Anti-immigration and racist discourse in social media

Abstract

This article assesses the strategies of anti-immigration actors on social media and the discursive construction of immigrants and refugees in user interaction on Facebook. It emphasizes the particular role of emotions in racist discourse and analyses how an open Facebook group generates and circulates anti-immigration and racist sentiments to a large audience. By analysing the general communicative features of the group, including user interaction, it demonstrates how anti-immigration and racist sentiments are moulded through interactivity between actors in an open digital space. Moreover, the article emphasizes that anti-immigration groups online can be understood as affective publics, in which racial expressions and overt racism are becoming increasingly normalized. It also argues that these publics must be taken into consideration when addressing the causes of anti-immigration and racist sentiments in contemporary societies.

Introduction

Anti-immigration sentiments are permeating public discourse across the global North – particularly subsequent to the ‘refugee crisis’ of 2015. These are reciprocally manifested in the current upsurge of far-right, including populist, politics throughout Europe and beyond. The implementation of harsher and illiberal policies on immigration and refugees mirrors these public sentiments. Concurrently, online publics play an increasingly important role in shaping public opinion on immigrants and refugees, yet ‘the dynamics of these groups are still poorly understood’ (Törnberg and Wahlström, 2019: 267). These online publics harbour citizens’ emotions of insecurity and generate anti-immigrant attitudes. They are mutually canalized by nationalist and far-right populist political actors. In short, commercial platforms such as Facebook provide spaces for xenophobic, racist and nationalistic discourse online, and they shape antagonistic (Farkas et al., 2018) attitudes towards immigrants. Moreover, through their large size, they affect mainstream discourses on immigration and refugees, and contribute to a normalization of previously marginalized types of utterances, attitudes and opinions. Anti-immigration groups and publics on commercial social networking services (SNSs) also seem to amplify xenophobic and racist attitudes among their participants. For example, in December 2017, a middle-aged woman from a small town in Sweden was convicted of hate speech. She had posted several abusive comments on Facebook targeting Somali immigrants. The woman was active in one of the largest public (Swedish) Facebook groups, ‘Stand up for Sweden’ (SufS) (‘Stå upp för Sverige’). When interrogated by the police, she explained that she felt ‘brainwashed’, and that the hateful jargon expressed by other members of the group motivated her to post comments that were more abusive than she intended (Larsson, 2018).

In order to understand how mundane use of social media contributes to growing anti-immigrant and racist sentiments in contemporary societies, the article analyses how immigrants and refugees are discursively constructed in the Facebook group ‘SufS’, and how racism is overtly and covertly expressed in user comments in the group. Drawing on both quantitative and qualitative content analysis, the article emphasizes the particular role of emotions in shaping racist discourse (Ahmed, 2004). Moreover, it discusses how affect drives the circulation of anti-immigration discourse online (Dean, 2010). It also argues that racist opinions and attitudes become normalized through the recontextualization of mainstream news (covering refugees/immigrants), and that the affective character of public comments triggers and augments racist attitudes. The article is structured as follows. First, it highlights recent discussions on social media and affect, providing a theoretical understanding of racism in social media as a form of affective network/public. The article then provides a brief overview of the Facebook group as a specific case, and its broader relevance for research on social media and racism. This is followed by a section on data collection and methodology. Finally, the analysis is conducted in two steps: first, using descriptive statistics to outline the overall communicative patterns in the group and then providing a more in-depth analysis of user comments.

