Who is our friend and who is our enemy? The enregisterment of tribalising digital discourse during the COVID-19 pandemic

: This study demonstrates how tribalism and nationalism, exacerbated by the COVID-19 pandemic, have given rise tothe enregisterment of a cultural discursive practice – dui ( 怼 ) – on Chinese social media. Based on a sample of social media posts collected through a combination of ‘ noticing ’ and systematic data scanning and archiving over a month, our analysis shows that dui shifted from a discursive practice of friendly teasing, a speci ﬁ c function that underpinned its rise of popularity prior to the onset of the pandemic, to an oppositional discursive practice characterised by wrangling, words of violence and a conversationalised journalistic discourse of refuting. We argue that COVID-19 has heightened Chinese netizens ’ sensitivities towards China ’ s perceived friends and enemies. This politicisation, together with the associated traditional discourses of national pride and humiliation, contributed to the emergence of dui as a tribalising discourse, which in turn has further destabilised and polarised the user community on Chinese social media and beyond. The study illustrates the mutually constitutive nature of the enregisterment of discursive practices and social relationships and the importance of bringing together discourse analytical and socio-cultural perspectives in analytical terms.


Introduction
'Who are our enemies?Who are our friends?' are questions posed by Mao Zedong nearly a century ago.He argued that the most important strategic question facing the Chinese Communist Party at that time was to know who its true allies were (Mao 1925).According to Dutton (2005: 4), the binary distinction between friend and enemy has ever since become a way of understanding China's politics as well as 'society and life which become fused in politics'.While it is possible that, as argued by Mouffe (2016), antagonistic relationship exists in all societies in the form of us versus them and friend versus enemy, Mao's concept of ally takes both domestic and external forms: 同志 tongzhi (comrade) for domestic allies and 友谊, youyi for foreign friendship, as in the 友谊宾馆 youyi bingguan (friendship hotels) and 友谊商店 youyi shangdian (friendship shops) that catered to foreigners in the early reform period (Nordin and Smith 2018).Importantly, this ontology of friendship departs from the Confucian emphasis on harmonious relationships to a more 'confrontational polemicisation' (Nordin and Smith 2018: 377) whereby dichotomised entities clash with and oppose each other.Following the outbreak of COVID-19, this friend-enemy dichotomy took on a new level of importance and became even more ossified, creating boundaries between groups, both domestically amongst the Chinese people around health practices such as lockdowns and quarantines, and externally with the rest of the world over vaccine nationalism and COVID-related policies and practices.In the present paper, we are interested in how these dichotomised group relationships have expediated the emergence of a tribalising digital discursive practice on Chinese social media embodied in term dui (怼).We employ the sociolinguistic notions of enregisterment and stance in our attempt to understand this emerging discursive practice, asking the following three questions: 1) what kinds of cultural discursive practices does dui embody?2) how did intensified nationalism and patriotism during the COVID-19 pandemic facilitate its enregisterment as a tribalising discourse?and 3) what impact has this discursive practice had on social relationships?
We will first provide a brief discussion of what we mean by tribalising discourse and make the case for the use of enregisterment and stance as the key analytical notions in our study.This is followed by a discussion of existing work on Chinese digital conflict discourse, with an emphasis on its relationship with group affiliation and networks.We then provide an account of our methodology and then analyse the changing discursive practice embodied in dui through the presentation of illustrative examples from our data.

Tribalising discourse, enregisterment and stance
We use the term 'tribalising discourse' to emphasise the mutually constitutive nature of oppositional discourse and group relationships.Tribalising discourse is oppositional discourse that sets out to discursively accomplish tribal epistemologies and at the same time serves as a communicative resource to reinforce the division between friends and enemies.The available studies on conflict and associated discourse events have been very helpful in understanding conflict discourse with a small 'd', i.e., language in use (Gee 2015).These include interactional features of conflict talk as situated, local interactional accomplishments (see a review of conflict talk in Leung 2002) and their variations and patterns in a range of social contexts where arguments occur in either interpersonal or institutional interactions (e.g., Grimshaw 1990;Zhu 2008).Evans et al. (2019) provide an overview of how manifestations and degrees of opposition can take different forms of participation, means, and spatio-temporality and how they are engendered or minimised through discourse.The types of opposition included in their review range from interpersonal impoliteness to hate speech, from institutional settings such as courts and airport control rooms to asylum gate-keeping and war zones, and from written and spoken interaction to digital communication.
