Social Media, Populism, and Migration

Several scholars have attributed high hopes to social media regarding their alleged ability to enable a nonhierarchical and freely accessible debate among the citizenship (Loader & Mercea, 2011; Shirky, 2011). Those hopes have culminated in theses such those describing the social web as being a ‘new public sphere’ (Castells, 2009, p. 125) as well as in expectations regarding its revitalizing potential for the ‘Habermas’s public sphere’ (Kruse, Norris, & Flinchum, 2018, p. 62). Yet, these assumptions are not uncontested, particularly in the light of socially mediated populism (Mazzoleni & Bracciale, 2018). Interestingly, research on populism in the social web is still an exception. The same is true for the populist permeation of the social media discourse on migration, as a highly topical issue. This study seeks to elaborate on this research gap by examining to what extent the Twitter debate on the Global Compact for Safe, Orderly, and Regular Migration (GCM) was permeated by populist content. For this purpose, almost 70,000 tweets on the most important Hashtags referring to the GCM that took place in Marrakesh in December 2018 were collected and the 500 widest-reaching tweets analysed in terms of their populist permeation. Against initial expectations, the empirical findings show that populist narratives did not dominate the Twitter debate on migration. However, the empirical results indicate that ordinary citizens play an important role in the creation and dissemination of populist content. It seems that the social web widens the public sphere, including those actors who do not communicate in accordance with the Habermasian conceptualization of it.


Introduction
It is widely assumed that social media has become an integral part of the public sphere (Castells, 2009;Papacharissi, 2009;Shirky, 2011). Castells (2009, p. 125), as one of the most prominent proponents of a network society, even speaks of the emergence of a 'new public sphere.' However, there are also critical voices (Kruse, Norris, & Flinchum, 2018) stating that citizens do not engage on social media in a way that matches the wellknown criteria from the concept of 'communicative rationality' as established by Habermas (1984Habermas ( , 1989. Already in the late 1990s and early 2000s, researchers were able to demonstrate that illiberal voices were more visible on the Internet than liberal ones (Hill & Hughes, 1998, p. 153) and hypothesized that the "growth of vir-tual counter-public spheres" (Downey & Fenton, 2003, p. 199) would lead to an increasing impact of rightist actors on the public debate. This article intends to contribute to the long-running discussion on the influence of digital media on the public sphere by analysing the populist permeation of social media discourse on migration. The focus on populism and migration seems reasonable since populist communication strategies have to be seen as the exact opposite of the Habermasian ideal of communicative rationality and thus as a challenge for the public sphere (Privitera, 2018). The populist challenge is especially noticeable in the public debate on migration, which has recently become a major issue (Heidenreich, Eberl, Lind, & Boomgaarden, 2020, p. 1261. While there are studies on the degree of populist pervasion of mass media (Rooduijn, 2014), as well as on the mass media coverage of migration (Eberl et al., 2018) and the social media discourse on migration (Heidenreich et al., 2020), no research concerning social media, populism, and migration has been conducted so far.
The main empirical objective of this article is to form a better understand the characteristic of the populist discourse concerning migration on social media. For the purpose of this article, the global Twitter discourse on the intergovernmental signing conference of the Global Compact for Safe, Orderly, and Regular Migration (GCM) in Marrakesh in December 2018 will be examined. This study will focus on Twitter due to its significance for the agendas of the mass media and thereby for the wider perception of political issues. It has been shown that mainstream news media frequently respond to trending topics on Twitter (Araujo & van der Meer, 2020, p. 634), which can be observed especially well concerning the mass media coverage of tweets by the US American President Donald Trump (Fuchs, 2017, pp. 53-54). The usage of Twitter as a source for news media is very common among journalists. One might even speak of "a reversed agenda-setting pattern" (Araujo & van der Meer, 2020, p. 647). Thus, debates on Twitter influence the public debate outside the Twitter realm.
