Passing on the “whistle”: Users’ creative engagement with platform censorship in an online relay campaign in China

Abstract

The advances and proliferation of social media technologies have not only empowered Chinese users with more opportunities for opinion expression and public participation, but also provided the censors with increasingly sophisticated means to monitor and control the public discourse. In this game of cat and mouse, Chinese censors and users are intertwined with each other as new forms of censorship and counter-censorship stimulate both sides to further develop tactics to compete with each other. Drawing on (social) media discourse analysis, this study examines the strategies users adopted in an online relay campaign to keep a censored article alive on WeChat Public Accounts. The analysis highlights three major forms of creative strategies including textual reproductions, multi-semiotic and multimodal reproductions, and technically encoded forms. It also underscores platform censorship as a particular level of censorship where the platform takes responsibility for removing sensitive or perceived sensitive posts and content. This study adds to the growing literature that examines creative strategies of censorship engagement and a more nuanced understanding of the multi-layered online censorship mechanism in China.

Keywords

creative engagement discursive strategies (social) media discourse analysis platform censorship whistle

Introduction

The proliferation of social media has expanded the channels for expressing opinions and political participation among Chinese people. Among them, the super-app WeChat and its associated functions such as WeChat Public Accounts (微信公众号) and Moments (朋友圈), microblogging services such as Sina Weibo (新浪微博), short-video sharing platforms such as Douyin (抖音), and discussion forums such as Tianya (天涯) and Zhihu (知乎), have constituted an important part of a virtual moderated public space (Gao, 2022; Gleiss, 2015; Wu, 2018; Wu and Montgomery, 2020; Yang, 2009). However, the positive effects brought by Chinese internet and social media technologies have been complicated by China’s political and social environment where censorship plays a significant role in constructing and shaping the public discourse, especially during important political and crisis events (Sun and Zhao, 2021; Wu and Fitzgerald, 2023). Online censorship in China has been entwined with the internet since its arrival and has developed into a multi-layered sophisticated apparatus employing both technological infrastructure and large amounts of human monitoring (Yang, 2016; Ye and Zhao, 2022). In light of this extensive censorship regime, arguably the most sophisticated censorship in the world, Chinese users develop techniques to avoid, challenge, circumvent and test the censorship systems by mobilizing a combination of technological, cultural, and language resources (Lee, 2016; Mina, 2014; Yang, 2009, 2016).

As the censorship systems have evolved and adapted to new forms of technology, a more sophisticated understanding of censorship and counter-censorship in China has emerged. Recent studies have moved away from conceptualizing it within a simple dichotomy of control and resistance or state repression and public resistance to explore the nuances and dynamics of an evolving censorship regime and users’ creative engagement with it (Han, 2015; Lee, 2016; Sun and Zhao, 2021; Yang, 2016; Ye and Zhao, 2022). However, while previous research has noted the complex layers and creativity in the interplay between users and censors, less attention has been paid to the role of “platform censorship’’ as a form of “delegated censorship” (Sun and Zhao, 2021). Within this layer of the censorship system, the platform is responsible for the content on the platform, rather than the state, and so the platform itself takes on the responsibility for censoring by removing sensitive content published on the platform. This then creates a peculiar ecology of platform censorship where the decisions made to remove content may be made for content posted on one platform but not necessarily on the others. At this layer of censorship, the way users engage with platform censors is shaped by the technological environment of the platform. That is, the forms of resistance, avoidance or simply delaying a post being identified as censorable and removed, are contained within the technological affordances of the platform.

This paper examines this form of contained censorship and engagement through the case of a published article that was briefly censored by the WeChat platform during which users mobilized a range of creative resources to keep the article alive on the platform. The case is interesting as similar content was still available on other platforms like Sina Blogging (新浪博客). Therefore, it was not a matter of the content of the censored article, but rather the decision made by the platform to engage in censoring actions. This led to a game of cat and mouse where WeChat users aimed to prolong the presence of the article or its representations before they were taken down. This case highlights both the role of “platform censorship” as a distinct component of the multi-leveled Chinese censorship system and the ingenuity of users in posting increasingly creative representations of the article drawing on technological, multi-semiotic, and cultural resources.

Chinese online censorship and users’ engagement with it

China is well-known for its sophisticated censorship regime, also referred to as “The Great Firewall” which filters “undesirable” information from outside China and blocks foreign websites such as Facebook, Twitter, and Google. Within the firewall, online censorship is executed mainly through technological forms and human power (King et al., 2013). While technological information control largely relies on software and algorithms to help identify and remove “sensitive” or prohibited content (Cobbe, 2021), human power refers to voluntary user reporting and employed human censors including internet police, internet monitors, and the “50-cent army” (五毛党) as human censors at various government levels but also delegated to media and technology companies and institutions to examine and delete content based on their given guidelines and regulations (Zhu and Fu, 2020). Chinese online censorship is complex, multilayered, and constantly evolving with three key players: government or state censorship, company or platform censorship, and self-censorship (Sun and Zhao, 2021). According to Sun and Zhao (2021), state censorship involves governmental sections such as The Ministry of Public Security (公安部) and The Central Cyberspace Affairs Commission (中央网络安全和信息化委员会) that monitor the whole internet at one level and then provincial, municipal, and local branches that help with the surveillance and management. As part of the delegated censorship framework, company or platform censorship is where the internet companies or social media platforms themselves accept the responsibilities delegated by state censorship and take charge of blocking or removing content published on their platforms. Lastly, self-censorship refers to the practice where users purposefully do not post or change, adapt or adjust their posts in order to avoid censorship. These three main systems constitute a layered texture of online censorship and responsibility, with state censorship being the hub of the system, company and platform censorship doing the actual work of moderating and removing content, and self-censorship as reactive practices to avoid trouble (Sun and Zhao, 2021).