Opposition is not just confined to interactions that can be described as conflict or disagreement.'The automatic, ritualized, knee-jerk use of opposition' can come to dominate whole societies (Tannen 2013: 180).Writing about the United States in the 1990s, Tannen attributed this culture of argumentation to an ethics of aggression that placed the highest value on attack and the liberal belief that opposition leads to truth.In the last few years, the war-like oppositional discourse described by Tannen has evolved into tribalising discourse amongst the growing political polarisation in the U.S., a phenomenon that was further fuelled by the COVID-19 pandemic (see Kerr et al. 2021 on political polarization in the response to COVID-19 in the United States).In showing loyalty towards friends and animosity towards enemies, tribalising discourse often works in concert with the psychology of fake news, in that one's group identity influences discernment of what is fake and what are facts (Oyserman and Dawson 2021).
To understand how the tribalising discourse routine this paper is concerned with has become a significant cultural mode of conduct on the Chinese internet, we turn to Agha's notion of enregisterment.Enregisterment refers to the 'processes and practices whereby performable signs become recognized (and regrouped) as belonging to distinct, differentially valorised semiotic registers by a population' (Agha 2007: 81).While the term so far has been predominantly used to understand how linguistic forms or variations become linked with places, personas or identities, Johnstone (2016) argues that enregisterment is not confined to ways of speaking such as pronouncing words, grammatical patterns or patterns of intonation.In fact, 'any potentially meaningful acts of any kind, at any level, can become semiotically linked with the culturally-relevant ways of acting and being that Agha calls registers ' (2015: 654).This applies to the discursive practice our study is concerned with.We are interested in finding out how a particular discursive practice comes to be associated with particular social practices and with persons who engage in such practices, as well as what the agents of this enregisterment are and under what interactional and historical conditions it occurs (Johnstone 2016).Additionally, we are interested in understanding how enregisterment in turn impacts social conditions and structures, following the belief that discourse and society is co-constitutive (e.g., Fairclough 1989;Foucault 1991).
To understand the dual purposes of tribalising discourse as an interactional accomplishment of opposition and an aggravating force for tribalism, we need an analytical notion that can offer a means of investigation both at the level of social interactions and at the level of social relationships and identities, and for this we turn to another sociolinguistic notion, that of stance.Stance describes the ways in which interlocutors relate to objects of interest and to others through ways of invoking systems of sociocultural value, assigning value to objects of interest, positioning social actions, and calibrating alignments between stance takers (Du Bois 2007: 139).Stance centres on relationships (cf.positioning, Jaffe 2009) and creates the relationships of speakers to their interlocutor, or to figures or ideas represented in the discourse (Kiesling 2011).It relates speakers/objects to each other through evaluation, positioning and alignment.Various components of stance have been proposed in the literature.The most relevant ones to our study are those proposed by Kiesling (2011), i.e., affect, alignment and investment stance.For Kiesling, affect stance is an interpersonal component, in which the discursive figure is liked or disliked, or distance or solidarity are signalled.Alignment stance is about how a speaker aligns or disaligns to an interlocutor (both epistemically and interactionally).It is achieved when interactants work together in moving the activity forward, even if the activity is about arguing against each other.Investment stance is about the extent to which a speaker defends their claims.These components are particularly helpful in understanding the kind of tribalising discourse embodied by dui.
The notion of stance has proven especially helpful in understanding how on-line commentators position themselves in their views and engage with others' views (Walton and Jaffe 2011;Zhu and Wei 2016).Digital communication has some built-in functionalities such as 'like', 'share', and 'block', which encourage people to evaluate each other's postings and thereby enable a public declaration of digital conflict on record (Graham and Hardaker 2017).Equally, other functionalities such as on-line commentaries also give people opportunities to display their stance, and thereby 'put stance-taking at the centre of the activity' (Walton and Jaffe 2011: 288).
In this paper, we will explore the enregisterment of a new tribalising discursive practice, dui (怼).Using the notions of stance-taking and enregisterment, we will examine what constitutes dui, what distinctive features it has that enables participants to recognise it as such, how it was mobilised and enregistered during the COVID-19 outbreak, and what consequences it has on social relationships.

Chinese conflict discourse and digital culture
In this section, we will provide a brief summary of studies on Chinese conflict discourse and the impact of such culturally relevant acts on group dynamics.A small number of available studies have identified some interactional features of conflict talk in Chinese such as the use of formulaic opposition markers (Kuo 1992), and strategies for opposition (Shen 2006).Others (e.g., Zhang et al. 2021) have observed cross-linguistic and cross-cultural differences in disagreement and mitigation, and still others have focused on disagreement strategies on the Chinese internet, including on online forums (e.g., Lee and Shum 2017;Shum and Lee 2013;Yang 2021) and via email (Rodríguez Velasco et al. 2021).