The following questions will guide the investigation: (RQ1) how strongly was the Twitter discourse on migration permeated with populist content during the signing conference on the GCM? Since this article deals with migration, in other words, with the core issue of rightwing populist actors (Jagers & Walgraave, 2007, p. 322;Lutz, 2018, p. 517), this article is especially interested in the number of right-wing populist tweets. If it emerges that there were (right-wing) populist tweets, the following question arise: (RQ2) which populist communication strategies were employed? Thus, this study sheds light on the degree of populist content in the online discourse regarding migration (content dimension: What is published?) and about populist communication strategies (strategic dimension: How is populist content published?). It needs to be stressed that this article is not solely interested in social media activities of political actors who are well-known for their populist attitude, but also in those of 'ordinary' citizens since their populist activities also have to be examined if one wants to fully understand populism on social media. Moreover, the concept of the public sphere requires the "equal and protected participation" (Kruse et al., 2018, p. 63) in public debates. Since social media is credited with free access and a non-hierarchical structure of debate, it seems to match those key features of the Habermas's public sphere. Even though the focus of this article is on social media and populism, it aims to make an empirical contribution to the primarily theoretic debate on the public sphere.
In Section 2, a brief explanation of the GCM is presented. The following part of the article intends to present a conceptualisation of the term 'populism,' as well as an overview of the current state of research on populism and social media. A specific focus is placed on the role of non-elite actors in the realm of populism since this aspect has been widely neglected by academic research. Before the empirical part of the study is described in detail (Section 5), I will give an insight into the research design of this article and its methods (Section 4). The article concludes with a summary of the main findings and an indication of further aspects to be examined (Section 6).

The Global Compact for Safe, Orderly, and Regular Migration
The GCM is seen as an extremely important compact in the realm of international migration policy since it is the "first negotiated United Nations document to address migration governance comprehensively and to receive wide support" (Bloom, 2019, p. 481). The GCM is the result of a UN Resolution-the so-called New York Declaration of 2016 (United Nations Human Rights, 2016). This declaration places the GCM in the wider context of the United Nations' 2030 Agenda for Sustainable Development. The New York Declaration launches a sequence of intergovernmental consultations and negotiations towards the development of a GCM. This subsequent process concluded with the adoption of the GCM by 164 UN member states at a signing conference in Morocco. The formal endorsement by the UN General Assembly took place on December 19th (International Organisation for Migration, 2019). The agenda, as well as the compact, stresses the potential of migration for enriching the societies of the recipient countries. The main objective of the GCM is the reduction of risk and vulnerability during the migration process by establishing an international framework. The final version of the GCM includes several mentions of the state's human rights obligations concerning migrants (Guild, Basaran, & Allinson, 2019;United Nations, 2020). The closing statement of the signing conference by Louis Arbor (Special Representative of the Secretary-General for International Migration) includes the notion that the GCM has to be seen as "a re-affirmation of the values and principles embodied in the UN Charter" and as a commitment "to safer and fairer ways of managing borders" (United Nations, 2018). Although the GCM is not legally binding, and although the principle of state sovereignty remains unaffected (Bloom, 2019, p. 482), its adoption was preceded and accompanied by an extensive and mostly adversarial debate in several countries which even lead to some governments to refuse to sign it (e.g., US, Hungary, Poland, Austria, Italy). The US government, for example, rejected signing the GCM with reference to its right to decide "how to secure its borders, and whom to admit for legal residency" (US Mission to the United Nations, 2018). Other governments made a similar argument. The potential loss of sovereignty seems to have influenced the debate, although the compact in its own right has given no cause for this fear.

