A Literature Review on Coevalness from Queer Temporality to Large Language Models
Introduction
On June 6, 2026, I woke up very early. I took a shower, put on my “gayest” outfit, with a giant pride flag draped over my shoulders as if it were a knight’s cape. I borrowed some makeup from my friend and applied a very simple, rough layer of eyeshadow. It did not make me look better, but I chose this unusual gesture as a symbol of my queer identity. That afternoon in Utrecht (GMT+2), I walked with rainbows and laughter, feeling as if I were living in the dream I had at 14 – when I came out to all of my friends, listened to Lady Gaga, and scrolled through the comment section of her song “Born This Way” to read the queer manifestos and imagine a beautiful, colourful future. At my second Pride Parade, I finally realised the queer dream I had held for eight years. It was a dream that had rescued me from the stagnant, oppressive Chinese reality that was behind the times. The Oudegracht was embracing the warm sun, while it was already late at night in China, GMT+8.
Behind the times?
This was one of the moments when I unconsciously used a term that implies a benchmark for evaluating the degree of “queerness”. “Behind the times” presupposes a single, linear timeline, on which some queer lives are validated as modern, while others are cast as backward or “traditional”. In Feeling Backward, Heather Love draws on postcolonial critiques of modernity and backwardness to show how the allegation of backwardness was extended to sexual and gender deviants – queers were seen as “a backward race”.[1] “Backward” names the unconscious feeling that shaped how I described my encounter with people who seemed to be “from another time”. Where does this kind of judgment come from, and where is it produced and circulated? These are the questions that my thesis aims to investigate.
In my thesis, I focus on this unconscious judgment and aim to examine the interaction between queer culture and queer subjects in China. More specifically, I investigate contemporary public discourse on queerness and queer people, as well as the queer people whose lives diverge from this discourse. In the contemporary media landscape, where AI has become a significant new medium, Large Language Models (LLMs) provide a lens for understanding how this discourse is encoded: by analysing their output, we can make certain inferences about their training corpora, which, in turn, reveal to some extent the general contours of social discourse.[2]
The provisional research questions of my thesis are: “How is the description of queer life in contemporary China constructed in the discourse of LLMs?” Furthermore, what is the connection between the broader socio-cultural context behind this construction and individuals’ understanding of queer life – for instance, my own, as someone who is both a queer subject and an observer? This paper establishes the academic background for this question: starting from discussions of temporality in queer theory, it reviews how temporality has been addressed in Chinese queer studies and identifies the gap left by existing research. This paper argues that, the texts analysed in existing studies of Chinese queerness all have authors, intentions and contexts, so the temporal benchmarks they set can be interrogated; by contrast, when narratives about Chinese queer life are generated by LLMs on a probabilistic basis, there is no author to question, and where their benchmarks come from has not yet been studied.
Queer Timezone
In the context of queer and postcolonial theory, a linear, progressive timeline is not a neutral measure, but is deeply embedded in multiple dimensions of power relations, serving as a tool for producing “others”. Therefore, in Chinese queer studies, whether the discussion starts from transcultural comparison or from debates within Chinese culture, the temporal dimension is essential to understanding queer subjectivity. In his book Time and the Other, Johannes Fabian coins the term “allochronism” – namely “denial of coevalness” – to describe the political nature of time: researchers tend to place the referents of research at a different point on a timeline of modern development, even though both in fact share the same present.[3] The “backwardness” discussed in this thesis, or in Fabian’s terms the “primitive”, is essentially a temporal concept: it is “a category, not an object, of Western thought”.[4] This insight provides the basic theoretical framework for this paper: how is time used to produce “others”? With this understanding, this chapter turns to the work of Jack Halberstam and Heather Love to explore how queer temporality is deployed within the queer community and in a larger social context, as well as how the queer community relates to its own history.
