Online discontent: comparing Western European far-right groups on Facebook

The Internet offers a powerful tool for far-right groups for reaching followers, connecting with like-minded groups, and spreading their ideology (Caiani and Parenti 2013; Zhou et al. 2005). Social media also enables ‘ordinary’ citizens to actively participate in online discussions and express their opinions (Zuckerman 2015).

The issues that far-right groups and their followers are concerned about – such as Islamisation, European integration and immigration – are hotly debated in almost all Western European countries. They use different channels to exert influence – ranging from institutionalized politics to street marches. Right-wing groups differ in their organization (Minkenberg 2003), the issues they emphasize and whether they take a moderate or hard line (Akkerman et al. 2016).

Research on the far-right has focused on the electoral channel – political parties – and certain segments of the discourse – traditional mass media and party programs (for recent reviews, see Golder 2016; Mudde 2016; Muis and Immerzeel 2017). Far-right groups and their supporters often make use of social media, but research into this topic is still scarce (Ernst et al.2017). This study addresses these two lacunas in the literature. It poses two interrelated questions: first, what is the relationship between political opportunities and type of online mobilization (parties vs non-institutionalized groups)? Second, what are the differences in the online discourse between parties and non-institutionalized groups? To answer these questions, we analysed data from Facebook pages of British, German, French and Dutch far-right groups.

Our results show that far-right mobilization manifests itself in different organizational forms depending on the political opportunities for the far-right. Closed political opportunities can explain why social movements are more prominent online than political parties in some countries. These findings are in line with Mudde’s (2016) observation that since 2015, we have seen a strong presence of non-party organizations offline in Germany and Britain, most notably Pegida and English Defence League (EDL).

Furthermore, our comparison of comments of online followers on a selection of far-right pages reveals that the discourse is shaped by the type of actor. Parties focus more strongly on anti-establishment populism, whereas non-institutionalized groups tend to address issues of immigration and Islam more strongly. Concerning the radicalness of rejecting non-natives, our findings indicate that the debate is generally more extreme on pages of non-institutionalized groups.

The core feature of far-right groups is exclusionary, ethno-nationalist xenophobia (Rydgren 2005). Far-right discourse is nativist, which refers to a combination of nationalism and xenophobia: emphasizing one’s own culture, traditions and nationality, and negatively portraying culturally deviant outgroups (Mudde 2007). Far-right groups can be extreme or radical. Extremist groups are anti-democratic, they go beyond the limits of the democratic political processes. Radical groups only question some of the constitutional foundations of liberal democracies (Golder 2016; Mudde 2007). Hence, the radical right is democratic, in that it accepts popular sovereignty and majority rule (Mudde 2016). Far-right actors are not necessarily anti-establishment or populist. Nevertheless, the combination is very likely (Greven 2016).

Far-right mobilization takes different organizational forms (Caiani et al. 2012; Hutter 2014; Minkenberg 2003). Parties have increasingly become movement-like, networked organizations (Chadwick & Stromer-Galley 2016; Della Porta et al.2017). On social media, ‘political communication of social movements, parties and ordinary citizens is meshed together much more fluidly than in a mass media setting with traditional gatekeepers’ (Chadwick 2013, as cited by Stier et al.2017: 1368). Content is often posted by followers (in the form of comments) instead of by groups (in the form of posts). Followers are likely to post similar content on different pages. Stier et al. (2017: 1381), for example, show that the same users are active on the pages of the party AfD and the movement Pegida. However, as comments are reactions to what is posted by page owners, discourse between pages varies. As Atton (2006) showed of the webforum of the British National Party, moderators might steer conversations or remove comments.

As we will explain, the type of actor is relevant for understanding differences in discursive positioning against non-natives and political elites. We distinguish parties (institutionalized) from non-institutionalized groups, which range from more organized social movements to looser online communities. In contrast to parties, movements do not take part in elections. They do have members and organize meetings or protests. Online communities are social aggregations with similar values and interests that only exist online (Shen et al.2006). The differences between non-institutionalized groups are sometimes fuzzy. Caren et al. (2012: 167) identify a form of activism that bridges movements and online communities: the social movement online community (SMOC). They define SMOC as ‘a sustained network of individuals who work to maintain an overlapping set of goals and identities tied to a social movement linked through quasi-public online discussions’.

To explain variation in far-right mobilization, it is not the grievances per se that matter – we assume that discontent about immigration is similar across Western Europe. Rather, it is whether far-right groups can act effectively on existing grievances within the opportunities offered by the political-institutional setting (cf. Koopmans et al. 2005; Muis 2015; Kincaid 2017).