Populism and Social Media
Despite the frequently mentioned ambiguity in terms of the populism concept (Müller, 2016, p. 2), one finds central elements that have been part of most definitions of the concept. One of the oldest, though most important elements of populism, is the "reference to an analogical basis which is the people" (Laclau, 1979, p. 165). The appeal to 'the people' as a very basic aspect of populism confronts researcher with the problem that 'the people' "is a concept without a defined theoretical status" (Laclau, 1979, p. 165). This might be one reason for the "ambiguity surrounding 'populism"' (Laclau, 1979, p. 165). Bearing this in mind, a meaningful conceptualization of populism, as a prerequisite for a valid operationalization (see Section 4), needs additional definition criteria. Thus, almost all conceptualizations of populism contain anti-elitism as a second very basic criterion. Anti-elitism is, in the words of Canovan (1999, p. 3), "directed not just at the political and economic establishments but also opinion-formers in the academy and the media." It seems important to understand that "populism challenges not only established power-holders but also elite values" (Canovan, 1999, p. 3). Mudde (2004, p. 543) sums up the aforementioned conceptualization fragments well when he states that populism "considers society to be ultimately separated into two homogeneous and antagonistic groups" and in which the socalled pure people are exploited by a corrupt elite. Thus, populists prefer politics to be "an expression of the volonté générale (general will) of the people." The last aspect refers to the alleged goal of populists, namely to ensure that the will of the people is enacted against the resistance of the so-called elite. Analysing the recent academic debate on populism, it seems fair to conclude that the aforementioned core elements of the long-standing discussion about a useful conceptualization of the term populism-people-centrism, anti-elitism, and restoring the sovereignty of the (good) people-are more or less consensual (Meléndez & Kaltwasser, 2019, p. 520). Those elements point out that "populist politics is not ordinary, routine politics," but politics with an "extra emotional ingredient," that "can turn politics into a campaign to save the country or to bring about a great renewal," as Canovan (1999, p. 6) put it. Right-wing populism can be understood as a specific characteristic of populism (Krämer, 2017(Krämer, , p. 1295. One of its special characteristics is its strong emphasis on 'the others' (Bobba, 2019, p. 16). This aspect refers to groups other than the elites who are suspected of endangering "people's values, identities and rights" (Bobba, 2019, p. 13), namely to "different ethnicities, cultures, or nations" (Krämer, 2017(Krämer, , p. 1296. Most of the research on populism has been conducted in terms of traditional mass media. Academic research on populism and mass media has revealed, among other things, an "affinity between a populist communication style and 'media logic"' (Sorensen, 2018, p. 2). Interestingly the same seems to hold for populism and the logic of social media. Bartlett (2014, p. 106) considers that populist actors would profit from the prevalence of said new digital media since their inherent logic would match perfectly with their communication style. There is plenty of research on the logic of social media, from which we know that "emotional, controversial, even violent content typical of much populist activism" (Mazzoleni & Bracciale, 2018, p. 3) is shared by users of those social networking sites much more frequently than messages with a balanced expression of opinion. This is why populist messages frequently have such a resonance on social media (Bobba, 2019, p. 11). Due to the horizontal network architecture of Twitter and other social media tools, messages can spread to huge publics: "In the context of Twitter, a handful of hops in a retweet chain is enough to reach a substantial audience and saturation is usually reached within one day" (Araujo & van der Meer, 2020, p. 636).
As journalists are especially interested in trending topics, populist news on social media is very likely to spill over in the traditional media system and thereby extend the reach of a tweet dramatically. Moreover, populist tweets often contain a higher news value, since journalist tend to select negative and sensational content, which has been well described by the theory of news selection (Galtung & Ruge, 1965;Lippmann, 1922;Östgaard, 1965). The key news factors, described in the theory of news values, also seem to be an important selection criterion for the public, as Araujo and van der Meer (2020, p. 637) put it with reference to Eilders (2006). Thus, populist messages on Twitter and other social media platforms are very likely to be exported and noticed beyond the boundaries of their origin (van Dijck & Poell, 2013). Araujo and van der Meer (2020, p. 647) witness a dynamic interplay between Twitter and traditional media. The connection between traditional media and digital media was prominently described by Chadwick (2017) who named this new communication eco-system a 'hybrid media system.' Especially "'elite' populist actors" (Krämer, 2017(Krämer, , p. 1294, such as political parties, know about the different media logics and the connection between the traditional and the new media and apply a populist communication style, not only because it matches their worldview, but also strategically to spread their messages on social media and traditional media. This strategy comes as no surprise, since populist actors, as Mazzoleni and Bracciale (2018, p. 3) put it, "have everywhere constantly relied on the visibility and the ensuing popularity assured by the coverage, both critical and supportive, of mainstream and popular media." It should be noted that social media is used by populist actors to "bypass the forceful power of mass media" (Mazzoleni & Bracciale, 2018, p. 3), as well as to influence their agenda (Casero-Ripollés, Sintes-Olivella, & Franch, 2017, p. 689).
