The Populism of Online Communities: Constructing the Boundary Between “Blameless” People and “Culpable” Others

Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

Social network sites (SNSs) play an important role in the spread of populist ideas as they empower different actors to communicate through ungated media channels. To provide an example, Trump's Twitter account is followed by 55+ million citizens, and one does not have to search long for content that opposes the ordinary people to the corrupt elites: "We have to repeal and replace #Obamacare! Look at what is doing to our people" and "Hillary is the most corrupt person to ever run for the presidency of the United States." Such "us versus them" constructions can further be illustrated by a Dutch Facebook message, posted in the midst of Greece's EU-related conflict in July 2015: "The government should spend the billions going to Greece in the healthcare of our own people." A day after message was posted on the Facebook community Save our Health Care, over 6,000 people liked it.
Moreover, 343 people replied to it, mostly by emphasizing how the EU and the national government are corrupt, evil, and deprive ordinary people of their wellbeing.
Populism entails the construction of a binary divide in society-in which the ordinary people are pitted against the corrupt elites and/or societal out-groups (e.g., Jagers & Walgrave, 2007). Although extant research has provided important insights into the populist ideas expressed by politicians and the media (e.g., Jagers & Walgrave, 2007), the online construction of populist worldviews by ordinary citizens and the democratic implications of such citizen-initiated populism remain understudied. Specifically, previous research has studied the populist audience by quantitatively measuring their populist attitudes (e.g., Hawkins, Riding, & Mudde, 2012) or by investigating individual-level effects of exposure to populist content (Müller et al., 2018). Still, extant literature has failed to demonstrate how populist sentiments are actually constructed by citizens themselves and how the technological affordances of SNSs may provide a discursive opportunity structure for the expression of online populist discourse (Ellison & boyd, 2013;Suler, 2004). Existing studies mostly focused on online populist communication by politicians, silencing the perspective of the "ordinary" citizens central to the definition of populism. It is therefore crucial to shift our focus to citizens as active meaning-making actors in populist discourse.
In order to understand the role of social media in providing a platform for the dissemination of populist viewpoints, I investigate how populist sentiments are shaped, shared and reinforced by Dutch citizens on Facebook. As a Dutch citizen, I am familiar with the right-wing rhetoric communicated by politicians, and frequently come across hostile radical right-wing sentiments on my own Facebook timeline. Based on the results of this study, I argue how citizens' populist constructions may be harmful for deliberative democracy. More specifically, online platforms are not an extension of public opinion, but promote hostile and xenophobic language. Such language has real-life consequences as diversity and truth are outweighed by conflict and identity-based coverage, which, in turn, augment polarized divides in society.
The Netherlands provides an interesting case, as right-wing populism has been electorally successful since 2002. The current right-wing populist Freedom Party (PVV) is the second largest party in government, and the radical right-wing ideas of Geert Wilders continue to spark debates across the globe. The results indicate that people blamed the elites for not representing the will of law-abiding citizens. Migrants and other societal out-groups were scapegoated for depriving the native people of material and cultural resources. The construction of these populist boundaries between "us" and "them" reflected extremely hostile and uncivil sentiments. Importantly, Facebook did not foster debates between citizens with differing viewpoints, but mostly provided a channel for one-sided hostile populist discourse.

The dynamics of inclusion and exclusion central to populist discourse
Defining populism In its essence, populism can be defined as a set of ideas that emphasize that society is ultimately separated by the "good" people versus "evil" others (e.g., Bonikowski, 2016;Mudde, 2004;Mudde & Rovira Kaltwasser, 2017;Taggart, 2000). As argued by Albertazzi and McDonnell (2008) populism cultivates a homogenous and virtuous in-group of people juxtaposed to the elites and/or dangerous others. These elites and others are accused of depriving the people of their "rights, values, prosperity, identity and/or voice" (Albertazzi & McDonnell, 2008, p. 3).
The people are for example seen as a unity of hardworking citizens, whose will should be central in political decision-making (e.g., Mudde, 2004;Müller, 2016). However, because of the elites' failed representation and self-interests, the people are deprived from what they morally deserve (Elchardus & Spruyt, 2016). Populist ideas thus articulate a divide between the "innocent" people and "culpable" outgroups (Krämer, 2014;Laclau, 2005). In this setting, populist actors present themselves as the only ones who really care for and represent the people (e.g., Müller, 2016).
