“Kaigewanxiao” (‘Just kidding’): the use of humor markers in Chinese media discourse

2.1 Metapragmatics and metapragmatic makers

The notion of metapragmatics, introduced by Silverstein (1976), has been defined in multiple ways (e.g., Bublitz and Hübler 2007; Caffi 2007). The definition by Caffi (2007) is an often-cited one, which characterizes metapragmatics as “the management of discourse, based on common knowledge and reflexivity” (p. 83). Following this definition, the current study discusses metapragmatics as the management of language devices to achieve certain interactional goals, which demonstrates a speaker’s reflexive awareness in interaction.

Caffi (1994) delineates three senses of metapragmatics. The first sense is concerned with the theoretical debate on pragmatics as a discipline; the second sense pertains to the reflections about pragmatic conditions; and the third sense focuses on a speaker’s competence that is reflective of their judgments of appropriateness of their own and other people’s communicative behaviors. Our study falls within the third sense of metapragmatics, which addresses a speaker’s competence in the process of assessing and managing the appropriateness of their language use.

Specifically, studies concerning the third sense of metapragmatics have progressed in two general directions (Parvaresh and Tavangar 2010). The first direction involves holistic explorations of metapragmatic markers across different interactional contexts (e.g., Culpeper and Haugh 2014). The second direction focuses on the in-depth functional analysis of specific types of metapragmatic markers or structures in particular settings (e.g., Liu 2022; Ran 2015). Our study aligns with the second direction, aiming to examine the interactional functions of the humor marker kaigewanxiao in Chinese media discourse.

2.2 Humor markers

In humor research, it has already been shown that conversational humor plays an important role in relationship management (Attardo 1994). Along this vein, a number of studies have examined the interactional functions of specific HMs across diverse settings (Haddington 2011; Senft 2018; Skalicky et al. 2015). More specifically, based on a corpus of face-to-face and telephone conversations in Finnish, Haddington (2011) investigates a reduplicate linguistic item called the ‘joke–joke’ structure. The author shows that speakers use this structure as a metacomment: (1) to withdraw from an immediately prior stance; and (2) to “disclaim a stance made in the first turn after a recipient’s repair initiation” (p. 158).

Examining five corpora covering a variety of settings, Skalicky et al. (2015) capture the functions of “just kidding” in American English as a resource for rapport management. They identify four major functions of the phrase: (1) “inoculation”, the pre-emptive use to inoculate the speaker against problematic reactions, typically following statements that are potentially offensive and preceding utterances provided by any co-participants; (2) “repair of failed humor”, occurring after an intervening turn, during which a speaker’s attempt at humor has received an undesired reaction; (3) “bid to return to serious”, which is a return to a serious frame subsequent to a non-serious statement; and (4) “setting up a new joke”, a function “subverting interlocutors’ expectations and extending the previous joke” (p. 23). Moreover, Senft’s (2018) research in the traditional Trobriand community shows that in their language Kilivila, if hearers signal that they may be offended by a certain speech act, speakers can recede from what they have said by labelling it as sopa wala (‘just joke’), thereby regulating the reactive behavior of an addressee. These studies suggest that the interactional work accomplished by using humor markers “goes beyond simply repairing misunderstood or unappreciated humour” (Haugh 2016: 121).

2.3 Humor markers in Chinese

In Mandarin Chinese, a general formulaic pattern for HMs is the predicate-object construction “V + (classifier) + wanxiao(hua)”, which does not require an explicit subject like “I” (Li and Yan 2010). The core morpheme, wanxiao (‘joke’), is a noun, with the optional suffix -hua (‘speech/talk’). The common verbs (kai, shuo, and jiang) all convey a meaning of ‘speak’ or ‘say’, while the optional classifiers ge or ju may be deployed. Notably, ge can serve a subjective diminutive role, minimizing the pragmatic force of the utterance (Ren 2013). In terms of sequential placement, these HMs typically occur immediately before or after – but rarely within – the marked humorous utterances (MHUs),^[1] and are consistently followed by a clear prosodic pause in speech (Li and Yan 2010).