Several studies have shown that group affiliation and identities are crucial to understanding the social-cultural role of disagreement in Chinese interactions.Through a detailed analysis of the sequential organisation of codeswitching, for instance, Zhu (2008) has shown how intergenerational bilingual conflict talk within migrant families is closely related to differences in values and identities.Others, however, such as Zhu (2014aZhu ( , 2014b)), have demonstrated that disagreement can also be employed as communicative strategies to build rapport among peers in the Chinese context, departing from the stereotypical view of Chinese as preferring conflict avoidance.Kádár (2019: 210) observes that politeness in Chinese culture hinges on 'the Chinese normative distinction between inside and outside relationships,' and the lack of connectedness to outsiders is an important motivating factor behind different discursive practices.At the same time, he argues, who is an insider and who is an outsider is not always straightforward.An outsider as a visiting guest lecturer to a Chinese university with the right connections could be treated with ultimate hospitality, while people may not feel the need to apologise for treading on a stranger's foot on a train.Such behaviour, he argues, is, in many ways, 'logical' in a country as vast as China.
There is a general observation both inside and outside of China, that 'digital discourse lends itself to conflict' (Graham 2019).In addition to the built-in functionalities that encourage the display of stance mentioned above, other possible factors influencing on-line conflictual behaviour are 'degrees of anonymity, the (perceived) lack of consequences for bad behaviour, perceptions of privacy, the scope of the audience, degrees of a/synchronicity (e.g.how quickly messages are received and responses are generated) and cross-cultural expectations/interpretations that exist on a global scale' (Graham 2019: 310-311).The cross-cultural factors are particularly relevant to our discussion here.Despite the growing global deterritorialisation accelerated by the internet and migration, group dynamics on the Chinese social media have been found to be different in some ways from those in the West, creating different platform cultures, many of them characterised by what Schneider (2018) has termed China's 'digital nationalism'.Yang et al. (2022), for example, have demonstrated how this digital nationalism, along with optimism, fuelled narratives of COVID-19 vaccination nationalism on the Douyin platform (a popular short video platform in China), turning the platform users into nationalist networked public.A comparative analysis by Yang and Vicari (2021) illustrates the different participatory dynamics between Weibo (a popular Chinese social media platform) and Twitter during the COVID-19 outbreak: Weibo, they observe, serves to 'unite Chinese people and promote public policies under the control of the government and the guidance of the mainstream media', whereas Twitter, with its emphasis on personal expression, often 'shows the influence of defined partisan political discourses' (p.493).In fact, Weibo has gone through several phrases of development, from the initial stage of user empowerment and active participation to the current stage of interest-based networks as it carefully navigates the tension between corporate and public interests (Han 2018(Han , 2019)).On the whole, there appears to be less conflict talk between the users who post on both Douyin and Weibo compared to Twitter, and users of these platforms seem to be more united toward the 'common enemy', i.e., 'forces from outside China', though this is not necessarily due to the platforms themselves but also reflects the overall political climate within China.
In summary, existing, albeit limited, research on Chinese conflict discourse has offered some insights on the strategies of communicating disagreement in face-to-face contexts and the socio-cultural motivations behind these strategies.While this research has identified some differences between disagreement in face-to-face contexts and those mediated through digital communication, with the latter allowing more direct acts of disagreement and showing different participatory dynamics, little attention has been paid to how Chinese social media platforms and, broadly speaking, Chinese digital culture, afford novel ways of expressing opposition and conflict and how these ways of speaking, unite or divide networked groups.

Methodology
The data collection for this study consisted of two stages.The first stage was 'noticing'.Both authors are bilingual Chinese and English speakers based in the UK and moderate users of Chinese social media (e.g., WeChat) for work and social purposes.During the first few months of the outbreak of COVID-19, we began to notice the frequent use of the character 怼 (dui) in social-media postings.We did not know what the character meant precisely, but we were intrigued by its pictophonetic components, which indicate 'opposing hearts'.We noticed that it was used frequently to describe the way an infectious diseases expert, Dr Zhang Wenhong spoke and engaged with the public.Later we also noticed its frequent use in the heated debates on social media around the news of the publication of Wuhan Diarya first person account of the lockdown of Wuhan at the beginning of the pandemic by the prizewinning writer Fang Fangin English and German in April 2020.Its usage grew over time as key events related to COVID-19 such as the WHO visit to China and the development and subsequent roll-out of vaccines took turns to become trending topics on social media.