Examples of the manner in which this binary opposition is made concrete in populist messages are the "law-abiding hardworking citizen" versus "profiting immigrants" (Krämer, 2014), or the "will of the ordinary people" versus the "corrupt elites" (e.g., Mudde, 2004). By constructing this Manichean outlook on society, populist rhetoric not only creates a moral distinction between the "good" and the "evil" (e.g., Bonikowski, 2016), it also implicitly blames the out-group for problems experienced by the ordinary people. In populism, the established political parties are held responsible for the corruption of politics (Taggart, 2000), and are blamed for failing to respond to the will of the common people (Laclau, 2005). Such sentiments represent the vertical component of populist out-group constructions (Jagers & Walgrave, 2007). In right-wing populism, horizontally opposed societal outgroups are also blamed for polluting the purity of the heartland or for depriving the in-group on a cultural, social or economic level (Müller, 2016). Populism's outgroup threat is however not necessarily nativist of right-wing: the extreme-rich, experts and intellectuals can also be held responsible for posing a threat to the ordinary people. Left-wing populism may be inclusionary in its conception of the ordinary people, but also excludes others from this in-group.

Populism and social identity theory
In line with the premises of social identity theory, individuals process information in a biased way congruent with their internalized group membership (Tajfel, 1978). Positive qualities, such as being hard-working, innocent, and law-abiding, are ascribed to the in-group. Negative qualities, such as being lazy, self-interested or corrupt, are ascribed to the out-group (Tajfel, 1978). Individuals thus positively distinguish their in-group based on a comparison to out-groups, a process of boundary constructions described as positive distinctiveness (Tajfel, 1978). Boundary constructions are thus central in populist identity discourse (e.g., Krämer, 2014). The conception of belonging to the in-group is shaped by defining who is not part of the people's in-group (e.g., Van Dijk, 2004). SNSs can create a discursive opportunity structure for individuals to compare their in-group to culpable out-groups. Here, the concept of the constitutive outside is relevant (e.g., Derrida, 2005). The elites and/or dangerous others are constructed as the outside, posing a threat to the ingroup of the people. It can be noted that others may be excluded on both a rightwing (nativist) and left-wing ideological basis. Hence, anti-immigrant discourse may tie in with populist sentiments on the right-wing whereas discourse targeted at the economic inclusion of the people versus the economic elites resonates with leftwing populism.

Populism in the Netherlands
It is important to situate populism in the political and societal setting of the Netherlands. Dutch populist political parties have been highly-successful from the early 2000s onwards. In the 2017 general elections, Wilders' right-wing populist PVV even became the second largest party. At the same time, new populist political parties have emerged, such as Verdonk's Proud of the Netherlands and Baudet's Forum for Democracy more recently. During the 2017 general elections, some analysts argued that mainstream politicians have adopted populist viewpoints in their rhetoric. In that sense, people who express populist worldviews on online media are familiar with populist framing.
The largest populist party in the Netherlands (the PVV) is a radical right-wing or "complete" populist party (e.g., Aalberg, Esser, Reinemann, Strömbäck, & de Vreese, 2017). Next to shifting blame to the "corrupt" national and European elites, the PVV constructs the Islam and immigration as the largest threats to ordinary Dutch people. Nativism and anti-immigration can thus be associated with the discourse of the PVV. As this right-wing populist party has been electorally successful, the right-wing populist sentiments may also resonate with citizens' populist attitudes on the demand-side. Facebook offers a platform for the expression of such sentiments, and citizens can directly share their populist attitudes with others, bypassing media elites. For this reason, it is important to assess how Dutch citizens are using their own Facebook accounts to express populist sentiments.
Shaping populist discourse in the context of social network sites SNSs, such as Facebook or Twitter, provide an important context for the construction of populist discourse (e.g., Engesser, Ernst, Esser, & Büchel, 2017). However, previous research exclusively focused on how populist politicians use SNSs to communicate their rhetoric to the electorate. Still, we know too little about how the processes of inclusion and exclusion central to populist discourse are shaped and experienced by ordinary citizens on social media.
Populism poses severe threats to democratic communication because it shapes a one-sided discourse revolving around the conflict between the people and the elites (also see Waisbord, 2018). Specifically, it may harm the principles of diversity, tolerance, reason, and truth-seeking. In online settings, ordinary people can communicate fact-free issue positions, and communicate the conflict between their in-group and others. As the online and fragmented media setting gives rise to a polarized climate where truth is less central than belonging to an imagined community that shares worldviews, the online expression of populism has important democratic implications (Ouellette & Banet-Weiser, 2018). It may for example give rise to the construction of an imagined community of nativist and xenophobic citizens, who are connected by their partisan views and empowered to express their identity in online mediascapes that are independent of space and time. Therefore, SNSs may create a discursive opportunity for anti-social and hostile expressions targeted at (marginalized) groups in society (e.g., Lowry, Zhang, Wang, & Siponen, 2016).
SNSs enable people to asynchronously forge linkages between people based on their shared interests or political views (Ellison & boyd, 2013). People who oppose the incumbent government, for example, can join Facebook pages that voice their political discontent. A key democratic implication of this affordance is that communication in like-minded online environments can place people into echo chambers or filter bubbles (Pariser, 2011). Filter bubbles are virtual spaces in which people are mostly exposed to congruent information. Because of their previous patterns of selective exposure, they filter out incongruent information so that their prior attitudes are not challenged but rather reinforced.