Previous studies on Chinese HMs like kaigewanxiao have primarily explored their general pragmatic and discourse functions using broad-coverage Chinese corpora encompassing diverse genres (e.g., Li 2012; Li and Yan 2010). The observed functions mainly include pre- or post-directing recipients’ understanding of the marked discourse, avoiding misunderstanding, enhancing the acceptability of potentially inappropriate utterances, and foregrounding a speaker’s humorous stance. However, their findings require refinement, as they do not systematically engage with the sequential positions and interactional functions of HMs, nor do they explain their role in relationship management within institutional settings. To address these gaps, this study examines the sequential positions, interactional functions and rapport managing function of the HM kaigewanxiao in Chinese media discourse.

3.1 Data collection

Our data were collected from the Chinese media database, Media Language Corpus (MLC),^[2] developed by the National Broadcast Media Language Resources Monitoring and Research Center of the Communication University of China.

With reference to Kádár and House (2020) where they discuss how pragmatically important terms can be studied in a bottom-up way in corpora, we first need to identify the HM that we intend to examine in this study. We selected the targeted HM for analysis based on frequency, a criterion verified by panels of native speakers of the languages under investigation (see to Culpeper 2010 for further information on the connection between conventionalization and frequency). Three native speakers of Mandarin Chinese were involved in the identification of HMs, specifically “to identify expressions with the closest meaning and pragmatic use” (Kádár and House 2020: 147). One can refer to a wide range of the possible expressions of HMs, such as: (1) kai wanxiao; (2) kai ge wanxiao; (3) kai ju wanxiao(hua); (4) shuo/jiang ju/ge wanxiao(hua); etc. To find the most frequently and widely used HM, we asked our three informants to list variations of HMs with varying styles and syntactic complexities. All our informants referred to kai ge wanxiao and kai wanxiao. A preliminary review of our corpus, however, indicated that kai ge wanxiao exhibits greater flexibility in its sequential positioning, and it was therefore chosen as the HM for analysis.

Then, following the sampling procedure of Kádár and House (2020), we collected an initial dataset of 200 examples featuring the identified HM through collecting 200 hits by randomly sampling batches of 10 examples in our corpus. Each instance was examined in its source context to ensure that it functioned as a metapragmatic HM. Invalid cases (e.g., propositional uses of kaigewanxiao as in Example 1) were excluded, and immediately replaced “with valid ones that occurred before the batches of 10 examples in the corpus” (p. 148).

3.2 Data analysis procedure

In line with our research questions, we carry out the data analysis relying on three levels:

1. Identifying the sequential positions of the HM. In order to grasp the sequential positions of the HM, we adopt the typology of positions of metapragmatic comments (Tanskanen 2007). Tanskanen (2007) has proposed three positions where metapragmatic comments occur: prospective, mid-message, and retrospective. Due to the high flexibility of the retrospective position, its two sub-types are discriminated based on Schegloff (2016), including next-beat^[3] (within the beat following the prior possible completion of talk) and post-other-talk (after some talk by another speaker), so as to attend to a finer level of granularity. This step provides a basis for the investigation of the interactional functions of the HM.

2. Identifying the pragmatic functions of the HM at the relevant sequential positions, along with the expanded context. During this procedure, we take into account inferences accessible to recipients which are licensed by the sequential turn designs (Schegloff 2007) and make reference to recipients’ responses (if any) to speakers’ utterances involving the HM.