The noticing spurred us to carry out a systematic investigation of dui as an opposing discursive practice in the context of COVID-19 and its relationship to group dynamics among Chinese internet users.In the second stage, together with a bilingual research assistant, we conducted an extensive process of data scanning and archiving of the use of dui in social media over the course of June 2021 (as described below) with three questions in mind: What is dui?How is it used on social media in COVID-related posting?And what kind of discursive practices does it embody?
We first looked into the etymology of dui and on-line commentaries about the word.We monitored the trending topics on Weibothe popular Chinese social media platform as discussed above, visiting the site eight or nine times each day during this stage.These trending topics included vaccine rates in various countries; vaccine efficacy, side effects, vaccine donations, and investigations into COVID-19's origins.When we noticed a topic related to controversies around COVID-19 which netizens themselves characterised as dui in their postings or headlines, we archived the original post and screenshot the comments section.We collected 19 trending topics and their ensuing comments.Meanwhile, we also tracked the use of dui in news reports and in other online discussion about COVID-19 using Baidu, the main search engine in China.We searched for 怼 itself as well as a number of collocations with 怼, for example 怼+病毒 (virus), 怼+疫苗 (vaccination), and 怼+溯源 (origin), to narrow our searches to COVID-19 related discussions.We identified 13 news reports which contained dui in the headlines and archived the news headlines together with their content into a database of about 28,000 words with accompanying images.
We then analysed all the examples of dui in our collection closely, focusing on how it was used in the specific contexts and what kind of discourse practices it embodied in terms of affect, alignment and investment stances.We first noticed that there were differences in the way dui was used in affect stance.In the examples often cited as a typical example of dui in netizen's metacommentaries prior to or at the start of COVID-19, dui conveys jocularity and solidarity.However, in many COVID-related examples in our collection, more than a year after the start of the COVID-19 pandemic, dui is used to signal and in fact to construct a sense of opposition of different degrees of intensity ranging from wrangling to words of violence.In the following sections, we will discuss the changing discursive practices of dui and the enregisterment of dui as a tribalising discursive practice in the context of rising nationalism and polarisation, which was worsened by the traumatic experience of the pandemic.We will first examine the emergence of dui as a discursive act of friendly teasing when it first became an internet buzzword and its usage as a communicative strategy by a prominent health communicator at the beginning of the pandemic and then explore the more tribalizing associations it developed later in the course of the pandemic.The examples were selected to demonstrate differences in affect stance and intensity of opposition and to meet the overall aim of understanding how the new meaning and form of dui became enregistered.
5 Verbal push back: the internet buzzword 怼 dui as a cultural act and communicative strategy of friendly teasing In the 2017, dui was one of the top 10 internet buzzwords in China (http://www.chinadaily.com.cn/interface/flipboard/1142846/2017-12-22/cd_35327887.html),and, by January 2018, it appeared in more than 460 trending topics on Weibo and in more than 170 million news headlines with about 250 million readers (Gong 2018).
The rise of dui as an internet buzzword is testament to its iconicity (its resemblance between its linguistic form and meaning) and the role of social media in expediating linguistic creativity.怼 is a pictophonetic character combining a radical '心' at the bottom (meaning heart) and a phonetic component '对', indicating the sound (dui) and in this case also extending the semantic meaning of the character (while 对 can be 'right' or 'okay', it can also mean 'oppose' or 'be opposite to').Combining the two parts together, the character symbolises 'opposing heart' and was defined by the first Chinese Character dictionary 说文解字/Shuowen Jiezi, compiled between 100 and 121 AD during the Han Dynasty as 'grudge'.The consensus among the available materials tracing the origin of the word is that despite some evidence of its usage in the early Classical Chinese literature, the character gradually disappeared during the Qing Dynasty (1644-1912) and had rarely been used in contemporary Chinese apart from its use in some dialects until it became an internet buzzword in recent years.In the Wuyang dialect in Henan province, dui is used as a 'magic verb' equivalent to the English word 'do' in a variety of contexts.In the Northeastern dialect, dui is pronounced with a third tone/falling-rising tone and means 'push with hands' or 'argue with words', opposing either physically or verbally.It has to be said, though, that with the colloquial nature of the dialects, there are some questions about whether dui is indeed represented by the character 怼 or other characters with similar sounds in these dialects.
The emergence of dui as an internet buzzword signifying a cultural act of lighthearted rebuking is generally attributed to a popular Chinese reality television programme, 真正男子汉 (Takes a Real Man), in which participants go through rigorous training.In one episode, a group leader frequently conversed in this style of rebuking when trying to make his team members work harder.Once he was heard making a metapragmatic comment, '大怼大进步;小怼小进步' (literally meaning 'big dui, big progress; small dui, small progress'), declaring that the intensity of dui is positively correlated to one's progress.The combination of the various factorsthe trending television programme, the existence of a number of user groups of local dialects which use variants of the character, and its transparent, pictophonetic writinghave all contributed to the spread of dui on the social media (Gong 2018).There is some anecdotal evidence suggesting that dui is especially popular among young people (Yu 2017), though this needs to be substantiated by further investigation.
