Amplification of regional discrimination on Chinese news portals: An affective critical discourse analysis

Regional discrimination is a significant social issue that leads to divided societies. In China, people from Henan Province, who are verbally abused by non-Henan users on the Internet, are often victims of regional discrimination. This article presents a case study of Chinese Internet users’ discriminatory practice against Henan people in the commentary sections of two major Chinese news portals – Tencent and NetEase. By advancing an affective critical discourse analysis approach with the assistance of content analysis, I analysed user comments on news reports that covered a news event relating to regional discrimination against Henan people. The analysis showed that Internet users’ discriminatory practice was notably amplified by the locative IP-address function in NetEase’s commentary section. The research findings shed light on the interplay between Internet users’ discursive practice and the technological architecture of interactive digital platforms in the context of regional discrimination.


Introduction
Across the world, discrimination has always been a significant issue influencing the harmony of societies. Existing literature has examined discriminations concerning gender, ethnicity and religion (Baker et al., 2013;Chang et al., 2018;KhosraviNik, 2014). However, much less scholarly attention has been paid to regional discrimination, which takes the form of discrimination against a social group based on its members' place of origin (Zhang, 2013). Regional discrimination is particularly apparent in China, and people from Henan Province, who are stereotypically portrayed by the media as 'dishonest', due to their disadvantageous socio-economic position, are commonly the victims. Such stereotypical media portrayals have notably influenced their fellow citizens' collective mentalities, often reflected as Henan citizens being verbally abused by people from outside of the province (Chen, 2016). The regional discrimination that Henan people suffer is not the only form of regional discrimination in China, but it is the most well-known type, given the public attention it attracts (Zeng, 2012).
With the emergence of interactive digital platforms, which enable Internet users to generate original content (Fuchs, 2015), regional discrimination has become a significant aspect of Chinese digital culture today. Popular news portals are important interactive digital platforms through which discrimination against Henan people is spread in Chinese society (Chen, 2016). News portals mainly aggregate original news content from the mass media, and their design often incorporates commentary sections, which engage Internet users by enabling them to comment on each news report (Qian, 2011). The extent to which news portals facilitate the spread of regional discrimination is evidenced by the large volumes of derogatory comments about Henan people circulated in the commentary sections of these sites (Chen, 2016).
Existing studies of Chinese digital culture tend to pay a great deal of attention to the Chinese Communist Party's (CCP) Internet censorship (Schneider, 2018;Stockmann and Luo, 2015). Yet, research shows that a diversity of voices is allowed on the Internet as long as the voices do not encourage civil disobedience. This is because the main purpose of the CCP's censorship is to prevent collective protests against the Party-State (King et al., 2013). Taking the form of conflicts between citizens themselves, regional discrimination is not of great concern to the government and is, therefore, not targeted by the censorship. In this way, the role of news portals becomes salient in the shaping of regional discrimination in Chinese digital culture.
Focusing on the technological architecture of news portals, this article explores the process through which Chinese Internet users' discriminatory practice against Henan people is affected in the commentary sections of these sites. To this end, I present a case study, analysing the user comments beneath the coverage of a news incident retrieved from two major news portals -Tencent and NetEase. In particular, NetEase's commentary section incorporates an IP-address function, which automatically displays Internet users' geographic location alongside their comments (Qian, 2011). No such function exists in Tencent's commentary section. By comparing Tencent and NetEase users' postings, the analysis uncovered how regional discrimination-related debates are amplified by the IP-address function. The findings shed light on how technological architecture modulates Internet users' discursive practices.

Literature review
Sociocultural contingence of regional discrimination Political economists tend to analyse regional discrimination in relation to the political engineering of Chinese people's regional identity (Afridi et al., 2015). The notion of regional identity constructs different social groups based on different places of origin (Raagmaa, 2002). Since the establishment of the CCP-led government, Chinese people's regional identity has been institutionalized by the hukou (household registration) system, which requires all citizens to register with the local authority where they were born (Zhang, 2018). A person's access to social welfare and benefits is restricted if they do not hold a local hukou. While the hukou system no longer completely constrains the geographic mobility of the population, by binding people with a hukou from the same province or city together at an institutional level, it has a knock-on effect on regional discrimination (Afridi et al., 2015). Yet, regional discrimination reflects a socioculturally contingent process that is also influenced by the shifting ethos in post-reform society (Chen, 2016;Zeng, 2012;Zhang, 2013). Since the late 1970s, the CCP has implemented market-oriented reform of the Chinese economy, leading to the acceptance of capitalist moralities by the general public (Goodman, 2014). The pursuit of wealth has become glorified, and one's social value has become measured by annual income. The acceptance of the capitalist moralities shifts the responsibility for social well-being from the government to the individual (Yang, 2017). In this process, poverty comes to be considered a consequence of laziness, leading to the emergence of discrimination against the poor. This is often regionally organized because of the imbalanced economic development within the Chinese territory (Afridi et al., 2015).
