Abstract
The COVID-19 pandemic has fueled heated discussions and intensified polarization in the exchange of opinions. Due to strict government restrictions, including a hard lockdown, much of this debate took place online. In this study, we examine how Czechs engaged in political discussions on Facebook during the pandemic. This research looks into the occurrence of incivility and intolerance and the similarities/contrasts with normatively desirable discursive features. We also focus on the perception of differences between various types of antinormative expressions among participants in online discussions and their roles. We use a mixed-method approach: a quantitative content analysis of comments on the Facebook pages of two of the Czech Republic’s most popular news media outlets (i.e., public service television, ČT24, and commercial television, TN.CZ) and two top Czech politicians (Andrej Babiš and Tomio Okamura); and qualitative interviews with 20 participants collected during the Springs of 2021. Data were downloaded by Facepager during the hard lockdown in the Czech Republic (March and April 2021), and the final sample consisted of 1,792 comments. We reveal that incivility, when expressed with a justified opinion, was less likely to directly attack other commenters within the thread. We also observe differences in the amount of incivility in politicians’ pages compared to the news media. Incivility has increased over time since the hard lockdown started. We address the implications of different conditions in which incivility and intolerance occurred and their impact on deliberative democracy.
Introduction
Antinormative expressions (i.e., incivility, intolerance, and hate speech) are often referred to mainly as threats to democracy and as factors in the declining quality of public discussions online (Friess et al., 2020; Papacharissi, 2004), either because they threaten the possibility of consensus or because they decrease participants’ willingness to participate (Yamamoto, 2020). Another concern raised about the impact of uncivil online communication relates to the reinforcement of uncivil behavior and its acceptance (Hmielowski et al., 2014; Rösner et al., 2016), which may lead to opinion polarization. On the other hand, scholars have argued that uncivil expressions can still beneficially contribute to the online public sphere and enrich the perspectives of people with diverse opinions despite the antagonistic elements of the communication style (Chen, 2017; Rossini, 2022). One of the reasons for this discrepancy might be the different types of incivility and the lack of conceptual consistency (Friess et al., 2020), raising different expectations and leading to various results.
This study contributes to this research agenda by building on the conceptual distinction between incivility and intolerance (Rossini, 2022). This approach allows us to differentiate between expressions that are inherently threatening to democratic values and those that are not—even though they may be perceived as offensive. Whereas incivility might be seen as an “antinormative” way of expressing personal opinions, intolerance is inherently problematic because it refers to forms of expressions that violate democratic values, such as inclusion and equality, by undermining individuals and groups, or denying their value in society (Rossini, 2022). Leveraging this approach, we examine the extent to which different types of antinormative discourse are prevalent in Czech discussions on Facebook.
Prior research has focused on either diagnosing how much antinormative discourse is present in online debates (Vergani et al., 2022), or understanding how people perceive different types of uncivil and intolerant discourse (Bormann, 2022; Kenski et al., 2020). In this study, we leverage a mixed-method approach, combining quantitative content analysis of Facebook data and qualitative interviews, to provide a nuanced account of both the prevalence of antinormative discourse and the perceptions of these expressions by people who frequently engage in online discussions, with a focus on online discussions on Facebook in the Czech Republic during the COVID-19 pandemic in 2021.
The content analysis allows us to examine the prevalence of incivility and intolerance, as well as their relationship with desirable discussion features such as disagreement and justified opinion expression. Through conducting interviews, we gain insight into the perspective of people who frequently engage in online discussions. This includes their perceptions of expressions as uncivil and how such perceptions affect their evaluation of other participants in discussions. The findings obtained through this analysis allow us to consider the differences between how researchers problematize antinormative discourse and the ways it affects people’s experiences with online discussions.
Furthermore, we examine incivility and intolerance in the context of the Czech Republic, which represents an interesting case because of the government’s controversial response to COVID-19. Despite early-stage measures, the situation rapidly changed during the Autumn of 2020 and the start of 2021. Chaotic government measures and the lack of hospital capacity were followed by the strictest measure to date: a hard lockdown in Spring 2021, which is the period upon which we focus. The Czech Republic quickly twisted from being “best in COVID” within the early stage of the pandemic to being “worst in COVID” during the subsequent waves (see Kaniok, 2023).