In recent years, the expression of hostile sentiments in online environments has become more societally accepted, even more so when people experience closeness to the social media environment and the source (e.g., Nekmat, Gower, Zhou, & Metzger, 2019). In light of these developments, people may feel less ashamed to share hostile and racist sentiments on Facebook, where they perceive they are surrounded by like-minded others. Hence, Facebook offers "community pages" where people can interact with citizens who share their views on certain political and societal issues. Such (perceived) homogeneity and community formation is less central to other social media, such as Twitter.
In the context of these affordances, two alternative perspectives on the deliberative nature of SNSs can be foregrounded. First, online communities may enable participants to freely express their opinions and identity-based boundaries to others in a decentralized communication context (Pingree, 2007). Each user can express his or her opinion, raise new questions, or respond to questions and opinions introduced by others. In other words, the affordances of the online context may foster argumentative discussions between participants, which may have positive ramifications for democratic deliberation (Stromer-Galley & Wichowski, 2011). Facebook may thus create an environment to construct and renegotiate identities-also in opposition to the "culpable other." This may however crystalize in anti-democratic, hostile sentiments. Alternatively, populist discourse on Facebook can depersonalize membership in groups, potentially resulting in uncivil expressions and negative stereotyping (Lowry et al., 2016). Disinhibition can be regarded as a dangerous feature of online environments, as it may provide a context for anti-social behavior targeted at specific groups in society. This critical perspective is in line with the understanding of populism as a one-sided and conflict centered discourse (Waisbord, 2018). The online spaces provide a niche for ordinary citizens to vent their frustration and distrust, rather than a platform to debate political issues and learn about the truth. One of the most harmful consequences of online populism is thus the dissemination and perseverance of misperceptions in a post-factual media setting (Harsin, 2018;Waisbord, 2018). Populism's focus on conflict rather than accuracy may harm the democratic principle of finding the truth in a diversified public sphere.
This tension between dangerous versus potentially constructive features of communication on SNSs is crucial for online populist discourse expressed on Facebook. Facebook can give "the people" a voice. On the pages, they can interact with both known and unknown others. Facebook further allows for multimodal communication, such as posting cartoons and manipulated images that ridicule the elites and immigrants. As a social media platform, Facebook is a more people-centered, and less elite focused social media platform, for example in comparison to Twitter.
Against this backdrop, the two-fold central research question of this article is: How are the dynamics of inclusion and exclusion shaping the construction of populist discourse on Facebook in the Netherlands, and what are the political consequences of populism in a fragmented online media environment?
I pose four more specific research questions that focus on different parts of the populist "us versus them" divide: How are Dutch citizens constructing conceptions of their selves as part of a social in-group of "the people" on Facebook ? (RQ1); How are Dutch citizens constructing boundaries between their in-group and the elites on Facebook? (RQ2); How are Dutch citizens constructing boundaries between their in-group and societal out-groups on Facebook? (RQ3) and to what extent and how are people expressing belonging to a homogenous in-group? (RQ4).

Data collection
Materials were selected on three different levels: On the level of Facebook communities, on the level of Facebook posts within these communities, and on the level of responses to these contributions. To ensure contributors' anonymity, names of the included communities, posts and contributors are not mentioned in the results. Two types of Facebook communities were selected. First, Facebook communities that reflected protest movements targeted at the establishment in the broadest sense were selected. These communities were used to vent disapproval of the current government and their policies. Other elitist threats, such as banks, traditional mass media and experts, were also discussed here. Second, Facebook communities that 152 focused on national identity were included. Here, the communities' self-presented aim was to share national and cultural symbols that form the core of Dutch identity, hereby expressing nostalgic sentiments and boundaries of the people's national identity. As these Facebook pages form a stage for native citizens to celebrate their heartland and to voice their discontent with others, these platforms should provide a context for the expression of identity and to shape boundaries between the ingroup and various out-groups.
Snowball sampling was used to select Facebook communities. This strategic informant sampling strategy was used to reveal hidden social forms based on the construction of a divide between "us" and "them." Although the sampled Facebook pages were publically accessible, community members can be difficult to identify without access to their network. The first community was found as a person in the researcher's network liked this page. By entering the personal page of this individual, similar pages were accessed. This sample was expanded by viewing the liked pages of these communities themselves. During the sampling procedure of communities, this study aimed for maximum variation of populist viewpoints.