The majority of the studies in the field of social media and populism deal with the communication of well-known populist politicians and political parties and in particular with right-wing populist actors (Arzheimer, 2015;Engesser, Ernst, Esser, & Büchel, 2016;Mazzoleni & Bracciale, 2018;Stier, Posch, Bleier, & Strohmaier, 2017). This focus is quite reasonable since the usage of populist communication strategies is most obvious in the context of professional politicians who are known for their populist attitude, such as Matteo Salvini from the rightist Italian party Lega Nord (Mazzoleni & Bracciale, 2018, p. 4). Right-wing politicians are particularly successful on social media, as several authors, including Bobba (2019) with regard to Italy, demonstrates. Ernst, Engesser, Büchel, Blassnig, andEsser (2017, p. 1356), have furthermore shown that "politicians belonging to right parties use populist communication strategies most frequently" (see also Jagers & Walgraave, 2007). Besides research on "'elite' populist actors" (Krämer, 2017(Krämer, , p. 1294, only minor attention has been given to "non-elite actors" (Krämer, 2017(Krämer, , p. 1294). Yet, most of the social media users are just average citizens. To the best of the authors' knowledge, so far only Hameleers andSchmuck (2017, p. 1425) distinguish "between populist politicians and ordinary citizens as communicators of populist messages" in an empirical article. The authors (Hameleers & Schmuck, 2017, p. 1428) examine the effects of populist social media messages on the attitude of Austrians and Netherlanders in terms of blaming the government and migrants. A distinction is made between the effects of politicians' messages and messages from ordinary citizens. The authors are able to demonstrate that participants of the study who identify with the sender of a populist message, accept this message and become themselves more populist (Hameleers & Schmuck, 2017, p. 1437). This finding is true for messages from politicians and ordinary citizens, which clearly indicates that theoretical as well as empirical knowledge on populist online activism of ordinary citizens is necessary in order to fully understand the populist penetration of today's (virtual) public sphere (Krämer, 2017(Krämer, , p. 1296. The need to fill this research gap has become even more urgent since the finding of Taggart (2018, p. 81), who demonstrates that a populist attitude is a potentially facilitating factor for many forms of political engagement. Anduiza, Guinjoan, and Rico (2019, p. 110) state that it is not distrust and frustration (Meléndez & Kaltwasser, 2019, p. 521) nor the feeling of being humiliated by the elites (Eberl & Seubert, 2017, pp. 4-5), but it is the framing of elites as evil forces that political engagement thrives on. Engagement against the hideous elite thus may become a moral imperative. Furthermore, it is plausible that the populist idea of a political system in which the people should express their will directly exerts a positive effect on the political engagement of the bearer of this idea. Consequently, populist attitudes seem not to facilitate conventional modes of political participation, like voting, but more direct forms, "such as petition signing and the online expression of political views" (Anduiza et al., 2019, p. 109). These findings indicate that citizens with a populist identity are particularly active in non-electoral times, which has also been shown to be true by Pirro and Portos (2020). Since Facebook, Twitter, and all the other newly emerged tools are easy-to-use, with cheap and wide-reaching information and communication technologies, they have to be seen as perfect instruments for professional and non-professional actors with a populist attitude to disseminate their political positions.
The findings presented so far indicate that the social web is far from being able to match the Habermasian criteria of communicative rationality. However, there are strong-albeit mostly theoretical-voices that speak in favour of the social web, as the new realm in which a nonhierarchical and truth-seeking discourse might take place (Loader & Mercea, 2011;Shirky, 2011). The debate on the deliberative potential of social media has not yet been finalized. Section 5 will shed some more light on this debate by examining the Twitter discussion on migration regarding its populist permeation. Before, an insight into the research design and methodology is given.

Sample
For the analysis, almost 70,000 tweets (including metadata, such as the sender's information) on the ten most important transnational hashtags on migration were mined (N = 69,153). The following hashtags have been streamed, using the R package streamR which simplifies access to the Twitter application programming interface (Barberá, 2018): #GlobalCompactMigration, #Global Compact, #UNMigrantCompact, #GCM, #Migration Compact, #CompactOnMigration, #Marrakesh Declaration, #ImmigrationPact, #MassMigration, and #MigrationPact. The data collection phase for Twitter communication on migration lasted for two weeks from December 5th, 2018 to December 19th, 2018, covering the period before the signing conference of the GCM in Marrakesh took place and the days following the conference. Since several tweets were simply re-tweets with (exactly) the same content, only the widest-reaching doublets were kept. Thus, the sample was reduced to 14,566 tweets.