Following four decades of rapid economic growth, the eastern provinces, where the annual income of local households is the highest in the country (National Bureau of Statistics of China, 2020), have become the most developed part of China. Against this backdrop, living in the eastern provinces becomes associated with a 'privileged' social status (Fleisher et al., 2010). This association has widened the gap between Chinese people from the east and those from the inland provinces, providing the socioeconomic grounding for regional discrimination today. Henan, in particular, is the most populous inland province, and more than half of its population is employed in the low-income rural economy (National Bureau of Statistics of China, 2020). For better lives, many Henan peasants move to eastern cities, becoming migrant workers employed in labour-intensive industries. Most Henan migrant workers continue to suffer from poor living conditions even after reallocating to eastern cities. The large numbers of Henan migrant workers, alongside the entire Henan population, have become the social group most vulnerable to regional discrimination. They are not only discriminated against in the job market outside of Henan Province but are also verbally abused by their fellow citizens from other parts of the country, especially those from the eastern regions (Chen, 2016).

Media discourse and regional discrimination
From a media studies perspective, existing scholarship suggests that the market-oriented media is responsible for the penetration of regional discrimination in post-reform China (Chen, 2016;Zeng, 2012;Zhang, 2013). Nowadays, state-owned media still monopolize the news publishing industry in China. However, the notable decrease in government subsidies since the 1990s has created large numbers of state-owned, yet market-oriented, media (Huang, 2016). These market-oriented media, which often share a similar, sensational journalistic practice with the tabloid press in the West, are more popular with the general public than their propaganda-oriented counterparts (Zhang, 2019). Focusing on the market-oriented media, Zhang (2013) conducted a textual analysis of Henanrelated news coverage. The research findings suggest that, by repeatedly promoting negative stereotypes of people from the province, the media encourage regional discrimination against Henan people. Such findings echo Zhou's (2004) observation of crime news reporting in China's TV industry, which unveils similar discriminatory media content concerning Henan people.
Highlighting the sensational journalistic practice of the market-oriented media, the research findings by Zhang (2013) and Zhou (2004) remind us of the role media discourse plays in the spread of regional discrimination in China. Discourse is a socially constitutive practice of language (Fairclough, 2003). Media discourse, which refers to the discursive practice of the mass media, is important to the construction of people's 'collective mentalities' (KhosraviNik, 2014: 512). According to van Dijk (2009), media discourse constantly influences the audience's perception of a social group by constructing stereotypes of the group. Downing and Husband's (2005) study of racial representation shows that stereotypes are characterized by constructing a social group as homogenous, without acknowledging variations in its members' individual qualities. These stereotypes often endure over time but do 'not so much aid understanding but aid misunderstanding', as they tend to misrepresent the social group (McGarty et al., 2002: 4).
'Dishonesty' is a representative stereotype applied to Henan people in Chinese media discourse. This stereotype is often found in the market-oriented media's fraud crime reporting (Chen, 2016). As Zhang (2013) notes, outside of Henan Province, the media usually emphasize the place of origin of Henan suspects by describing the fraud suspects as 'Henan fraudsters'. Such a referential instance portrays the dishonest characteristics of a small number of fraud suspects as a quality shared by the entire Henan population. It constructs a 'dishonest' stereotype of Henan people that has tarnished their reputation within the country.

Limits of a critical discourse analysis method
The existing literature on regional discrimination is largely in line with the paradigm of critical discourse analysis (CDA) research. CDA is a text-driven research method, which scrutinizes how language use is framed within specific sociocultural contexts (Fairclough, 2003). This method examines the power relations established in the process of textual production (Baker et al., 2013). Fairclough (2003), the co-founder of CDA, provides a three-dimensional model to analyse the linguistic elements of media content at the textual, discursive and sociocultural levels, respectively. As van Dijk (2009) notes, CDA is a suitable methodological approach for the study of biases in media discourse. By using the CDA method, KhosraviNik (2014) scrutinizes the linguistic features of five influential news reports on refugees in the United Kingdom. This research resonates with the study by Baker et al. (2013), uncovering how biased British newspapers, in their coverage of immigration issues, relate migrants to negative stereotypes, such as asylum seekers.