The contributions of this research are as follows. First, it contributes to a growing body of research that is focused on understanding the role of antinormative expression in political communication (Chen, 2017; Muddiman, 2017; Rossini, 2022) in an under-researched area of Central and East Europe (CEE). The Czech Republic is a post-communist country, that had limited access to the Western world. The prevalence of incivility and intolerance might differ from Western countries for a few reasons. For instance, Czechia has lower tolerance of immigrants (Prokop, 2019), which might mean that xenophobic or racist comments could be more prevalent than scholars have previously found studying the West. Second, the context of a global pandemic might also affect how people express themselves, as well as how they may engage with antinormative discourse. The controversies around the government’s handling of the pandemic and the politicization of the public health crises might be associated with emotional reactions amid restricted freedoms and economic uncertainty. Third, our mixed-methods approach allows us to consider the implications of uncivil and intolerant expressions from the standpoint of those who engage in Facebook discussion arenas. Combining qualitative and quantitative data is unique within this research agenda, and helps us close blind spots and expand knowledge of previous studies using either qualitative or quantitative methodology, but rarely both.
Incivility and Intolerance and Cross-Cutting Discussions
Competing views on the role and implications of uncivil or hostile discussions for democracy are rooted in different conceptualizations (Chen et al., 2019), which may cause discrepancies among comparisons across study results. Many authors approach incivility as a violation of the expectations of discussion norms in specific cultural contexts (Chen, 2017; Mutz, 2006). However, there is variation in what is considered uncivil. For instance, Chen (2017) includes in the definition of incivility profanity, insulting language, name-calling, and hate speech. To capture various types and expressions of incivility, some studies (Coe et al., 2014; Kenski et al., 2020; Rossini, 2022) also included additional attributes, such as lying, accusations, pejorative speech, and aspersions. While the attributes of incivility may vary, some scholars have argued for a distinction between incivility—more broadly understood under the lenses of politeness violations—and forms of expression that are democratically undesirable or threatening (Muddiman, 2017; Papacharissi, 2004; Rossini, 2022). In particular, there is an increasing tendency to distinguish between less and more harmful types of antinormative expressions, such as intolerance and hate speech (see Karunungan, 2023; Oz & Nurumov, 2022; Rossini, 2022). Rossini (2022) proposes a conceptualization that treats incivility as a matter of tone that follows a “politeness” standpoint, while intolerance refers to expressions that are inherently problematic in their substance. As such, expressions of intolerance directly threaten equality in society, attack people’s rights, and discriminate against particular individuals and groups. This approach allows us to understand the conditions under which different types of antinormative expressions emerge in online discussions, as well as the extent to which their occurrence may undermine democratic values.
The debate about incivility is part of broader debates about the value of cross-cutting discussions. From a deliberative standpoint, heterogeneous debates must be rooted in the consideration of, and respect toward, opposing views (Mutz, 2006). From this perspective, aggressive communication that disrespects or disregards opposing views is seen as uncivil. Despite the normative expectations about how disagreement should be expressed, scholars have highlighted that respectful discussions based on various points of view are relatively rare (Mutz, 2006). Intolerance, on the contrary, appears to emerge in more homogeneous opinion environments (Wojcieszak, 2010). Without diverse perspectives, participants in homogeneous opinion enclaves may become more extreme in their views, facilitating intolerance (Rossini, 2022). Considering that prior research has suggested that incivility and intolerance are associated with distinctive contextual settings, at least in terms of diversity of opinions, we investigate whether these relationships hold in the context of Facebook discussions about the COVID-19 pandemic in the Czech Republic. We formulate the following hypotheses:
Understanding the pervasiveness of antinormative expressions in cross-cutting discussions is particularly important in the context of a polarized issue because incivility may contribute to increased negative perceptions, potentially leading to more negative emotions and opinion polarization (see Hiaeshutter-Rice & Hawkins, 2022; Hwang et al., 2016). Research on the perceptions of incivility has primarily used surveys (Kenski et al., 2020; Mutz, 2015) or experiments (Hwang et al., 2016; Muddiman, 2017) to understand how participants rate different types of expressions and have considered how demographics, attitudes, and behaviors may explain these perceptions. However, this body of research provides limited insight into the consequences of those perceptions, as well as into people’s experiences with antinormative discourse, more broadly. This may be because experiences with incivility are challenging to measure using quantitative approaches. To shed light on how people who engage in online discussions experience and react to antinormative expressions in cross-cutting discussions, we ask:
Determinants of Incivility
We examined two crucial characteristics that potentially determine the role of incivility in conversations, namely justified opinion expression and the target of uncivil comments. Despite broad normative claims that incivility is not compatible with productive political conversation, scholars have argued that incivility is often used as a form of political expression (Herbst, 2010; Rossini, 2022). The acknowledgment that incivility is related to desirable discursive traits has led scholars to argue that incivility may serve an important role in enabling political communication (Chen et al., 2019; Rossini, 2022). One of the key characteristics of deliberation is reasoned opinion expression, that is when participants provide reasons to justify their perspectives (Rossini, 2022; Stromer-Galley, 2007). Justified opinion expression is important because it fosters awareness and understanding between the proponents of opposing views (Rossini, 2022). As such, online discussions may allow participants and bystanders to gain awareness of arguments associated with different perspectives (see also Chen et al., 2019).
