 2009: 4). Unlike this first definition which was rooted in evolutionary theory, the second category of bisexual was ‘used to describe a combination of masculinity and femininity in an individual’ (Macdowall, 2009: 4). This second usage comports with contemporary notions of androgyny and first came into use in the early 20th century. At around the same time, bisexual comes to signify sexual and romantic attraction to male and female identified individuals (Macdowall, 2009). Based on the evidence from the corpus and MacDowall (2009), the term bisexual can therefore be used to signify a combination of characteristics that are based on the binary oppositions of male/female, masculine/feminine and heterosexual/homosexual. Evidence from the corpus, however, demonstrates that, while there is consistent semantic variation between 1957 and 1990, it is bisexuality as a sexual identity that comes to dominate the language of The Times.

This shift in meaning becomes evident in the data between 1968 and 1978. During this time, the ratio of the lexeme bisexual to denote a sexual identity increases from 31% (1957–1967) to 66% just a decade later. The remaining cases fall roughly into the same categories discussed above, with the majority of usage referring to ‘bisexual clothing and fashion’ as well as people whom we would now likely characterize as intersex or transgender. ² In addition to these examples, there are also two concordance lines that contain a disambiguation of ‘androgynous’ through the use of ‘bisexual’ in parentheses (Concordance 4). In the first concordance line, the author is referring to the dock of a space station which is ‘androgynous’, while in the second, the author refers to the potential androgyny of the Judaeo-Christian god. The use of bisexual to disambiguate androgynous is interesting because it suggests that bisexual was falling out of common use and being replaced by terms that are more familiar by contemporary standards. While there are only two examples of bisexual being used to clarify a word that had come to replace it, the decreasing frequency with which bisexual was being used to signify intersex, transgender, androgynous or unisex is evidenced further in the next section of the corpus.

Concordance 4: Bisexual as disambiguation for androgynous

OPEN IN VIEWER

draw up preliminary plans for a so-called androgynous (	bisexual	) system “that is neither male nor female in design”
“he”. Biblical images of a female or an androgynous (i.e.	bisexual	) deity put forward by Dr Morton included the pillar of

The roots of bisexual erasure – HIV/AIDS and conflation through inclusion

Between 1979 and 1990, the use of bisexual to refer strictly to a sexual identity grew from 66% to 92%. This demonstrates that, during this era, bisexual largely ceased being used in the female/male and feminine/masculine paradigms and came to almost exclusively represent a third space in the homosexual/heterosexual binary. While there are likely to be other factors involved in this shifting lexical meaning, there is considerable evidence that, during the 1980s, the HIV/AIDS crisis played a significant role in how bisexuality figured in the discourse of The Times. After removing non-contemporary uses of bisexual and controlling for duplicates, there were a remaining 227 tokens of the lexeme bisexual. Of these, 52% were used in the discussion of HIV/AIDS.

An examination of the concordance lines shows that bisexual people were represented in two distinct, but not always discrete ways. First, bisexual people tended to be folded into discursive groupings with other ‘high-risk groups’ (Concordance 5). The corpus presents examples in which bisexual people would be listed with people described in the corpus as ‘homosexuals … intravenous drug users, haemophiliacs’, ‘prostitutes’ and ‘the very promiscuous’. Aside from haemophiliacs, these groups of individuals were not only described as having higher rates of infection, but were also presented as culpable for their HIV status as well as the growing epidemic.

Concordance 5: bisexual people with other ‘high-risk groups’

OPEN IN VIEWER

the great majority are male homosexuals or	bisexuals	, or injecting drug misusers, but in about 5
at risk include practising homosexuals,	bisexuals	intravenous drug addicts, haemophiliacs
men with the disease, 1,226 are homosexual or	bisexual	; 85 haemophiliacs; 43 intravenous drug
used to treat their condition. Homosexuals,	bisexuals	and intravenous drug abusers continue to
the main risk groups were still homosexuals,	bisexuals	and drug addicts. The figures showed that of

While bisexual people existed in the same discursive environment as other ‘high-risk groups’, it was with homosexuals that they were most closely aligned. In fact, 55% of the bisexual tokens returned in the search were paired with homosexual through the use of conjunctions such as and, or and the symbol, ‘/’. The frequency of such usage suggests that, for the first time, bisexual consistently collocates with homosexual and, in effect, contributes to the notion of bisexual erasure (Angelides, 2001). In other words, as bisexual becomes folded into a discursive category with homosexual, bisexuality as a distinct identity and group becomes compromised and, indeed, conflated with homosexuality more broadly.

Concordance 6: Bisexual people represented as a ‘bridge’ for HIV infection.