The existing CDA method has merits. However, emphasizing a sociocultural interpretation (Chang et al., 2018;Peng, 2020;Xu and Tan, 2020), this method traditionally does not pay due attention to the nuanced interplay between discourse and the technological architecture that facilitates the discursive practice in the digital age (Bouvier and Machin, 2018). Nowadays, the global diffusion of interactive digital technologies has created a participatory digital culture, encouraging Internet users to share original content that reflects their thoughts and ideas on various interactive digital platforms (boyd, 2014). This digital culture has redefined and restructured mediated communication by supporting a nonlinear, decentralized process of textual production (KhosraviNik, 2018). In this process, the assemblage of Internet users is politically and socioculturally pre-structured, often leading to polarized viewpoints (Papacharissi, 2015). A better understanding of this polarization is required to unpack the dynamics between Internet users' postings and the design of interactive digital platforms.
Specific to the Chinese context, regional discrimination is increasingly caught up in technologydiscourse entanglements. This is evidenced by Internet users' verbal abuse of disadvantaged Henan people on interactive digital platforms, such as the commentary sections of news portals (Chen, 2016). The formation of this phenomenon is inevitably influenced by the technological architecture of the commentary sections, which facilitates Internet users' discursive practice of this kind. Under these circumstances, a renewed search for a methodological approach is necessary to fully account for the technology-discourse entanglements that shape regional discrimination in the digital age.

An affective CDA approach
Under the CDA rubric, an emerging body of scholarship has incorporated the notion of affect to address the interplay between the production of user-generated content and the technological architecture of the platforms (Glapka, 2019;KhosraviNik, 2018). According to Deleuze (1988), affect is defined as the 'outcome of the encounter between entities', and it describes how the 'entities are affected by these encounters' (Ash, 2015: 84). Human emotions can be affected by the body's encounters with its surroundings (Papacharissi, 2015). For instance, while watching a film, audiences may be affected by music, which leads to their affective responses, such as emotional changes (Atkinson and Kennedy, 2019). In this way, affect opens up the affected to new capacities for action (Johanssen and Garrisi, 2019).
Two concepts, 'material thresholds' and 'associated milieus' (Ash, 2015: 84), help to articulate the process through which people's everyday practice is affected in their use of technology. The material thresholds refer to the limits that determine the potential affects a technology can generate. These are shaped by the design of the technology, which defines its capacity for enhancing affective production (Ash et al., 2018). However, affect cannot work 'outside of an […] ecological context' (Ash, 2015: 85). The associated milieus describe the ecological contexts in which the generation and transmission of affect are facilitated (Protevi, 2009). For instance, the research by Ash et al. (2018) shows that web designers may intentionally endow a website with the material threshold, that is, the design, to influence Internet users' actions when they interact with the website via the user interface. Yet, this material threshold can only make sense when Internet users are socially motivated to browse the website in specific scenarios.
Affect theory is often understood to be an approach that primarily analyses the pre-discursive process of bodily experience (Ash, 2015). Yet, Wetherell (2012: 7) argues that there is an affective-discursive loop, describing the intersections of people, affective experience and discursive practice. In general, one's actions of discourse are intended to affect others with whom one engages in certain modes of interaction (Johanssen and Garrisi, 2019). An affective process can articulate 'how actions and discourses affect participants and how those affects stimulate other discursive mechanisms' (Kelsey, 2015: 5). As such, the concept of affect is compatible with CDA studies. An affective CDA method, which provides an account of the affective-discursive loop of Internet users' content generation on interactive digital platforms, helps overcome the sociocultural determinism embedded in traditional CDA research.