Previous research on Chinese online censorship has explored the ways censorship works, the targets of censorship, and the strategies developed by users to circumvent censorship (Hassid, 2020; King et al., 2014; Yang, 2016; Zhang et al., 2022). With a better understanding of the ecology of Chinese censorship, an emerging research strand focuses on the layered and interconnected components of the system including the role of local sectors, private actors, and social media platforms (Han, 2015; Luo and Li, 2022; Sun and Zhao, 2021; Tai and Fu, 2020; Ye and Zhao, 2022). For example, Sun and Zhao (2021) examined how platform censorship worked in China through an experiment of running a digital political publication on WeChat Public Accounts and conducting interviews with Chinese officials. They argued that Chinese online censorship is a form of delegated censorship that outsources the responsibilities to internet companies and platforms and that delegated censorship varies across different platforms, under different circumstances, at different times, to different people and players, and employs different methods. Adding a further dimension to the picture of delegated censorship, Luo and Li (2022) proposed the notion of participatory censorship where censorship becomes collective work joined by social media users and organizations through accusatory reporting, a rewarded practice of reporting sensitive or “inappropriate” posts. The study highlighted how community members interpreted and navigated the content posted to the platform based on their interpretation of the censorship policies and participated in reporting posts that they felt should be removed.

Another prominent research strand explores users’ creative engagement with the evolving Chinese online censorship and the strategies developed by users to circumvent it (Xu and Feng, 2015; Yang, 2016; Ye and Zhao, 2022; Zhang et al., 2022). Users’ tactics and strategies in coping with online censorship somehow resonate with the notion of “digital hidden transcripts” (Yang, 2009: 60) which refers to different digital forms of power resistance. Yang (2009) argued that there were four main forms of digital hidden transcripts, namely technical means, online guerilla war, organizational creativity, and linguistic. Technical means uses software or applications to evade the firewall or filtering functions; online guerilla war means when a website or forum is shut down, users quickly start a new one; linguistic refers to symbolic devices users developed to replace censorable content to circumvent filtering; organizational creativity refers to hidden forms of online activities such as online secret meetings for activist events. Recent studies have further expanded the repertoire of users’ techniques and strategies to deal with online censorship. For example, Wu and Fitzgerald (2021a) drew upon the crisis event of the 2015 Tianjin explosions and examined the discursive techniques employed by users to misdirect censors – such as quotation, allusion, and irony – to express criticism of the local government’s handling of the explosions and corruption that made the explosions worse. The analysis showed how Chinese online censorship is a constantly evolving interplay between users, technology and cultural forms, as well as between different layers of government (state and local governments) and human censors. Through the lens of Sensitive Word Culture, Ye and Zhao (2022) explored users’ perception of online censorship, and their discourse strategies – “evading detection” and “expanding interpretability” – to circumvent censorship. Broadly termed as recoding, evading detection deploys images, emojis, special symbols, or splitting words to evade algorithmic detection while expanding interpretability uses metaphors, puns, or homophones to create ambiguous meanings of the sensitive words. Interestingly, this study challenged some established understanding of users’ negative perceptions of Chinese online censorship by showing that rather than seeing online censorship simply as government repression, many respondents viewed internet censorship as protection and subtle negotiation.

These studies have provided a richer understanding of how online censorship works in China’s digital space and how users react to and engage with it. However, among this emerging scholarship, a less explored level of censorship is social media platform censorship where the creative engagement between social media users of the platform and the censorship policies and mechanisms employed by the platform plays out. On the one hand, users engage with platform censorship in a particular way that employs the technical affordances of the platform as well as engaging with not only the mechanical but also with the human censors. On the other hand, the interplay between platform users and platform censorship gives rise to a carnivalesque and creative environment where both parties develop innovative means to compete with each other. It is within this context that this study explores the multi-staged and dynamic interplay between censors and users through a case study of an online “relay” campaign over a news article published during the early outbreak of COVID-19 in China which was briefly censored on the WeChat platform. This study draws on a (social) media discourse approach (Fitzgerald et al., 2022; Montgomery, 2007; Wu and Fitzgerald, 2021b; Wu and Montgomery, 2023) to explore the technological, cultural, and multi-semiotic and multimodal resources users employed to keep the article alive, reposting the article in various forms as users engaged in a cat-and-mouse game with the platform censors.