Racism and social media

Social media, including SNSs, are not only digital spaces for entertainment, connectivity and interpersonal communication (van Dijck, 2013) but also platforms for political discourse and identity making (Papacharissi, 2015). In order to understand the increase in anti-immigrant and racist attitudes, we need to recognize social media as socio-technical systems which play an increasingly important role ‘in shaping social relations, including those pertaining to race and racism’ (Farkas et al., 2018: 2). Alongside other (more frequently cited) explanatory factors such as socio-economic conditions, social unrest and failed integration policies, digital communication must also be acknowledged when assessing the impact of anti-immigrant and racist political actors in society. SNSs are systems with different sets of technical affordances that can be used by racist and far-right actors when constructing contentious publics online (Alvares and Dahlgren, 2016). One of the most common anti-immigrant strategies is to remediate and recontextualize mainstream news on issues related to immigration and immigrants. These news stories tend to focus on crime or other negative topics, such as public unrest, cultural misunderstandings, social problems and economic costs (Ekman, 2019). The practices of recontextualization and reframing are often executed through a few textual amendments: using certain naming strategies, selective extraction, reformulation of paragraphs or omission of explanatory factors in the story. Recontextualization means that seemingly ‘neutral’ news items transmute into news pervaded by an anti-immigration or racist agenda. Krzyżanowski and Ledin (2017) argue that these stories often include both ‘civil framing’, for example, emphasizing public safety, and ‘uncivil’ frames, such as reductive culturalist explanations – that is, that immigrants behave in certain ways because of their ‘culture’ or ‘religion’. The perpetual flow of ‘bad news’ on immigrants/refugees (Philo et al., 2013) is maintained through circulation on huge social media platforms, particularly Facebook and Twitter (Ekman, 2019). By embedding and circulating material published on anti-immigration websites or social media platforms such as Facebook and Twitter, anti-immigrant actors can reach a vast number of users with limited resources. Both the circulation and production of anti-immigrant content depend on the online labour of ordinary users (Fuchs, 2017). For example, Facebook users provide anti-immigration actors with raw material for stories by sharing mainstream news, circulating narratives relating to refugees and immigrants, or articulating experiences, and in return share the ‘news’ produced by anti-immigration websites. These user practices are not necessarily intentional; oblivious users share and disseminate information from anti-immigration actors without reflecting on the reliability or political objective of the source. In parallel, anti-immigration actors exploit a growing distrust towards mainstream media, particularly regarding immigration reporting (Andersson, 2017). This can partly be explained by a mounting commercialization (including sensationalism, click-bait news, unverified stories, exaggerations, etc.) of mainstream news online, alongside a fast-growing information-extensive social media system which is also more fragmented than older media systems (Enli and Rosenberg, 2018).

Rather than focusing on explicit forms of racism and the strategies of racist and far-right actors only, however, we also need to recognize what Sharma (2018) defines as ambient racism in everyday articulations on social media. Ambient racism occurs at the micro-level of Internet communication, in user comments, tweets and affective responses circulating between users. In open-networked publics such as Facebook groups, racism is connected with and produced at both the micro and the macro level. Through the constant articulation and circulation of messages between ‘ordinary’ users and racist and anti-immigrant political actors, mundane forms of racial and racist expressions and far-right content become conflated. Social media publics facilitate a seamless flow of communication that does not distinguish between the intentional strategies of far-right actors and the affective responses of everyday users. So, anti-immigrant actors benefit from the constant flow of information in social networks, where endless affective responses to single events, news flows and shared experiences circulate on platforms such as Facebook. The commercial architecture of social media also shapes the production of racism in different ways (Matamoros-Fernández, 2017). User policies are vague or not enforced, moderation is ‘outsourced, decontextualized’ and ‘the algorithmic clustering of social media content and groups’ can benefit anti-immigration actors (Nikunen, 2018: 13). The recent years’ controversies centred on Facebook’s role in distributing political misinformation (Merrill and Åkerlund, 2018) expose how the platform can be utilized for undemocratic, including racist, purposes.