As interactive digital platforms, news portals incorporate a commentary section, allowing Internet users to generate original content beneath each news report published on a website (Qian, 2011). This commentary section constitutes an important material threshold, which turns a website into an affective network in which affect is circulated through the exchange of comments between Internet users. A particular form of affect, namely attentive affect, is notable in influencing Internet users' discursive practice in the commentary section of a news portal. Attention is the result of capturing and holding one's perception or the flow of consciousness (Stiegler, 2010). As Stiegler (2010) suggests, people are living in a world in which their attention is constantly modulated by a complex combination of factors in their everyday encounters with surroundings. These factors continue to influence people's conscious and unconscious actions (Ash et al., 2018). By incorporating what is referred to as 'floor-building', the commentary section of a news portal visualizes an original commentand its responsesin a 'block' of dialogue (Qian, 2011). Internet users can reply to a comment by referencing it. A 'block' of dialogue expands as more Internet users participate in the discussion. This 'floor-building' feature has the capacity for generating attentive affect, which invites one to pay attention to others' comments and reply to them. Given that a news portal is accessible to all, Internet users may easily engage in conversations with others across the country.
The commentary section of NetEase, in particular, supports a unique IP-address function, which displays Internet users' geographic location alongside their comments (Qian, 2011). This material threshold facilitates the generation of locality-specific attentive affect. It encourages Internet users to pay particular attention to each other's location by presenting this information as a salient aspect of their participation in the commentary section. Such a mechanism has self-representation implications (Schwartz and Halegoua, 2015). Self-representation involves the conscious or unconscious use of signs and symbols in one's identity articulation (boyd, 2014). On interactive digital platforms, Internet users enact their identities by using available digital symbols. This digitalized self-representation led to the assumption that Internet users' identity could be disconnected from their geographic locations (Schwartz and Halegoua, 2015). This assumption has, however, collapsed in the era in which locative services are widely used. By using Global Positioning System (GPS) or IP-address recognition technologies to detect a terminal device on a digital map, locative services create an affirmative interaction between user-generated content and the place at which it is generated (de Souza e Silva and Frith, 2012). These locative services have added a locative dimension to Internet users' self-representation.
The IP-address function provided by NetEase is certainly different from the locative function of social media platforms. Based on IP-address recognition technologies, this function displays the locative information of a user, which only specifies the province and/or the city in which s/he is currently situated. This locative information is automatically attached to a user's comment (Qian, 2011), meaning that it does not constitute a customized performance of identity. However, with the locative information being made a salient feature that characterizes their digital profiles in the commentary section, the IPaddress function may encourage Internet users to recognize each other based on an assumption that their place of origin corresponds to their current geographic location. This affective process possibly modulates these Internet users' self-categorization, shaping their discursive practice in the commentary section at various levels, depending on the scenario. It reflects the information-seeking strategies Internet users develop to cope with mediated communication, revealing how they actively 'acquire information about others', and how 'this information, in turn, affects [their] relationships with and impressions of targeted others' in the digital age (Ramirez et al., 2002: 214).

Research methods
The present research specifically analyses the affective modulation of Internet users' regional discriminatory practice occurring in the commentary sections of Chinese news portals. I conducted a case study, collecting original data based on a news incident, which provoked public attention to regional discrimination in 2016. The incident related to the police force's anti-fraud campaign in Henan Province. According to news coverage of the campaign, more than 100 suspects were charged for impersonating military personnel and using this false identity to organize telecom fraud crimes, with almost all of them holding a hukou from Shangcai County in Henan Province. 1 I sampled 10 news reports (6 from Tencent and 4 from NetEase), which comprised the complete collection of Tencent and NetEase's coverage of the incident, and local authorities' investigation and/or reactions. In particular, the incident fed into discriminatory discourses about Henan people because the suspects were accused of being 'fraudsters', thus, connecting with the long-existing, negative stereotype of people from the province. This feature of the incident made the news coverage eye-catching among Internet users. It created an opportunity to retrieve various types and a huge number of comments below the sampled articles, providing swathes of empirical data for analysis. All user comments beneath the 10 news reports were retrieved from the commentary sections of Tencent and NetEase, resulting in a total of 7792 comments collected.
An affective CDA approach was employed to analyse all the sampled comments, focusing on the intersections of Internet users' postings, sociocultural contexts and the design of the commentary sections of Tencent and NetEase. In particular, the news reports retrieved from each news portal on the same dates show high levels of homogeneity. Yet, NetEase's commentary section incorporates a locative IP-address function, which does not exist on Tencent. A comparison between Tencent and NetEase users' comments helps uncover the affectivity of locative information manifested in their discursive practice. Given the size of the sampled comments, quantitative content analysis (CA) techniques were also incorporated in the analysis to minimize the possibility of cherry-picking data (Baker et al., 2013). To this end, I calculated the frequency of each word used in Internet users' postings and coded the sampled comments based on a thematical categorization of the words. The coding results were compared to identify the most salient differences between Tencent and NetEase users' postings. The CA process uncovered the general trends in the data set, paving the way for implementing the affective CDA method at the next stage.