We probe this relationship in the context of the pandemic, as online debates during periods of crises (e.g., measles epidemic) may foster emotional polarization and a lack of willingness to convince the other side, and well-reasoned comments may be less frequent (Vochocová et al., 2022). Based on the findings of Rossini’s (2022) study, it is anticipated that in a homogeneous environment, individuals may not be willing to persuade others with differing opinions by justifying their own views. As a result, we formulate the following hypotheses.
Despite the assumption that justified opinions are important even if they are expressed with uncivil tone, we explore how participants in online discussions interpret these expressions, and whether they value antinormative opinion justification in the context of disagreement. To capture the wider context of uncivil divergent opinion exchange from a qualitative standpoint, we ask:
Incivility that targets personal attributes and character is generally perceived as more uncivil than attacks toward political arguments (Muddiman, 2017). However, there is limited insight into how antinormative discourse may focus on different targets online (see Oz & Nurumov, 2022; Rossini, 2022). Considering whether incivility and intolerance target other participants is important because this may also affect the dynamic of discussions. For instance, positive reactions (e.g., “up votes”) to uncivil comments may trigger other uncivil comments (Shmargad et al., 2022). This might also be the reason why intolerance is most likely to be spread in a homogeneous opinion setting (Rossini, 2022), as the risk of disapproval is lower.
While prior work has not found consistent associations between expressions of incivility and interpersonal targets, with some suggesting a negative relationship (Rossini, 2022) and others with insignificant results (Oz & Nurumov, 2022), it is possible that the heightened polarization around COVID-19 affects such dynamics. In particular, deeply divided opinions about the government response may lead citizens to be less accepting of other views (Kaniok, 2023), and more willing to react with incivility. This is important because interpersonal incivility may discourage further participation in online discussions (see Kenski et al., 2020). In this context, we ask:
Prior research suggests that antinormative expressions are perceived differently based on individual-level factors, including personal characteristics, such as gender, conflict orientation, and political preferences (Bormann, 2022; Kenski et al., 2020). This makes it challenging to unequivocally determine its effects (Chen, 2017; Herbst, 2010; Kenski et al., 2020). Although perceptions of incivility from participants in online discussions may differ, there is an agreement that extreme expressions, for example, violent threats or wishing death upon someone, are considered serious violations of discussion norms (Bormann, 2022). However, existing research has not considered the extent to which antinormative discussions may shape participants’ experiences. To fill this gap, we explore the conditions under which participants perceive certain expressions to be uncivil or intolerant and how it is related to their discussion experience.
Online Discussions During the COVID-19 Pandemic
For two years, the COVID-19 pandemic was one of the most controversial topics in the Czech Republic, similar to other countries (Eisele et al., 2022). Heated reactions and polarization are frequent responses to societal and health crises, and COVID-19 was no exception (see Hiaeshutter-Rice & Hawkins, 2022; Vochocová et al., 2022). The unpredictable character of the pandemic and its impact on several spheres of everyday life, in combination with chaotic responses from the government, fueled the spread of disinformation, hate speech, and personal attacks in online environments (see Malova, 2021; Vergani et al., 2022). During this period, social media became an important venue for public discussion. People shared their opinions on a range of topics, from health protection, governmental action, and vaccination safety (see Malova, 2021). These controversies created fertile ground for heated arguments between vaxxers and anti-vaxxers, and between those who supported government restrictions and those who rejected them.
At the end of March 2021, the Czech government announced a “hard lockdown.” People could not cross the borders of national subregions without permission. The measure was meant to last until the end of March but it was prolonged until 10 April due to the increasing number of infected people and the number of hospitalized people. This period was characterized by chaotic governmental decisions and ongoing instability in the Ministry of Health (Government of the Czech Republic, 2020; Kaniok, 2023). We might expect increasing dissatisfaction about the situation and a thus higher level of both incivility and intolerance among comments over time. Thus, we test:
Different Places for Political Discussions
Social media allows for multiple discussion environments due to its affordances, like pages and groups, but few studies adopting a comparative stance have mainly focused on different pages related to the same actor. Research has primarily focused on news organizations, or politicians, but rarely both (see Humprecht et al., 2020; Oz & Nurumov, 2022). Previous studies suggest that different Facebook pages may affect discursive characteristics, such as the level of antinormative expressions (Humprecht et al., 2020; Su et al., 2018). Likewise, specific communication strategies generated by political representatives, such as incivility (Rega & Marchetti, 2021), and the particular topics of the news articles (Coe et al., 2014), may consequently lead to more heated comment exchanges, including uncivil expressions. Moreover, data shows that in the case of news media outlets, especially right-wing media trigger more antinormative expressions (Humprecht et al., 2020; Su et al., 2018).