In his book In a Queer Time and Place, Halberstam theorises “queer time” as a term “for those specific models of temporality that emerge within postmodernism once one leaves the temporal frames of bourgeois reproduction and family, longevity, risk/safety, and inheritance.”[5] In other words, the so-called normal timeline that serves as a benchmark is heteronormative. Halberstam also criticises Dennis Altman’s “Global Gay” model, which equates Western gay identity with modernity and relegates Asian and Latin American same-sex desires to tradition.[6] This suggests that allochronism operates not only between the West and its “primitive” others, but also within the queer community along the divide between Global North/South. More importantly, Halberstam connects temporality with spatiality, showing how temporal and spatial narratives reinforce each other. He coins the term “metronormativity” to describe narratives in which moving from the rural to the urban is equated with coming out of the closet, so that spatial displacement is interpreted as psychological and historical progress.[7] For example, in his reading of the Brandon Teena case, the transgender youth Brandon murdered in rural Nebraska, occupies, in popular narratives and in the imagination of urban audiences, the position held by so-called primitives in colonial anthropology.[8] Drawing on Fabian, Halberstam argues that Brandon is cast as living on a different timescale from the modern queer. Halberstam’s discussion of temporality and spatiality is crucial to this autoethnographic paper. Having grown up in a small city in northern China, I have had queer experiences that are bound up with the temporal narratives attached to a “backward” geographical location. Now living in Utrecht, however, I have myself moved along the metronormative route – from a “backward” small city to a “modern” metropolis. The concept of metronormativity, therefore, provides a theoretical perspective for exploring my positionality as an observer.
If I now occupy a “modern” position on this timeline, how should I, in the present of physical time, look back at a queer reality that is cast as lagging behind? On this question, Love advocates a viewpoint that we should develop the politics of the past.[9] If Halberstam reveals how queer lives are differently temporalised within a larger social context, Love turns her critique towards queer history itself. As mentioned earlier, Love argues that the concept of modernity relies on others marked as backward, and that queer people were seen across the twentieth century as a backward race.[10] In response to this stigma, the gay liberation movement sought to transmute shame into pride through a narrative of liberation.[11] However, Love repeatedly reminds us that this shift from past humiliation to present or future affirmation means “ignoring the past as past; it also makes it harder to see the persistence of the past in the present”,[12] “shame lives on in pride, and pride can easily turn back into shame”.[13] Therefore, the politics of the past attends to the shame and other negative feelings that cannot be erased even during present-day pride, and refuses to leave behind those who cannot keep up in the name of progress. From this perspective, the progressive narrative of gay liberation itself denies coevalness to its own past, in the way Fabian described. Following Love’s backward turn, this paper re-examines the figure of the “lagging” queer from the perspective of coevalness, asking what attitude I should adopt towards the queer narratives and subjects left behind once the progress narratives are dismantled.
In summary, the work of these three scholars collectively reveals the exclusionary nature of linear, progressive time: Fabian provides the analytical framework of allochronism, Halberstam extends it along the axes of queer time and space, and Love turns it inward, towards the queer community’s own history. However, the empirical material of all three is centred on the West. When the global development narrative of the LGBTQ+ communities is placed in the Chinese context, do Chinese queers – cast as lagging behind the West, and as “local” in relation to the metropolis – face a double allochronism?
When “Queerness” Intertwined with “Chineseness”
The previous chapter established that the core of allochronism lies in the setting of a temporal benchmark. When this framework enters the Chinese context, the benchmark takes on a concrete form. As Lisa Rofel argues in Desiring China, Altman’s account of the “global gay” arranges Asia’s various sex/gender orders on a continuum from tradition to modernity.[14] Forms that approach a “Western-style” model as modern and relegate the rest to tradition. Although Altman admitted that cultural specificity coexisted with Western queer, he denied their coevalness.[15] In Bao Hongwei’s article “When Queer Theory Speaks Chinese”,[16] he identifies three key moments in queer theory’s entry into the PRC: the appropriation of the term tongzhi – which represents the attempt by activists and scholars in the Sinosphere to construct a Chinese queer epistemology; the publication of Li Yinhe’s translated anthology of Ku’er Lilun (Queer Theory) in 2000 – which marks the debut of queer theory in mainland Chinese academic discourse; and the “Sailor Moon Warrior Lala” online debate of 2011-2012 – which marks the symbolic separation between gay identity politics and queer politics in China.[17] Bao’s three moments serve here not as objects of analysis, but as an entry point into the temporal assumptions of the theorists who wrote within them. Focusing on the first two moments that have identifiable authors whose benchmarks can be traced, this chapter asks: in narratives of Chinese queerness, whose present serves as the standard against which time is measured? And can this benchmark be moved?