CA measures and results
The collected comments were generally relevant to the news coverage, revealing Chinese Internet users' opinions on the fraud incident and its aftermath. However, while some Internet users condemned the criminal conduct of the individual suspects involved, many others widened the target of their criticism to the entire Henan population. The wordlist showed that negative words or phrases, such as 'fraud' (Tencent: 253 hits; NetEase: 1307 hits) and the 'headquarter [of fraud gangs]' (Tencent: 65 hits; NetEase: 391 hits), were frequently mentioned in both Tencent and NetEase users' postings. These negative words and phrases were often deliberately used alongside those relating to Henan people (Tencent: 281 hits; NetEase: 2722 hits). Such a collocation facilitated the portrayal of Henan people as a homogenized social group, whose members share the same 'dishonest' characteristics and are responsible for all fraud-related crimes in the country (see Figures 1 and 2 for typical examples). These discriminatory comments were indicative of the existing, biased evaluation of Henan people's characteristics in Chinese society.
As the research was interested in the differences between Tencent and NetEase users' postings, I coded the sampled comments, based on an independent variabletheir posting website and three dependent variableswhether a comment mentioned words or phrases relating to (a) Henan, (b) non-Henan regions or (c) the most developed regions. The independent variable was determined by the news portal from which the comment was retrieved (Tencent and NetEase). A total of 995 comments were collected from Tencent, while the rest 6797 comments belonged to NetEase. The first dependent variable was coded based on whether a comment involved reference to Henan Province (yes = 1; no = 0). A total of 2363 (30.3%) comments were coded as involving mentions of Henan Province. The second dependent variable was determined by whether a comment mentioned any word or phrase relating to a place outside of Henan (yes = 1; no = 0). A total of 1127 (14.5%) comments belonged to the non-Henan-related category. The last dependent variable was coded for mentions of any place specifically located in the most developed part of China (yes = 1; no = 0), and 593 (7.6%) comments fell into this category.
Based on the coding results, cross-tabulation tests and χ 2 tests of independence were performed to determine whether differences existed between Tencent and NetEase users' postings. As can be seen in Figure 3, the tests showed that about one-third of the sampled NetEase user comments mentioned Henan-related words or phrases, and this figure was about 5% more than that of the user comments retrieved from Tencent (χ 2 = 9.962; df = 1; p < 0.01). More notably, 16% of the sampled NetEase  user comments mentioned words or phrases relating to non-Henan provinces or cities, and this figure was four times higher than that retrieved from Tencent's commentary section (χ 2 = 106.448; df = 1; p < 0.01). The difference was even greater when the compared figures were percentages (NetEase: 8.62%; Tencent: 0.71%) of comments involving mentions of any place in the most developed, eastern region (χ 2 = 77.390; df = 1; p < 0.01). The tests showed statistically significant effects between the independent variable and the three dependent variables, suggesting that the observed differences between NetEase and Tencent users' lexical choices did not occur by chance.
In the data set, the ratio of user comments retrieved from Tencent to those from NetEase was not perfectly balanced because there were fewer comments generated by Tencent users on this topic. Yet, Tencent's news portal is generally more popular than NetEase (Fuchs, 2015). The paradoxically higher number of comments collected from NetEase revealed that Internet users' postings on the website were stimulated. The notable differences between Tencent and NetEase users' postings in terms of the volume and the lexical choice can be explained by the fact that the commentary section of NetEase has an IP-address function while that of Tencent does not. The detailed process through which such differences were established is articulated below.

Affective CDA analytical discussion
Based on the CA results, an affective CDA approach was adopted to unpack the process through which Internet users' discursive practice was modulated in the commentary sections of the news portals. The affective modulation concerned three aspects, including biased media discourse, the interactive design of news portals and the IP-address function unique to NetEase.