We can assume that pages managed by distinctive actors may foster discussions with different levels of incivility for a number of reasons, including perceived credibility, the actor issue-stance and own communicative style, as well as distinct approaches to comment moderation—as page administrators can decide their own moderation approaches on Facebook. To assess the role of distinctive discussion arenas during the pandemic, we intentionally focused on key actors and their public profiles on Facebook: media—ČT24 (public service medium) and TN.CZ (commercial medium); and politicians—Andrej Babiš (prime minister at the time, populist, center to center-right) and Tomio Okamura (government opposition, populist, right-wing extremism). Comparison between government and opposition was motivated by the different approaches toward measures that the opposition party often criticized. Thus, we ask:
Methodology
This study is built on a mixed-method approach that combines semi-structured interviews and a quantitative content analysis to gather a more nuanced understanding of how incivility and intolerance are expressed, and how they may affect those who actively participate in online discussions.
Study 1: Incivility and Intolerance in Comments About COVID-19 on Facebook—Content Analysis
Data for this study were collected from four Facebook pages: the two most popular Czech news media Facebook pages– a public service television channel (ČT24, 749,399 followers 1 ) and a commercial television channel (TN.CZ, 328,165 followers) — and two top Czech politicians—Andrej Babiš (prime minister at the time, 272,957 followers) and Tomio Okamura (leader of an opposition party, 274,040 followers). Data was downloaded using Facepager (Jünger & Keyling, 2019) during a period of hard lockdown in 2021 in the Czech Republic, from the 1 March 2021 to 10 April 2021.
The initial data collection yielded a total of 2,436 posts. 2 Posts were categorized based on whether they contained information about COVID-19 (N = 1,110). Then, 70 posts per Facebook page were randomly selected. They were, first sorted according to reactions and comments, from highest to lowest, (N = 825,047). Given the high number of comments, we implemented stratified random sampling to proportionally represent comments that yielded different levels of engagement (i.e., the sum of total interactions: likes and comments). For every Facebook post, comments were first sorted from highest to lowest reactions. Then, we proportionally divided each set of comments into seven sections. After this, we randomly chose one comment from each section. This resulted in seven comments per post for 70 Facebook posts for each actor (490 total comments per actor). This created a subsample of 1,960 comments. Finally, comments were coded by relevance (N = 1,792 relevant comments). We excluded irrelevant comments that contained spam or advertisements and comments that included only name tags, one-word expressions, interjections, emojis, and words without any meaning.
We constructed two dependent variables — incivility and intolerance — and five independent variables: Facebook pages, comment date, target interaction, disagreement, and argument opinions. The control variables were established as the number of interactions (i.e., likes and comments on each comment). Intercoder agreement was measured with approximately 7% of the sample. Due to the low occurrence of intolerance in the sample, reliability was tested with approximately 12% of the sample. 3 Two trained coders coded the data. Reliability was measured using Krippendorff’s Alpha (Table 1).
Reliability test for dependent and independent variables.
We adapted the codebook proposed by Rossini (2022) 4 for the Czech context. All independent and dependent variables were inspired by her study. In addition to the content analysis categories, we include comment date as an independent variable to account for the timing of the discussions. This variable was constructed based on the original comment date (retrieved from Facepager), as a count variable as days go since lockdown.
Codebook Categories
Incivility
Following Rossini (2022) and Coe et al. (2014), expressions of incivility include the use of profane or vulgar language, personal attacks, lying and aspersions, and attacks against arguments or perspectives (see examples in Supplementary Material). While these subtypes were originally coded separately, they were combined under a binary measure of incivility due to the low occurrence of particular subtypes and low reliability across distinguishing between these sub-categories.
Intolerance
We adopted Rossini’s (2022) definition of intolerance as “discourse that threatens democratic pluralism and values” (p. 6). Coders distinguished between the following sub-categories of intolerance: xenophobia, racism, hate speech, violent threats, religious intolerance, and offensive stereotyping, and attacks against sexual preferences and gender, and economic status (see examples in Supplementary Material). All sub-categories were conflated to create a binary measure of intolerance for analysis.