The political opportunity structure (POS) is typically defined by the openness of the political system, the stability of elite alignments, the presence of allies among the elite and state repression (McAdam et al.1996). Scholars applying the concept specifically to far-right parties have pointed at the electoral system and at ‘political space’ left open by political competitors (Koopmans et al. 2005; Muis and Immerzeel 2017).

Strong far-right parties are more likely in an open political context. In closed contexts, far-right groups have to find alternative ways to mobilize and express their grievances, resulting in more non-parliamentary action groups (Hutter 2014). In addition to joining social movement organizations, people can establish or join online communities when they feel that the political arena neglects their concerns (Awan 2016). The political opportunity model assumes that actors make a rational choice among alternatives. For instance, street violence is a relatively costly strategy, which is mitigated where less costly alternatives are available (Koopmans 1996). The electoral channel seems to effectively substitute for street activity and violence (Minkenberg 2003; but see Jäckle and König 2017). Thus, we expect:

Furthermore, we expect that open political opportunities will yield a more concentrated network, structured around a prominent institutionalized actor. Most non-parliamentary groups will establish links with this actor since they ‘treat parties as their champions in the electoral arena’ (Arzheimer 2015: 8). More extreme movements strategically link to prominent right-wing parties, as these might serve the ‘hard-core groups in their recruitment efforts’ (Burris et al.2000: 323). In contrast, closed political opportunities result in a segmented network, with several prominent non-institutionalized actors. Consequently, we predict that:

Far-right groups differ in the issues they emphasize and the extremity of their frames. Cultural ‘outgroups’ (nativism) and political elites (populism) can be frequently discussed or not, depending on the type of group. Movements tend to articulate issues that are left unaddressed by institutional actors, and vice versa (Hutter 2014: 26). Moreover, non-institutionalized groups tend to be more ideologically outspoken and single-issue focused than its party variants (Minkenberg 2003). Parties need to address a broader spectrum of issues than social movements to attract broader support (Przeworski and Sprague 1986). Additionally, far-right parties often downplay controversial issues. Constraints imposed by cultural norms are more relevant for groups that mobilize via the parliamentary channel (Minkenberg 2003). Therefore, we expect far-right parties to focus more on anti-establishment populism, rather than on the exclusion of non-natives (Jagers and Walgrave 2007; Stier et al.2017). Hence:

Far-right groups furthermore differ in their extremity of discourse (Maan et al.2017). As we pointed out, parties aim to appeal to a broader support base. The expectation is that parties adopt a more moderate discourse, specifically on nativism, than non-party actors (Golder 2016; Rydgren 2005). Ideological extremism tends to decline when groups become more entrenched in the electoral channel (Minkenberg 2003; Piven and Cloward 1979). Groups without formal membership or structures tend to adopt more extreme positions (Goodwin et al.2012: 6) and more activist forms of mobilization online (Caiani and Parenti 2013: 107). Consequently, far-right parties often distance themselves from movements. In Austria, for instance, the FPÖ considers integration and assimilation possible, while this is unacceptable for the Identitarian Movement (Maan et al.2017). We expect thus:

The countries included in this study – The United Kingdom, Germany, France and The Netherlands – vary in party systems and standing of far-right parties (cf. Engesser et al.2017).

Table 1 illustrates the electoral successes of far-right parties in each country when we collected our data.

According to Kitschelt (2007: 1191) ‘the institutional configurations most unfavourable to the rise of radical right-wing parties exist in Britain’. The First Past the Post electoral system limits possibilities to achieve electoral success. Far-right parties have historically been isolated by hostility from mainstream parties and have failed to gain and consolidate many parliamentary seats (Ford and Goodwin 2014).

In Germany, a threshold hampers niche parties in gaining electoral success. Moreover, mainstream parties refuse to form coalitions with far-right parties (Backes and Mudde 2000). At the time that we collected our data, parties, such as the NPD, Die Republikaner and the AfD, were relatively marginal (Arzheimer 2015). In the 2013 General Election, the AfD did not pass the electoral threshold.

The French two-round majoritarian electoral system is more forgiving to niche parties than the British electoral system, but still hinders the electoral success of the far-right on the national level. Potential allies have always rejected any electoral agreement with the far-right (Caiani and Parenti 2013). The Dutch electoral system of proportional representation with a low threshold offers parties such as the PVV the most favourable opportunities to gain electoral success (Akkerman et al.2016; Muis 2015). We thus conclude that during the time of our research, political opportunities for the far-right were closed in the UK, open in the Netherlands, and intermediate in France and Germany.