Data and methods

WeChat Public Account, also known as WeChat Official Account (微信公众号), is a function launched by WeChat in 2012 that allows WeChat users to create, publish, and share content in verbal, audio, and visual forms, to follow or subscribe to others’ accounts, and to engage with their audience. Public accounts come in two main forms, namely service accounts and subscription accounts. While service accounts are service oriented that provide adjunct customer services such as banking and network services, subscription accounts focus on content creating and sharing – for instance, news organizations or individual users can set up their accounts and push their content in words, audio, pictures, and videos to their subscribers and followers. For example, Figure 1 below shows the interface of the public account of People that published the controversial article.

Figure 1.

A screenshot of the public account of People.

The boxes beside the interface (added by the author) briefly describe the basic features of a subscription account, including the title, the publishing house, its forms of content, and its recent publications. This study focuses on the form of subscription accounts where users created and published different translations and forms of reproductions of the censored article entitled “发哨子的人” (The person who handed out the whistle).

Data

The data consist of a magazine article that was briefly censored on WeChat along with over 50 different versions and transformations of the article that were posted on WeChat before the article was uncensored. The original article entitled “发哨子的人” (the person who handed out the whistle) was published by People, a Chinese magazine run by a state-sanctioned publisher, in its March issue “Wuhan Doctors” on its WeChat Public Account on March 10, 2020. The article was based on an interview with Dr. Ai Fen, in which she recalled how she discovered the coronavirus and tried to alert her colleagues and friends but was harshly censured by the leaders and director of the hospital for spreading “rumors” and causing public panic. Within 4 hours, the article received over 100 thousand “reads.” It was removed quickly partly due to its political sensitivity. In response to this, Chinese social media users initiated and participated in an online relay campaign of the article, also recoded as the “whistle,” by translating and transforming it into different textual, semiotic, and technical forms posted on WeChat public accounts. Interestingly, while the original article and its various versions were removed from WeChat, the article was reposted to other platforms and websites and remained accessible, indicating that the censoring was a platform action rather than a governmental one. The censoring on WeChat was also brief as the article posted and censored on WeChat on March 10 became available again the next day, March 11, 2020. This event was chosen for examination for two reasons. First, it illustrates a further dimension of the intertwined interplay between censorship and users by highlighting the role of platforms in the censorship mechanism. Secondly, this event provides a particularly rich example of evolving creativity in mobilizing both technology and multi-semiotic resources in the cat-and-mouse game of Chinese social media censorship and creating an ephemeral but spectacular show.

To collect the data, the author searched for different translations and transformations of the article through two steps. Firstly, the author searched the keywords “吹哨子的人” (whistleblowers) and “发哨子的人” (The person who handed out the whistle) in the search box of WeChat and then sorted the results by selecting the type of “文章” (articles) because relevant data are in the form of articles. Among the filtered results, there were three articles that summarized the most popular versions of the original article, see Table 1 below.

Table 1.

Overview of the sources of the clues to the different versions.

Public accounts	Title	Number of versions covered	Posting time
廖信忠	今天是微信公号诞生以来最荒谬的一天	26	March 12, 2020
致和致中文化传承	吹哨子的人，近100个翻译版本！至少30亿人次看了	36	March 13, 2020
流放地	全网接力不同文字和符号版《发哨子的人》	16	March 11, 2020

Secondly, following the clues offered in the three articles including the titles of those versions and the Public Accounts that have published them, and, wherever possible, the researcher traced these back to the original posts published by their WeChat Public Accounts. By the time of data collection (around June 2020), over 80 versions had been created and the author was able to collect 50 of them including mostly screenshots and some original posts because most of them have been removed by WeChat together with the Public Accounts that posted them. The collected 50 versions include 17 translations into different languages such as English, Hebrew, and Elvish, 12 multi-semiotic and multimodal representations such as emojis, cartoons, and calligraphy, and 21 technologically transformed versions such as QR codes, morse codes, and DNA sequences. It is these representations that the analysis focuses on below.

Methods

This study adopts a (social) media discourse approach (Montgomery, 2007; Wu and Fitzgerald, 2021b; Wu and Montgomery, 2023) that examines discourse as social actions and social practices that take place within or across media platforms. This form of analysis draws upon a range of pragmatic, discursive, sociolinguistic, and sociological tools designed and configured for the analysis of mediated forms of communication and the interplay of technology and cultural resources. It also draws upon forms of multimodal discourse analysis and social semiotics (Djonov and van Leeuwen, 2018; Poulsen and Kvåle, 2018) that have been developed in line with the emergent focus on the creative technological, linguistic, and cultural resources users employ in the unique environment and ecology of Chinese social media to examine how meanings are created and negotiated in a particular social and cultural context. The approach has been applied and adapted to a range of forms of mediated forms of communication including traditional broadcast talk and digital media (Hsiao, 2022; Tolson, 2010). It is useful in exploring how users mobilize multi-semiotic and multimodal resources provided by platforms to perform social actions, enact identities, create meaning, and communicate in the Chinese digital space (Wu and Fitzgerald, 2023; Teng and Chan, 2022). In this study, WeChat Public Accounts provide their users with a wide range of technological affordances and semiotic resources to create and share multimodal content and participate in a networked movement of reproducing and circulating the article and the message contained in the article. The analysis below begins by examining the more conventional ways of representing the text of the article through technology and translation to the more innovative ways where the text became increasingly unrecognizable and inaccessible and eventually became a collective memory prompt mobilized through a semiotic sign.