Social media thrive on the production and circulation of ‘affect as a binding technique’ (Dean, 2010: 95). Affect can be understood as the encountering of emotions circulating between subjects in a communicative space. Affect is produced and distributed in the relation between subjects and bodies (Clough, 2008). It is shaped in communicative interactions and provides meaning, attention and a ‘little surplus enjoyment’ to users (Dean, 2010: 95). Engagement on social media platforms generates feelings of belonging and mutual recognition, but also of exclusion; these reciprocal emotions drive the circuits of communication, that is, constitute the gratification of user labour. Papacharissi (2015: 125) argues that affective publics are ‘networked public formations that are mobilized and connected or disconnected through expressions of sentiments’. She dismisses the idea that online racist groups can serve as examples of affective publics, since they suppress deliberation and pluralism (Papacharissi, 2015: 122). Nevertheless, affective publics can also be built around uncivil expressions, including racism and hate. In the American presidential election campaign of 2016, part of Trump’s ‘fan base also deployed racist, sexist, and anti-Semitic memes and tropes developed by “alt-right” white nationalists . . . develop[ing] a sort of weaponized, automated affective public’ (Karpf, 2017: 201–2). Consequently, social media publics with anti-immigrant and racist sentiments may not advance deliberation or pluralism, but they are nevertheless constituted on a mutual circulation of affective discourse and social connectivity. They assemble storytelling practices, real-life experiences and affective responses that build not only on sentiments of disbelief, resentment, fear and hate but also on understanding, recognition and in-group solidarity (Ekman, 2019). Ahmed (2004: 117) argues that emotions such as insecurity, fear, hostility or hatred align bodies of individuals within the network (creating a sense of belonging), while excluding other bodies. Communication in racist networks creates in-group and out-group identification. The construction of a common nation, national identity or ‘culture’ relies on the exclusion of ‘others’ based on fear (Ahmed, 2004). Anti-immigration and racist networks define these identities, cultures and behaviours through the continual construction of the threatening other (i.e. immigrants/refugees). A recurrent element in constructing the ‘immigrant’ is to depict violence against women as an ‘immigrant problem’ (Miles and Brown, 2003: 52). This has become particularly evident after the recent forced migration into Europe, where anti-immigration sentiments play on variations on the theme of male refugees as sexists, rapists and concurrently as (d)emasculated cowards (Ekman, 2018; Rettberg and Gajjala, 2016). In addition to this, people who express solidarity with refugees are depicted as naive, mentally disabled and sexually frustrated (Devlin and Grant, 2017).

Finally, since social media communication is characterized by affect and the circulation of emotions, it can also amplify expressions and articulations between users (Farkas et al., 2018). Fekete (2014) describes how racist ideas travel ‘from the fringe to the mainstream and back again’, a process that she defines as ‘cumulative racism’. It builds on the idea that racist views become acceptable when mainstream discourses increasingly target minorities and immigrants in various ways. Reciprocally, the increase of racist ideas from the fringe has an equivalent effect on mainstream discourses. Thus, this could be defined as a relational process between the mainstream and the far-right, in which the boundaries of publicly acceptable discourse are continually being pushed forward. In the following analysis, the idea of cumulative racism is related to user interaction in online networked publics. In order to understand how racism is shaped and fomented in public online comments and conversations, the article examines how sentiments (aimed at refugees and migrants) trigger and nourish racist discourse, which becomes amplified when circulating between users on Facebook.

The case: ‘SufS’

In order to assess the role of commercial social media in producing and circulating anti-immigration and racist discourse, the article uses the Facebook group ‘SufS’ as a case study. Its ‘position’ as a huge Swedish anti-immigrant forum is particularly interesting given the fact that Sweden has a long-standing history of open multiculturalism, including liberal policies on refugees and immigration. During the ‘refugee crisis’ of 2015, Sweden was a nation in the European Union that allowed the most (per capita) refugees to seek asylum, but from a Swedish perspective, the events of 2015 also led to harsher political attitudes towards immigration. This finally led, in November 2015, to Sweden ending its internationally recognized generous immigration policies. In the years following the refugee crisis, positive attitudes towards immigration and immigrants have decreased among the public, even if people with negative attitudes still are in the minority (Ahmadi et al., 2016). The growing negative attitudes towards immigration and immigrants are unmistakably visible online, including on commercial platforms (Ekman, 2018, 2019; Merrill and Åkerlund, 2018; Törnberg and Wahlström, 2018). ‘SufS’ was launched on 5 February 2017, as ‘Stand up for Peter Springare’ (‘Stå upp för Peter Springare’) in support of a criticized police officer. The stated aim of the group is ‘to support the Swedish Police in the struggle against rising crime and reveal the weakness politicians show towards the increasingly unsuccessful integration’ (Wahlström and Törnberg, 2019). SufS is a huge and open group that as of December 2017 had 171,000 members, making it one of the largest Swedish-language groups on Facebook. As noted by Törnberg and Wahlström (2018), the threshold for participation in an open discussion group on Facebook is low, but this does not diminish its importance as an arena for public discourse and political identity formation. The group has a significant impact on mainstream news diffusion online. Half of the 40 most-circulated mainstream news items during the period of 1 January to 18 April 2018 were shared on SufS (Dahlberg, 2018). Posts in the group frequently come under scrutiny for violating hate speech legislation, and recently the group’s moderator was convicted of hate speech because of comments published by other users (Rapp, 2019). However, according to the group’s own description, it forbids racism (Merrill and Åkerlund, 2018). The group is highly active, with thousands of posts published every day. In addition, each post generates a vast amount of user comments, creating a lively, multi-flow communication.