The inclusion of new Facebook communities stopped when the point of theoretical saturation was achieved. At this point, the inclusion of new communities did not reveal additional insights into people's populist frames of reference. More specifically, a sixth, seventh and eight community was sampled and analyzed, but although these communities differed in content and popularity, the analyses did not reveal novel insights into the ways in which people constructed the populist opposition between "us" and "them." The point of saturation was established by constantly comparing the themes resulting from the analyzed material to the analyses of new materials. The sample, which resulted from a cyclic-iterative process of data collection and analysis, consisted of five Facebook communities.
Within each community, ten posts were selected. The inclusion of additional posts did not reveal novel insights: theoretical saturation on the level of posts was achieved by comparing the analyses of new posts to the themes found in the analyses of the previous ten posts. The selection procedure aimed for maximum variation by selecting posts with varying dates of publication that dealt with various topics. The sample reflects viewpoints on many societal and political developments that resonated in public discourse over the past few years. If available, the first 10 responses to each included post were selected. For posts with less than 10 responses, all available ones were included. In most cases, no more than 15 responses were found. In total, five communities, 50 posts and 362 responses were included in the analysis (72 responses for each community on average). Beyond topical variation, the included communities also reflected maximum variation in terms of popular support. One highly popular community with 186,000 members was included. A community with a relatively low number of members was also included (1,600). The other three communities reflected positions between these extremes, counting 18,000, 17,000 and 30,000 members.
By means of triangulation, and as an additional check for saturation, the content analysis was also conducted on a similar Facebook community in the United States. For this community, 10 posts and 101 responses were sampled and analyzed. In the cyclic-iterative process of data collection and analysis, the results from the U.S. sample were compared with the results from the Dutch sample. The analyses of the U.S. data did not result in novel themes or an enrichment of the already established themes in the Dutch data set.

Analysis
The data were analyzed according to the coding steps explicated in the grounded theory approach (Glaser & Strauss, 1967). Three subsequent phases of coding were conducted: open coding, axial coding, and focused coding. During the first step of open coding, sensitizing concepts related to the central components of populist discourse and social identity theory were used: in-group and out-group constructions, dramatization, emotionalization, hostility, crisis perceptions, constructions of the heartland, anti-establishment, us against them constructions (e.g., Mudde, 2004;Taggart, 2000). During axial coding, a matrix consisting of core categories was constructed. These categories are: boundary constructions, blame attributions (causal interpretation), sentiments/style of communication, in-group homogeneity, treatment recommendations and moral evaluations. The analysis distinguished between dominant and challenging positions. Finally, during the process of focused coding, the categories were interpreted as the main themes discussed in the results.
Two coders independently coded a randomly selected sample of ten posts and 72 responses. To establish consensus among researchers, emerging themes were reviewed for similarities and differences. Although these checks did not aim for complete agreement between researchers, they assured that independent investigators familiar with the sampled data could agree on the interpretation of the emerging themes (Potter & Wetherell, 1987). In addition to peer debriefing, member checks were conducted by inviting four members of the sampled communities to review the themes. All members agreed on the core themes found in this study. The thick descriptions provided in the result section and the closeness to the raw data provide a further reassurance of the validity and reliability of the qualitative data analyses reported here.

Results
Constructing membership to the in-group: consolidating a sense of "the people" People expressed closeness to their in-group's community on two different levels: socio-politically and cultural-symbolically. On a socio-political level, members expressed their closeness to the ordinary Dutch people's will, which should be central in political decision-making. However, they experienced that the majority of the ordinary people is silenced by the elites who refuse to listen: "The people did not want to be a member of the EU, but they did not listen to us." The people expressed belonging to a powerless and silenced community. Empowerment of the ordinary people as in-group thus played a pivotal role in the construction of community on a socio-political level. Marking the boundary between the people's will and the elites' failed representation related to the expression of a political image of the self in relation to the distant, unresponsive elites.
People further demonstrated empowerment by expressing how their in-group should have the agency to decide on the limits of toleration. Some foreign elements can be allowed in the heartland, but if too many foreigners will enter, the heartland will be polluted severely: "I prefer having only one Dutch people. I don't have problems with normal foreigners who work and behave." Hence, people expressed clear limits to the acceptance of others: "they" have to obey the rules of the in-group and they should not receive anything more than the absolute minimum to get by. An important recurring theme in people's responses to messages about immigration is that the power to tolerate should be reserved for Dutch people. The elites should not be trusted with the power to tolerate. This perspective is in conflict with democratic communication's principle of fostering diversity in viewpoints (Waisbord, 2018).
Morally, in-group attachment was based on the perception of a shared set of "superior" cultural symbols in relation to the "inferior" traditions of others. To express the perceived homogeneity of the people and their shared history, the online setting was frequently used to share visuals and videos related to important moments in the past of the Netherlands. Some examples are pictures of large families having dinner together, or old streets in city centers without "foreign" shops. People relied on these multimodal forms of communication to illustrate the distinction between the glorious past of ordinary Dutch citizens and the current symbolic and cultural deprivation caused by the elites.