Data analysis

When the article was deleted from its Public Account, users first deployed conventional ways to reproduce the original article including methods such as converting the text into an image or PDF file, changing the layout of the article, and using homophone to replace the sensitive words or expressions that share similar pronunciations to circumvent censorship (Yang and Jiang, 2015; Ye and Zhao, 2022). While these forms of reproduction are used widely and are adequate for evading non-human censors as they avoid keyword filtering and whole text screening, not unexpectedly, these did not work for the censorship mechanism of the WeChat platform where human censors play a significant role in this system (Sun and Zhao, 2021). To keep the relay alive in the face of human censors, Chinese users came up with ever more innovative and creative techniques. The focus is then, not on the initial reproduction of the text as an image or PDF format but on the subsequent forms of reproduction as they became increasingly unrecognizable to the original article yet maintained the connection to the article through forms of semiotic representation. This section focuses on three innovative types of strategies WeChat users created in this relay campaign.

Creative textual reproductions

The first level of reproduction involves creative manipulation of the text of the article, such as disrupting the structure of the original article and translating the article into various Chinese dialects and a wide range of languages including English, German, Hebrew, Klingon, and Elvish.

The first technique of textual reproduction was to disrupt the structure of the original article which involved reverting or disorganizing the order of the Chinese characters. While both reversion and disorganization follow the sequential order of the paragraphs and structure of the original article, the tactic of reversing the text placed the paragraphs in an inverted order where the paragraphs start from the end to the beginning, as shown in Example 1 in Figure 2. Readers need to read it from the end of the paragraph in order to make sense of it.

Figure 2.

Examples of textual reproductions 1.

In Example 2 in Figure 2, the tactic of disorganization represented the article by shuffling the syntactic order of clusters of Chinese characters in a seemingly random way but not affecting the interpretation of the semantics. The paragraph in Example 2 was taken from a version entitled “哨发人子的.” Although the characters were rearranged out of order, Chinese readers could understand that “哨发人子的” in the title means “发哨子的人” (the person who handed out the whistle), and that “明不炎肺” in the first line of the paragraph means “不明肺炎” (unexplained pneumonia) because reading and understanding the main meaning does not require that the reader stick close to every Chinese character.

The second technique of textual reproduction was to translate the article into various languages and dialects. This includes translations within the Chinese language system such as Chinese dialects, classical Chinese, and Chinese pinyin, a romanization system for standard Mandarin Chinese. For instance, Example 3 in Figure 3 is an extract that translates the original article into written Cantonese that combines characters from classical Chinese, a traditional style of written Chinese that is different from modern spoken forms of Chinese, and new characters developed by Cantonese speakers. In a different manner, Example 4 in Figure 3 displays a pinyin version in which the author switched to a different system that turned the Chinese characters into their Romanized form which also makes it less recognizable from the censors because the algorithmic censorship mechanism only recognizes Chinese characters.

Figure 3.

Examples of textual reproductions 2.

In these examples, to ensure the article was identifiable to its readers, the translations provided some clues to the main idea of the censored article. For example, the Cantonese version kept the cover of Doctor Ai from the original article while the pinyin version replaced the cover picture of Doctor Ai with an image of a whistle to retain the symbolic meaning of the article and the event. At this stage, the textual reproductions are still intelligible to a wide range of Chinese readers who can read and speak Chinese and have some knowledge of traditional Chinese and Chinese dialects. However, while designed to circumvent keyword filtering and whole-text screening, these forms of reproduction were deleted within a few hours, suggesting that human censors were engaged in removing such posts.

Chinese users then switched to translations into other language systems ranging from more widely used languages such as English and German, to minority languages such as Hebrew, and Vietnamese, and even invented languages such as Klingon and Elvish. For instance, Example 5 in Figure 4 shows an extract from the English version that provides a literal translation of the original article. Notably, the visual part of this version inserted the cover picture of Doctor Ai in which the blank space was filled with a quotation from Dr Ai “早知道有今天, 我管他批评不批评, 老子到处说” (If I had known this was happening, I would have spread the words everywhere. I wouldn’t care about the criticism from the hospital management board) and the first paragraph of the original article serving as important clues for readers to identify this as a version of the article.

Figure 4.

Examples of textual reproductions 3.