Measuring the Populist Permeation of the Twitter Discourse
Since populism has to be seen as a complex construct, automated methods should not be used for coding the populist content of a tweet. Thus, most empirical studies on social media and populism work with manual coding (Ernst et al., 2017). This article will follow in these (methodological) footsteps. As it is hardly feasible to code each tweet on migration, an indicator is needed to identify the level of populism in the Twitter debate on migration: the populist permeation of the 500 most popular tweets in the sub-sample of 14,566 tweets. The popularity of a tweet is measured by its number of re-tweets and appreciations (so-called 'favoritings'). The share of tweets that can be classified as populist will be investigated (RQ1). If it emerges that there are populist tweets, the question arises: Which populist communication strategies have been used (RQ2). In addition, an automatically conducted word pair analysis of the initial sample of almost 70,000 tweets will be applied to check the validity of the results from the manual coding procedure.
To classify populist tweets, a content analysis coding schema is used that was established by Ernst et al. (2017). See Table 1 in the Supplementary File for this article, which is available online. The schema consists of nine sub-categories of populism or rather communication strategies, which represent the three core dimensions of populism: (1) anti-elitism, (2) people-centrism and (3) restoring the sovereignty of the people. Due to the article's special interest in the occurrence of rightwing populism in the context of the migration debate on Twitter, an anti-migration dimension is added to the coding schema that consists of a modified set of the anti-elitist sub-categories. By using that schema, one or more than one sub-dimension of populism can be assigned to each tweet (unit of analysis). Since the existence of a single dimension hardly indicates that any particular tweet is populist, only those tweets that contain at least one communication strategy from at least two of the three core dimensions of populism (anti-elitism, people-centrism, restoring the sovereignty of the people) are coded as populist, furthermore, tweets need to additionally contain at least one anti-migration statement to be coded as right-wing populist. Figure 1 illustrates the coding approach. The first statement is coded as an anti-elitist statement, since it calls the Irish government treacherous (discrediting the elite) and the second statement as people-centrist, as it states a monolithic people and stresses the people's virtues. Because there are only two populist dimensions in this tweet, it is coded as a populist one, but not as a right-wing populist one due to the lack of an anti-migration statement. It should be noted that the coding was only conducted by one coder. This surely has to be seen as a methodological shortcoming.

Bot or Not
In the academic, as well as in the public debate the activities and the influence of fake profiles on Twitter are subject of controversial debate. This is why many academic researchers use bot detection applications. Yet, frequently used programs to detect bots have recently been criticised by Twitter for their "extremely limited approach" (Roth & Pickles, 2020), which may cause "false negatives (i.e., bots being classified as humans) and false positives (i.e., humans being classified as bots)," as Rauchfleisch and Kaiser (2020, p. 1) put it. With regard to inauthentic content from human Twitter users (e.g., state-driven trolls), the situation becomes even more complex. Yet, there is recent research indicating that "that their effect on social platforms was minor" (Zannettou et al., 2019, p. 218). Interestingly, journalists, who are very important disseminators of Twitter content, seem unfamiliar with 'fact-checking and verification services'-despite the controversies on the malicious use of automation on Twitter and the alleged influence of trolls. The same seems to be true for the public (Varol & Uluturk, 2020, p. 98). It would seem to follow that neither the majority of professional opinion-makers nor its recipients distinguish between authentic and inauthentic content on Twitter. This comes as no surprise since it is actually not that easy to distinguish between authentic and inauthentic activity. Taking all the above arguments into account, an examination of tweets regarding their authenticity (was this particular tweet created by a bot, a troll or an authentic user?) is not conducted. However, it needs to be stressed that the empirical results of this study potentially might have been influenced by inauthentic content. This has to be taken into account when reading the following section.