Data were retrieved from Facebook using Netvizz (Rieder 2013). We mapped the online networks using a snowball technique (cf. Caiani and Parenti 2013). We first gathered the direct links for the following initial sample of Facebook pages: BNP; UKIP; Britain First; Infidels of Britain; I am proud to be British (United Kingdom); Alternative für Deutschland; Identitaire Bewegung; Pegida; Wir sind das Volk (Germany); Génération Identitaire; Bloc Identitaire; Patriotes de France; Front National (France); Nederland mijn Vaderland, Wij zijn Tegen het nieuwe regeerakkoord; PVV aanhangers; PVV; Identitair verzet (the Netherlands). The selection was based on earlier research and by searching for pages within Facebook using keywords. Subsequently, the links of pages that were discovered though the crawling of this sample were also gathered to generate a bigger network and identify less well-known groups.

The networks contained several pages that might fit the far-right ideology in specific contexts, such as homophobic pages in the French network, and animal rights and anti-paedophilia pages in the British network. We excluded these pages, irrelevant pages (such as those of brands) and pages from different countries, as to allow a better comparison between the networks in the different contexts.

Network analyses were performed using Gephi (Bastian et al. 2009). To identify influential groups in the network (H1), we calculated the number of likes that a Facebook page receives from other pages in the network – the in-degree (Caiani and Parenti 2013).

A cluster analysis was performed to map the clusters making up the far-right network and reveal the segmentation of the network (H2) (Zhou et al. 2005; Caiani and Parenti 2013). As our hypotheses indicate, we are mainly interested in which type of actor (institutionalized versus non-institutionalized) dominates the online network. To further substantiate our network analysis, we also provide the main topics these groups focus on. Using the picture, banner and info-section on the Facebook pages, groups were categorized as focusing on immigration, Islam, nationalism, national- or European anti-establishment populism, or on supporting a far-right party or politician. Pages addressing multiple issues were categorized as general. The page Ban the Burka in UK was, for example, categorized as a page about Islam. A random selection of 55 pages was hand-coded by both authors to assess the intercoder reliability (Krippendorf’s alpha: 0.96).

For the discourse analyses, five pages – at least one party and a variety of non-institutionalized actors – from each country were selected. For each actor-type, we selected the most popular pages, based on likes of users. One prominent page – the English Defence League – could unfortunately not be included in the selection for technical reasons.¹ For all pages, except for the extremely active Pegida and Nederland mijn Vaderland, data covering a year (March 2014–March 2015) were gathered (Table 2). Because of this long timespan, we expect that specific events do not significantly influence the findings. Most pages included in this analysis have now either been permanently removed, or have been reinstalled, thus making earlier posts inaccessible.

We analysed comments instead of posts, as posts often solely consist of images, links or videos. Furthermore, comments are posted by ‘ordinary people’ and reflect more diverse opinions than posts. Automated text analyses were done using Python’s Natural Language Toolkit (cf. Bird et al.2009).² Comments were pre-processed by removing punctualization, uppercases and stopwords.

To analyse the salience of topics (H3), frequency analyses were performed. We used thematic word clusters in English, French, German and Dutch around the issues of nativism (immigration, Islam) and anti-elitist populism (the national and European political elite). Clusters contained synonyms and misspellings, such as ‘immigrants’, ‘foreigners’ and ‘foreners’ for the immigration cluster.³ We selected words for the clusters after manually reading a random selection of 100 comments from which words related to these issues were identified.

To analyse the extremeness of claims by far-right groups (H4), the most often occurring words in comments containing cultural ‘outgroups’ and political elites were analysed. In addition, we carried out qualitative analysis, consisting of the authors reading 50 comments from each page.

We first describe the networks in the four countries. Subsequently, the prominence of actors in each network is linked to the POS in which these operate. The French and German online networks are much bigger (810 and 476 pages) than the Dutch and British networks (69 and 225 pages). This is important to note, since the network size influences the maximum in-degrees.

We expected that a closed political structure, such as in Britain, would result in non-parliamentary groups being more prominent in the online network (H1). The street movement EDL indeed represents the most ‘visible’ online actor for expressing far-right views. We expected parliamentary groups to be more prominent in online networks in open political contexts, such as the Netherlands. The Dutch network, however, consists mainly of online communities and action groups. Nevertheless, several Dutch groups clearly revolve around support for Geert Wilders – the attitudes and identities of followers are tied to this far-right party. Analogous to Caren et al. (2012), we could conceptualize this form of online activism as ‘a political party online community’. Germany and France have an intermediate POS. In both countries, the online networks have a combination of both parties and movements.