OPEN IN VIEWER

sneeze. It seems that the semen of infectious	bisexual	men could be a danger to female partners.
be as well to avoid intercourse with a	bisexual	or the very promiscuous. 09.28.85 As
presumably being transmitted to women by	bisexual	men. There is some concern that prostitutes
are believed to have caught the virus from	bisexual	or drug abusing men. “The increasing
to know, the truth about their partner’s	bisexuality	, or past history of drug-taking. As one New
active women, especially those who have	bisexual	partners, to avoid becoming pregnant. In a
envisions New York after Aids has become a	bisexual	plague; Talk Radio (Public Theatre) exposes
discovered that she had contracted Aids from a	bisexual	lover some years previously. Her husband man
contracted Aids after a one-night stand with a	bisexual	. A teenager who thought it could not happen to
exposure to the virus was sexual contact with a	bisexual	. Britain may have 1,800 new Aids patients in

In the 33 years between 1957 and 1990, the term bisexual came to signify something very different from whence it had begun. It had transitioned from having various meanings to primarily signifying a fixed sexual identity at the margins of society. This marginal nature meant that, in their vulnerability to the HIV/AIDS crisis, they became discursively represented as the bridge between deviant populations associated with the epidemic and the broader heterosexual population. Discourses of a discrete bisexual sexual identity could therefore be argued to have both contributed to their formation as a discrete identity category associated with sexual practice while simultaneously contributing to their erasure within the broader homosexual community.

Fictional and temporal displacement of bisexuality

While the previous three sub-corpora have demonstrated how the term bisexual changed significantly between 1957 and 1990, there was one way in which bisexual representation remained consistent, and indeed, has continued to be until 2017. Namely, The Times has consistently represented bisexuality in discourses of fiction and the past. One of the ways in which this fictional and temporal displacement of bisexual people is made evident is through their ubiquity in the ‘Reviews’ section of The Times. The TDA divides articles into 8 sections and 25 sub-sections including subject areas such as ‘Reviews’, ‘News’ and ‘Editorials/Leaders’. While one might expect bisexuality to be featured in several sections, bisexual people are overwhelmingly featured in the ‘Reviews’ section and, largely, as characters in theatre productions, film and television (Table 2). Between 1957 and 1967, the lexeme bisexual occurred in the ‘Reviews’ section 50% of the time. Between 1968 and 1978, this percentage dropped to 42% and then, as a result of bisexuality and HIV/AIDS discourses, it fell further to 22% between 1979 and 1990 as bisexual became largely featured in the ‘News’, which now comprised 42% of its mentions.

OPEN IN VIEWER

Table 2.

Distribution of bisexual by publication section.

Years	Sections in The Times
	Reviews	News	Arts and Entertainment	Other
1957–1967	50%	38%	0%	12%
1968–1978	54%	23%	4%	18%
1979–1990	24%	42%	20%	14%
1991–2003	31%	21%	21%	30%
2004–2017	N/A	N/A	N/A	N/A

Nevertheless, while the language of bisexuality eventually moved out of the ‘Reviews’ section of The Times, discourses surrounding bisexuality continued to be relegated to the realm of fiction and of the past. That is to say that, if bisexuality was discussed in reality, it was either referred to as a sexual practice that had occurred in someone’s life but no longer did at the time of writing – for example, ‘ If I ever have grandchildren, I’ll probably make them listen to stories about when I was bisexual ’; an ongoing practice in someone’s life who was no longer living – for example, ‘Michael Redgrave was a bisexual and an alcoholic’; or in the far distant past – for example, ‘(Julius) Caesar himself was bisexual’.

This fictional and temporal displacement of bisexual individuals continues to be salient in the sub-corpora dating between 1991–2003 and 2004–2017. Due to the size of these sub-corpora, the ‘Word Sketch’ function was used to identify collocates that would likely indicate how bisexuality was being represented in The Times. The 1991–2003 corpus returned 56 types and 540 tokens that collocated with bisexual. Through a close reading of the concordance lines in each of these groupings, it became evident that bisexual was frequently used in discourses that dealt with the past or in fiction.

