Migrant Racialization on Twitter during a border and a pandemic crisis.

This work examines how the outbreak of the COVID-19 pandemic reshaped the migration debate on Twitter. Through co-hashtag network analysis, time-frequency and content analysis, it shows that the pandemic was related with positive (humanitarian) and negative (threat) stances about migration. The positive side focused on the need to protect refugees stranded at camps in Greece from COVID-19. The negative focused on the Greek-Turkish land-border crisis (Evros crisis), using COVID-19 to reinforce migrants as racialized others. These findings fit the problematization of positive and negative migrant representations in the Global north as Eurocentric. In the case of camps, refugees fit well within the victim/helpless frame, justifying humanitarianism, this time on health grounds. Regarding the border crisis, refugees also fit the Eurocentric frame of violent/male/inferior other who could spread a deadly virus. Overall, COVID-19 intertwined with migration in Twitter debates, reinforcing the racialized, Eurocentric representational field on migrants from the Global south.

Introduction

The study of the relation between traditional media and migration is a well-travelled trail (Cohen, 2002; Cooper et al., 2020; Xu, 2020), which is increasingly focusing on social media and migration (Ekman, 2018; Lee and Nerghes, 2018; Ozduzen et al., 2020; Rettberg and Gajjala, 2016) and their intersections (Pöyhtäri et al., 2019; Siapera et al., 2018). The representational field, offline and online, around migrants and refugees in the Global north moves between securitization and humanitarianism (Siapera et al., 2018) shaped by Eurocentrism and orientalism (Avraamidou, 2020; Xu, 2020). As argued, media remain trapped in post-colonial mentalities representing Europe as civilized or generous and its former colonies as underdeveloped and weak (Cioban, 2016). There is a representational paradox of in/visibility and silence (Nikunen, 2019: 411), because even when present in news and other journalistic genres, migrants and refugees remain absent. With the onset of the COVID-19 pandemic, media engaged anew in processes of migrant othering like distinguishing between the deserving Muslim and the undeserving, culturally alien migrant not complying with social distancing (Poole and Williamson, 2021). Simultaneously, disinformation and misinformation about COVID-19 in mainstream and social media went hand-in-hand with hate speech and stigmatizing migrant communities (UN, 2020; Yücel, 2021) and certain ethnicities (e.g. demoting references to the virus as ‘Chinese flu’, Carter and Sanford, 2020), and fake news of ‘others’, as either immune or disease carriers (Pelizza, 2020). The stakes of this exclusionary mediated discourse are grave, particularly during a pandemic, ranging from furthering social inequalities to exacerbating tensions, increasing human rights violations, including that of life. This work examines the ways the outbreak of the pandemic in Europe related with the migration debate on the microblog Twitter and how and if it remained within the two conflicting poles that dominate traditional and social media representations of migrants from the Global south.

Delineating the nexus between COVID-19, an infectious disease which people carry across borders, and human migration from a communication perspective can offer deep insights to our understanding of how a health crisis affects meanings and representations of migration and the social media participation around them. In so doing, the study draws from two bodies of literature: critical social media and digital sociology, and critical migration studies. Specifically, it concurs with critical approaches that recognize the role of social media in the reproduction of restrictive ideologies (Fuchs, 2017) and with the argument offered by critical migration studies that refugees and migrants are positioned in European contexts of securitization as ‘victim/pariah, to be “saved” and “suspected”’ (Kyriakides, 2017: 1).

Twitter is ideal to address the study's aim, as it is a space of contestation, where meaning-making processes take place (Lindgren, 2017) and which has been repeatedly found to amplify racism (Cisneros and Nakayama, 2015), antimigrant stances (Avraamidou et al., 2021, Ozduzen et al., 2020), nationalism, fascism (Fuchs, 2017, 2019), antisemitism (Gantt, 2017; Jakubowicz, 2017) and white supremacist ideologies (Siapera, 2019). Notably, the abundance of relevant content led the social media giant to suspend accounts of white supremacists (Dearden, 2020), which of course did not bring about an ideological katharsis, as relevant policies only lead to exclusionary voices migrating to other platforms (Urman and Katz, 2020).