Contextualized by the media discourse
Media discourse often represents a contextual pre-structure of Internet users' postings. In the present incident, the news reports published by NetEase and Tencent showed a level of homogeneity in their discourse strategy. They employed a referential strategy that invoked the longexisting, negative stereotype to describe Henan people. As KhosraviNik (2014) notes, referential strategies are typically used by the media in their stereotyping of a particular social group. In this instance, in both the news titles and the news stories, the stereotyping of Henan people as 'dishonest' was realized by explicitly referring to the fraud suspects as 'of Henan origin'. The extracts below, which were retrieved from the news reports published on Tencent and NetEase, provided a good illustration of this referential strategy. Tencent: 起底河南诈骗村: 村民扮军人行骗 每家存千万 Translations: The discovery of a Henan fraud village: villagers defraud by acting as soldiers; each household has saved more than 10 million. (Tencent,26th September 2016) NetEase: 河南一诈骗村村民扮军人行骗 Translations: A Henan fraud village: villagers defraud by acting as soldiers; each household has saved more than 10 million. (NetEase, 26th September 2016) As previously mentioned, news portals often aggregate stories from the same information sources, such as newspapers and news agencies. This leads to their often-homogenized news coverage on the same incident. Yet, in the present case, the original reports, sourced from traditional media, detailed the fraud suspects' place of origin by naming the village from where they came, rather than by referring to the province of Henan. Interestingly, in both the Tencent and NetEase's articles, 'Henan fraud village' was used to replace the name of the village in their re-edited versions. This formed a referential strategy that highlighted 'Henan' as a key element of the story. As Baker et al. (2013) note, referential strategies are often deployed in association with juxtaposition techniques that place two items next to each other for contrasting or complementary effect. In the present incident, the juxtaposition of 'Henan' and 'Fraud' was discovered in both titles above. This juxtaposition technique, alongside the referential strategy employed, proposed an association between local people and 'organized fraud crime'. It appeared to the Internet users that this incident was not an isolated case but served as a window to the 'habitual dishonesty' embedded in the characteristics of all residents living there.
Inter-province population mobility has increased in China today (Zhang, 2018), but the historically constructed divide between different regions has shaped place of origin into an important aspect of people's identities (Afridi et al., 2015). With reference to this regional identity, non-Henan users may easily accept the negative stereotype of Henan people because they do not consider people from the province as 'one of their own'. In the present incident, the referential strategy and juxtaposition technique employed by the editors at both news portals invoked this region-based self-identification of Chinese Internet users. It discursively encouraged non-Henan users, who already accepted the negative stereotype of Henan people, to share their biased evaluation of the characteristics of the entire Henan population.

Encouraged in the interaction
Certainly, the interactive dynamics of text production occurring in the commentary section of a news portal provides Internet users with the opportunity to challenge biased media discourse. In particular, discriminatory comments about Henan people can be offensive to those users who believe that their fellow citizens deserve fair treatment. While different Internet users may hold very different perspectives, one comment may snowball quickly, turning into a continuous and often heated debate among users. Figure 4 shows a cluster of Internet user comments retrieved from Tencent. User A-10 initiated the dialogue by suggesting that 'Henan fraudsters are most well-known (chuming)'. Deploying the adjective -'well-known', which is largely a neutral term in the Chinese language, user A-10's commentary was furnished with a seemingly unbiased stance. However, the user's textual production in the commentary section was contextualized against the regionally discriminatory news report. It, therefore, became an extension of the prejudicial media discourse, pointing towards a lexical association between Henan and fraudulence that is highly offensive to people from Henan Province. User A-10's assertion was immediately rejected by user B-10, who struck back with a personal attack against user A-10. The arguments continued as user A-10 verbally assaulted B-10 and her or his family in return.
Internet users are often unknown to each other prior to their encounters in a news portal commentary section. The anonymity determines that they are not completely bound by social norms to which people normally adhere, such as being polite to each other (boyd, 2014). Under these circumstances, the debates between Internet users can escalate aggressively, becoming a heated exchange of words due to one side's use of abusive language. Certainly, not all sampled comments involved using swear words. Some Internet users pointed out that the judgement of Henan people's characteristics could not be based on the fact that a small number of them were involved in isolated incidents (e.g. Tencent user C-7 in Figure 5). However, these comments were often marginalized by a throng of comments involving highly abusive language use.