The First Moment – Tongzhi: Reversing the Timeline
In 1989, the organisers of the first Hong Kong Gay and Lesbian Film Festival adopted Tongzhi as the Chinese equivalent of “gay and lesbian”, on the ground that “gay” and “lesbian” were Western constructs with specific histories. Similarly, the manifesto of the 1996 Chinese Tongzhi Conference argued that the Western les-bi-gay movement, built on confrontational politics and individualism, might not offer the best path to tongzhi liberation in family-centred Chinese societies.[18] Chou Wah-shan is one of the most representative scholars of this moment. In his book Tongzhi,[19] Chou not only refuses the Western temporal benchmark, but also attempts to reverse the colonial timeline. In his view, modernisation is not progress, but rather the end of a long history of cultural tolerance of same-sex eroticism in China: what China‘s modernity imported was precisely the “outdated” Western mental illness model that the West itself was abandoning.[20] Going further, he suggests that there is no necessity to repeat the Western path since mainland China “has not been implicated in the straight-gay dichotomy.”[21] Therefore, Chou refuses to translate tongzhi as “comrade”, “gay”, or “queer”, lest it be “reduced to an earlier form of the U.S. gay movement”.[22] In these arguments, Chou seeks not only coevalness but also to question who is ahead, attempting to reverse the temporal narrative of queer liberation. In practice, Chou proposes “Coming Home” as an alternative to the Western narrative of Coming Out: whereas the latter follows a linear sequence from the closet to the public visibility, in which “out” implies “leaving the family, parents, and culture”, the former draws the same-sex relationship into the family kinship.[23] Yet for all their critical force, a reversed timeline is still a single line: the reversal remains within the logic of allochronism rather than moving beyond it.
The Second Moment – Ku’er: Accepting and Critiquing the Benchmark
Where Chou reverses the timeline, the second moment accepts it. In translating “queer” Li Yinhe adopted ku’er (literally “cool kid”), a term already in use in Taiwan and Hong Kong, which “loses the stigmatised history and subversive meanings embedded in the English term”.[24] For Li, queer theory was less an anti-normative critique than a means to accelerate the “gradual formation of a more rational and more civilised sexual order” in China.[25] This optimism rests on the same linear, progressive timeline that underpins allochronism, and Rofel and Bao reveal the temporal structure beneath it.
Rofel describes a postsocialist allegory in which Maoism is blamed for suppressing “natural” humanity, while “postsocialist reforms are portrayed, in contrast, as setting human nature free.”[26] Bao further argues that the emergence of gay identity in China is part of a postsocialist imagination of modernity, which is based on the “rejection of China’s socialist legacy and an uncritical endorsement of neoliberal capitalism”.[27] Rofel and Bao share with Chou a refusal to equate “opening up” with liberation. Where they differ is in how they treat the past. Chou appeals to a cultural tradition interrupted by modernisation.[28] Bao, on the other hand, turns to the forgotten socialist past: questioning the “repressive hypothesis”: he shows that same-sex intimacy in that period was not defined by repression alone, and argues that different modes of sexual subjectivity coexist rather than one replacing another.[29] Rofel does not deny that Chinese sexual experience has its own specificities, but she rejects explanations that invoke an ahistorical “Chinese culture” stripped of its socialist past, insisting instead that these differences are shaped by China’s colonial and socialist histories.[30]
In summary, these two moments outline a trajectory: from reversing the timeline to accepting the benchmark, to critiquing it by recovering the multiple pasts that the progress narrative erases. Notably, these texts all have identifiable authors, intentions, and context, thus the benchmarks they set can be traced. But when narratives about queer people in China are probabilistically generated by Large Language Models trained on large-scale corpora, where do their benchmarks come from, and how can they be observed?
The Queer Digital Clock
Generative artificial intelligence may serve as an infrastructure for writing the progressive timeline into public narrative. Bender et al. point out that “from initial participation in Internet fora, to continued presence there, to the collection and finally the filtering of training data,” at each step “current practice privileges the hegemonic viewpoint”.[31] Read through Fabian, this hegemonic viewpoint is “the present of the producer of anthropological discourse”[32] in a new form: the training corpus. When LLMs become new producers of queer narratives, the question posed in Chapter 2, namely whose present serves as the benchmark, returns with a new object. Therefore, this chapter aims to reconsider existing studies of queer AI bias from the perspective of temporality and coevalness.