The collection of data was done in a two-step process. First, in order to get an overview of the posts in the group, including the communicative patterns, a quantitative content analysis was carried out (Riffe et al., 2014). The sample includes posts published between 3 November and 5 December 2017. The data collection is based on a systematic random sampling using a skip interval (1/10) until 300 posts (N) were obtained. The algorithms of Facebook (at that time) positioned posts that received a new comment at the top of the feed. Due to the high level of user activity in the group and the vast number of posts published every day, posts are continuously moving up and down in the hierarchy, depending on user interaction. Two student coders collected and coded the data, and in order to obtain descriptive statistics, the analysis covered seven dimensions. Facebook posts are frequently centred on sharing material from other sources, so external sources were coded for. In order to assess the overall characteristics of the posts, they were coded for topics, problems addressed and person(s)/subjects(s) addressed, and also for the presence of overt racism (if the post included racist slurs, etc.) Since racist and anti-immigration attitudes online are more common among male users (Ekman, 2018), the gender of the poster was coded for. Finally, three forms of interactivity (reactions, shares and comments) were assessed. The descriptive statistics provide an overview of the content, including what kind of characteristics drives the circulation of, and attention to, messages in the group. A sample of 30 units was coded by the two coders for inter-reliability measurement, yielding an inter-coder reliability level of .96 (Riffe et al., 2014).

Second, in order to analyse how racism and anti-immigration attitudes are moulded through communicative interaction among users, the analyses include a qualitative assessment of user comments on two posts published in the group. The two were strategically selected to highlight both similarities and differences; both included ‘immigrants/refugees’ and ‘crime’, but they covered a trivial and a severe event, respectively. The qualitative approach draws on critical discourse studies and assesses the socio-political dimension of language use in everyday communication on the micro level of SNS (KhosraviNik, 2017). The analysis was conducted in three steps. First, all the comments (n = 541 and n = 110) were retrieved manually and read through. Second, comments were thematically structured, yet keeping the original clusters of user conversation/interaction intact (reply comments are ordered chronologically below the ‘original’ comment and are here called responses). Third, the comments were analysed more closely, assessing the construction of ‘immigrants/refugees’ at the referential, predicational and argumentational levels (Reisigl and Wodak, 2016). The FB group is open and does not require membership or approval to read the posts. This means the data are publicly available. It also implies that users posting in the group are aware that their posts are available to the public. Nevertheless, for ethical reasons, user names are omitted in the analysis.

Results: Communication in ‘SufS’

In the subsequent section, an overview of descriptive statistics obtained from the sample (N = 300) will be presented. A majority of the posts circulate and comment on current news events, whether published by traditional news media or coming from far-right actors. An active link to an external source is contained in 51% of posts (n = 152), and of these 42% include links to traditional news media services and 29% links to far-right sites. Immigration is included as the main topic in 54% of all the posts, and 35% include ‘immigrant/s’ as the main subject/object. This can be compared to the second-most common subject/object, which is ‘politicians’, present in 16% of the posts. The most commonly addressed problem is ‘crime’ (27%), followed by ‘welfare problems’ (22%), ‘immigrant culture’ (19%) and ‘demographics’ (7%). When assessing the only demographic factor that can be extracted from Facebook profiles, we see that the vast majority of the posts are published by male profiles (75%). The preponderance of male users in the group corresponds with previous findings on digital communication and racism (Ekman, 2018; Merrill and Åkerlund, 2018). When assessing the presence of racism, the result shows that 21% of the posts include overt racism (such as racist slurs, etc.). In posts including overt racism, ‘immigrant(s)’ are the most common subject/object (73%). Worth highlighting is that Facebook clearly does not uphold its policy of prohibiting racism on the platform.