Target of Interaction
This category captures whether there is a clear dialogical interaction among commenters. It is coded using three mutually exclusive sub-categories: (1) Replies meant to capture comments that replied to another person within the thread. This did not always have to include the tagged name of the previous commenter. (2) Reaction to Facebook posts used in cases where there is clear evidence for a connection between the comment and Facebook post (i.e., text, video, and image). (3) Generic opinion expression was coded for questionable examples (e.g., “what are you saying you did not experience this period”) or if commenters expressed their opinion without tagging someone else. Those expressions are unrelated to Facebook posts (e.g., “politicians are corrupt”).
Disagreement
This variable captures whether there was explicit disagreement in a discussion. It considers only comments that directly reply to others (see “Target of Interaction”). This includes disagreement expressed politely, and potentially includes an acknowledgment of other arguments (e.g., “you are right, but . . .”). It is a binary category (i.e., 0—an absence of disagreement, 1—the presence of disagreement).
Opinion Expression and Argumentation
This variable distinguishes between different levels of argument development. Following prior studies (Rossini, 2022; Stromer-Galley, 2007), this variable does not consider the quality of the argumentation. Rather, it focuses on the presence of different types of arguments. It was coded as follows: 0—no opinion expression, which refers to neutral comments that do not include any opinion toward other discussants, topics, Facebook posts, or the discussion itself (e.g., “was it voluntary to be part of StB” 5 ?); 1– simple opinion expression, which refers to opinions (e.g., “vaccination is bad for our body”) that are not justified; and 2—justified opinion expression, which presents an elaboration or justification (e.g., “vaccination is bad for our body as it might impact our immune system fatally”). The explanation could be based on motivations, examples, stories, or analogies.
Study 2: Users’ Perspective on Incivility and Intolerance in Discussions About COVID-19 on Facebook—Semi-Structured Interviews
We conducted semi-structured interviews in Spring (March-April) 2021. The trainer interviewers compiled data from participants based on the following selection criteria: more or less active engagement in discussions on Facebook and participation in online discussions about the COVID-19 crisis topic in discussions. We implemented a snowball sampling method approach to get contacts for participants. Sampling for qualitative study has no overlap with the content analysis. All participants signed an informed consent form before participating in an online or face-to-face recorded interview, which lasted approximately 60 minutes. While looking for participants, we strived for variability within the sociodemographic characteristics. The sample also varied within levels of engagement in online discussions as participants with various frequencies of engagement in online debates were involved. In the final sample of 20 participants, ages ranged from 24 to 56, and we captured different participants according to their education level, household, marital status, residence (size of cities/villages), and gender (female: 45%).
The interview script included the following three thematic sections: (1) usage of social media with a focus on Facebook, including the frequency of usage and motivations; (2) online discussions, including engagement, motivations to participate, and barriers in participation; and (3) personalization and the environment on social media, which focused on cross-cutting discussions, selective activities (e.g., unfriending), and experience with incivility and intolerance online. Though interviews included rich information about online discussions and social media usage, we intentionally focused on statements about various types of incivility, disagreement, argumentation within the online environment, and the perception of other discussants.
After data collection, interviews were recorded, transcribed, anonymized, and analyzed by four trained coders (including, and led by, the first author). Intercoder reliability was reached through several annual meetings and discussions during the ongoing coding process and analysis. For the purpose of coding and thematic data analysis (Braun & Clarke, 2006), we used ATLAS.ti. The main themes connected to engagement in online discussions were coded and reviewed after the initial phase. In the second phase, we intentionally focused on the themes related to experiences with incivility and intolerance in various contexts, participation in counter-attitudinal opinion exchanges, the perception of other participants in Facebook discussions, and their prescribed characteristics.
Results
Study 1: Incivility and Intolerance in Comments About COVID-19 on Facebook—Content Analysis
Before answering the research questions and testing the hypothesis, we present the descriptive information (see Figure 1). First, in the sample of 1,792 Facebook comments, uncivil comments dominate (62.6%), and intolerant ones are the minority (8.1%) (see Figure 1). Second, comments with simple opinion expressions prevail and make up more than half of the comments. At the same time, only 4.9% of comments do not include any opinion expression. Third, we observe that disagreement and the absence of disagreement are split almost evenly within comment replies. It is important to highlight that disagreement was coded only for comment replies, which means that the comment was directed toward a particular person within the comment thread; thus, the subsample includes 697 Facebook comments.
The distribution of incivility and intolerance in justified opinion expression (n = 1,792) and disagreement (n = 697).