This relational component of community construction can further be illustrated by the following quotes: "How about leaving our country? The ones who are not respecting our traditions do not have the right to live here" and "Go away with these backwards traditions." In a similar vein, people who identified with Islam were regarded as an out-group that should have less rights than native Christian people, as exemplified by the following quote from a post: "Christian Dutch people have morally more rights than Islamic newcomers, because Christians have participated in building this country for decades." An important theme articulated in the words of commenters is the belief that the nation has been built by their ancestors, which is used to legitimize the perception that the in-group has more rights than foreign people.
To answer RQ1, the construction of boundaries between the in-group and others related to different concepts of the self, varying from identification with the general will of ordinary citizens to nationally and culturally superior images of the native people. These self-concepts were consolidated by positively distinguishing the in-group from out-groups. Images were used to mark the contrast between the past and present. The interactive context of SNSs allowed people to be aware of likeminded community members and to share their belonging to a perceived homogenous in-group opposed to elites and foreign elements constructed as a threat to the well-being of the people. The online mediascapes did not offer room for a diversified identity discourse, and, albeit multifaceted, promoted specific partisan identities. Although people may encounter more multifaceted identities in offline discussions, where people with different ideological leanings may participate in the exchange of arguments, the highly politicized leanings of the online populist communities leave little room for the negotiation of meaning.
Marking the central distinction between the ordinary people and the elites The opposition between the ordinary people and the elites was constructed by shifting blame from the powerless people to the corrupt elites. More specifically, people emphasized their opposition to the elites by referring to the discrepancy between the people's will and the elites who are only acting on behalf of their own will: "These self-interested politicians in The Hague are not doing anything on behalf of our will." The moral boundary between the elites and the common people was interpreted in terms of large distances between the people and the elites. This distance was illustrated by using terms such as "they who sit in their ivory tower" or the "politicians in The Hague" when referring to the political elites. Clearly, the politicians were framed as living in a world isolated from the ordinary men on the street. In addition, people frequently voiced the opinion that the distant elites do not understand the real problems facing the common people: "Rutte [the prime minister], you do not understand what is really going on for ordinary people. I would say, get lost." The elites were also perceived as a lying, deceiving, and corrupt other: "I don't believe it. Again, lies and deception from this cabinet." In support of Suler's (2004) notion of disinhibition in online spaces and the salience of threats and dangers in online spaces identified in extant research (e.g., Lowry et al., 2016), people's opposition to the elites reflected a harsh and cruel style of communication. When referring to the establishment, people frequently expressed feelings of anger targeted against specific politicians. The following quote illustrate this personal way of expressing anger: "Just wait until we kick Rutte [the prime minister] and his criminal organization onto the streets." The aggressive tone of people responding to Facebook posts crystalized into death wishes to the prime minister: "I hope he will break his neck" or "It would be a fantastic day if someone would hit this nation's traitor with a bullet through his head." By expressing fear and anger toward the elites, the boundary between the in-group of the blameless people and the out-group of the culpable elites was emotionally consolidated. Offline, people may experience a stronger boundary to vent such hostile and aggressive language. The homogeneity and perceived anonymity of the Facebook communities may promote the extremist language of "keyboard warriors" who can avoid direct consequences for their hostile online behavior.
Populist Facebook users frequently used cartoons to mark the distance between the people and the elites. For example, people posted a picture of the primeminister and added a long nose in the visual to depict him as part of the "lying elites." Some posts only consisted of a single cartoon with a text balloon. The replies to these posts were, however, mostly textual. In any case, the online context allowed people to upload an image that engaged people with similar views to vent their shared distance to the failing elites.
Online anti-elitist populist discourse was targeted at the violent elimination of specific politicians belonging to the elitist out-group. The specific expression of the conflict between the people and the elites reached beyond the borders of freedom of speech, shifting to hate speech and online incivility. Moreover, the hostile expressions contained fact-free and generalized attributions of blame to the elites. In that sense, the absence of checked facts and rational argumentation indicates that online populism may catalyze the public sphere's shift to post-factual relativism (Harsin, 2018;Van Aelst et al., 2017).

Ways of excluding societal out-groups
Horizontal oppositions to societal out-groups were based on an exclusionist relationship between native citizens and foreign others living in the heartland, most saliently immigrants and refugees. This out-group threat was culturally rooted in xenophobia and economically grounded in welfare chauvinism. Moreover, and corresponding to more left-wing sentiments, people excluded out-groups on a nonnational basis: extreme rich people and the "culture of greed" were held responsible for augmenting the gap between the ordinary people and the extreme rich minorities in the Netherlands. The exclusion of these different groups has a certain hierarchy: foreign elements that were dehumanized were seen as most distant from the people. The extreme-rich, such as managers of banks and CEOs of large corporations, were distant from the lifeworld of the ordinary people, but not excluded in extremely hostile ways. People with different religious and cultural backgrounds were seen as most distant from the "true" people whereas economically excluded others were closest to the people. These out-group exclusions are discussed in a hierarchical way below.