Within the dui sample we collected, this discursive practice of dui as friendly teasing is often associated with the communicative strategy used by an influential health communicator, Dr Zhang Wenhong, following the outbreak of COVID-19.Dr Zhang is widely regarded as the face of China's COVID battle, or 'China's Dr Fauci'.His way of speaking has been described as 'adroit' by The Guardian's Chinese Affairs correspondent Ni (2021).In the same article, this adroitness is attributed to a number of factors such as his sense of humour, his use of metaphors, and his common sense.On Chinese social media, this adroitness was often summed up using the word dui.In Q&A sessions with journalists and the public, for example, Dr Zhang often teased about 'silly' health behaviours among the general public.In Example 1, for instance, when invited to comment on the vaccination uptake, instead of providing a medical justification for vaccinations, Zhang jokingly asserts that knowing too much only makes people over-think.He draws a comparison with occasions where people take vaccinations at birth when they do not 'know' anything.This friendly, witty 'tellingoff' is characterised as dui in the blog title.
Example 1 '出生时你什么都不懂, 很多疫苗打就打了。 现在你犹豫其实是因为懂得太多了' (When you were born, you knew nothing and were vaccinated for many things.You are hesitant now because you know too much.)(Cited from a blog with the title of 'Are you still hesitating to take up vaccine?Zhang Wenhong dui humorously: You know too much' with the publication date of 2021-05-28 (还在犹豫打不打疫苗？ 张文宏风趣开"怼"： "你懂得太多了……"; Source: https://www.163.com/dy/article/GB3DDGA90514CFJG.html) In Example 2, Dr Zhang talks about transmission among close contacts, drawing from the experience of the 2013 bird flu outbreak in Shanghai, when patients ended up Who is our friend and who is our enemy?
passing the virus to their mothers, but not their husbands.He teases his listeners about the nature of 'intimate' relationships.
Example 2 '大多数感染给自己的母亲， 却没有感染给自己的老公。 我对爱情产生了怀疑。' (Most of infected patients passed the virus to their mothers, but not their husbands.I'm beginning to doubt the romance.)(Cited from a blog with the title of 'Dui those within China and dui outside China, Dr Zhang Wenhong with your sharp tongue, please stop exuding this damn charm!' with the publication date of 2020-03-28 (怼完国内怼国外， 毒舌张文宏医生， 请你停止散发这该 死的魅力！) Source: https://www.sohu.com/a/383896200_478426)These examples illustrate how dui was used as an effective communicative strategy by a public health communicator.However, a close analysis of other samples in our collection point to a different kind of discursive practice of dui, where a strong, less 'playful' oppositional stance is expressed by speakers.In the following section, we will investigate how dui became enregistered as a tribalising discursive practice.Partly as a result of circulation of its enregistered meaning and form through metapragmatic commentary, two new meanings/functions of dui appeared, influenced by nationalistic discourses: dui as words of violence that polarise social relationships, and dui as a discursive strategy of conversationalisation in storytelling.We will use the three stance components proposed in Kiesling (2011), i.e., interpersonal affect stance, alignment stance and investment stance to unpack these emerging cultural meanings of dui.
6 Enregisterment of dui as a tribalising discursive practice

Circulation of enregistered meaning and form through metapragmatic commentaries
Enregistered meaning and form can be circulated through metapragmatic commentaries as well as framing (Johnstone 2016).In our data, we noticed that a wealth of metapragmatic commentary about 'how to dui' are readily available on social media, complete with memes and emojis.Some blogs even illustrate how to dui in languages other than Chinese.A blogger (referred to as UD here), for instance, posted a long piece on Zhihu (知乎), a popular Chinese question-and-answer website comparable to Quora, sharing his experience of how to dui in English on Twitter in March, 2020, during the time when the social media in the West was flooded by speculations about the origins of the COVID-19 virus and hate speech towards China.The piece is written as an introductory guide on how to dui in English with hostile, anti-China 'overseas netizens' (外国网友, literally meaning foreign internet friends) or 'foreigners' (老外, meaning 'old outsider', a colloquial and friendly term for non-Chinese).It starts with a categorisation of three types of dui which are often used together.These are: 'reasoning', 'quarrelling through all means' (swearing or jokes with no logic required) and 'sarcasm'.The blogger does not mention teasing at all.In Example 3, UD talks about the importance of following one's feelings and achieving the overall effect of wrangling over worrying about (grammatical) details.In other words, understanding others' stance and displaying one's own stance are more important than concerns about the technicalities of language.