The first mode of allochronism is absence. Drawing on markedness theory, which recognises “the normative assumptions buried in the way we do or do not append modifiers”,[33] Gillespie posed unmarked narrative prompts multiple times to each of four generative AI tools and found that they “tend to reproduce normative identities and narratives, rarely representing less common arrangements and perspectives”.[34] In nearly two hundred love stories, only one was explicitly queer; in the holidays on which couples quarrelled with their parents, 139 of 154 were Thanksgiving – Gillespie calls this “the tyranny of the unmarked”.[35] In Fabian’s terms, the unmarked is the present of the producer: it needs no marker because it is itself the benchmark; furthermore, as the holidays suggest, that present is not only heterosexual but American.
Marking the prompt seems the obvious remedy: as Gillespie concedes, a request for two men falling in love can receive the appropriate answer exactly “every time”.[36] But marking not only makes the Other visible, but also decides where the Other is placed. Vicsek et al. mark not the story but the speaker: they prompted ChatGPT and Bard with homophobic statements with a user background: Russia, Saudi Arabia, Orthodox Christian or Conservative Muslim.[37] Support for LGBTQ+ people declined, especially when a religious background was given.[38] They referred to this tendency as “cultural relativistic logic” – the adaptation of responses to match the societal or religious norms specific to each context.[39] A reply such as “It is understandable that as a conservative Muslim, you may adhere to certain religious teachings” sounds empathetic,[40] but it treats the user’s view as the expression of a fixed culture rather than a contemporary position open to dispute. As Vicsek et al. themselves mentioned, relativism “tends to portray culture as consensual”,[41] which touches upon how Altman treats Asian cultures as “timeless, bounded, homogeneous, and unchanging”.[42] The questioner is thus placed in another time. As an alternative, Vicsek et al. advocate a universal human rights logic, in which cultural differences are respected only when they do not result in human rights violations, and an LLM’s support for LGBTQ+ people “remains steadfast across various cultural contexts”.[43] However, this does not escape this problem either. If applied without reflection, it may slide back onto Altman’s single linear timeline, on which the Other can only catch up with the West. This AI debate thus repeats the disagreement between tongzhi and ku’er in the previous chapter, in which neither side moves beyond the single timeline.
In conclusion, the two studies reveal two modes of allochronism in LLMs: when unmarked, queerness is absent from the default present; when marked, it is placed in another, possibly a “primitive” time. However, neither study examines China. Moreover, when these studies touch on “time”, they only ask whether the model lags behind social progress, without questioning whose present defines that progress. What clock LLMs set when narrating queer life in China remains to be studied.
Conclusion
Starting from the phrase “behind the times” that I blurted out on the day of the parade, this paper has summarised a limited body of existing literature to further ask where such a judgment comes from. Chapter 1 draws on Fabian’s concept of allochronism, showing that linear, progressive time is not a neutral measure but a mechanism for producing “Others”, and follows Halberstam and Love as they extend this concept into queer contexts along the dimensions of space and history respectively. Chapter 2 brings this discussion into the Chinese context: Chou tries to reverse the timeline, Li Yinhe accepts its benchmark, and Rofel and Bao critique it by recovering the multiple pasts it ignores. Then, Chapter 3 re-reads existing studies of queer AI bias from a temporal perspective. The benchmarks discussed in Chapters 1 and 2 can all be traced to concrete authors, intentions and contexts, and can therefore be questioned. By contrast, LLM narratives are generated probabilistically, and their benchmark appears as a neutral default. However, existing research on AI bias has not specifically examined mainland China, and it only asks whether the model lags behind social progress, without asking whose present that progress is based on. Do Chinese queers face a double allochronism? What “clock” does LLM set in the narrative of queer lives in China? These are the questions that my thesis will discuss.
Moreover, recognising this temporality does not mean one can escape from it. Having traced these benchmarks, I do not stand outside the timeline either. The “behind the times” I blurted out on the day of the parade, and the impulse to reverse it to which I return below, both come from the same position. This position starts from a small city in northern China and follows the metronormative route to the metropolis. As Love said, “shame lives on in pride”,[44] similarly, the feeling of being “behind” does not disappear simply because it has been named. How this position shapes my observations will be elaborated in the autoethnographic section of this thesis.