3. Using the concepts of Rapport Management (Spencer-Oatey 2000), we analyze why the MHUs put rapport at stake and how the use of the HM dynamically engages in rapport management, such as the management of face wants, sociality rights and interactional goals in the ongoing interactions. The use of HM addresses “quality face” when showing a speaker’s concern for their own or others’ qualities, such as competence, abilities and appearance, and “identity face” when demonstrating a speaker’s awareness for their own or others’ “social identities or roles, e.g., as group leader, valued customer, close friend” (Spencer-Oatey 2002: 540). Additionally, it deals with interactional goals when it involves the management of “the specific task and/or relational goals that people may have when they interact with each other” (Spencer-Oatey 2008: 14). During this qualitative process, we consider the basic information (e.g., identity, relationship, age, etc.) and the immediate responses (if any) of participants to ensure the HMs are not examined in isolation.

The data were annotated and coded by the four authors independently. On each level of coding, the authors were provided with sets of options, including the typology of positions of metapragmatic expressions, the taxonomy of functions proposed by Skalicky et al. (2015), and the notions from the framework of Rapport Management (Spencer-Oatey 2000). The individual coding results were discussed until agreement was reached. It should be mentioned that because our dataset consisted of text data, where interactional subtleties such as paralinguistic traits are rarely made apparent, we did not follow conversation analytic conventions during the analytical process but focused primarily on contextualization cues and language suggestive of humor or attempts at humor. Moreover, in order to be consistent and discuss each instance conveniently, we have formatted the extracts in the analysis section to include turn numbers, add emphasis (bold type and italics) to the HMs, and underline the MHUs.

4.1 Prefiguration of an upcoming stance

The HM kaigewanxiao at prospective position serves as a prefiguration of an upcoming stance. It is employed when the speaker presumes his/her upcoming utterances might cause negative emotions in the target recipient. By predefining such potentially problematic events in a humorous manner, the HM reduces their salience as an interactional cue (Hewitt and Stokes 1975). Instances with this function are coded from MLC when a speaker states kaigewanxiao prior to any potentially playful or non-serious statement that is “offensive or otherwise rife with the potential for misinterpretation” (Skalicky et al. 2015: 23). Consider example (2), in which the HM kaigewanxiao prefigures an upcoming statement that is potentially offensive.

In example (2), the guest Huang begins by describing the Iceland dishes, which are likely to be perceived as unusual and illegitimate by most Chinese people (01). Then, the host Luyu explicitly questions the legitimacy of consuming such food (02). In response, Huang justifies this practice by referencing the whaling traditions of Iceland and other nations. Before offering the potentially offensive analogy that “they were Viking pirates”,^[4] he pre-emptively frames it with kaigewanxiao (03). This use of the HM at the prospective position serves as an overt signal for the audience to interpret the upcoming statement not as a factual accusation but as a joke, thereby prefiguring a non-serious stance.

As a public figure, Huang would be aware of the risks of calling Icelanders Viking pirates, especially in a public interview given its wide-reaching audience. Icelanders here are the targeted overhearers who are not present but are the target of Huang’s MHU. Without the HM kaigewanxiao to prefigure his upcoming utterance as non-serious, his stance could threaten Icelanders’ group identity face (Spencer-Oatey 2005) by associating them with piracy rather than a legitimate profession. Therefore, the HM at prospective position reflects his metapragmatic awareness of rapport management with the targeted overhearers.

4.2 Adjustment

In our corpus, when the HM is put at the mid-message position, it serves to realize the adjustment function (Levelt 1983), which indicates a speaker’s awareness of the inappropriateness or risk of his/her ongoing utterance and his/her efforts to make adjustment midway through the utterance. Instances with this function are coded when the HM kaigewanxiao occurs midway through a non-serious statement. Consider example (3).

In example (3), as the commentator Yang launches a criticism (01), he suspends his turn after ‘I think this prediction’ to insert the HM kaigewanxiao before producing the evaluative punchline (‘is very irresponsible’). Subsequently, he provides a justification for his criticism, that is, almost all people over the age of 30 might not live till 2084 to see the prediction realized. The host Shui Junyi’s response (02), ‘but we must have confidence’, shows that he aligns with his comment, while orienting toward a positive communal outlook.