The appropriateness of Twitter relates also to its wide use during crises to receive and spread information in real-time. The events investigated essentially constitute ‘crises’ at Europe's borders: the Greek-Turkish land-border crisis in the area of Evros river (Evros crisis) and the COVID-19 pandemic which started almost simultaneously early 2020. “End of February 2020,” Turkey suspended its infamous 2016 migration deal with the European Union (EU), foreseeing the return of all irregular migrants crossing from Turkey to Greek islands. Subsequently, numerous people attempted to cross from Turkey to EU Member States Greece and Bulgaria. The Greek government cited the pandemic and hermetically closed its borders, eventually suspending asylum procedures. Paramilitarists joined security forces at the borders with right-wing extremists from across Europe. On various media and social media, each side of this new border crisis accused the other of weaponizing refugees and human rights abuses. EC President Ursula von der Leyen visited the area 3 March 2020, declaring Greece as Europe's shield. On February 26, 2020, Greece confirmed its first COVID-19 infection. During March 2020, most EU Member States entered lockdowns, shutting their borders, to varying degrees, to human passage.

By combining critical social media approaches and critical migration studies in examining Twitter participation and meanings around the border and the pandemic crises, we are able to expose well-embedded understandings that otherwise could go unseen, and to expose the predominance of restrictive ideologies and the role of social media in their reproduction.

Predominant representations of migration in traditional and social media: between threat and humanitarianism

Migrants and refugees remain mostly invisible in the traditional media of the Global north, as even when they gain media attention, their voices are rarely heard (Georgiou, 2018). Still, there is a spectrum of positive and negative representations: the deserving/underserving migrant and/or refugee; the threatening/vulnerable migrant or refugee and the migrant or refugee as an asset/burden to the host society. In all, migrants appear as a largely de-humanized outgroup (Chouliaraki and Zaborowski, 2017). These representations reproduce colonial logics, as media in the West ‘never escape the orientalist presuppositions that seem to perpetuate a Western imagination of the refugee as a voiceless ‘other’ (Malik, 1996, cited in Chouliaraki and Stolic, 2017).

The literature on negative media coverage in the Global north is prolific. Media associate migrants with security threats (KhosraviNik et al., 2012; Lynn and Lea, 2003), cultural threats (Cohen, 2002; Kadianaki et al., 2018), economic threats (Bauder, 2005; KhosraviNik et al., 2012) and health threats (Ana, 1999), constructing them as a deviant group (Cohen, 2002). The economic threat relates to associating increased migration to unemployment in the reception population and with receiving social benefits at the expense of the host societies (KhosraviNik et al., 2012). Security threats relate increasingly with terrorism and other forms of criminality (Holmes and Castaneda, 2016; KhosraviNik et al., 2012), and they are enhanced by sensationalist language (Holmes and Castaneda, 2016). Specifically, media link different types of crime to different ethnicities: Arabs or Middle Easterners are cast as terrorists in Europe and the US (Cooper et al., 2020), and Latinos as deceitful in the US (Chavez, 2001). This is the typical way of migrant racialization in the media. Media were also found to associate migrants with diseases, equating them to an illness (Ana, 1999) or accusing them of spreading diseases (Cisneros, 2008; Henry and Tator, 2002; Leudar et al., 2008) and therefore dehumanizing them.

Positive media coverage relates to the promotion of humanitarianism and utilitarianism. Specifically, humanitarianism in media concerns the promotion of care and compassion to migrants (Chouliaraki et al., 2017); UK media emphasized justice for migrants when their human rights were overwhelmingly violated (KhosraviNik et al., 2012), and in Canada, migrants were represented as needed in some economic sectors, but still unwanted (Bauder, 2005). During the 2015 refugee crisis, national media in Europe stressed generosity towards refugees (Dahlgren, 2016; Kadianaki et al., 2018), and UK and US media emphasized migrant contributions to the economy while underlining the alarming numbers of people drowning in the Mediterranean Sea to raise sympathy (Avraamidou, 2020). However, by depicting migrants merely as sufferers or as economic assets, media contribute towards legitimizing discrimination against those not fitting in these frames (Kadianaki et al., 2018), like sketching economic migrants as undeserving for not fitting the refugee frame (Kyriakides, 2017). The routine mobilization of celebrities to promote humanitarianism has been problematized for reproducing neo-colonial logics related to gender norms (Hopkins, 2018) and following market logics (Chouliaraki, 2013).