While translations into more widely used languages may still attempt to help readers understand the original article, translations into obscure and invented languages seemed to make it more of a competition between users and human censors. For instance, Example 6 in Figure 4 above shows a version that includes an invented language, Elvish, and a reference to the well-known saying by the Star Trek character Dr Spock “Live long and prosper.” Despite the fact that most people would not understand or read Elvish, the author hid the clue in the Chinese title by using an abbreviated form “发哨人” to stand for its full form “发哨子的人” (both forms mean the person who handed out the whistle). The author included an explanation of what language this version was written in following the abbreviated title in brackets and a brief description of the Elvish language before the translation. Notably, these forms of translation were not produced to be read but more like a gesture to challenge the censors of the WeChat platform. This challenge to the censors is strengthened in the highlighted text in the black box below the title which reads “生生不息, 繁荣昌盛” (Live long and Prosper). These eight Chinese characters in English are one of the most well-known quotes from the Star Trek series and films and are often said by the Vulcan character Spock, though deliberately or unconsciously attributed to the Klingon language invented for Star Trek. Here, the saying is recontextualized from a sci-fi greeting to being attached to the article such that “live long and prosper” becomes the encouraged social action of successfully relaying the article into the future. Here then, the cat-and-mouse game has escalated: it is not about whether the posts can evade censorship but rather for how long they can stay posted, “alive,” before being taken down.

Creative multi-semiotic and multimodal reproductions

As the textual reproductions were removed, users adopted multi-semiotic and multimodal resources to reproduce and circulate the article. This signaled a new level of creativity in mobilizing multiple modes and sign systems to create translations and representations of the original article in emojis, sign language, braille, cartoon, morse, and different styles of Chinese ancient art and calligraphy. For instance, the cartoon version in Example 7 and the emoji version in Example 8 in Figure 5 both combined verbal and visual means to reproduce the original article.

Figure 5.

Examples of multi-semiotic and multimodal reproductions 1.

These examples drop the structure and the verbal mode of the original article. In the cartoon version, the author created important scenes according to the interview with Dr Ai through visual representations. For instance, the extract in Example 7 above illustrated what happened on the day by using stick drawing – a drawing style that uses a few lines to represent the outline of a scene – when an infected patient was admitted to the hospital where Dr Ai worked and a dialogue between Dr Ai and her colleagues. While the drawing recreated the outline of the scene, the verbal text in the extract highlights the key information such as the South China Seafood Market where most early cases of COVID-19 were reported, the symptoms of the infected patients, the initial name of the virus, and the main characters of the story including Dr Ai, her colleagues, and the infected patients. Different from the visual storytelling in the cartoon version, the emoji version in Example 8 requires more effort in the meaning-making of pictographic symbols. Using a smart keyboard incorporating emojis installed on their mobile phones, users can choose emojis to replace Chinese characters and expressions. For instance, in the first line in Example 8, the emoji corresponds to the input of “日” which can mean both “day” and “the sun,” while the emoji corresponds to the input of “爱” which means “love.” The ambiguous nature of emojis renders them highly context-specific which in turn, enables an open interpretation of them (Zappavigna and Logi, 2021). However, with enough background knowledge of the article and commonly seen collocations of those words, users would be able to decode as “day” when it follows year and month. However, decoding involves two steps: first, decode it as “爱” (pinyin: ai, meaning love) and second, extend it to “艾” (pinyin: ai, a family name in China), and when coming before the Chinese character “芬” (pinyin: fen), users can successfully understand that it is the name of Dr Ai Fen.

While the cartoon version and emoji version are still comprehensible to their readers, other visual forms like Chinese ancient art and calligraphy make it hardly recognizable. For instance, Examples 9 and 10 in Figure 6 below show two different styles of Chinese calligraphy in which the original article was rewritten. The calligraphy style used in Example 9 is called “小篆” (pinyin: xiao zhuan), a form of ancient Chinese writing style dated back to the Qin Dynasty (BC 221), while the calligraphy style used in Example 10 is called “天书” (pinyin: tian shu) or “云篆” (pinyin: yun zhuan), a form of ancient Chinese writing style developed by Chinese Taoism and is known for its mystery and intelligibility.

Figure 6.

Examples of multi-semiotic and multimodal reproductions 2.

By reproducing the article in calligraphy, the use of artistic forms was introduced into the game. This is somehow reflected in the subheading to the left of the name of the Public Account in Example 10 in Figure 6 that says “老子到处说” (I’m going to spread the words everywhere), a quoted speech from Dr Ai’s interview. On the one hand, it works with the title “发叫居的人” (the person who handed out the whistle) in classical Chinese to provide a clue that this is a new version of the censored article. On the other hand, it also shows the author’s determination to spread the “words,” the article in this case. This is expressed even more strongly in the line above the inserted calligraphy in Example 9 in Figure 6, saying “语言可以被删除和泯灭, 但思想和记忆长存” (text can be deleted and destroyed but thoughts and memories will last). At this point, the event had evolved into an open contest between users and the platform censors with users knowing the censors would keep deleting their recreated versions and censors knowing that users would go ahead of them and create even more. However, this also indicates the next level of creativity in the semiotic representation of the article. Whereas in the previous section, the call was to keep the article alive through reposting, the shift to less recognizable and less comprehensible representations that require sophisticated interpretation also means that the text of the article is no longer the thing to keep alive, but by posting forms of representation that “point to” the article keeps the article alive in users’ thoughts and memories, a shared understanding among users. That is, the article is kept alive beyond the text itself as it is held in users’ memory and requires only semiotic prompts to keep the article alive. This transformation into semiotic prompts characterizes the final stage where the article and text were rendered completely absent and replaced with semiotic prompts that were designed to remind the reader of the article.

