Who endorses online hate? The roles of ideology and knowledge in South Korean perceptions of anti-Chinese slurs

1.3.1 Social identity theory (SIT)

Social identity theory (SIT), first proposed by Tajfel and Turner in 1979, provides a lens for a basic understanding of the anti-Chinese sentiment. Tajfel (1972) defined social identity as an “individual’s knowledge that he belongs to certain social groups together with some emotional and value significance to him of this group membership” (Tajfel 1972, p. 292). Social groups provide their members with a “shared identity” which makes them evaluate themselves in the context of the group. Also, social identity is a critical aspect to make distinctions between ingroups from outgroups (Hogg 2016).

To differentiate ingroups from outgroups, members rely on “subjective belief structures,” which refers to “members’ beliefs about the nature of the relationship between their group and a specific outgroup” (Hogg 2016, p. 7). To apply this to the anti-Chinese sentiment and slurs observed so far, ingroup members tend to share their own subjective belief structures that contain negative perceptions and prejudice, to differentiate themselves from China or Chinese people. The reason many Korean people call Chinese people “jjang-ggae” may be associated with this strategy to differentiate Chinese people from Koreans. As a lot of online communities in South Korea are formed based on specific interests and tilted toward specific political ideologies, we can assume that users of online communities can experience and solidify their ingroup identity with other users who share similar interests and political ideologies.

Drawing from Social Identity Theory, the strength of an individual’s identification with a group is a key predictor of their willingness to engage in in-group-favoring and out-group-derogating behaviors (Ellemers et al. 2002; Tajfel and Turner 1979). Specifically, the theory posits a moderating role for the strength of group identification; individuals who are more strongly identified are more psychologically sensitive to intergroup dynamics and more motivated to protect the group’s positive distinctiveness, which is expected to amplify their reactions to intergroup events (Hogg 2016). In the context of online communities, this study operationalizes the theoretical idea of strength of identification (Scheepers and Ellemers 2019) through the concept of users’ “emotional attachment” to their community. Emotional attachment does not fully capture all facets of social identity, but it provides a measurable proxy for how strongly individuals identify with their online groups. Accordingly, emotional attachment is expected to moderate the relationship between online engagement and outgroup attitudes, leading to the following hypotheses:

1.3.2 The integrated threat theory

Together with social identity theory, this study explains the anti-outgroup sentiment in South Korea by utilizing the integrated threat theory of prejudice. According to the theory, there are four types of threats that cause prejudice: realistic threats, symbolic threats, intergroup anxiety, and negative stereotypes(Stephan and Stephan 2000).

Realistic Threats. Realistic threats refer to a danger to “the physical or material well-being of the ingroup or its members (e.g. their health),” which threaten “the very existence of the ingroup” (Stephan and Stephan 2000, p. 25). These threats include “perceived” threats. Perception of threat, regardless of the existence of the actual threat, can cause prejudice. In the current study’s context, China’s military, political, and economic expansion, and dominance, together with air pollution issues can be categorized into these threats for Koreans. Even if there is no evidence or actual behavior from China, what Koreans perceive as a threat can trigger prejudice against China and the Chinese people.

Symbolic Threats. Symbolic threats have an association with “perceived group differences in morale, values, standards, beliefs, and attitudes” (Stephan and Stephan 2000, p. 25). These pose “threats to the worldview of the ingroup,” and “any of the central values held by the ingroup.” As Ysseldyk et al. (2010) note, powerful belief systems, such as religion or political ideology, function as social identities; therefore, challenges to those beliefs, particularly when originating from a perceived out-group, are perceived not just as disagreements, but as threats to the self and the group. For Koreans, China’s attempt to revise history and distort Korean cultural heritages such as kimchi (traditional Korean food) or hanbok (traditional Korean outfit) can function as symbolic threats. Additionally, the political and ideological differences between the two countries can also be contributing factors to symbolic threats.

Intergroup Anxiety. South Koreans can feel threatened by China because “they are concerned about negative outcomes” (Stephan and Stephan 2000, p. 27) from intergroup interactions with China. This idea of intergroup anxiety posits that such levels of anxious expectation would be expressed explicitly and be associated with prejudice directly. South Koreans’ heightened anxiety about interacting with China or Chinese people and negative expectations about them can be categorized into this.

Negative Stereotypes. According to the theory, “almost all outgroup stereotypes embody threats to the ingroup” (Stephan and Stephan 2000, p. 27) since ingroup members expect the behaviors of the stereotyped group based on such stereotypes. Once the stereotypes are negative, the ingroup members would make a negative judgment accordingly. Koreans’ negative stereotypes about China and its people regardless of their actual experiences or interactions, solidified through online communities, can be explained by the negative stereotypes.