The average number of reactions (likes) per post is (M = 633/Mdn = 413) and the average number of shares is (M = 95/Mdn = 51); 57% of the posts generated more than 50 comments. When assessing what kind of content generates attention and circulation on Facebook, the results show that posts including the topic ‘immigration/immigrants’ generate the highest average number of reactions (M = 694/Mdn = 456) and posts including ‘immigrant(s)’ as the main subject/object generate the highest average number of shares (M = 116/Mdn = 52). Posts including the topic ‘economics’ generate the highest average number of shares (M = 195/Mdn = 91), and when relating economics to subject/object, the result (albeit not statistically significant) shows that the combination of the topics ‘economics’ and ‘immigrant/s’ as main subjects/objects yields the most shares. The addressed ‘problem’ of politicians and mainstream media ‘lying’ to the public generates both the most reactions (M = 935/Mdn = 652) and the most shares (M = 190/Mdn = 74). To conclude, posts including immigration/immigrants, and reports on crime drive the attention to and circulation of content in the group. Economics, that is, reports on costs of immigration/immigrants, plays an important role in generating circulation; hence ‘traditional’ anti-immigration discourse prevails. Also worth noting is the relative impact of posts accusing politicians and mainstream media of ‘lying’; thus, expressions of distrust towards social institutions underpin user engagement.

Analysis: Facebook comments as cumulative racism

Social media services such as Facebook are built on the content generated by their users, but they also advance networking practices such as reacting, sharing and commenting. Connectedness or ‘platformed sociality’ (van Dijck, 2013) is what constitutes the very fabric of Facebook. The practices of user responses draw attention to the profile (individual) who publishes content. In return, the responding user can draw attention to the reaction, comment and so on. Consequently, the affordances of Facebook facilitate connectivity, which in turn can transmute into a sense of recognition. User engagement with ‘uncivil’ publics can accelerate racist discourse through communicative interaction between users (e.g. Farkas et al., 2018). In addition, online group dynamics can amplify antagonistic and even violent attitudes and behaviours (Wahlström and Törnberg, 2019). In order to understand how user interaction contributes to negative attitudes and racism towards refugees, the subsequent section analyses user comments in two posts from the sample. Both posts cover the topic of crime in relation to alleged refugees/immigrants as subject/object, but depict a trivial and severe event, respectively. The analysis is structured to demonstrate how the interaction between user profiles develops.

Post 1: Sheep’s throats slashed in pizzeria

On 18 November 2017, a Facebook profile posts an embedded link to a news story published on the national public service broadcaster Swedish Television’s (SVT) news site. The news story is almost 2 years old at the time. The post includes two images from the news piece. The first depicts two dead sheep lying on a bloodstained floor, and the second shows a long bloody knife on a windowsill. The headline of the article states ‘Sheep’s throats slashed in pizzeria’. The original article is a news story about an alleged illegal slaughter that took place in the back room of a local pizzeria in the Swedish town of Falkenberg. The two animals had been killed by the use of traditional halal methods, but without the use of anaesthesia, making the slaughter illegal according to Swedish law. At the time of data sampling, the post had 2300 reactions, 542 comments and 513 shares. The FB profile had added his own statement to the linked news article: ‘Fucking bastards’ (user post). User comments start by demanding accountability for the actions described in the news article. Comments express emotions of distrust and resentment aimed at governing politicians, mainstream news media and the ‘politically correct left’.