Example 3 from those occasions when you are engaged in debates with reasons, there is no need to understand what is said by the other party word for word, particularly when you have already started dissing.Most of the time you just rely on your feelings.You may not even understand what is said by the other party, but you can still feel hostility.Similarly, as long as you can achieve the same effect (hostility), that will do.There is no point of poring over grammar.Source: https://zhuanlan.zhihu.com/p/113324142)Example 4 is an instance of UD's own engagement with dui on Twitter in the context of a conflict over the origins of the virus, illustrating how he follows his own advice.The exchange is characterised by the use of derogatory language, nationalistic sentiments and personal attacks.Both parties try to belittle each other's affiliations and opinions.UD attempts to discredit the other twitter user's opinions by insinuating that they are being paid, while the latter responds with an even more serious accusation that UD is a Chinese agent spreading propaganda.Their views are polarised and tribalizing and both parties are determined to defend their nations.There are internet slangs and abbreviations commonly used in online conversations (e.g., LMAO-laughing my ass off, Pfff-expressing dismissal or contempt), insults (morons), and sarcasm (Wait do you get a free test for this?).There are also some loaded cultural references: bat soupa myth circulated in early days of the outbreak of the Coronavirus that bat soup caused the Wuhan Virus; and sharing a mic with Rudy Gobertthe incident in which basketball player Rudy Gobert jokingly touched all the mics in a post-game conference in November, 2020 to joke about the Who is our friend and who is our enemy?
Coronavirus and then tested positive soon after).Both sides are invested in the claims they put forward.Despite their opposing views, they mirror each other in their discursive practices and collaboratively build the momentum of dui.
Example 4 (Note: The highlights are original.Red: the tweet by the blogger himself; blue: the opponent.The last tweet is highlighted in blue in error by the blogger.Source: https://zhuanlan.zhihu.com/p/113324142)

Words of violence in on-line commentaries: Wuhan Diary
Wuhan Diary was written and self-published on social media between January and March 2020.The author, Fang Fang, an established writer, provided a daily account of the life of the people in the city during the Wuhan lock down.Despite the early popularity of the diary (hailed as 'the people's voice'), Fang Fang was criticised as a 'China traitor' on Chinese social media after the news broke that the Diary would be published in English and German by two prestigious publishers (HarperCollins and Hoffmann und Campe Verlag respectively).The multilingual publication was supposedly intended to 'amplify the voice of courage', according to its English translator, Michael Berry (2020).
However, in the context of increasing tensions between China and the West, intensified by the spread of COVID-19, and the rising hostility towards China, the publication quickly became politicised and triggered tribalising discourses and violence on Chinese social media.The credibility of the Diary was questioned, and Fang Fang's motives were scrutinised.Fang Fang defended herself and argued against her vocal critics, including fellow writers, public figures and netizens on social media.These acts of arguing back were frequently referred to as dui on social media.For example, in a question-and-answer platform, Toutiao (头条), a competitor to Zhihu, a web user posted a charged question in May 2020 (Example 5).Starting with a statement of the problem, i.e., 'Fang Fang has been dui'ing netizens up to now'.
In another blog, a blogger provides several examples of how her fellow writers, public figures and netizens have, in turn dui'ed Fang Fang to her defeat.The title of the blog post summarises the blogger's main message, i.e., netizens could not stand Fang Fang's complaints any longer and they kept dui'ing her until she gave in (Example 6).
Example 6 最近作家方方又开始闹腾了， 有网友看不去了， 直接怼得她服 (Recently, the writer Fang Fang began to make trouble again, and some netizens couldn't bear it any longer, and dui'ed her Who is our friend and who is our enemy?directly to give in.Source: https://baijiahao.baidu.com/s?id=1689320612857194823&wfr=spider& for=pc) The blog post provides examples of exchanges of dui between Fang Fang and a netizen under the name of 'Zheng**' in the comment section.Zheng** rants: '你应该庆幸自己还能笑！ 其实你真的该把全家都移民出去， 否则你怎忍心看见自己的家 人因为你就天天生活在唾弃之中呢？ 他们为什么要因为你承担这份耻辱呢？' (You should count yourself lucky that you can still smile.Actually, you should move your whole family overseas.Otherwise, how could you bear watching your family drowning in the hatred and spite every day?Why should they bear this shame because of you?') To this, Fang Fang replies: ll help this IT guy, Zheng xx, forward his comment.Let's trust that 郑 is a real name.From this name, we could tell how small his world is.A frog living at the bottom of the well can only hear its own croaking.Even its imagination suffers.')Zheng's comments are words of violence, both highly charged and very personal, an example of an extreme affective stance, but also containing very little substance.They set out to upset Fang Fang at all costs, an invitation to 'investment' in Kiesling's terms.Fang Fang chooses not to fully engage with Zheng's personal comments, i.e., not to fully invest, but to satirise the netizen's semi-anonymous name.