The Internet and its technologies offered new spaces for heated debates over migration in which antimigrant stances prevailed. For example, the far-right on Twitter represented migrants as cultural threats to Europe (Froio and Ganesh, 2019), and Brexit debates, again on Twitter, proliferated antimigrant positions in the UK (Miller et al., 2016). The 2015 so-called refugee crisis boosted scholarly interest in these debates, and studies showed that the securitization and humanitarianism debate dominated Twitter at the time (Siapera et al., 2018). Tweets using refugee-related hashtags tended to be more positive, whereas those using migrant-related hashtags were more negative (Nerghes and Lee, 2018). Positive hashtags such as #safepassage and #humanrights promoted support to refugees – particularly children – whereas hashtags such as #islamisttheproblem and #refugeesNotWelcome promoted exclusion, associating migrants with Islamic terrorism (Gualda and Rebollo, 2016).

Migrant representations moved also across a spectrum of positive and negative orientalist frames with the onset of the pandemic (Poole and Williamson, 2021). For example, UK media acknowledged the role of migrant and non-migrant Muslims in the National Health Service, while still speculating that many would not comply with social distancing due to their culture, which was a ‘neo-orientalist’, Islamophobic framework (Poole and Williamson, 2021). As argued, ‘The mediatization of Asian-ness’ as contagion has been a large part of the information environment surrounding the 2020 COVID-19 pandemic (Kuo et al., 2020). But this, at least in the US, was challenged in social media via antiracist and anticapitalist feminist positions (Ibid.). Racializing discourses of migrant immunity (e.g. ‘black immunity’, Pelizza, 2020) and as contagious dominated. German media on Facebook accused migrants of spreading the virus to locals (Boberg et al., 2020), while fake news stories that ethnic minorities are immune were diffused through traditional and social media in Italy (De Nardi and Phillips, 2021). Again in Italy, but also in the UK, the radical right used Twitter to spread antimigrant hate speech as part of a wider populist agenda (Caiani et al., 2021).

Unlike traditional media, social media provide tools for citizen coordination (Fuchs, 2017) and facilitate practical migrant solidarity (Dahlgren, 2016). Still, they routinely exhibit an abundance of antimigrant stances and ideologies which may be explicit or implicit (Siapera, 2019).

The study’s approach

This study is interested in how one crisis (border) may feed understandings and practices of the other (pandemic), and vice versa, on social media. In so doing, it rejects deterministic approaches to the Internet and its technologies, including social media, recognizing that, while they have a democratic potential (Papacharissi, 2002), they still play a significant role in the reproduction of restrictive ideologies (Fuchs, 2017). Specifically, the co-existence of antithetical representations concerning migration (e.g. humanitarianism vs. threat) in social and traditional media in various contexts in the Global north does not imply that the debate on migration is democratized. By contrast, the democratic or emancipatory potential of social media to actually challenge the status-quo is limited (Fuchs, 2017) and similar to that of traditional media, because of structural constraints (Hall, 1982). Both traditional media and social media function in specific divisive and unequal social structures in which racism is embedded. As argued ‘notwithstanding the many efficiencies that digital technologies and media have brought to migrants and racialised people, because these technologies have emerged in the context of techno-capitalism within a neoliberal paradigm, they end up not only reproducing existing racial inequalities but also generating new ones’ (Siapera, 2019, p. 104).

The study also concurs with the argument that refugees and migrants are positioned in European contexts of securitization as ‘victim/pariah, to be “saved” and “suspected”’ (Kyriakides, 2017: 1). European colonialism played a significant role in racializing colonial subjects (De Genova, 2017; Fanon, 1967), and in the reproduction of the ideology of racism. Racism is centred around the assumption of a deterministic relationship between a group and certain negative (real or assumed) characteristics, which justify its unequal treatment (Banton and Miles, 1996: 310). Even associating migrants with diseases relates to colonial logics of the contaminated other (McFarlane, 2008). Progressively, culture became a homologue of race to exclude others (Malik, 1996). Overall, discourses on migration exemplify the complexity of race-nation-ethnicity and their co-articulation with gender, sexuality, and class (Rattansi, 2005). Therefore, in delineating the social media-migration nexus, the concept of racialization is useful, as it refers to the ‘signification of some biological characteristic(s) as the criterion by which a collectivity may be identified…. [T] he collectivity is represented as having a natural, unchanging origin and status, and therefore as being inherently different.’ (Kyriakides et al., 2019: 3).