An analysis of words that were ‘modifiers of “bisexual”’ showed that 67% of the examples discussed bisexuality and bisexual individuals within discourses of the past. These included the ancient Gaels who were described as ‘partly bisexual’, Leonardo Da Vinci who was ‘almost certainly bisexual’ and Lord Byron’s half-sister, Augusta, who was ‘probably bisexual herself’ (alluding to Lord Byron’s sexuality as well). On the other hand, while past was one of the highest frequency collocates in the ‘modified by “bisexual”’ section of the ‘Word Sketch’ (Figure 2), discourses dealing with fiction were more significant than discourses of the past. For instance, writers at The Times frequently discussed ‘bisexual girlfriends’, ‘bisexual love triangles’ and, of course, ‘bisexual characters’ all within films, television and theatre. Beyond such bisexual tropes as ‘triangle’, however, fictional bisexual characters also included bisexual ‘barrister(s)’, ‘queen(s)’ and two instances of a bisexual ‘cameraman’. In fact, 46% of the collocates that were modified by bisexual were used in the discussion of fiction, a ratio that was followed by discourses of HIV/AIDS at 19% and the past at 15%.

OPEN IN VIEWER

Figure 2.

Nouns and verbs modified by ‘bisexual’ (1991–2003).

This trend was complemented by the grouping, ‘verbs before “bisexual”’, which showed that the lemma be was the most significant collocate of bisexual. Of the 79 tokens, 42 used ‘was’ or ‘were’ to discuss someone’s bisexual past or a bisexual person who lived in the past. In fact, the only grouping that revealed a slightly different story was collocates that were connected through the conjunctions ‘and/or’ (Figure 3). There were still numerous examples of fictionalized bisexual people, including ‘black, bisexual warrior(s)’ and ‘dwar(ves)’, ‘boozy, bisexual mother(s)’ and ‘aunt(s)’ and a ‘feisty, bisexual loner’. The majority of collocates, however, folded bisexual within discourses concerning homosexuals and gay people – a discursive strategy that conflates the unique identity of bisexual people with the larger community of (in this corpus) mostly homosexual and gay men.

OPEN IN VIEWER

Figure 3.

‘bisexual’ and/or (1991–2003).

The 2004–2017 corpus further supports this observation, demonstrating how bisexual has increasingly become locked in as one component in the broader LGBTQ community. In a ‘Word Sketch’ of bisexual, 50% of the 2064 collocates demonstrated that bisexual collocates primarily with lesbian, gay, transgender, queer and intersex, thus becoming simply one letter in the abbreviated community labels LGB, LGBT or LGBTQI. Such erasure through inclusion/conflation was further demonstrated as another 124 examples of bisexual were incorporated within the language of men’s sexual identities and practices more broadly, for example, ‘gay or bisexual men’ or ‘homosexual or bisexual men’ (Figure 4).

OPEN IN VIEWER

Figure 4.

‘Nouns and verbs modified by bisexual’ and ‘bisexual and/or’ (2004–2017).

Where bisexuality was discussed as an identity or practice in its own right, it continued to be relegated to fictional discourses and the historical past tense. An analysis of words that modified bisexual revealed how 41% were used to describe people who had lived in the past or who had practised bisexuality in the past, but no longer did. These examples also used words that suggested scepticism, as the articles seemed to question whether their subjects were ‘possibly bisexual’, ‘perhaps bisexual’ or ‘probably bisexual’. This suggests that, in The Times, bisexuality is rarely accepted as a fact of one’s identity, but is rather a suspicion of past actions that are open to question and subject to scrutiny (Concordance 7).

Concordance 7: Skepticism of bisexuality.

OPEN IN VIEWER

are Georgiana, the gambling-addicted, possibly	bisexual	18th-century duchess and the early Victorian duke who
whether MacMillan was homosexual, or perhaps	bisexual	There is some suggestion that he might have been
a morbid, sadomasochistic psychopath, probably	bisexual	who employed young men to flog him and slept with

Examples such as ‘flamboyantly bisexual’ – also in the ‘modifiers of “bisexual”’ category – also indicate a level of judgement when considering bisexual identity. A Word Sketch of flamboyant and flamboyantly in the British National Corpus (BNC) reveal that their subjects are often negative and would therefore benefit from being less flamboyant and more covert, such as flamboyant ‘voyeurs’, ‘exhibitionists’, ‘charlatans’ and ‘homosexual(s)’. This negative discourse prosody would also suggest that one’s bisexuality is something that ought to be kept secret as a source of shame and certainly not to be displayed ‘flamboyantly’. Such negative representations of bisexual people were echoed in fictional discourses as well. There were certainly associations with bisexual that appeared neutral, such as ‘teachers’, ‘writers’ and ‘characters’ but, echoing the evaluative discourses in the 1980s, there continued to be collocates which suggested the inherent promiscuity and depravity associated with bisexual people (Johnson, 2016). In the ‘Word Sketch’ grouping that shows collocates connected to bisexual through and/or, ‘voracious’, ‘promiscuous’, ‘manipulative’ and ‘psychotic’ were all notable suggesting that bisexual people, if represented as such, are likely to be perceived as such as well.