To test hypotheses 1-5, we ran two binary logistic regressions (see Table 2) with incivility and intolerance as dependent variables. The model for incivility was explained satisfyingly by the study variables (Nagelkerke R = 0.20). For intolerance, the explanatory power was lower (Nagelkerke R = 0.10). We found a positive relationship between incivility and disagreement (H1). Comments including disagreement are more than two times more likely to include incivility. This is the strongest predictor of incivility in the model. Considering that disagreement was only coded in comment replies, this suggests that people tend to be more uncivil when they react to comments that are not in line with their opinions. Conversely, intolerance (H2) has a negative, but not significant, association with disagreement.
Incivility and intolerance: characteristics of their occurrence (n = 1,792).
OR: odds ratio; CI: confidence interval.
p ⩽ .05, **p ⩽ .01, ***p ⩽ .001.
Note. The effect for intolerance and the independent variable Opinion Expression and Argumentation at the “no opinion” level is 0 due to the absence of cases.
Turning our attention to comment replies (i.e., when participants directly respond to another commenter), we find that they are negatively associated with both incivility and intolerance (RQ3), indicating that antinormative expressions are mainly directed toward the Facebook post or expressed in generic opinions, and less used when participants engage with others in a discussion. Confirming H3, the connection between justified opinion expression and incivility is positive. Contrary to our expectations, intolerance turns out to be also positively related to justified opinion expression (H4). We find partial support for H5 as there was an increase in uncivil comments as they went by since the start of the hard lockdown, but for intolerance, we did not find significant results. The distribution of uncivil and intolerant comments can be seen in Figure 2.
The distribution of incivility and intolerance during March (N = 1,423) and April 2021 (N = 369).
Finally, we found that political pages were associated with incivility, but media pages were not (RQ5). Unsurprisingly, the Facebook public page of opposition leader Tomio Okamura was less likely to feature uncivil comments than the page of Andrej Babiš. There is no significant difference between TN.CZ (i.e., commercial medium) and ČT24 (i.e., public service medium). For intolerance, we only found a significant relationship for TN.CZ, when compared to the reference category, Andrej Babiš. According to Bonferroni post hoc tests (see Supplementary Material), there is a significant difference between Tomio Okamura’s Facebook page and all other Facebook discussion arenas, which have a higher occurrence of incivility contrary to that of the political leader of the oppositional party.
Study 2: Users’ Perspective on Incivility and Intolerance in Discussions About COVID-19 on Facebook—Semi-Structured Interviews
Interview participants described uncivil expressions as a communication strategy adopted by those on the “other side,” and perceived as more aggressive and vulgar (RQ1). In addition, participants extrapolate their perceptions of how others communicate to make harsh judgments of their personal characteristics, describing uncivil users from the other side as less intelligent or educated and more extreme. Their discussion comments are perceived to have grammatical errors, disinformation, and argumentative mistakes.
People threaten writers in comments with violence. But it is maybe extreme. I won’t try to understand those people. It says something about their intellect (Evženie, female, 52): They are absolutely less intelligent, maybe bored with life and perhaps just angry, like I am. At the same time, I would say that they are often mentally a little unbalanced. (Jarmila, female, 45)
These perceptions strengthen participants’ unwillingness to engage in heterogeneous debates. Some interviewees describe whole groups and communities as hostile and thus not “worthy” discussion partners.
Antinormative expressions also cause the negative judgment of other people’s opinions and arguments (RQ2). Vulgar expressions and personal attacks overshadow the arguments used in the conversation, and negative feelings prevail in the retrospective evaluation of the discussion experience. Constructive and interesting discussions are often contrasted with discussions that are full of aggression, attacks, and rudeness and often follow a reaction in the form of leaving the debate: I would not block someone just because of a different political preference, but when it turns out that someone starts to threaten someone or is disgustingly rude, I do not want to see it anymore or read it. (Gvendolína, female, 24)
Participants also describe polarizing discussions about the vaccines, with participants being divided between vaxxers versus anti-vaxxers, as problematic because participants are often aggressive toward the other side (Čeněk, male, 56). Heterogeneous discussions are generally seen as challenging, and participants feel discouraged from engaging with counter-attitudinal opinions when these conversations include aggressive and hostile behavior (Lubomír, male, 38).