Based on the assumptions of the integrated threat theory, I posit that such phenomena of severe anti-Chinese sentiment and heavy use of slurs have a relation to the threat perception that Koreans possess. Shamoa-Nir and Razpurker-Apfeld (2020) found that subtle religious reminders (“primes”) increased Jewish-Israeli participants’ perception of threat from Muslims. This heightened sense of threat – not the prime itself – then directly predicted more negative attitudes toward the outgroup. Their findings are highly relevant as they suggest that the hostile content within online communities may function as a continuous “prime,” activating the same psychological process of threat perception to fuel intergroup animosity.

1.4.1 Political ideology

The relationship between political ideology and perceptions of outgroups in online spaces is complex and multifaceted. Rains et al. (2017) found that American conservatives exhibited less uncivil behavior when interacting with perceived in-group members in online discussions. This suggests that ideological alignment may influence the expression and perception of incivility. Gubitz (2022) observed that political ideology influenced sensitivity to incivility, with individuals showing greater sensitivity to uncivil comments targeting their in-group or co-partisans. This finding highlights the potential influence of political ideology on shaping perceptions of anti-outgroup content. However, Muddiman (2021) found no significant differences between Democrats and Republicans in their reactions to incivility, highlighting the complex nature of this relationship.

To add depth to these mixed findings, research in Western political psychology has sought to identify the underlying personality traits and motivations associated with ideology that predict prejudice. A prominent framework is the dual-process motivational model, which links conservatism to two key traits: Right-Wing Authoritarianism (RWA), a preference for social order and cohesion, and Social Dominance Orientation (SDO), a preference for group-based hierarchy. Duckitt and Sibley (2009) demonstrate that the perception of the world typically activates RWA as a dangerous place (predicting prejudice against threatening groups), while a view of the world activates SDO as a competitive jungle (predicting prejudice against groups seen as inferior or competing for resources). This provides a deeper psychological mechanism for why conservatism may be linked to negative outgroup attitudes, connecting ideological labels to fundamental motivations regarding group status and threat.

Adding further complexity, recent research highlights how psychological nuance and context can moderate these ideological responses. For instance, Borinca et al. (2022) found that while conservatives may engage in the dehumanization of immigrants, they can nonetheless respond positively to interventions like imagined intergroup contact, especially when emotions are engaged. This suggests that ideological rigidity does not necessarily block responsiveness to interventions. Conversely, intergroup emotions can be counterintuitive in crisis contexts. Borinca et al. (2023) found that even individuals low in prejudice may respond with negative emotions to an outgroup offering help during a crisis, complicating the simplistic expectation that perceived threat and ideology jointly shape intergroup emotions. These studies underscore the value of examining not only general sentiment but also the perception of specific, emotionally charged content, such as slurs, as these may reveal the nuanced ways in which ideology operates.

Based on the theoretical expectation that conservatism is linked to a greater sensitivity to out-group threats, yet acknowledging the contextual nuances highlighted by the literature, this study proposes the following hypotheses to test these relationships in the South Korean context.

1.4.2 Political knowledge

Political knowledge has been shown to play a crucial role in shaping online behavior and attitudes towards outgroups. Carpini and Keeter (1996) argued that political knowledge is fundamental to the functioning of democracy, influencing citizens’ ability to form coherent and stable opinions on political matters. This suggests that individuals with higher levels of political knowledge may be better equipped to critically evaluate anti-outgroup content.

Jordan (2024) found that lower levels of political knowledge were associated with a higher likelihood of political extremism. This finding implies that individuals with less political knowledge may be more susceptible to anti-outgroup narratives and less able to critically assess such content. Zaller (1992) proposed that political awareness, closely related to political knowledge, influences how individuals receive and process political information. This theory suggests that those with higher levels of political knowledge may be more discerning in their consumption and interpretation of anti-outgroup content in online communities.

However, research on “motivated reasoning” in the U.S. adds an important layer of nuance to this relationship. While knowledge can foster tolerance, it can also provide highly partisan individuals with the cognitive tools to more effectively defend their pre-existing biases. For instance, Taber and Lodge (2006) found that politically knowledgeable participants were more likely than less knowledgeable participants to counterargue information that contradicted their beliefs and to accept information that supported them uncritically. This suggests that political knowledge is not a neutral tool, but it can be used as a weapon to protect one’s partisan identity. This adds a crucial comparative perspective, suggesting that while knowledge may reduce prejudice based on simple stereotypes, its effect could be different in highly polarized, identity-based conflicts.