Technically encoded means

With versions of the article being detected and deleted by the WeChat platform censors, users developed more innovative methods that encoded and stored the article in technical and technological texts including web links, bar codes, QR codes, computer codes, blockchain, and DNA sequences. At this stage, the focus shifted from simply translating and disseminating the actual content of the article to encrypting the article to see how long it would survive before it was removed. For instance, Examples 11 and 12 in Figure 7 show that the article has been encoded into a bar code and QR code respectively. This means that the article is not immediately accessible as the reader would need to scan the bar code or QR code to access the article.

Figure 7.

Examples of technically encoded forms 1.

Notably, the paratext in the bar code in Example 11 and the verbal texts above the QR code in Example 12 employ different figurative speeches to express the authors’ opinions about whistleblowers and platform censorship. The barcode version used indirection where the verbal texts were hidden and encrypted in the code. In a closer examination of the bar code, two lines of words in Chinese and English become apparent which say “吹哨人法案” and its English translation “whistleblower protection act,” suggesting that this action is part of a grassroots action under an invented legal policy. In a different vein, the verbal text in the QR code version in Example 12 envisions a scenario of an art exhibition where the QR code becomes a piece of artwork shown to the audience. The author also includes a metaphor that compares “摄像头” (camera) to censorship that monitors the artwork. Here, the QR code facing directly at the camera (“被摄像头对准”) expresses the author’s aspiration for it to become a piece of valuable and safeguarded artwork that symbolically capures the intricate interplay between Chinese users and online censorship.

The removal of these versions then sparked another round of creativity with users building and positing multi-layered encrypted versions that involved multiple steps of decoding. For instance, the blank version in Example 13 in Figure 8 demonstrates that the article can be encoded in a blank space where the article becomes invisible. In the blank space, there is a line of instructions “Click + long press + Control A + Control C” in small font to help interested readers access the article. Going even further, Example 14 in Figure 8 shows a version that has gone through four steps of encoding. The original article signified by its publication People (人物) was firstly converted into Unicode (“\u4eba\u7269”) and secondly into hexadecimal number (“4eba7269”), and thirdly to quarternary system (“1032232213021221”), and finally into letters “ATCG” by converting the numbers “0123” into letters “ATCG” (“TAGCCGCCTGACTCCT”). Therefore, to access the original article, users need to decode the instructions provided by the author in reverse order. Below the image of a flow diagram, the author commented on saving the article through a joke. They called this multi-layered encrypted version a DNA sequence which alludes to a popular Chinese internet expression “刻进DNA,” literary translated as “carved into one’s DNA,” meaning that it is now an indelible part of the body and is able to pass onto the next generation. At the end of the comment, the author made another playful suggestion that involved transforming the article into a DNA sequence and preserving it within one’s intestinal flora, accompanied by a doge emoji which often implies sarcasm and playfulness in the Chinese online communication (Zhou, 2021).

Figure 8.

Examples of technically encoded forms 2.

Looking at those technically encoded versions, we can see the details and clues carefully designed by the authors. Their titles deployed various techniques to hide the clues and meanwhile, avoid keyword searching and filtering. For instance, the QR code version in Example 12 in Figure 7 placed its title in reverse order “术艺的子哨发” (the art of giving out the whistle), which should read from right to left in order to get the meaning, and metaphorized social media users’ discursive act of reproducing and circulating the article as an art of giving out whistles. Adopting abbreviation, the titles “接力哨” (relaying the whistle) in Example 11 and “哨” (whistle) in Example 13 in Figure 8 used a single Chinese character “哨” (whistle) to stand for the original article “发口哨的人” (the person who handed out the whistle). In addition to abbreviation, the title “发□肖人— —DNA序列版” in Example 14 in Figure 8 made use of Chinese characters (“发,” “肖,” “人”), spaces between the characters, and symbols (square symbol “□” to replace the Chinese radical “口”) to represent an abbreviated title “发哨人” (the person who handed out the whistle). In this way, although the article is not visible in those technically encoded forms, the audience could still understand them as part of the relay campaign. Moreover, we can also see a shift from keeping the article and text alive in the first section of analysis through keeping the article alive in different forms of semiotic signs to this point where the article is indelibly imprinted into human genetics, meaning that it cannot be removed even if forgotten.

Discussion and conclusion

This study further examines and contributes to understanding the creative and innovative ways that Chinese social media users employ to engage with the unique environment of Chinese internet censorship, social media technologies, and digital culture. Focusing on an online campaign where users reproduced and circulated a briefly censored article on the WeChat platform, the analysis examined how users openly engaged with both algorithmic censorship and human censors in order to post representations of the article before being removed. Drawing on analytic tools from (social) media discourse analysis, the analysis identified three major forms of strategies that users developed to challenge platform censorship including textual reproductions, multi-semiotic and multimodal reproductions, and technically encoded forms. While textual reproductions involve disrupting the structure of the original article and translating the article into different dialects and languages, multi-semiotic and multimodal means reproducing the original text into various Chinese visual-verbal forms and art forms, and technically encoded forms refer to ways of storing the original text into multi-layered encrypted forms. The analysis highlights that through these strategies, the original text moves from being recognizable to unrecognizable, from verbal text to visual symbols, and finally to symbolic representations that rely on a shared scheme of understanding and reference. Interestingly, as the posts became increasingly esoteric, moving away from recognizable texts into both semiotic prompts and multi-layered encryption, the article transcended its original form and became part of users’ collective memory, thereby circumventing censorship where any semiotic form can be employed as a memory prompt.