Amplified by the IP-address function
Following a similar pre-structured affective process, an exchange of swear words was observed among the sampled user comments circulated on both Tencent and NetEase. However, it emerged from the CA process that many NetEase users' comments involve the use of words or phrases relating to regions outside of Henan. Scrutinizing the detailed use of these phrases, the analysis noted a locative sensitivity in the debates relating to the characteristics of Henan people. As previously mentioned, the NetEase commentary section includes a unique IP-address function, which is not supported on Tencent (Qian, 2011). This function allows each NetEase user to see the location from where their peers generate a comment. Figure 6 captured a representative scenario, in which Internet users' discursive practice was modulated by the locative nature of the IP-address function. According to the user location verified by the IP-address function, two of the commenters were based in Henan, while the other three were from the more-developed eastern regions (Beijing, Shanghai and Guangdong). In this cluster of comments, opposing groups were formed between commenters, who seemed to be, respectively, from Henan and eastern regions, following an attackand-counterattack pattern.
The commentary cycle was initiated by Guangdong-based user A-3, who emphasized that Henan is the 'keyword' for this fraud news. Based upon an incident involving Henan fraud gangs, such a discursive practice involved a predicational association, which implied a connection between frauds and Henan people in general. This discursive orientation of the comment was implicit but intentional. It diverted the focus of other Internet users from the news incident to the negative stereotype of Henan people, limiting the predictions and expectations of fellow commenters. A-3's comment received an immediate rejection from Henan-based user B-3, who argued that Guangdong Province has the highest crime rate in the country. B-3's claim was then dismissed by a seemingly Shanghai-based user -C-3, who called for scrutiny over the demographics of criminals, suggesting that most of the fraudsters captured in Guangdong were originally from Henan. Ironically, this call was exploited by user D-3, who appeared to be from Henan and claimed that it is Shanghai gangs, who are roaming the country and committing crimes everywhere.
The debates on the crime rate continued, as user E-3, who appeared to be from Beijing, joined in and supported others' discriminatory claims by asserting that Henan people were notorious across the country and their dishonesty was officially recognized by the police force. Such an assertion provoked Henan-based user D-3 to return to the 'battle'. D-3 retorted using abusive language, by describing people from Shanghai and Beijing, where C-3 and E-3 seemed to be from, as 'disgusting' and 'making people sick'. As a consequence, Henan-and eastern-region-based users were polarized into two camps. The interactions between the two camps followed a blockbuilding pattern, with comments overlaid first within, and then across groups. The abusive language was used as a response to the initial abusive language throughout the cycle. The arguments were, therefore, turned into an exchange of swear words, which detracted from the original debate.
In the debates on Henan people's characteristics, opposing groups often form as soon as the regional identity of the participants is invoked. Yet, the locative information, recognised by the IP-address function embedded in NetEase's commentary section, encourages these participants to pay attention to each other's current location. It reveals that the attentive affect of the locative information may open up new capacities for Internet users' action by affecting how they recognize each other. In the above scenario, while the Internet users, who appeared to be from the eastern region, circulated a series of biased opinions on the characteristics of the entire Henan population, seemingly Henan-based users were unified to counterattack in self-defence. In particular, the discourse strategy that these Henan-based users deployed was influenced by their opponents' locative information made available by the IP-address function. Rather than attacking individual characteristics, as often found in Tencent commentary section, NetEase's Henan-based users tended to generate similarly biased comments against the whole population of the cities or provinces where their opponents seemed to be from. This discursive practice was indeed affectively modulated by how their opponents' regional identity was visually represented. As the above analysis revealed, the IP-address function automatically locates NetEase users' geographical position by recognizing the IP addresses of the devices on which they access the Internet. NetEase users' current location is revealed as soon as they generate a comment. This IP-address function merely displays the place where a NetEase user's body is temporarily situated. Yet, the connotation of this locative information, as a symbolic sign, is shaped within Chinese society, where people's regional identity has been economically and sociopolitically constructed to be prominent. Displayed as the most notable digital sign of a user's presence in the commentary section, the attentive affectivity of the locative information is activated. This affectivity encourages Internet users to recognize each other through their geographic location displayed on their digital profiles. It provokes users to form regionally organized camps, potentially modulating how regional discrimination is practised in the commentary section. This affective process reflects how Internet users strategically seek information about each other, which, in turn, influences the interaction between them in mediated communication (Ramirez et al., 2002).