On a cultural level, people expressed ethnocentric sentiments by reflecting on their belonging to a nativist community that should be free of polluting elements. One post, for example, depicted two Rubik's cubes next to each other: one in which the puzzle was solved, which referred to a mono-nationalist outlook of society. The other puzzle was left unsolved: all colors next to each other illustrated the multicultural society. Responses to this post clearly articulated people's preference for a society with only the culture of the native people. This nationalist utopia of the populist heartland is exemplified by the following quote: "Everything needs to go back. This will give us more air to breathe. Their houses will need to be disinfected, but we will do this with pleasure." This response not only reflects an exclusionist perception of identity, it also relates to xenophobia in the sense that foreign elements are unwanted as they pollute the purity of the nation. This "polluting other" was a recurring theme in responses to posts in the online communities. Some responses to posts reflected dehumanization of others. For example, one response to a post blaming boat refugees for their own situation contains the following comment: "It's just like killing a fly, you won't notice. Because there are so many of them." The other was thus not only excluded from the common people, he or she was also constructed as not belonging to the same species, as the following quotes in related comments further illustrate: "The police is just as afraid of these monkeys as the government," "these dogs," "the rats." People expressed their symbolic and economic opposition to societal out-group in extremely hostile ways. People for example intended to eliminate foreigners, such as the following words used to respond to a post on refugees who enter Europe on boats indicate: "Are we not allowed to shoot at these boats?" People further excluded foreign others based on their assumed lack of respect for the native's cultural traditions. The dramatization of issues prevalent in the posts and responses can be illustrated by the following quote found in a response on a seemingly innocent post depicting a picture of a Dutch traditional meal: "They are not respecting our rights. They keep on doing it and they laugh at us." This fear of a foreign invasion was triggered by a post on an everyday national tradition.
The threats cultivated on a cultural-symbolic level emphasized that dangerous others were far away from the ordinary people. Exclusions on an economic level were less extreme and mainly related to sentiments of relative deprivation partially caused by the corrupt elites. People constructed the extreme-rich minority as a depriving other, but did not use extremely uncivil language to mark the boundary between "us" and "them." Moreover, although the extreme rich minority deprived the ordinary people of their wealth, the "corrupt" elites were eventually held responsible for allowing these others to profit. Against this backdrop, the exclusion of others on a (radical) right-wing level were extremely hostile, whereas the exclusion of out-groups on an economic level placed "the other" less far-removed from the people.
To answer RQ3, the pervasive community on SNSs provided a context allowing individuals to mark a causal and moral boundary between "their" people and societal out-groups perceived as substantially different and culpable. Extending social identity theory, people compared their native in-group to out-groups on both cultural-symbolic and economic levels. As the societal out-group was accused of depriving the people from their "superior" cultural symbols, marking the boundary between "us" and "them" was central to people's online construction of the self. The xenophobic and nativist online populist expressions shape an uncivil discourse, which is fueled by the depersonalized and homogenous online setting. Out-groups that were excluded on an economic level were less hostile, and did not reflect a dehumanizing discourse. In that sense, the findings point to a hierarchy of outgroup constructions: foreign "polluting" elements are most dangerous and most distant from the people, whereas "profiteers" and people part of the "culture of greed" were excluded less explicitly-as the elites were eventually seen as the culprits.
Here, we can also distinguish populist discourse from nationalist or nativist sentiments. As argued by De Cleen and Stavrakakis (2017) populism refers to an ingroup of the "ordinary" people, whereas nationalism refers to the nation (and not the ordinary people per se) as the imagined community of the people. People commenting on Facebook lost their connection to the real nation. Hence, this imagined community does no longer exist because the corrupt elites and dangerous others deprived Dutch people of their cultural, economic and political needs. One Facebook community, for example, is called "This Netherlands is no longer my nation." In expressing the relative deprivation of the people, and the distance to out-groups, people expressed that they no longer feel connected to their nation, and rather identify on the bases of shared threats to the homogenous ordinary people.
Challenging the populist master-frame Can we say that all sentiments expressed on the analyzed SNSs contained the ideational core of populism? Most posts indeed framed societal issues in the binary opposition of the pure, blameless people versus the evil others. Although these stances were shared, accepted, and reinforced in most comments, some people resisted the dominant populist frame in their responses to posts. Still, only for one relatively extreme post proposing to kill boat refugees, counter-arguing outweighed agreement. As the following quote illustrates, the populist simplification of societal issues in black and white terms was not accepted by all people: "This is nonsense. Get informed before you spread this society poisoning crap into the world. Otherwise, people may think this is true." This response clearly emphasizes that fact-free posts are harming public opinion.