‘Europe's borders’, both the material and the symbolic site of the events investigated, are heavily surveilled through traditional military means and data-centric technologies that curb rights (Sánchez-Monedero and Dencik, 2020). They are places of systematic violations of racialized others’ right to life (Brian and Laczko, 2014), reflecting and reproducing the concurrent European apartheid, or Fortress Europe (Ponzanesi and Blaagaard, 2011). Therefore, Europe's borders resemble the North American border, specifically in militarized zones, which Walia (2013) conceptualized as border imperialism where oppression consists of killings, torture and arbitrary arrests.

Methodology and methods of analysis

This study was initially guided by an open research question, namely of how the beginning of the COVID-19 pandemic and the Greek-Turkey land-border crisis intersected on Twitter.[1] This open question sought for complexity to provide a comprehensive understanding of the phenomenon studied. Therefore, the methodological approach was shaped by the need to respond to the gamut of data, where they were generated, and their type. Some of the literature, especially on the 2015 so-called European refugee crisis and Twitter, has applied a big data approach interested mostly in the affordances of the medium applying social sentiments analysis (Öztürk and Ayvaz, 2018), social network analysis and other statistical analyses (Siapera et al., 2018) and sentiment with social network analysis (Ferra and Nguyen, 2017). Other studies focused on qualitative analysis (Bozdag and Smets, 2017) and discourse analysis (Bennett, 2018; Gallego et al., 2017; Rettberg and Gajjala, 2016). Although each approach offers insight into different aspects of the crisis-Twitter nexus, the affordances of Twitter are shadowed in the second approach, which emphasizes content, and exemplified in the first, which overshadow the content. To avoid this limitation, we follow Lindgren (2017) in conducting a well-designed, theoretically informed methodological bricolage interested in the form and the content of the debate. Additionally, whereas the scholarly trend is to focus on one event/crisis, we take a cross-event approach (Eriksson and Lindgren, 2018), focusing on a migration and a pandemic crisis. Therefore, this study has a more descriptive element, consisting of co-hashtag network analysis, hashtag analysis, time frequencies analysis and quantitative content analysis, and a qualitative element, consisting of thematic analysis on boundary making. Both aim at telling something about the meaning-making process on Twitter about COVID-19 and migration. In so doing, our broader question is split into the following inter-related research questions: To meet its aims, the study adopted a network perspective or sensibility, as we are ‘exploring culture in formation’ (Markham and Lindgren, 2014: 3). The study concurs with Eriksson and Lindgren (2018) that ‘[h]ashtags and their ecologies are user-created, just like discourse and culture more generally, and must be approached with similar methods.’ (p. 10). When hashtags co-occur, then their relationship becomes part of the users’ meaning-making process within the Twitter discourse.

How and in what ways have the outbreak of the COVID-19 pandemic and migration inter-related on Twitter during the Evros crisis?

When did the COVID-19-related hashtags become more prominent in terms of quantity?

When and how was COVID-19 used to exclude and/or include migrants and refugees (positive/negative way)?

In what negative ways has the content of migration debates on Twitter intertwined with COVID-19?

We recognize a set of limitations. First, Twitter is but one of many social media, and definitely not the most popular one or with the most diversified pool of users (e.g. 70% of users are male, Omnicore, 2021). It is also overtly studied (e.g., it is the most studied platform in race/racism studies (Matamoros-Fernández and Farkas, 2021). Therefore, focusing only on Twitter admittedly offers a glimpse of the debate in social media, while a combination of platforms would have highlighted the spectrum of the debate and advance diversity in the field. This limitation can be addressed by future works. Second, the issue of language is both a limitation of our sampling method and of our analysis. Inevitably, due to our fluency in two languages, our results concern mainly tweets in English and Greek. Third, the sampling method as a whole related to the availability of data from Twitter. Using the streaming API to collect data with its inherent limitations arguably gave us a representative sample (Siapera et al., 2018), although it did not provide the entire universe of data. Fourth, focus on hashtags (SNA) has limitations, as we excluded other relevant tweets that did not use hashtags. Nevertheless, as discussed, we are confident that, by using multiple methods (traditional and digital), we were able to address the particular research aim and overcome the limitations of conducting merely co-hashtag network analysis (Tufekci, 2014). In terms of ethics, we protect user anonymity and privacy (Eriksson and Lindgren, 2018).

To summarize, our research questions were answered through co-hashtag network analysis, hashtag analysis, timeline frequency distribution and quantitative and qualitative content analysis. Their combination offers insights into the study's overarching aim about whether and how the pandemic was used to re-produce or re-shape pre-existing representations of migrants.