Empirical Findings: The Populist Permeation of the Twitter Dialogue on Migration
First of all, it is interesting to note that all of the 500 widest-reaching tweets are re-tweets. Most of them refer to professional political actors. Yet, only 12 tweets match the applied operationalization of a populist tweet (2.40 percent) and only two the conceptualization of right-wing populism. Additionally, it should be noted that their reach has to be considered as rather low since only one of those tweets initiated more than 1,000 retweets and favoritings (see Table 2 and Table 3 in the Supplementary File). Nevertheless, it is interesting to note that 10 of those 14 (right-wing) populist tweets contain at least one anti-elitist and one people-centrist statement. This indicates that anti-elitism and people- centrism is the most frequently used populist combination in the discourse on migration. Furthermore, we found both dimensions occur most frequently (Figure 2), yet, in most cases not combined. This finding supports the classical conceptualization of populism, in which antielitism and people-centrism have been seen as the two very basic elements of populism (Canovan, 1999;Laclau, 1979). By taking a closer look at the anti-elitist dimension, one finds that blaming and discrediting elites occur the most. Regarding the people-centrist dimension, stating the existence of a monolithic people has to be seen as the most prominent communication strategy.
A prominent example of a populist tweet is one created by the leader of the UK Independence Party in Wales stating that "any politician who willingly signs up to the #MigrationCompact has betrayed their people" (Shelly, 2018). This tweet contains the two core elements of populism: anti-elitism (all politicians that sign the GCM are fraudsters) and people-centrism (the people will be betrayed). The description of elites as being treasonous to the people, who are described as sharing common feelings, desires and opinions on the GCM seems to be a popular strategy of populist actors in the realm of the Twitter debate on the GCM since this accusation can be found in several other populist tweets. There is talk about, e.g., the "trea-sonous…Government" (O'Neill, 2018), government acting "is against its own people" (Martin, 2018), "High Treason against" (Walters, 2018) the people, betrayal of "our children" (Sabhat28, 2018), or the people who are united by their rejection of the GCM (see Table 2 and Table 3 in the Supplementary File).
Furthermore, some Twitter users allege "a conspiracy of silence" (Gormally, 2018) of the "corporate media" (O'Neill, 2018) and politics and as being "on blackout mode" (O'Neill, 2018) with regard to the GCM. It is alleged that "the media and political establishment so far have refused to inform or consult with the…people" (Gormally, 2018). In addition, it is alleged that the GCM prohibits criticism of migration ("under this pact, stories like this will be banned"; Shields, 2018). These findings support the thesis of Canovan (1999, p. 3), who states that populism not only challenges political actors but also those actors, who allegedly disseminate elite values against the will and the basic interests of the people. Moreover, toying with the human fear of the unknown and the stranger is detected. One author writes that by signing the GCM "#ISIS terrorists" (Walters, 2018) will be imported and another that it "could extinguish" (Tenacious, 2018) the identity of his country. The GCM is also described as a possible threat to the sovereignty of the people and as a compact that "dictate(s) immigration policy and removes sovereignty" (Greine, 2018), as well as a contract that gives "control of our borders and sovereignty to a corrupt unelected bureaucracy" (Orchard, 2018). The two right-wing populist tweets contain as additional elements anti-migration strategies (see Table 3 in the Supplementary File), namely the (implicit) description of migrants as "non-citizens" (Shields, 2018), welfare recipients and "terrorists" (Walters, 2018  tweets indicate that there has indeed been a populist "campaign to save the country" (Canovan, 1999, p. 6). Yet, it needs to be stressed (again) that not many tweets were found that contained at least two populist statements representing at least two different core dimensions of populism and almost none that additionally contained at least one anti-migration statementas already mentioned. However, it is interesting to note that 13.6 percent of the tweets analysed contain at least one populist sub-category and thereby a soft form of populism, as Mazzoleni and Bracciale (2018, p. 7) have called social media messages with "only one populist reference." This finding is in line with the finding of Ernst et al. (2017) who found slightly more than ten percent of their social media sample contained at least one populist reference.
The results of an automated analysis of the most frequently used word-pairs in the overall sample of almost 70,000 tweets support the result of the manual examination of the sub-sample (N = 14,566 tweets), because no word-pair was found that would have indicated that the discourse on migration was strongly permeated by populist narratives (see Table 4 in the Supplementary File). Yet, the results indicate that the Twitter discourse on migration during the signing conference was predominantly critical against the compact, since the fifth, sixth, and seventh-most frequently used word pairs refer to a video of demonstrators against the GCM in Brusselsone of the most popular tweets in the sample (Figure 3). Furthermore, the word pair 'protest rt' occurs among the top 10 word-pairs, indicating that tweets showing protest against the GCM occur very frequently (rt means 're-tweet').