Based on the literature suggesting that political knowledge fosters more tolerant and stable attitudes yet acknowledging the nuance that it can also fuel motivated reasoning, this study proposes the following hypotheses to test the primary expected relationship in the South Korean context.

1.4.3 Education

While the relationship between education and anti-outgroup attitudes is not always consistent, previous research has suggested links between education and online behavior. Vargo and Hopp (2017) found that lower levels of education were associated with a higher likelihood of engaging in online incivility. Min and Shen (2023) observed that individuals with higher levels of education tend to be more discerning and sensitive about incivility. These findings highlight the complex relationship between education, political knowledge, and online behavior, warranting further investigation in the context of outgroup perceptions.

Also, individuals with higher levels of education are less likely to believe in conspiracy theories (Duplaga 2020; van Prooijen 2017). These findings imply that people with higher levels of education are more discerning and sensitive about incivility (Min and Shen 2023) and they are more capable of critical thinking than less educated (McPeck 2016).

A foundational study on intergroup attitudes in Western contexts by Quillian (1995) offers a detailed examination of the role of education. In his analysis of prejudice across 12 European countries, education was treated as a key individual-level independent variable, measured in years of schooling. The study’s results confirmed a statistically significant negative relationship. Respondents with higher levels of education expressed less racial and anti-immigrant prejudice. The study also found that the prejudice-reducing effect of education was more substantial in high-threat contexts (i.e., countries with a higher percentage of non-EEC immigrants). This suggests that while education is generally associated with greater tolerance, its role as a buffer against prejudice may become even more important when intergroup tensions are high.

Drawing on established literature that suggests education can act as a buffer against prejudice, particularly in high-threat situations, this study proposes the following hypotheses to test this relationship in the South Korean online context.

2.6.1 Engagement in the online communities

At the beginning of the survey, respondents were asked whether they use online communities or not. The survey continued only with the respondents who use online communities. They were requested to indicate which online communities they are engaging with. The frequency of using online communities was measured by asking how much they use online communities: more than once a day; once every 2 days; once every 3–4 days; once a week; and once every 10 days or less. The respondents’ participation was measured by asking how often they read posts, reply to posts, press ‘like’, share posts to others, post my own, and meet other users offline on a 7-point Likert scale (never to always). The alpha was 0.85. The two items converged to create an engagement scale (M = 20.87, SD = 7.25).

2.6.2 Political ideology

Political ideology was measured by asking Which of the following do you think is your political orientation? on a 7-point Likert scale from 1 (Liberal) to 7 (Conservative). Most people indicated their political position toward the center (M = 4.04, SD = 1.20).

2.6.3 Political knowledge

Political knowledge was measured using four items: How long is the term of office of the National Assembly of Korea? How long is the term of office of the Korean president? Please select which government branches are of ‘separation of powers’ and How many seats are currently in the National Assembly of Korea? The total number of right answers was coded to create the political knowledge scale (M = 3.07, SD = 0.90).

2.6.4 Education

The respondents were asked to indicate their level of education: elementary school graduate; middle school graduate; high school graduate; college graduate; and graduate school or above. 70 % and 11 % of the respondents reported they are respectively college graduates (n = 210) and graduate school or above (n = 33), consisting 81 % of the total respondents. 18.7 % (n = 56) graduated from high school.

2.7.1 Anti-outgroup sentiment

The feeling thermometer is a widely used tool to capture “respondents’ feelings about a given person, group, or issue” (Lavrakas 2008, p. 276). In this study, the feeling thermometer of 0 (very cold) to 100 (very warm) was used to measure how the respondents think about China and Chinese people. Also, the respondents were asked to evaluate their own country – South Korea – and three countries that are politically close to them – North Korea, the United States, and Japan – and their people to compare the results. The respondents reported the highest feeling thermometer towards their ingroup, South Korea, and its people (M = 66.97, SD = 22.37). The United States was the second favorite (M = 59.81, SD = 18.45), followed by Japan (M = 44.50, SD = 23.33) and North Korea (M = 40.30, SD = 24.90). The feeling thermometer of China was the coldest (M = 32.53, SD = 24.40), less than half of South Korea’s thermometer, indicating that China is the farthest outgroup to South Koreans.

In addition, they were asked to evaluate those countries and the people by using eight trait adjectives: violent (reverse coded); dishonest (reverse coded); unintelligent (reverse coded); friendly; arrogant (reverse coded); kind; and inferior (reverse coded), using a 7-point Likert scale, 1 (never) to 7 (extremely) (Velasco González et al. 2008). South Korea scored the highest evaluation (α = 0.86, M = 4.47, SD = 0.99), followed by the US (α = 0.78, M = 4.00, SD = 0.79), Japan (α = 0.86, M = 3.97, SD = 1.04), North Korea (α = 0.86, M = 3.40, SD = 0.1.05). China obtained the lowest score (α = 0.86, M = 2.96, SD = 0.97).