In the following exchange (Example 7), Zheng** unpicks Fang Fang's arguments, provides counter-arguments point to point, and brings the focus back to the controversial issues he believes are at the heart of the debateloyalty to one's motherland, the credibility of the diary, and the author's motives.He retorts that a) it is his real name (which cannot be true), b) his world is not small, as he chooses justice and conscience; c) even if he lives at the bottom of the well, he loves his country; d) the croaking is actually from Fang Fang, and e) he accepts that his imagination is not as good as Fang Fang, as Fang Fang is very good at making things up.
In her response, Fang Fang acknowledges that Zheng's counter-argument is well composed and tactful, before refuting Zheng's accusation about the credibility of the diary and her identity in her response.Netizen: It is my real name.I have nothing to fear, not like you, like a fly hiding behind a fake name.My network is not small, because I choose justice and conscience.Even if I live at the bottom of a well, I love this land deeply and passionately.I did hear frog croak, but that is from you; my imagination is not as rich as you, as you are good at making things up while I rely on my observation and thinking.@Fang Fang.
Fang Fang: This paragraph is better written than the previous one.We have different values, and what you see and infer will be different.There is no need to fight over these.Time will tell.My novel is made-up, but my diary in the pandemic is not made-up.Since you are open-minded, you can search (and find the evidence).I hope that you can do what you say: choose justice and conscience, and think independently.By the way, my name is a pen name, not a fake one.My real name is available on the website.
In these exchanges described as dui, we see words of violence and strong expressions of emotion in the users' stance-taking.The publication of the diary in other languages Who is our friend and who is our enemy?had been politicised amongst the political tension.It les to hardening of stance and rising nationalism on the part of many internet users against the 'enemy'the 'Western Powers' (西方列强国家), thereby redrawing the boundaries between friends and enemies according to values and worldviews that are conflated with nationalism.The 'anti-diary side' in these exchanges is taking an extreme affective stance, trying to force Fang Fang to invest her time and energy in responding to their attacks.Fang Fang chooses not to engage directly with the tribalizing discourse (which could also be seen as a subtle form of dui from her side) appearing to ignore her attackers, which seems to aggravate some attackers even more.She does nevertheless respond to the less personal attacks in a more matter-of-fact kind of way, sustaining the ritualistic performance of dui with the opposition in the public domain.

A discursive strategy of conversationalisation in story-telling
The tendency to adopt colloquial language features in media and other public discourse is described as conversationalisation, a phenomenon that restructures the boundary between public and private orders of discourse (Fairclough 1994).In our analysis, we noticed that dui appears frequently in COVID-19 related news headlines on blog sites.It has become an 'enregistered emblem' (Agha 2007: 235) of a conversationalised journalistic storytelling style in which two parties refute each other, and direct speech quotes are used to support the opposing stance.
In June 2021 various countries rolled out vaccination programmes, and there was a considerable amount of negative press about China's vaccination efforts in the West.The WHO's visit to China at that time also reignited speculation about the origins of the COVID-19 virus.As part of our data gathering, we have collated a list of COVID-19 related news items from Chinese online news sources with dui in the headlines during this time (Table 1).All the parties involved in these headlines apart from one (No.13) are pro-Chinese speakers who are portrayed as 'successfully dui'ing' foreign governments, media or officials.
First, we noticed that dui is used in the headlines to key who are involved in opposition, and how the opposition plays out (e.g., 'reciprocally' in the Headline No. 3 or 'angrily' in No. 5 in Table 1).Three headlines (Nos 2, 5, and 11) also contain direct conversational quotes, further conversationalising the story-telling.For example, in the second headline, WHO experts were quoted as saying 'You are so shameless!'.In another example (No. 5), the Chinese side dui'ed: the US side by saying 'This is not a playground where you can come and leave when you want to!' Direct quotations also appeared in other forms, such as tweets, which serve not just to prove the authenticity of the story, but also to provide examples of the practice of 'dui'ing'.Take as an example the second news story published on sina.com.cn with the headline: 'WHO experts visiting Wuhan dui New York Times angrily'.It reports that a  Daszak's stance was echoed by another panel member, Thea Fischer, who copied his style of writing in her tweet.Her message was translated into Chinese in the news report and accompanied by a screenshot of the tweet, presumably to convey the authenticity of the message.Daszak then reinforced his stance with another tweet which finishes with the phrase, 'shame on you@nytimes'.This rally of stance-taking on both a local level (claim, counter claim, another counter claim and conclusion) and the macro level of stance between China and the Western powers is brought together through dui as a discursive strategy of conversationalisation in storytelling.