Data collection and methods of analysis

Step 1 Data collection & co-hashtag network analysis

We collected tweets using a combination of relevant keywords (e.g. Greece asylum, Greece border(s), Greece migrant(s) and Turkey migrant) through DMI-TCAT (Borra and Rieder, 2014) from February 26, 2020 (suspension of migration deal and first COVID-19 case in Greece) to April 10 2020, when the volume of tweets diminished significantly. This resulted in 1,940,701 tweets and 504,004 distinct users. Co-hashtag analysis was performed on co-occurring hashtags, tweeted at least 100 times, resulting in Figure 1.[i]

Figure 1.

Co-hashtag analysis entire network: COVID-19 nodes in green (29,868 nodes, 57,326 edges).

Step 2 COVID-19 social network through co-occurring hashtags

Within the collected tweets, we found 306 COVID-19 related hashtags (used at least twice) and used them as keywords to form a COVID-19 subset of 126,850 tweets. We opted for keywords rather than hashtags, because we intended to study tweet content and hashtags. Co-hashtag analysis (see Figure 2) resulted in a network containing 282 nodes (one hashtag per node) with 7935 edges (connections between two hashtags). The resulting network contains hashtags used at least 100 times, co-occurring at least once. Using the modularity class algorithm in GEPHI, we identified seven communities.

Figure 2.

Co-hashtag COVID-19/ migration communities.

Step 3 Frequency distribution of COVID-19 keywords

We used the 306 keywords to make a timeline to show how many times per day COVID-19 keywords and/or hashtags were used in a single tweet (see Figure 5).

Figure 5.

Daily frequency of tweets referring to COVID-19.

Step 4 Quantitative and qualitative content analysis

To understand the meanings that linked COVID-19 and migration, we performed content analysis on a purposefully selected Twitter data-set. Building on time-frequency distribution analysis and social network co-hashtag analysis, we formulated RQ3. To address it, we focused on the first three days for which we had data and for which we expected that they would mostly cover Evros (28 February to 01 March 2020), and on the 16 March 2020 peak. Period A amounted to 4310 tweets and Period B to 7058 tweets, which we assigned two broad codes: (1) stance on migrants/refugees (positive, negative, neutral, non-available); and (2) event/topic (Evros, Moria, camps, migration in general, borders in general). Tweets were coded as negative when they opted for migrant exclusion (e.g. migrants carry coronavirus or migrants are invading Greece) and positive when they defended migrants (e.g. migrants do not carry COVID or migrant communities are under threat because of the pandemic). Finally, tweets were coded as neutral when no clear stance, negative or positive, was adopted and non-relevant when the tweet was irrelevant to the topic. For the qualitative analysis, to keep the sample to a manageable size, we focused on the 100 most retweeted negative tweets (50 from each period) in Greek and English. Then, in line with the RQ4, we identified three themes about negative migrant representations in relation to COVID-19.

Analysis

Social network analysis of co-occurring hashtags in the entire data set

Social network analysis (SNA) of the entire data set showed a distinct coronavirus-related community of co-occurring hashtags (#coronavirus, #covid19) (Figure 1, purple nodes/edges) consistently used together with place-related hashtags associated with Greece and refugees (#lesbos, #moria – the refugee camp in Lesvos). It also showed the co-occurrence of the pro-migrant hasthtags like #leavenoonebehind and #noborders. The debate within this community appeared to concern primarily the Greek refugee camps from a rather positive perspective.

Social network analysis of co-occurring hashtags in the COVID-19 data set

The co-hashtag network, resulting from all the COVID-19 keywords, is highly connected, with an average degree of 56,277 per node.

Some nodes are more inter-connected, producing a community. For example, #moria-#leavenoonebehind-#lesbos belong to the same community and are connected with relatively thick edges (See Figure 2, purple). A total of seven communities were produced with a modularity score of 0.347, indicating blurred boundaries. Figure 3 illustrates the four most prominent communities, showing nodes with PageRank higher than 0.0037 l.

Figure 3.

Co-Hashtag Covid-19 network.