Furthermore, we hypothesized that open political structures yield dense networks, structured around prominent far-right parties (H2). The Dutch network, however, lacks a dominant institutional actor around which it revolves. The French network is centralized around FN. We expected more segmented networks in closed political contexts, in which a dominant far-right party is less likely. This is confirmed in the German context, which consists of many demarcated clusters, with few intra-group links between parties and movements. However, the British network is rather dense and structured around the movement EDL.

After describing the discourse of far-right groups in each country, we draw a conclusion on whether similarities and variations are related to the type of group. Table 4 presents the number of times words related to the national political elite, Europe, Islam and immigration are mentioned per 10,000 words. Tables 5– 8 show word co-occurrences of these ‘targets’ of the far-right for a selection of pages. Combined with our qualitative analysis, they show how strongly non-natives and the political establishments are rejected by online followers of the far-right.

On pages of the movements Infidels of Britain and Britain First, the Islam is clearly the most salient issue. In contrast, the comments on the UKIP page suggest that its followers focus more on the political elite. The national elite and Europe are discussed 120 and 131 times per 10,000 words, respectively (Table 4). Islam (24) and immigration (68) are mentioned less often. The issue of populism is also remarkably salient on the community page Proud to be British.

In contrast with our expectation, our results suggest that the British National Party deem nativism, rather than anti-elitism more important. Even though the party has adopted a more moderate discourse over time, it maintained its core ideology of racial prejudice (Edwards 2012). The discourse seems to be more extreme on the pages of Infidels of Britain and Britain First than on the other pages. This is visible in the kind of words that are most often used in relation to Islam and immigrants (Table 5), such as terrorists, fuck and scum. Allah is referred to as ‘evil and Satan in disguise’. References are made to bomb, detaining and deporting Muslims. Islam itself is referred to as cult, curse and plague. For example:⁴

Comments on the German pages of the movements Pegida and Identitäre Bewegung both mention issues related to Islam most often, respectively, 94 and 64 times per 10,000 words. Alternatively, the comments on the AfD page mainly address the European (65) and national political elite (47). Islam (1) and immigrants (14) are mentioned less often. Interestingly, both populism and nativism are salient on the community pages Ich bin Patriot aber kein Nazi and Wir sind alle Deutsch.

In line with the focus of German citizenship on ‘origin’ (Koopmans et al.2005), Islam and immigration are frequently discussed in relation to incompatibility with German nationality and Christian traditions (Table 6). One follower of Alternative für Deutschland for example argues: ‘we are part of the Christian tradition (…) Do you want to betray our past to open the gates to Islam?!’

Nativism is clearly most extreme in the discussions on the pages Pegida and Identitaire Bewegung. Users express a ‘fear of Islamisation of Germany’. Some comments are even calling for killing ‘all followers of Islam’ and accuse the government of genocide of their ‘own people’. Paradoxically, this follower of Pegida criticizes the ‘dangerous ideology’ of Islam by associating it with Nazism:

Interestingly, we hardly notice any differences between pages in terms of the extremeness of discourse. The French online far-right debate about Islam emphasizes ‘adaptability’ to French culture and stresses that public manifestations of religion should not be allowed. Moreover, Islam is frequently referred to as a cancer, fascism, and Muslims are sometimes referred to as paedophiles. The political elite is blamed for letting in immigrants and the Islam in. Some express the fear of a future war against Islam in Europe. Community pages are often less extreme in their frames.

Lastly, if we turn our attention to Dutch far-right pages, it is remarkable that anti-elitist criticism is the most salient topic on almost all of them. On the unofficial PVV page, for instance, the political elite is mentioned 146 times per 10,000 words, which is more often than immigrants (24) and Islam (45). The movement Identitair Verzet slightly differs from this observation. The comments mention Islam much more often than on the other pages (138).

The Dutch discourse on immigration is mostly related to socio-economic topics, criminality and healthcare. Moreover, critique on the Islam is generally discussed by emphasizing the lack of adaptation to Dutch norms and values. In this respect, the anti-Islam movement Identitair Verzet again differs from the other pages because the discourse on the Islam is rather extreme, for example:

As expected (H3), nativism is less salient in the online discourse of far-right parties than of non-institutionalized groups. Followers of UKIP, FN and AfD generally mention the political elite relatively often compared to movements and communities and focus less on Islam and immigrants.

The Dutch case differs from this picture, since discontent on Dutch far-right online communities also strongly revolves around anti-elitism. The open political opportunities for the far-right in the Netherlands could perhaps explain this. Wilders has considerable electoral support. Consequently, his anti-Islam stance figures prominently in the traditional media and political debate. At the same time, however, all established parties ignore or criticize the PVV.