The location of Internet users detected by NetEase's IP-address function may not always accurately reflect their regional identity. In the data set, there were instances in which users appeared to be from Henan but claimed to be from other regions and circulated biased comments against Henan people. In addition, there were users, who seemed to be from eastern regions, but who proactively defended Henan people in their posting. The confusion in these instances was most likely caused by the increased inter-province mobility in Chinese society today. However, the analysis undeniably highlighted the particular attention to each other's locative information that NetEase users paid, whether or not this geographic location accurately reflected their regional identity. The locative sensitivity of NetEase users' discursive practice is manifest, causing the intensifying of the discriminatory practice against Henan people and leading Henan people to similarly use others' geographic locations as a discursive resource in opposing discrimination. This locative sensitivity has, to a certain extent, widened the divide between people coming from different provinces and holding different perspectives, as well as escalating regional discriminatory practice in a news portal's commentary section to an extreme. It provides a window to the relationship between the amplification of regional discrimination and the presence or absence of an IP-address function in the commentary section of a news portal.

Conclusion
This article has uncovered how the locative design of interactive digital platforms influences Internet users' discursive practice, highlighting the pivotal role the IP-address function plays in regional discrimination occurring in the commentary sections of Chinese news portals. Existing literature on regional discrimination tends to focus on biased media discourse following the paradigm of traditional CDA research (Zeng, 2012;Zhang, 2013). This scholarship often overlooks how technological architecture modulates Internet users' discursive practice. Different from the mass media, interactive digital platforms are not an 'ideological machine in a traditional sense' (KhosraviNik, 2018: 11). By allowing Internet users to generate original content, these platforms are automatically controlled by their design whereby human factors do not directly influence the regimentation (Papacharissi, 2015). With the technology-discourse entanglements in mind, the present study advances an affective CDA approach to articulating how regional discrimination is amplified on Chinese news portals in relation to the affective modulation of Internet users' discursive practice caused by their locative design. The research findings facilitate an understanding of the affective-discursive loop embedded in Internet users' postings.
The affective modulation of Internet users' discursive practice is not independent of sociopolitical contexts . The commentary sections of Chinese news portals are subject to the CCP's Internet censorship, evidenced by the fact that their launching companies are required by law to allow user-generated content circulated on the sites to undergo the government's 'computer security inspections' (NetEase, 2019: 18). Yet, the main purpose of the CCP's Internet censorship is to prevent collective protests against the Party-State (King et al., 2013), meaning that regional discrimination is not targeted by this censorship system. Under these circumstances, the amplification of regional discrimination concerning the design of NetEase's news portal may not be considered as independent of the high-tech company's profit-driven daily operation.
As Fuchs (2015) notes, Chinese high-tech companies have the autonomy in designing their interactive digital platforms, and their daily operation of the platforms is similar to that of their counterparts in Western capitalism. Something specific to NetEase is that its annual report reveals that a significant proportion of the company's revenue comes from news portal-based advertisings (NetEase, 2019). The profitability of the advertising business relies on the news portal's capacity for attracting Internet users' engagement. Echoing the analysis of web design by Ash et al. (2018), one may speculate that the incorporation of the IP-address function in NetEase's commentary section, which has evidently increased Internet users' postings on the news portal, represents a strategy that NetEase adopts to serve its profitability. As one of the company's web designers admitted in his blog, NetEase is indeed aware of the amplification of regional discrimination caused by the IP-address function of its commentary section (Qian, 2011). However, facing increasingly intensive market competition, NetEase shows no intention of effectively tackling this collateral consequence. The company is, in part, responsible for the escalation of regional discrimination on the Chinese Internet.
The present research was conducted in 2016 when news portals were the most important channel through which Chinese Internet users access news. However, China's social media ecology has been undergoing rapid change in recent years (Keane and Fung, 2018), with the emergence and spread of algorithm-controlled platforms having reshaped how the news publishing industry operates in the country (Schneider, 2018), despite news portals remain a popular information channel for many Internet users. 3 As such, the findings of this research should be considered as a snapshot of the technology-discourse entanglements established in regional discrimination in the early 2010s, rather than a grand narrative of the socio-technological complexities behind the phenomenon. Furthermore, without interacting with Tencent and NetEase users, the study does not fully account for the scenarios in which sampled comments may have been generated by ghost accounts or by Internet users currently living in places with which they do not identify. However, on a methodological note, this study makes important, original contributions to the application of CDA in the digital context by shedding new light on an affective-discursive loop reflected in Internet users' regional discriminatory practice. It has the potential for inspiring further research to examine the interplay between Internet users' postings and the technological architecture of interactive digital platforms on which the discursive practice is facilitated, as well as its potential impacts on discrimination that takes other forms, such as sexism and racism, in the digital age.