Pro-refugee community (25.89%, purple colour)

This community appears expressing a positive attitude towards refugees through co-occurring hashtags. It mainly does this through explicitly positive hashtags (e.g. #leavenoonebehind, #refugeeswelcome, #restorehealthcare2refugees) and hashtags referencing NGOs that support refugees, like Doctors Without Borders (#msf). The word refugee (in various languages) is preferred reiterating a positive disposition. This community is probably German-affiliated, because hashtags are often written in German and because German actors are made as hashtags (e.g. German Minister of Interior, #seehofer). The place of attention is Greece, particularly the Greek Aegean island that hosts refugees at the infamous Moria camp (#Lesbos, #Moria).

Europe-related hashtags are neutral (#europa), evaluative, demanding Europe to act, (#europemustact) and critical (#shameoneu). Neutral hashtags concern coronavirus-related (#covid19, #corona) and country-specific (#coronavirusdeutschland). Other COVID-19-related hashtags are: #socialdistancing, #who, #triage, #mindestabstand (social distancing in German) and #leavealighton.

The hashtags #anonymousnews and #propagandavirus relate to stances that deny or question the severity of COVID-19. Finally, there is the #coronakrise representing the pandemic as a crisis. This community uses hashtags that do not relate directly to the Evros crisis except for the pro-refugee, #refugeesunderattack, which follows the pattern of a popular hashtag used in the #IStandWithGreece community, #greeceunderattack (Avraamidou et al., 2021). But it reverses the argument that Greece was under attack, claiming that it was refugees who were under attack.

The most frequently co-occurring hashtags in this community are #Moria-#leavenoonebehind-#lesbos-#covid19-#corona. The two which co-occurred the most were #moria and #leavenoonebehind (12,591 times), followed by #moria and #lesbos (10,301 times), #moria and #covid19 (6162 times), #moria and #corona (5562 times) and #lesbos and #leavenoonebehind (5147 times). Co-occurring hashtags reflect a demand directed mainly at Germany and the EU to protect refugees on the Greek islands, particularly in Moria, from COVID-19. The following is an indicative tweet of this community: ‘#COVID19 affects everyone! While many of us are #SocialDistancingNow 25.000 refugees in #Moria are locked in together. Safe lives! Act now! #LeaveNoOneBehind’ (March 17, 2020).

#IStandwithGreece community (28.37%, blue)

The hashtag with the biggest betweenness centrality score in this community is #IStandWithGreece (1365.74), associated with the Evros crisis. This community has a plethora of actor-related and evaluative hashtags compared to the pro-refugee community. Evaluative hashtags consist of five anti-Turkey, seven pro-Greece, one anti-migrant, one anti-Muslim and one pro-migrant. The prevalence of co-occurring pro-Greece and anti-Turkey hashtags reflects its position concerning the Evros crisis. The co-occurrence of the anti-Muslim #jihadi_virus hashtag is evidence of a negative disposition towards Muslim migrants but which, on its own, does not allow generalizations that it is an anti-Muslim community. Figure 4 shows that it co-occurred with #Greece_under_attack, #coronavirus and #Greece. Although it is a rather weak link compared to other co-occurring hashtags (e.g. weighted degree is 420 compared to #coronavirus with a weighted degree of 42,858), this is a COVID-related hashtag, which links the Evros events with Islamist terrorism. Multiple hashtags in Greek are also noted along with hashtags about Greek politics showing that Greek Twitter users were active in this community.

Figure 4.

#Jihadi_virus (node sizes, edges and labels scaled for visibility purposes).

The following is an example of co-occurring hashtags mainly from this community: ‘#IStandWithGreece #StandWithGreece #GreeceUnderAttack #Greece_under_attacκ #GreeceDefendsEurope #COVID–19 #COVID19greece #Covid_19 #ΜΕΝΟΥΜΕΣΠΙΤΙ #μɛνουμɛ_σπιτι #μɛνουμɛστοσπιτι.’ (March 17, 2020). The tweet contains just hashtags; the Evros-crisis hashtags are clearly pro-Greek, representing it under attack but also as Europe's defender. The COVID hashtags – some in Greek – support social distancing. Yet, no explanations are given about how the two events were related.