In this example, the deployment of the HM at the mid-message position is indicative of Yang’s online monitoring of his own utterance. It serves as an overt signal to the host and the audience to interpret his ongoing criticism not as a direct attack, but as a non-serious remark, constituting a punctual adjustment upon perceiving the potential face-threat. Had kaigewanxiao not been employed, the comment could have more directly threatened the quality face of the IMD World Competitiveness Center – the targeted overhearers – by bluntly criticizing the irresponsibility of the prediction made by this authoritative international organization. Taken together, this use of the HM exemplifies a form of metapragmatic awareness of rapport management with the targeted overhearers.

4.3 Retraction from an immediately prior stance

The HM in this section occurs at the next-beat position, one subtype of the retrospective position, serving to retract from a speaker’s immediately prior stance (Haddington 2011). In our data, when speakers anticipate that their prior stance might be seen as offensive or inappropriate, they often use HMs to retract from the prior stance by clarifying their non-serious intent. We code instances when a speaker deploys the HM immediately after a non-serious statement and also before any responses from the interlocutors (Skalicky et al. 2015). Example (4) is a telling case, in which the HM kaigewanxiao serves to retract from an immediately prior stance.

This example begins with Vice President Chen explaining that it is part of their corporate culture to be on first-name terms with each other instead of using a more formal title (01). Gu shows affiliation by saying that it sounds a little distant to call Chen Zong (02).^[5] Chen affirms that nobody addresses him like that in the company (03), evoking Wang’s curiosity about whether the female colleagues also use his first name (04). Following Chen’s reassurance (05) and the host Gu’s prompting (06), Wang addresses the vice president by his first name with a teasing offer of coffee (07). This tease, however, is immediately followed by kaigewanxiao at the next-beat position; this HM retracts the prior stance, frames it explicitly as non-serious, and thereby leaves no chance for a response from Chen.

If kaigewanxiao were not added by Wang after her tease (07), her utterance could threaten Chen’s identity face (Spencer-Oatey 2000). Specifically, it could imply disrespect for his status, suggesting she feels entitled to use his first name and tease him without being a company insider. By deploying the HM to retract from her prior tease, Wang demonstrates concern for Chen’s identity face as a vice president. Therefore, the use of kaigewanxiao reflects the speaker’s awareness in maintaining rapport with Chen, who is both the addressee and the target of the tease.

4.4 Repair of a failed humor

In the previous two sections, the HMs are used in the same turn as the trouble source before getting any response from the co-participants. In this section, the HM is situated at post-other-talk position, another subtype of retrospective position. It is uttered after a speaker’s attempt at humor fails or a co-participant shows misunderstanding of the speaker’s non-serious utterance. In this way, the HM is employed at post-other-talk position to repair the failed humor (Skalicky et al. 2015) or set aside the co-participant’s misunderstanding (Haddington 2011; Schegloff 1992). Example (5) is a case in point.

In line 01, the guest Zhao exaggerates his symptom of food impaction, implying that it is very easy for food to get trapped between his teeth. The host Shengbo further highlights Zhao’s symptom of food impaction by dubbing him Super ‘Saiyan’ (02), which evokes Zhao’s affiliative response (03). ‘Saiyan’ here is an appellation from a Japanese anime that shares a similar pronunciation with the word saiya (‘food impaction’) in Chinese. The host Jiajia then shifts the focus to Zhao’s long pinky nail (04). This remark is treated as a breach of taste and propriety by both Zhao and Shengbo (05, 06), and is facetiously amplified by Shengbo’s vivid imagery (06) and Fang’s mock praise (07). Zhao then declares that their program will be complained about for undermining the image of the guests (08). Faced with these disaffiliative responses, Jiajia produces the HM kaigewanxiao, prefaced by two negations bushi (‘no’), to repair her prior utterance by recontextualizing it as non-serious, and subsequently shifts to a serious frame to account for her prior statement (09).