Examining the Twitter profiles of the senders of the 12 populists' tweets, it seems like those users are ordinary citizens (see Tables 2 and 3 in the Supplementary  File). Among other indicators (no obvious affiliation to a certain political party; no self-description as being a politician etc.) the moderate number of followers (range: 144-6,565) supports the assumption that we are indeed referring to normal citizens and not famous people or well-known professional politicians. This observation is interesting, as it indicates that ordinary citizens can be quite influential on social media. Moreover, it shows that populist communication in the digital sphere needs not to be the exclusive domain of professional politicians or political organizations, but also the matter of ordinary citizens. Yet, one has to recognize that those ordinary citizens did not create their own content but retweeted that of other users. Interestingly, 5 of the 12 populist tweets were initially created by ordinary citizens. Furthermore, the tweets of professional political actors, such as the leader of the UK Independence Party in Wales, were more re-tweeted and favorited than those of normal citizens. Nevertheless, it is striking to find populist tweets from ordinary citizens that were disseminated with the help of other ordinary citizens. Those non-professional actors seem to act as important disseminators for populist content and as creators of populist narratives. Almost the same is true in case of the two right-wing populist tweets. Both were created by non-elite actors with a rather low number of followers and both were re-tweets. However, the re-tweeted users were no ordinary citizens but a (former) Irish journalist and a nationalist party from Canada.

Conclusion
In light of the aforementioned results, the remarkable and maybe surprising result is that the Twitter discourse on migration during the period of the signing conference on the GCM has not been dominated by (right-wing) populist narratives (RQ1). Although the debate on the GCM must be seen as predominately critical against the compact in general and against migration in particular, only very few (right-wing) populist tweets were found among the most popular tweets. Yet, it is interesting to note that most of the few (right-wing) populist tweets contain anti-elitist and people-centrist statements. It seems like those dimensions are, as previous work has already indicated, the most important ones in the realm of online populist communication (RQ2). The populist (re-)tweets identified were published by ordinary citizens and not by professional politicians. Furthermore, half of the populist tweets were initially created by other ordinary citizens. Yet, one-third of the populist tweets identified were originally created by professional political actors, indicating that ordinary citizens also seem to be very important disseminators of (right-wing) populist content created by professional politicians.
Finding populist content among the widest-reaching tweets demonstrates that while the populist penetration of the discourse on migration on social media might be less strong than expected, professional, as well as non-professional populist actors might influence the attitudes of a large number of people. We have to keep in mind that tweets might create populist attitudes among their recipients (Hameleers & Schmuck, 2017, p. 1437. Moreover, journalists looking for a good story (Chadwick, 2017) might have used those few (right-wing) populist tweets to indicate a larger critical assessment of the GCM on social media. Thus, even when the empirical results reveal only a few populist contents and even when some of this content might be created by bots or trolls, this content might have been very influential. Interestingly, the findings indicate that populist online activism goes beyond the organized groups, movements, etc., of civil society. We might face a new type of political activism making use of the social web to spread (right-wing) populist narratives that do not match the assumptions being made in the concept of the public sphere in terms of an alleged civil character. It follows that social media indeed seems to provide free access for the citizenship and thus the opportunity to participate in political issues, such as in the case of the GCM. However, such kind of populist participation can hardly be seen as enriching the public sphere, but rather as an element of its corruption. This finding is in line with previous empirical work on the role of social media for the concept of the public sphere (Kruse et al., 2018, p. 82).
Yet, it needs to be emphasized that the presented findings are limited to Twitter. Against the empirical data presented by Ernst et al. (2017Ernst et al. ( , p. 1357, one might assume that the migration dialogue on Facebook is penetrated to a greater degree by populist statements. Unfortunately (at least from a scientific point of view), Facebook has shut down most parts of its application programming interface. Thus, it is virtually impossible to determine whether the presented findings are also true for the migration discourse on Facebook. One option would be the (re-)examination of already queried Facebook data under the aspects of this article. Furthermore, there is a huge research gap in the usage of other social media tools than Twitter and Facebook by populist actors, such as Telegram, WhatsApp, and Instagram. In particular, Telegram and WhatsApp seem to be interesting cases, as it has already been shown that those tools were extensively (mis)used by right-wing actors (Davis & Straubhaar, 2020;Urman & Katz, 2020).