While the study contributes to a further exploration of the potential of social media as a powerful channel for opinion expression and public participation within the context of internet censorship, it also highlights the creative and carnivalesque environment of Chinese social media (Fitzgerald et al., 2022; Wu and Fitzgerald, 2023), especially the explicit and creative engagement between users and platform censorship. As the Chinese censorship apparatus is never revealed to the public, users often react to it based on their personal experiences and accumulated knowledge of it (Luo and Li, 2022). Throughout the relay campaign, users were aware that both algorithms and human censors were involved. The evolving interplay between users and censors resembled a game of cat and mouse where users challenged and sometimes teased the censors who found and removed their posts. As this game took place at the level of platform censorship – the article was available on other platforms like Sina Blogging, it suggests that this form of engagement is less serious and can be more playful as it engages with the platform censors who are also familiar with both the affordances of the platform and the reproduction forms users may adopt. Therefore, the playful and creative engagement with platform censorship does not fall into the previous dyadic understanding of a suppressing state and a resisting public. Rather, it involves more nuanced complexities in both the operation of censorship – multi-layered censorship and various parties and players from governmental sectors, companies, and individuals, and users’ engagement with it including resistance, creativity, and playfulness. This somehow attests to Ye and Zhao’s (2022) argument that users’ discourse strategies in coping with Chinese internet censorship are embedded in users’ everyday resistance and everyday politics. Indeed, as many Chinese users have perceived censorship as part of their everyday communication and their public participation in social and political discussions, they have developed ways to live and cope with it.

The interplay between WeChat users and platform censorship also points to a better understanding of this layer of censorship within the overall system and mechanisms. Firstly, WeChat platform censorship relied on both algorithms to detect censorable articles containing keywords and employed human censors to review those articles and make the decision of whether to remove them from the platform (Sun and Zhao, 2021). While algorithms can quickly identify censorable content, decisions by human censors take hours and sometimes days to make, which is roughly the life span of the posts that managed to get published online but were eventually deleted. In this case study, most representations of the “whistle” were removed within hours. Notably, the data for this study did not include perspectives from the WeChat censorship staff. What is revealed from the data is that those representations were alive for a few hours and that the shift of users’ strategies from more conventional to more creative ones was triggered by the involvement of WeChat censorship staff. In response to censors’ increasingly stricter monitoring and censoring practices, users developed ever more innovative and sophisticated ways to keep the “whistle” alive as long as possible. Secondly, the study shows that social platform censorship as a form of delegated censorship operates differently and sometimes more strictly than governmental censorship. This is reflected in the fact that the article Whistle was censored just for a day and became available again the next day on WeChat, and while it was censored on WeChat, it was accessible through Chinese search engines and on websites such as Sina Blogging. This suggests that the censoring act was performed at the level of social platforms without the intervention of government censorship because, if governmental censorship were involved, the article would have also been removed from the whole Chinese internet.

Finally, what is also apparent in users’ engagement with platform censorship is that the technological affordances of a platform provide a wealth of creative resources for multi-semiotic forms of communication. In this case, the reproduction of the text moves from words to images to pure symbolic signs, and from being recognizable to being deliberately unrecognizable and hidden beneath layers of code. While it was possible to trace back the layers of coding to find the article, it was unnecessary as ultimately it was the existence of the “game” and users engaging in the “game” that kept the “whistle” alive beyond any particular individual posting. While this points to a further level of understanding of the complexities of Chinese internet censorship, the creative use of technologically mediated semiotic resources and cultural elements in the wider Chinese social media ecology requires ever more sophisticated forms of discourse analysis to analytically examine emerging and evolving forms of meaning-making in the digital environment.

Footnotes

Acknowledgements

The author would like to acknowledge Richard Fitzgerald for his analytic insights and contribution to this article. She would also like to thank the two reviewers for their useful comments in improving the article.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is supported by The Philosophy and Social Sciences Planning Projects of Guangdong Province “广东省哲学社会科学规划2020年度青年项目”[grant number GD20YXW03] and the Start-up Fund of BNU-HKBU United International College [grant number UICR0700008-22].

ORCID iD

Xiaoping Wu

References

Cobbe

(2021) Algorithmic censorship by social platforms: Power and resistance. Philosophy & Technology 34(3): 739–766.

Djonov

van Leeuwen

(2018) Social media as semiotic technology and social practice: The case of researchgate’s design and its potential to transform social practice. Social Semiotics 28(5): 641–664.