This is supported by the government, media and courts. And the PC left. Disgusting. (User comment A)

Löfven’s voting herds. (User comment B)

And it does not look particularly hygienic!!!! What does the food legislation have to say! (User comment C)

They look the other way. (User response C)

‘Löfven’ refers to the current Swedish prime minister. The post advances the idea that the generous immigration policies of the government during the ‘refugee crisis’ were merely a strategy to gain future voters. This is a recurring argument within anti-immigration discourse. Comments also reduce the presumed suspects in the article to the level of thieves, although there are no reports of theft in the article.

They can’t be sane *devil emoji*. (User comment A)

They have probably stolen the sheep *angry emoji*. (User comment B)

They should be prosecuted for animal abuse and theft, the sheep are probably stolen from the farmers. (User comment C)

Yes, this is murder, assault, theft, animal abuse, etc. This is what we’re paying for. (User response C)

The comments articulate victimization through binary oppositions, that is producing ‘us’ as victims of the actions and behaviour of immigrants and refugees. ‘We’re paying’ for this, is part of the economic discursive logic that emphasizes the financial burden resulting from acceptance of immigrants and asylum seekers. The comments naturalize difference by stressing that the behaviour of the suspects is an inherent part of their culture and religion.

An old Swedish tradition don’t you think?! . . . Or, is this perhaps an act committed by new Swedes? (User comment A)

Those fucking bastards, one thing is sure, they aren’t Swedish. (User response A)

You grieve with these animals, damn multiculturalism. (User comment B)

Muslims, the only thing they eat *devil emoji*. (User comment C)

Immigrant culture and religion are given meaning by the classification of difference. ‘New Swedes’ and ‘Muslims’ become objects of referential distinction to native Swedes and non-Muslims. Boundaries are constructed through articulations of symbolic differences; that is, ‘this is not the behaviour of us’. Eventually, the comments become more vile, and finally the subjects of the news story become completely dehumanized.

Yes. To slaughter your own sexual partner in this gruesome way. (User comment A)

Halal slaughter the owners . . . turn them into crayfish bait. (User comment B)

You should treat animals kindly. (User response B)

Yes you’re right; we can’t feed the crayfish such trash. (User comment B response)

Cockroaches, crawl back to where you came from. (User comment C)

Pigs! (User comment D)

Pigs are not disgusting, they are nice and kind animals, call the idiots monsters instead. (User response D)

In the comments and responses, the subjects are deprived of human qualities through the association with animal behaviour. Depictions of them as ‘cockroaches’ and ‘crayfish bait’ are predicated on the necessity to eradicate vermin. Thus, the subjects are symbolically excluded from having any human qualities. The use of animals and vermin as metaphors mirrors historical references in war propaganda and genocide, in which similar signifying practices have legitimized atrocities against civilians and ethnic groups. Moreover, the fantasy of a deviant sexuality, predicated on the bodies of the other, is another element in constructing a symbolic order of acceptance and non-acceptance, and of exclusion.

Post 2: ‘One of the suspects: “I recorded to prove it was a voluntary act”’

On 29 November 2017, a Facebook profile posts a link to a news article published the previous day on SVT’s online news site. The headline reads ‘One of the suspects: “I recorded to prove it was a voluntary act”’. The linked article describes five men accused of a gang rape of a 30-year-old woman in a Stockholm suburb (known for its large immigrant population). The story does not reveal any information about the suspects, but their identities have been disclosed on various far-right blogs and websites. All five suspects are immigrants, and the case receives excessive public attention and outrage, both due to the cruelty of the actions and because the suspects were found not guilty in court (Bjarnefors, 2017). At the time of data sampling, the post had 592 reactions, 110 comments and 80 shares. In the post, the poster writes, ‘some pleasant reading *devil emoji*’ (user post). User comments begin by stressing the accountability of the government and public institutions for the actions described in the news article.