With regards to the perceptions of various types of incivility and intolerance (RQ4), interviewees appear to have different thresholds for what is and is not acceptable. Whereas some perceive vulgarity as problematic, others do not mind profanities in online discussions. Resilience toward incivility seems connected with personal experiences, such as one’s own use of profanities, as well as their active engagement in online environments, with frequent commenters being more tolerant toward incivility. Overall, participants reflect that online discussions are generally more aggressive than offline communication. Tolerance of aggressive and uncivil communication is also justified by the importance of “not giving up” on cross-cutting discussions, even if they are emotionally charged. Patrik (male, 28) describes how he tries to go through the attacks and strive for mutual understanding in debates. But, when people use unclear arguments without explanation, he keeps his distance from those conversations online: (. . .) you find people, who react disgustingly, (and it) offends you. I have a method for it, and I do not pay attention to it that much anymore (. . .). (Alžběta, female, 45)
Participants were generally more negative toward personal attacks. It is clear that most people draw a line between attacks toward individual people and attacks that are more generic expressions (i.e., they are not aimed at anyone in particular). In particular, participants describe ad hominem attacks, as inappropriate, or even a conversation stopper. Instead of making an effort at argumentation, it will turn into personal (attacks) and some insults: So I simply wrote to this person that we just ended, that I am not interested when they scold me, so bye! I ended it because it makes no sense to continue in such a discussion. (Jonáš, male, 30)
While there are varied levels of acceptance of different types of incivility, intolerance appears to be more universally perceived as unacceptable. For instance, violent threats directed against minorities (e.g., LGBTQ +) and women, racism, and more extreme personal violence (e.g., wishing death upon somebody), are unequivocally described as unacceptable, fueling negative feelings and often leading to blocking or unfriending. There was unanimous condemnation of intolerant behaviors. The line between more serious and less serious attacks seems to be fairly clear for participants in discussions, as evidenced by the harsher reactions and perceptions of intolerance vis-a-vis more mixed reactions to incivility: When someone writes something like, I have already mentioned that they wish me to die from COVID and so on, it stays in my mind for a couple of days. Actually, it has stayed in my mind until now. (Jarmila, female, 45)
Discussion
The COVID-19 crisis and the hard restrictions fueled emotional reactions from citizens across the globe. Our findings suggest that this situation was reflected in highly negative and emotional online discourse, with 62.6% of our sample containing some form of incivility—a much higher rate than prior studies have observed (Coe et al., 2014; Oh et al., 2021; Rossini, 2022). This might be due to the inclusion of Facebook profiles of politicians in the sample, while most prior studies focused on news media outlets (Coe et al., 2014; Rossini, 2022). It is also possible that politicians are less concerned with moderating antinormative discourse than news outlets, or dedicate fewer resources to content moderation. In addition, the unprecedented COVID-19 crisis time may also be a contributing factor, as people had limited opportunities to engage in discussions face-to-face, and may have been more active online (McClain et al., 2021). The finding that incivility increased over the time of hard lockdown in Czechia also suggests that citizens may have been impatient and potentially reacted more negatively to having restricted freedoms. Another contributing factor is the declining trust in institutions and political representatives in Czechia (Pospěch, 2021). Dissatisfaction with politicians’ reactions could potentially fuel increased incivility in an online environment, particularly as a reaction to their online posts.
In contrast with heightened incivility, we found only 8.2% of intolerance, which is in line with previous studies that distinguished between more and less harmful types of antinormative expressions (Karunungan, 2023; Oz & Nurumov, 2022; Rossini, 2022). As such, our study provides further evidence that differentiating between uncivil and intolerant discourse allows for a deeper understanding of the extent to which discourse that threatens democratic norms and values is present in online spheres.
Importantly, our qualitative interviews suggest that participants differentiate between incivility and intolerance, as well as between different types of uncivil expressions. In line with survey-based work, interpersonal attacks are seen as unacceptable (see also Muddiman, 2017), but profanities are somewhat tolerated. In contrast, violent attacks and threats are unanimously perceived as unacceptable behavior, and they often result in unfriending or blocking. In general, participants’ perceptions aligned with data-driven typology between less (uncivil) and more (intolerant) harmful antinormative expressions (see Rossini, 2022). Moreover, their various responses to vulgar words and personal attacks highlight the role of personal characteristics, experiences (e.g., frequency of engagement in online discussions), and behaviors, as shown in the quantitative work (see Bormann, 2022; Hmielowski et al., 2014; Kenski et al., 2020).
In line with prior research, we found a positive association between incivility and justified opinion expression, which may contribute to the substantiveness of discussions and potentially support the awareness and understanding of different viewpoints (see Chen et al., 2019; Yamamoto, 2020). As suggested by Rösner et al. (2016), it is possible that exposure to incivility does not decline the quality of the arguments of other discussants, which supports the assumption that incivility is not an inherently toxic feature of online discussions and it does not necessarily have negative consequences. This is a perspective supported by some participants in our qualitative interviews. Some engage in heated debates despite incivility because they feel it is important to discuss with “those on the other side,” although it is important to highlight in that many participants said that incivility decreases their willingness to listen to counter-attitudinal arguments and to engage with them—potentially deepening perceptions of polarization in society (Hwang et al., 2016). Even though we found that intolerant comments were also supported by justified opinions contrary to previous study results (Rossini, 2022), it is apparent from qualitative interviews that participants perceive intolerant comments as unacceptable no matter how arguments are justified. These qualitative insights provide further nuance to studies on perceptions of incivility, which demonstrate that individual-level characteristics matter in how people experience and react to antinormative behavior.