Another way in which people disagreed with the dominant "us against them" distinction was by attributing responsibility to populist individuals. As the following quote illustrates, other people who complain too much while not trying to solve any of the problems themselves are the target of frustration: "I am getting so tired of this. Instead of complaining, you can better invest your energy in advancing your chances on the labor market." Despite these few examples, the populist master frame emphasizing the binary divide between the pure people and the polluting others dominated. To answer RQ4, in the context of SNSs, people expressed a pervasive awareness of a homogeneous native people, whose will should be central in politics and whose cultural symbols were superior in comparison to those of negatively stereotyped out-groups. These perceptions of the native, political, and cultural self were not challenged but reinforced by group members. The online populist spaces did not offer a platform for the exchange of opposing viewpoints, but rather reflected a one-sided and conflict-oriented discourse that circumvents any argumentative debate. Citizens that aimed to counter fact-free accusations to elitist actors or marginalized groups by checking the claims against facts or by offering conflicting evidence were disregarded and framed as ignorant and "lying" dishonest others. The prevalence of mis-and disinformation is fueled by the one-sided nature of post-factual online populist spaces.

Discussion
Despite the growing prevalence of populist expressions by citizens on SNSs, citizens' online populist discourse has not been studied yet. At the same time, it has been argued that the mass self-communication of people who express themselves online plays an important role in the formation of political opinions (Papacharissi, 2010;Pingree, 2007). Online communication enhances the construction of communities and spark the mobilization of citizens with similar political interests (Habermas, 2006). Such online communication may have crucial ramifications for democratic communication (Waisbord, 2018). Against this backdrop, this study aimed to provide an in-depth understanding of how populist discourse is expressed by citizens online, and to what extent these expressions stimulate or limit a well-informed, argumentative and critical online debate.
In the context of SNSs, the populist construction of identity entailed economic, cultural, and political components. People expressed their belonging to community by morally and causally articulating their opposition to threatening others. A positive self-concept was constructed by constructing external causes for the people's problems (Tajfel, 1978). The multifaceted nature of the construction of an elitist out-group corresponds with populism theory that has highlighted that populism does not only target political elites, but also other institutions blamed for the deprivation of the people (e.g., Bonikowski, 2016). The homogeneity and superiority of the self was not contested, but rather reinforced. Online populist discourse created a specific niche of multidimensional identifications whereas it did not allow for a diversified negotiation of identities on political, economic or cultural dimensions. Online populist communication is thus inherently anti-democratic: it does not allow for an exchange of different viewpoints. It promotes the issue positions of a relatively small group of disenchanted individuals that claim to speak on behalf of the majority of the people. An implication is that immigrants, religious minorities and refugees are marginalized. They are seen as inferior, and their needs and political interests are silenced by populist Facebook users that claim that they should not have a voice in society. This study has revealed important power discrepancies between the populist Facebook users allowed to share their hostile sentiments online, and people excluded from this in-group.
The expression of such power discrepancies can have real-life implications. The identity frames salient in online populist communication can prime mobilization (Simon & Klandermans, 2001). Previous research has indicated that the cultivation of in-group deprivation and a severe politicized out-group threat motivates people to mobilize politically (Simon & Klandermans, 2001). Against this backdrop, the populist construction of a binary divide between the "good" citizens and the "evil" and depriving other may motivate people to take action, for example by protesting together or voting for populist political parties. Although the online setting may involve typically disinterested people into politics, the actual consequences of spreading right-wing populist discourse may further increase polarized divides in society: populist citizens become further apart from the political establishment, and groups in society that do not identify with the ordinary, deprived people are excluded.
By looking at citizens as communicators, this article extends the field's growing interest in online populist communication from the perspective of populist politicians (e.g., Engesser et al., 2017). Ordinary citizens are also empowered to circumvent the elitist mass-media, gaining access to communities of like-minded other citizens to co-construct identification with a deprived group of ordinary citizens (Elchardus & Spruyt, 2016). Online populist communities are largely devoid of diverse viewpoints. Populist discourse on social media is highly emotionalized and conflict-centered without offering evidence or arguments to back up these sentiments. In this case, online populist discourse gives rise to a post-factual discourse where sentiments and belonging and excluding outweighs truth-seeking and making accurate judgements. Populist communication is thus at odds with the principles of balanced reporting that includes different groups in society-which is especially problematic in the multicultural setting of the Netherlands.