Parties and non-institutionalized groups also differ in the extremeness of the discursive practices of their followers. As expected, we find that British and German movements employ indeed a more extreme nativist discourse than parties (H4). In France, however, differences in extremeness between the different types of groups are not as evident. FN and the Identitarian Movements seem to hardly differ from each other in this respect, whereas French community pages are often less extreme in their frames than both FN and movement pages.

This article investigated the networks and discourses of far-right Facebook pages in Western Europe. Our findings show that limited political opportunities for far-right parties yield more online engagement in protest politics in Britain, and that intermediate political opportunities in Germany and France result in online networks with both parties and movements. The open POS is not reflected in the Dutch online far-right network, which consists mainly of online communities.

The clustering within the far-right networks reflects the offline realm to some extent: Germany has a fragmented network composed of small separated clusters, including different far-right parties that have some success at the regional level; the French online network is relatively cohesive and centralized around a party. Overall, however, the structure of the networks cannot be clearly linked to offline political opportunities for the far-right. Most notably, in contrast with what we expected, the closed British political context has a dense, centralized online network structured around a prominent movement (EDL). Allen (2011: 286) notes that its exceptional, inclusive strategy led to a broad, ‘multicultural’ coalition: ‘unlike other far-right organizations, the EDL is proud to recognize and proclaim its diversity’.

Secondly, our research analysed the discursive practices of supporters of far-right groups. Our findings show that the comments on pages of far-right groups clearly differ between types of actors. Online followers of parties discuss the national and European political elite more often and focus less on immigration and Islam than social movements and communities. Furthermore, discourse on party-related pages is generally less extreme than on pages of movements. Thus, the distinction between institutionalized and non-institutionalized groups is relevant. Additionally, our study highlights that there are different types of non-institutionalized groups. Except for the French case, community pages were generally more anti-elitist compared to social movement pages. Moreover, community pages were often less extreme in their discursive practices than movements.

Sometimes the division between parties, movements and communities cannot be easily made. Different types of groups sometimes overlap: for instance, some community pages focused on supporting a specific party or leader (e.g. PVV supporters) and some groups are so-called ‘party-movements’ (e.g. Britain First). Further research could focus more on the overlap and differences between types of online far-right groups.

Our comparison of far-right online groups across four European countries serves as a robustness check to determine whether our findings hold in different contexts. No claims are made about how well the four countries represent all Western democracies, but given their variation in terms of far-right mobilization and political-institutional context, the findings could perhaps be generalizable beyond the four countries at hand.

All studies have shortcomings and our study is no exception. Both quantitative and qualitative content analyses have their limitations. First, although automated text analyses are very reliable, frequent co-occurring words do not provide a deeper understanding of discursive practices. For instance, sarcasm is difficult to detect. Therefore, we also conducted a qualitative analysis and manually read comments. The downside of this method is that we did not manually code comments. Further research could incorporate qualitative and automated analyses more systematically.

Far-right online mobilization might differ between platforms, depending on legal regulations about the harmful speech. Further research could extend cross-group and cross-national comparisons to other platforms. Due to privacy restrictions, we only selected public pages and ignored closed Facebook groups (Cleland et al.2017). Moreover, like private discussions at home, closed groups cannot be considered as belonging to the ‘public’ discourse we studied in this paper.

Finally, our sample of followers of far-right groups might not represent the views of all far-right supporters, or of offline membership. This is not necessarily a weakness. The opinions and statements of far-right Facebook members are interesting themselves. Nowadays, people increasingly acquire their opinions via social media. The online world is not a separate world, but part of the real world, influencing offline interactions and decisions (Awan 2016; Golder and Macy 2014). Online political involvement of far-right supporters generally extends offline, where they ‘are far more likely to get involved in political activism compared with the general public’ (Bartlett et al. 2011: 22). Facebook, thus, ‘acts as a bridge to offline activism rather than a replacement’ (2011: 40). Studying far-right groups online is, therefore, an important and valuable addition to studying them offline.

No potential conflict of interest was reported by the authors.

Ofra Klein is a PhD researcher at the European University Institute. She previously studied Social Sciences and Digital Humanities. Ofra worked as a research assistant at Vrije Universiteit Amsterdam and the Berkman Klein Center for Internet and Society at Harvard University. Her research interests are in the field of right-wing populism.

Jasper Muis is an assistant professor at the Sociology Department of the Vrije Universiteit Amsterdam. His research interests include right-wing populism and protest behaviour. His dissertation (2012) about the rise of right-wing populist Pim Fortuyn in the Netherlands received the prestigious Dutch Research Prize of the Praemium Erasmianum Foundation.