Returning to the sentence that this paper began: while the Oudegracht was embracing the warm sun, “it was already late at night in China, GMT+8.” The word “already” put China ahead on the clock, since GMT+8 enters a new day six hours earlier than GMT+2. In the same paragraph, however, “behind the times” sent China back into the past. My first impulse was to persist with the former in order to deny the latter, and to insist that China is in fact ahead – which repeats Chou’s reversal of the timeline, and a reversed timeline is still a single line. When writing this essay, a few days later, at the end of October, the Netherlands will turn its clocks back for one hour, and the time difference with China will grow from six to seven. This made me think: is GMT+8 running faster, or has the Netherlands stepped back an hour? – a question that still asks who is ahead, again. The time zone is a human measurement, Greenwich became the zero point of world time only through an international agreement in 1884. The fact is that physical time never stops running. No matter whether I understand this or not, on the last Sunday of October my phone will still turn back an hour itself, without my consent or even my notice. I cannot live outside the time zone.
Examining temporality neither means abandoning the clock – which none of us can do – nor means searching for, or creating a neutral one. It is fundamental because it reminds us that every clock is constructed, and it is necessary to continue to ask whose present is its reference.
[1] Heather Love, Feeling Backward: Loss and the Politics of Queer History, 1st Harvard University Press paperback ed. (Cambridge, MA: Harvard University Press, 2007), 5-6.
[2] Nicholas Buttrick, “Studying Large Language Models as Compression Algorithms for Human Culture,” Trends in Cognitive Sciences 28, no. 3 (2024): 187–89, https://doi.org/10.1016/j.tics.2024.01.001.
[3] Johannes Fabian and Matti Bunzl, Time and the Other: How Anthropology Makes Its Object (New York: Columbia University Press, 2014), 31.
[4] Ibid., 18.
[5] Jack Halberstam, In a Queer Time and Place: Transgender Bodies, Subcultural Lives (New York: New York University Press, 2005), Chapter 1.
[6] Halberstam, In a Queer Time and Place, Chapter 2.
[7] Ibid.
[8] Ibid.
[9] Love, Feeling Backward, 21.
[11] Ibid., 28.
[12] Ibid., 19.
[13] Ibid., 28.
[14] Lisa Rofel, Desiring China: Experiments in Neoliberalism, Sexuality, and Public Culture (Durham, NC: Duke University Press, 2007), 89-90, https://doi.org/10.1515/9780822389903.
[15] Ibid., 90.
[16] Hongwei Bao, “When Queer Theory Speaks Chinese,” in Routledge Handbook of Chinese Gender & Sexuality, ed. Jamie J. Zhao and Hongwei Bao (Abingdon, Oxon: Routledge, 2024), 165-179.
[17] Ibid., 166.
[18] Bao, “When Queer Theory Speaks Chinese”, 167.
[19] Wah-Shan Chou, Tongzhi: Politics of Same-Sex Eroticism in Chinese Societies (Routledge, 2013).
[20] Ibid., Chapter 1.
[21] Ibid., Chapter 8.
[22] Ibid., Chapter 8.
[23] Ibid., Chapter 7.
[24] Hongwei Bao, “When Queer Theory Speaks Chinese,” 170.
[25] Ibid, 171.
[26] Rofel, Desiring China, 12.
[27] Hongwei Bao, Queer China: Lesbian and Gay Literature and Visual Culture under Postsocialism (New Delhi: Routledge India, 2020),Chapter 1, https://doi.org/10.4324/9781003027898.
[28] Chou, Tongzhi, Chapter 8.
[29] Bao, Queer China, Chapter 1.
[30] Rofel, Desiring China, 101.
[31] Emily M. Bender et al., “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜,” in Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (2021), 614.
[32] Fabian and Bunzl, Time and the Other, 31.
[33] Tarleton Gillespie, “Generative AI and the Politics of Visibility,” Big Data & Society 11, no. 2 (2024), 3, https://doi.org/10.1177/20539517241252131.
[34] Ibid., 1.
[35] Ibid., 4-5.
[36] Ibid., 9.
[37] Lilla Vicsek et al., “Cross-Cultural Challenges in Generative AI: Addressing Homophobia in Diverse Sociocultural Contexts,” Big Data & Society 12, no. 4 (2025), https://doi.org/10.1177/20539517251396069.
[38] Ibid., 8.
[39] Ibid., 10.
[40] Ibid., 7.
[41] Ibid., 4.
[42] Rofel, Desiring China, 91.
[43] Vicsek et al., “Cross-Cultural Challenges in Generative AI,” 4.
[44] Love, Feeling Backward, 28.
Leave a Reply