Conclusion
This article examines the enregisterment of tribalising discourse on Chinese social media during the key events in the COVID-19 pandemic, using the analytical notions of enregisterment and stance-taking.Over the course of the pandemic, dui (怼) has expanded from a discursive practice of teasing/subtle dressing down (the nature of the practice which made it popular in the first place) to a discursive practice of tribalism, a way of drawing boundaries between 'friends' and 'enemies'.It has become associated with digital words of violence and a discursive strategy of conversationalised journalistic storytelling, characterised by sarcasm, hostility, and collaborative momentum building on the part of both parties.As demonstrated in our analysis, this new tribalising discursive practice differs significantly from its original practice in its engagement with affect and investment stance.While 'dui as teasing' works on solidarity and group bonds, 'dui as words of violence' brings hostility out into the open.By conversationalising story telling in news reports, dui creates space for speakers' own words/voices, thus lending credibility to its claims and counter-claims.These new meanings and forms of dui are circulated through metapragmatic commentaries and a tacit agreement on both parties to work together to sustain the ritualistic, combative performance of dui interactionally, no matter how epistemically opposed they may be.The enregisterment of dui as a tribalising discursive practice is triggered by a number of socio-cultural factors.One of them is the politicisation of health practices and beliefs.The COVID-19 pandemic has heightened Chinese netizens' awareness of and sensitivity towards friends and enemies domestically and externally.Together with the associated narratives of national pride and national humiliation dominant in the country's foreign policy (Shen and Guo 2013), these sentiments have destabilised and polarised Chinese social, digital networks.Dui provides a direct means of conflict or disagreement, allowing parties to oppose each other in the 'automatic, ritualised, knee-jerk' fashion described by Tannen (2013) in her critiques of the argument culture.In understanding dui as the cultural act of intensified opposition, the study contributes to the demystification of the stereotypical view of the face-loving and consensus-seeking Chinese culture (e.g., Chen and Starosta 1997;Huang and Zhao 2021;Shi and Fu 2015) and points to digital culture as a new site of investigation where direct acts of disagreement and different participatory dynamics operate.
Our study also provides new insights into the role of language in managing relationships in the context of an increasingly ossified dichotomy of friend versus enemy.Tribalising discourse, while motivated by tribal epistemologies, has expediated the hardening of division between friends and enemies.Dui can be a high-stakes practice with serious consequences for the group dynamics of those involved, who are, as they converse, both shaping and being shaped by social structures.Thus, the study demonstrates the mutually constitutive nature of discursive practices and social relationships and the importance of bringing together discourse analytical and socio-cultural perspectives in understanding tribalising discourse.

Example 7 (
Source: https://baijiahao.baidu.com/s?id=1689320612857194823&wfr=spider&for=pc)Translation: number of WHO experts named and shamed the New York Times for their deliberately misleading reports on China's reporting data.The news story starts with the main message of the article published in the New York Times on 12 June, 2021, which claimed that the WHO experts visiting Wuhan complained about China's unwillingness to share the virus data.To refute this claim, the article provided a screenshot of a tweet (Example 8) from Peter Daszak, a panel member of the WHO and the president of EcoHealth Alliance.The tweet says unequivocally that the claim was not in line with his experience while he was on the WHO mission.The tweet was translated into Chinese in the news report (Example 9).Daszak's tweet in English demonstrates a strong investment stance with the key words (NOT, DID) capitalised.The intensity is mirrored in the Chinese translation through the use of adjective 的确 (meaning indeed).Example 8 (Source: http://news.sina.com.cn/o/2021-02-15/doc-ikftpnny6885394.shtml)Example 9 '这不是我在世卫组织任务中的经历。 作为动物及环境工作组的负责人， 我感到中国同行 们值得信任， 感受到他们的坦率。 我们的确获得了关键的新数据， 也增加了有关病毒传 播途径的了解。' (See Example 8 for the tweet in the original English language, emphasis ours; Source: http://news.sina.com.cn/o/2021-02-15/doc-ikftpnny6885394.shtml)

Table  :
List of COVID- related news items with dui in headlines.