Each country’s feeling thermometer score was converted by multiplying 0.07, then converged with the evaluation score to create a general sentiment scale. South Korea scored 4.57 (SD = 1.07), followed by the United States (M = 4.10, SD = 0.89), Japan (M = 3.54, SD = 1.18), and North Korea (M = 3.10, SD = 1.22). China scored the lowest, 2.62 (SD = 1.18).

2.7.2 Consumption of slurs against the outgroup

The existing slurs about China, “jjang-ggae (짱깨)”; “The only good Chinese is a dead Chinese (착짱죽짱/착한 짱깨는 죽은 짱깨)”; and “Nanjing Grand Festival (난징대축제)” were presented to the respondents. The respondents were asked whether they had seen these slurs in online communities and whether they had any experience in using them in person. 71.3 % of the respondents reported that they have seen the slurs (n = 214) and 10 % of the respondents reported that they have used such slurs (n = 30).

2.7.3 Perceptions of slurs against the outgroup

The respondents selected whatever they felt when they saw the slurs: nothing; pleasure; unpleasant; funny; and sorry. The respondents’ positive perceptions of the anti-Chinese slurs (pleasure and funny) and negative perceptions of the slurs (unpleasant and sorry) were coded separately for analysis. 64.3 % perceived the slurs as negative (n = 193) and 13.7 % perceived the slurs as positive (n = 41).

2.8.1 Threat perception

The respondents were asked to pick one country that poses the greatest threat to South Korea among four countries, China, North Korea, Japan, and the United States. A large number of people selected North Korea (41.3 %, n = 124), and China (35.3 %, n = 106). 19.7 % of people selected Japan (n = 59) and only 3 % perceived the United States as the greatest threat (n = 9).

For those who selected China as the greatest threat, two additional questions were given. One question asked which area is mostly being threatened by China: security; economy; environment; democracy; culture; and history. Another question asked about the degree of the threat to each area, using a 7-point Likert scale (none to extremely). 42.5 % of people who perceive China as the greatest threat selected the economy as the most threatened area (n = 45). Security (24.5 %, n = 26) and environment (17 %, n = 18) followed, indicating that the realistic threat is the most salient threat to Koreans.

2.8.2 Emotional attachment to the online communities

First, the respondents reported how much they feel a sense of belonging on a 7-point Likert scale, 1 (none) to 7 (extremely). Their mean sense of belonging was 4.13 (SD = 1.47).

Their feeling thermometer toward users of the online communities was measured from 0 (very cold) to 100 (very warm). Their mean feeling thermometer was 56.38 (SD = 21.80).

Their evaluations of other users of the communities were measured on a 7-point Likert scale by asking how much the users are; trustworthy; respectable; useful; friendly; and (have) a lot to learn from. The five items converged to create an emotional attachment scale (α = 0.90, M = 4.37, SD = 0.97).

The feeling thermometer score was converted by multiplying 0.07, then converged with the sense of belonging score and evaluation score to create a general emotional attachment scale (M = 4.15, SD = 1.12).

3.3.1 Model 1: predicting anti-Chinese sentiment

Hypothesis 5a (H5a) predicted that a lower level of political knowledge would be associated with higher anti-outgroup sentiment. As shown in Table 1, the results indicate that political knowledge was a significant negative predictor of anti-outgroup sentiment (β = −0.17, p = 0.005). This means that as political knowledge increased, anti-outgroup sentiment toward China decreased. Therefore, H5a was supported. Hypothesis 6a (H6a) predicted that a lower level of education would be associated with higher anti-outgroup sentiment. As shown in Table 1, education was not a statistically significant predictor of anti-outgroup sentiment when controlling for other variables (β = 0.05, p = 0.431). Therefore, H6a was not supported.

3.3.2 Model 2: predicting positive slur perception

Hypothesis 5b (H5b) predicted that a lower level of political knowledge would be associated with more positive perceptions of anti-outgroup slurs. The results in Table 1 show that political knowledge was a significant negative predictor of positive slur perception (β = −0.16, p = 0.008). As political knowledge increased, the likelihood of perceiving slurs positively decreased. H5b was supported. Hypothesis 6b (H6b) predicted that a lower level of education would be associated with more positive perceptions of anti-outgroup slurs. As shown in Table 1, education was not a significant predictor of positive slur perception (β = −0.03, p = 0.683).