We also find a positive relationship between disagreement and incivility in comment replies, providing further evidence that incivility emerges in heterogeneous opinion exchanges (Rossini, 2022). Our qualitative interviews bring in-depth insight into how incivility amid disagreement may affect participants’ perceptions of those on the “other side”: several interviewees referred to “uncivil” discussants as having lower education and being unworthy of engagement, which may lead to the dismissal of counter-perspectives. This finding is echoed by prior research that suggested that, despite incivility being effective in raising attention and awareness (Mutz, 2015), it may also increase the negative perceptions of the “out-group” (Hwang et al., 2016). At the same time, intolerance is not affected by disagreement, which might be explained by its link with prejudices, which are instead more likely to emerge in homogeneous discursive settings (Rossini, 2022; Wojcieszak, 2010).
Importantly, uncivil and intolerant comments are generally less likely to be present in direct replies and are instead reactions to the topic or content of Facebook posts—echoing prior research (Rossini, 2022). Whereas incivility toward specific people is often perceived as problematic (see Muddiman, 2017), as suggested by our qualitative interviews, incivility as a rhetorical asset might not undermine the otherwise beneficial outcomes of engaging in political communication (Rossini, 2022).
We considered that political debates on Facebook might be affected by distinct actors that shared information about COVID-19. However, this was only true when considering politicians’ Facebook pages, not media outlets. A potential explanation may be grounded in the perception that political actors are the ones making the decisions, whereas the media amplifies those decisions without being directly held responsible for the measures. Moreover, Czech media are not clearly distinguishable in political orientation, unlike the United States (Humprecht et al., 2020). This might de-emphasize the roles of various media types. The government faced criticism from citizens and the opposition political parties (see Kaniok, 2023), which might heighten incivility in the comments sections. Notably, most uncivil comments were reactions to the post and not directed at other discussants. This is in line with citizens’ higher emotionality toward political actors and decision-makers than the general mention of policy information (see Eisele et al., 2022), which was, in our case study, spread by news media outlets. Despite Tomio Okamura’s extreme right-wing rhetoric and his intense opposition to the government, comments on his Facebook page were less uncivil than those on Andrej Babiš’s page, whose politics is less radical and more center-oriented. This might be explained by the more homogeneous environment related to Okamura’s systematic strategy to delete comments and ban participants in discussions that were against his opinions (Cemper, 2018).
Our study has limitations. First, our insights are limited to the context of the pandemic in Czechia. Second, our analysis does not include comments deleted by users or moderated by page administrators or Facebook itself, which may lead to fewer uncivil and intolerant comments. This is specifically relevant in the case of Tomio Okamura and his comment-moderation strategy. Finally, the qualitative results cannot be extrapolated beyond the interviewees’ experiences, and are not representative of the Czech online population. Despite these limitations, our study is the first to examine the impact of antinormative expressions by combining two methodological approaches which allowed us to verify our assumptions based on content analysis through the qualitative insight of participants in online discussions.
The contributions of this article can be summarized as follows. First, we provide further evidence that differentiating between more and less harmful expressions is important to interpret online political discussions, and demonstrate that participants in those discussions have nuanced interpretations of antinormative expressions. As such, we contribute to refuting the idea of unconditional negative effects for incivility beyond the Western context. Second, our qualitative insights reveal a potentially detrimental effect of incivility: namely, leading people to extrapolate from people’s opinions to make negative judgments about their character, potentially widening perceptions of polarization and discouraging engagement in heterogeneous debates. Future work may further explore these dynamics by investigating discussion participation and withdrawal across different discursive contexts.
Supplemental Material
sj-docx-1-sms-10.1177_20563051231207848 – Supplemental material for Incivility and Intolerance in COVID-19 Discussions on Facebook
Supplemental material, sj-docx-1-sms-10.1177_20563051231207848 for Incivility and Intolerance in COVID-19 Discussions on Facebook by Martina Novotná, Alena Macková and Patrícia Rossini in Social Media + Society
Footnotes
Correction (November 2023):
Article updated online to correct the percentage of the occurrence of incivility from 59% to 62.6% in the Discussion section on page 9.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research was funded by the project Political polarization in the Czech Republic: The case of multi-party system (grant no. GA19-24724S) of the Czech Science Foundation.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