Fitzgerald

Sandel

(2022) Chinese social media: Technology, culture and creativity. Discourse, Context & Media.

Gao

(2022) The online activism of mock translanguaging: Language style, celebrity persona, and social class in China. Discourse, Context & Media 47: 100608.

Gleiss

(2015) Speaking up for the suffering (br) other: Weibo activism, discursive struggles, and minimal politics in China. Media, Culture & Society 37(4): 513–529.

Han

(2015) Defending the authoritarian regime online: China’s ‘voluntary fifty-cent army.’ China Quarterly 224: 1006–1025.

Hassid

(2020) Censorship, the media, and the market in China. Journal of Chinese Political Science 25(2): 285–309.

Hsiao

(2022) Authenticating discourses of ‘being oneself’ on monetary-motivated livestreams. Discourse, Context & Media 47: 100592.

King

Pan

Roberts

(2013) How censorship in China allows government criticism but silences collective expression. American Political Science Review 107(2): 326–343.

10.

King

Pan

Roberts

(2014) Reverse-engineering censorship in China: Randomized experimentation and participant observation. Science 345(6199): 1251722.

11.

Lee

(2016) Surviving online censorship in China: Three satirical tactics and their impact. China Quarterly 228: 1061–1080.

12.

Luo

(2022) Participatory censorship: How online fandom community facilitates authoritarian rule. New Media & Society: 1–19.

13.

Mina

(2014) Batman, pandaman and the blind man: A case study in social change memes and internet censorship in China. Journal of Visual Culture 13(3): 359–375.

14.

Montgomery

(2007) The Discourse of Broadcast News: A Linguistic Approach. London and New York: Routledge.

15.

Poulsen

Kvåle

(2018) Studying social media as semiotic technology: A social semiotic multimodal framework. Social Semiotics 28(5): 700–717.

16.

Sun

Zhao

(2021) Delegated censorship: The dynamic, layered, and multistage information control regime in China. Politics & Society 50(2): 191–221.

17.

Tai

(2020) Specificity, conflict, and focal point: A systematic investigation into social media censorship in China. Journal of Communication 70(6): 842–867.

18.

Teng

Chan

(2022) Collective colouring in danmu comments on Bilibili. Discourse, Context & Media 45: 100577.

19.

Tolson

(2010) A new authenticity? Communicative practices on YouTube. Critical Discourse Studies 7(4): 277–289.

20.

(2018) Discursive strategies of resistance on Weibo: A case study of the 2015 Tianjin explosions in China. Discourse, Context & Media 26: 64–73.

21.

Fitzgerald

(2021a) “Hidden in plain sight”: Expressing political criticism on Chinese social media. Discourse Studies 23(3): 365–385.

22.

Fitzgerald

(2021b) Reaching for the stars: DingTalk and the Multi-platform creativity of a ‘one-star’ campaign on Chinese social media. Discourse, Context & Media 44.

23.

Fitzgerald

(2023) Internet memes and the mobilization of a “One-China” cyber nationalist campaign: The case of the 2016 Diba Expedition to Taiwan. Social Semiotics: 1–23.

24.

Montgomery

(2020) Witnessing in crisis contexts in the social media age: The case of the 2015 Tianjin blasts on Weibo. Media, Culture & Society 42(5): 675–691.

25.

Montgomery

(2023) The microblogging discourse of disasters: Twitter and Weibo in action in the aftermath of two major industrial accidents. Social Semiotics 33(4): 731–749.

26.

Feng

(2015) Networked creativity on the censored web 2.0: Chinese users’ Twitter-based activities on the issue of Internet censorship. Journal of Contemporary Eastern Asia 14(1): 23–43.

27.

Yang

(2016) Rethinking China’s Internet censorship: The practice of recoding and the politics of visibility. New Media and Society 18(7): 1364–1381.

28.

Yang

(2009) The Power of the Internet in China: Citizen Activism Online. New York: Columbia University Press.

29.

Yang

Jiang

(2015) The networked practice of online political satire in China: Between ritual and resistance. International Communication Gazette 77(3): 215–231.

30.

Zhao

(2022) ‘I know it’s sensitive’: Internet censorship, recoding, and the sensitive word culture in China. Discourse, Context and Media 51: 100666.

31.

Zappavigna

Logi

(2021) Emoji in social media discourse about working from home [Formula presented]. Discourse, Context and Media 44: 100543.

32.

Zhang

Gearhart

Perlmutter

(2022) Avoiding online censorship through ‘fatty’ memes: How Chinese social media users talk about North Korea. Global Media and Communication 18(2): 199–218.

33.

Zhou

(2021) China’s beloved doge emojis don’t mean what they seem. Vice. Available at: https://www.vice.com/en/article/93y588/china-doge-emoji-dogecoin-memes#:~:text=Moreoftenthannot%2Cthedogeisused,arerushingtotrademarktheirowndogedesigns (accessed 12 April 2023).

34.

Zhu

(2020) Speaking up or staying silent? Examining the influences of censorship and behavioral contagion on opinion (non-) expression in China. New Media & Society 23(12): 3634–3655.