Jiajia’s use of kaigewanxiao thus serves a dual function: repairing the failed humor and shifting from a non-serious to a serious frame (see Section 4.4.2). If she did not clarify her non-serious intent through the HM kaigewanxiao and not turn to the account in a serious frame subsequently, her initial tease (04) risked casting Zhao as slovenly, severely damaging his identity face in front of the audience. Taken together, the use of the HM kaigewanxiao at post-other-talk position demonstrates the speaker’s metapragmatic awareness to save or remedy the rapport with the interlocutor Zhao by publicly repairing a failed humor.

4.5 Shift from non-serious to serious

In our corpus, the use of kaigewanxiao at the next-beat or post-other-talk position, when followed by statements in a serious frame, serves dual functions: it not only retracts an immediately prior stance or does repair work but also marks a shift from a non-serious to a serious frame (Schegloff 2001). This shift steers the discourse back to a serious footing, a function also observed in the English HM just kidding (Skalicky et al. 2015). We code instances with this function when an interlocutor states kaigewanxiao to retract from an immediately prior stance or repair a failed humor and then shifts to a serious frame. As illustrated in example (6), the HM at next-beat position serves as both a retraction from an immediately prior stance and a shift from a non-serious to a serious frame.

At the beginning of example (6), the host asks Bai for his overall opinion on the reason for the outflow of the consumption of luxury goods (01). In response (02), Bai first criticizes some people for getting extravagant goods before getting rich (which is omitted). Then he turns to ascribe the reason of consumption outflow to the high domestic taxes on luxury goods. To indirectly satirize this phenomenon, he uses hyperbole: there are so many Chinese buying luxury goods in America that all Americans could understand the three Chinese sentences which demonstrate Chinese people’s preference for buying cheap luxury goods in America. However, he immediately deploys the HM kaigewanxiao to clarify that he is being non-serious.

The deployment of the HM at the next-beat position here serves a dual function. First, it retracts the prior sarcastic stance. If kaigewanxiao were not used, the utterance might be seen as an attack on the overhearing audience who purchase luxury goods abroad and thus threaten their quality face (i.e., being economical and patriotic) and their equity rights (i.e., to be autonomous). Second, the HM shifts the discourse back to a serious frame, where Bai offers a policy suggestion on reducing the domestic tax on luxury goods. In this way, Bai attends to the interactional goal of their program (e.g., to make an impersonal comment on the issue of Chinese consumption), which is a key component of rapport management. Thus, by using the HM to shift from a non-serious to a serious frame in the same turn, Bai ensures that the rapport with the overhearing audience is preserved.

4.6 Setting up a new joke

In the previous section, we have demonstrated how the HM at the next-beat position serves to shift from a non-serious to a serious frame. In this section, we show how the HM at the same position functions to subvert recipients’ expectations and extend the previous joke or to set up a new joke. Example (7) is a telling case.

This example begins with the host asking if Guo and Yu are both ‘big bosses’ (01). Guo initially seems to confirm this as he says, ‘You’ve realized it’ (02). However, he follows his utterance with the HM kaiwanxiao, ostensibly to avoid sounding arrogant. Here, Guo’s use of the HM at the next-beat position aligns with the function to retract from an immediately prior stance; it becomes unnecessary for the recipients to take his words seriously after knowing he is ‘just kidding’. However, Guo then subverts this expectation by launching a new, enhanced tease – contrasting himself as a ‘small boss’ with the ‘big boss’ Yu Qian, from whom others can borrow money.

As already suggested, a recipient may expect the HM to signal a retraction of a risky joke. Instead, the usage in this excerpt flouts the standard retraction function; “it tricks the hearer into expecting a downgrade but works to enhance the original quality of the humor” (Skalicky et al. 2015). Through the lens of rapport management, it can be assumed that Guo’s recipients initially interpret his application of the HM as driven by his face sensitivity (in this case, to avoid being labeled as arrogant by the co-participants and the audience). Yet, Guo uses the HM kaiwanxiao to set up a new joke and to playfully manage rapport with the co-participants and the audience by deploying unexpected humor.