How dangerous are the populist constructions in online communities? As populist expressions may seem close to authoritarianism, extremism, racism, and fascism, important differences in the context of the disinhibiting potential of online spaces need to be emphasized here. People only voiced a desire to purify and destruct, while their actual behaviors merely consisted of gathering at public placed to voice their discontent using "soft" weapons of slogans on banners. Still, it must be noted here that the expression of right-wing populist discourse on SNSs reflected hostile sentiments targeted at specific vulnerable groups in society.
As people's online behavior is not isolated from their behaviors offline, the hostile expressions found on SNSs can also spill over to harm targeted at out-groups in the physical world. Online expressions may fuel negative sentiments towards refugees and migrants in real-life, and may strengthen people's opinion that most other Dutch people share these ideas. The overall tolerance of different cultural norms and values of Dutch society is at risk: if the preference for a monocultural nation becomes widespread and normalized, people's openness to other cultures and traditions is harmed.
Moreover, deliberative democracy is at risk. People placed themselves into populist filter bubbles (Pariser, 2011). Although some citizens challenged the dominant populist worldviews, people who counter-argued were always attacked by the community members. People that defended their populist viewpoints thus acted as gatekeepers who only engaged with similar viewpoints and rejected opposing viewpoints. As a long-term consequence, these spaces foster polarization in society: those with populist attitudes share their hostile sentiments with like-minded and supportive others. In the protected online communities, right-wing populist sentiments are accepted and normalized. People that oppose such viewpoints, in contrast, stay away from these online communities. If they do participate, they are excluded in a hostile way, which may augment their opposition to these groups in society-avoiding any constructive debate.
The connection of online right-wing populist discourse to xenophobic, racist, and hostile sentiments is not intended to overstretch the concept of populism. In contrary, these findings are in line with the literature that has argued that populism is a "thin cored ideology" that can be supplemented with a plethora of host ideologies on both the left-wing and right-wing of the political spectrum (e.g., Mudde, 2004;Mudde & Rovira Kaltwasser, 2017). Extending this literature, the online expression of populist discourse enriched the "thin" ideology of a societal divide between the people and culpable others by articulating hostile, xenophobic and racist sentiments targeted at the excluded out-groups.
The question on the feasibility of potential treatment recommendations remains. If we regard populist, hostile and nativist sentiments as dangerous, it can be argued that media and society should take some responsibility to counter the potential negative effects of online populist expressions. The results of this study indicate that perceived relative deprivation provides a context for populist discourse (e.g., Elchardus & Spruyt, 2016). Deprivation was expressed on different levels (i.e., economically and culturally), but the out-group was universally seen as coming from outside the nation (immigrants and asylum seekers). Specifically, people expressed the sentiment that their in-group of the blameless hard-working citizens is worse off than these groups. This salient in-group threat is caused by the unresponsive and corrupt elites and the profiting others. Acknowledging these sentiments, and relativizing this in-group threat may provide an important starting points for antidotes. Confronting citizens with actual numbers on their position relative to other groups in society may alleviate the magnitude of the in-group threat. At the same time, it is important that mainstream political parties show that they do understand the concerns that people experience, hereby partially "stealing the thunder" of populist political parties that claim issue ownership on voicing the grievances of the ordinary people. Social media, such as Facebook, should take some responsibility in setting boundaries to the freedom of expressions on their platform by flagging hostile language. A critique of this perspective may be that people of the scientific community are censuring online communication by removing messages with an (extreme) right-wing leaning. However, people are free to express they oppose immigration or the government's policy, as long as these expressions are civil and devoid of any racism. The same applies to extreme left-wing discourse: people should be free to share any issue position they may have, but the marginalization of groups does not belong on social media. This study has some limitations that can be addressed in future research. First, the selection of online communities may have biased the findings as all content is to some extent populist and nativist. As populist viewpoints are also frequently articulated in commentary spaces on mainstream news websites and online communities on all sorts of topics, future research may also analyze other online content that is less prone to be right-wing populist, for example providing insights into leftwing populist sentiments targeted at the economic elites. However, as this foundational study aimed to provide insights into the ways in which people construct populist discourse, the focus on most likely cases provided the richest and most suitable data for the aims of this study. Another limitation concerns the focus on a single SNS: Facebook. Although it may be argued that the communities on Facebook are not really "imagined" as most users know all members of their network (Ellison & boyd, 2013), I studied publically accessible pages that people can join to follow and share interpretations of like-minded, but not necessarily physically-known others.
Despite these limitations, this study has advanced our understanding of populist and nativist discourse online. The recent electoral successes of populism can be placed in the context of the discursive opportunity structure provided by SNSs. Political populism strongly resonates with the populist sentiments expressed by citizens online, which indicates a political parallelism between the political realm and public opinion (Hallin & Mancini, 2004). As populist movements are on the rise throughout the globe, it is important to understand how SNSs provide a platform for ordinary citizens to reinforce, construct and disseminate populist ideas that resonate with major political and social shifts across the globe.