Löfven’s pension deliverers. (User comment A)

As if rape, assault and filming it aren’t enough! The stupid bastards admit to selling narcotics, on prime time TV! Totally fucking crazy that this stupid bastard is not locked up! What the hell are the law enforcement and judicial systems doing??????. (User comment B)

The comments immediately establish that the alleged perpetrators are immigrants, and in the subsequent comments, various forms of derogatory naming take place. ‘Bastards’ and ‘fucking darkies’, alongside other racist slurs and appeals to ‘throw them out’, are recurrent comments and responses. User comments also associate the sexual violence used by the suspects with all refugees (cf. Horsti, 2017), and users predicate ‘culture’ as the explanatory factor for the deeds committed by the suspects. The construction of an out-group culture also centres sexual deviance and predatory conduct as intrinsic parts of this immigrant culture. Comments also become more vicious, with both implicit and explicit calls for violence against the suspects.

All newly arrived [immigrants] should get an inflatable doll *flirty emoji*. (User comment A)

Of course, this is just an ordinary cocaine deal, paid for with a blowjob. In the new Sweden. (User comment B)

Don’t they cut off thieves’ hands in their culture? Since they stole something from this woman, let’s give them some culture. (User comment C)

Sweden should adapt to Islam, let’s cut off the limb they stole with. (User comment D)

Fucking perverts *three angry emojis and a pistol emoji*. (User comment E)

Finally, comments are saturated with dehumanizing statements. One comment actually uses the word explicitly: ‘I dehumanize these pricks’ (user comment A). Other users call for ‘lobotomy’, because ‘castration doesn’t work. They’ll get it up anyway *devil emoji*’ (user comment B). One comment contains an image depicting a White man loading a gun. The comment states ‘there aren’t enough guns being fired’ (user comment C). The call for violence as justified ‘self-defence’ is a recurrent trope in racist discourse (Ekman, 2018); this is particularly evident in relation to events involving immigrant/refugee violence against women. The analysis shows that communicative interaction triggers and augments racist utterances, and that the social interaction instigates users to post increasingly repugnant comments. The comments and responses form a pattern (though not linearly), where demands of accountability, aimed at politicians, mainstream media, public institutions and so on are the initial stage. By projecting ‘us’ (Swedes, Whites, etc.) as victims of the actions and behaviour of immigration and immigrants, the comments construct in-group victimization. Moreover, by producing out-group identities through endless negative stereotypes, immigrants and refugees are reduced to the deeds and crimes committed (or not) by individuals. The posts, and the comments they generate, produce an imaginary ‘other’ – the sexually perverted, barbaric, violent and cunning perpetrator (Horsti, 2017), thus naturalizing the acts and morality of individuals as ineradicable aspects of a certain culture, religion or ethnic group. At the end of the line, comments dehumanize immigrants and refugees through the articulation of symbolic boundaries and discursive violence. Immigrants and refugees are symbolically excluded as human beings and thereby are constructed as legitimate targets of violence. Interesting to note are the similarities between the comments to the two posts. Even though they depicted very different events, the comments and responses developed in a similar pattern, differing only in terms of aggressiveness.

Concluding remarks

It is evident that large and open Facebook groups can constitute ‘weaponized, automated affective public[s]’ (Karpf, 2017: 201–202). Moreover, since the socio-technical infrastructure of social media shapes racism in new ways (Matamoros-Fernández, 2017), it can be effectively utilized by anti-immigration and racist actors. Racist discourses rely on affect to gain attention and circulation, meaning that such publics as ‘SufS’ facilitate affective communication through social architecture and technical affordances. The group establishes a space where user interaction amplifies racist comments and responses published by Facebook users, turning racial antagonism (Farkas et al., 2018) into discursive violence. Consequently, open Facebook posts and the ensuing user interaction push the boundaries of what is publicly acceptable language on topics (supposedly) related to migration and migrants. By taking advantage of commercial social media, anti-immigration actors gradually normalize previously unacceptable attitudes and utterances, and as recent research suggests, ‘radical right-wing sentiments on social media may instigate and/or facilitate violent (anti-immigrant) political action’ (Törnberg and Wahlström, 2019: 2). This also suggests that racist discourses on social media do not exist in a vacuum; they are manifested in the actions of people participating in the production and circulation of racism online.