Geopolitical community (18.79%, orange)

Half (21 out of 52) of the co-occurring hashtags in this community referred to 10 countries (Turkey, Greece, France, Syria, Russia, USA, China, Iran, Libya, Germany) or their leaders (e.g. Putin). The hashtag with the biggest betweenness centrality score in this community is #greece. People on the move are also made as hashtags: refugee (5 times), migrant (2) and asylum seeker (1). Evaluative hashtags include, #greeceisunderattack, #greeceattacksrefugees and #humanrightsrefugee. From co-occurring hashtags, we infer that focus is on international actors related particularly to the Syrian crisis and the Evros crisis's actors, Greece and Turkey. Two international bodies are used as hashtags, NATO and the EU, pointing to a discussion on the geopolitical side of the crisis. An ambiguous hashtag is #christianity, which co-occurred with #greeceunderattack, #covid19, #coronavirusuk, #coronavirussa, #coronavirus and #coronaoutbreak.

Greek pro-refugee community (13.83%, green)

In this community, most hashtags referred to Greece, the island Lesbos or a camp in Athens, and to pro-refugee initiatives like #antireport, an alternative pro-migrant group. The hashtag with the biggest betweenness centrality score in this community is #covid_19 followed by #refugeesgr. As there are also pro-migrant hashtags like #noborders, this appears to be a community of co-occurring hashtags which are Greek-centred and in solidarity with migrants and refugees, demanding the evacuation of Greek camps because of COVID-19.

Time-frequencies and distribution of positive versus negative stances on migrants

The timeline of COVID-19 keyword occurrences indicated that COVID-19 appeared throughout the reporting period to a varying extent. A relatively significant peak of COVID-19 tweets was 16 March 2020 (7058 tweets), when a fire broke out at Moria camp. Figure 6 shows the number of tweets using COVID-19 keywords/hashtags across the investigated period.

Figure 6.

Distribution of Negative & Positive Tweets.

The content analysis on the first three days following the Turkish decision to suspend the deal (Period A, 4309 tweets), and 16 March 2020 (Period B, 7058 tweets), when a fire broke out at Moria, confirmed positive and negative stances towards migrants on Twitter (Figure 6). Specifically, in period A, 3991 tweets concerned Evros, of which 1256 were negative and 92 were positive towards migrants; 204 tweets concerned camps like Moria, of which 32 were negative and 159 were positive. For period B, 4974 tweets concerned Moria and 5370 tweets concerned refugee camps in general, of which 534 were negative and 4306 positive towards migrants. In period B, 786 tweets were about Evros, 641 of which are negative towards migrants and 20 positive.

Qualitative study: three analytical themes

The qualitative study centred on COVID-19 uses to exclude migrants and refugees and is organized around three themes (Table 1). The first examines the representation of migrants as a health threat in addition to other threats. The second discusses the role of COVID-19 hashtags that build an ambient threat around migrants, and the third theme downplays COVID-19 as a threat while exacerbating the threat of migration.

Table 1.

Migrant othering and COVID-19.

Theme	Description	Example
Migrants as health threat	In/direct association Migrants & COVID-19: e.g. Direct threat: Migrants as spreaders vs. us; Burden threat: Sick migrants as burden during a health crisis	RT @---: One pho 1 million words. Borders of Greece  =  borders of Europe The other Europeans what are they doing to help? Thousands of illegals who may have the various #Coronavid19 want to invade. (February 29, 2020)
Ambient threat	Migrants & COVID-19: loose, ambient association	#IStandWithGreece #ΕΒΡΟΣ #COVID19 #Turkey #GreeceUnderAttack #migrants Are we saying, good month? (March 1, 2020)
COVID-19 denial	Migrant threat compared to COVID-19 threat: e.g., Migrants are a real threat, COVID-19 is fake/less severe	Watching German fakenews today and ups no refugee crisis at the Greece border is found! ‘Suddenly death horse’. Fakenews drive CoVid-19 foreward. Media cartell don't report only brainwashing people with deepstate propaganda. (March 13, 2020)

Migrants as health threat

Migrants in this theme were represented as COVID-19 spreaders and as a burden to the health sector, as they could get sick and need attention at times of particular pressure due to the pandemic. For example, Evros migrants were recurrently stigmatized as potential or actual carriers of COVID-19. The following tweet is one example: ‘ERDOGAN---> WTF????? OPENS HIS BORDERS AND FORCES 4 MILLION SYRIAN REFUGEES TOWARDS EUROPE BECAUSE HIS WAR IN SYRIA IS GOING DOWN THE SEWER & RUSSIA BOMBED HIS TROOPS. JERK. EUROPE CANNOT TAKE THEM THEY HAVE CORONAVIRUS.’ (February 28, 2020). According to the tweet, Turkish President Erdogan is responsible for forcing millions of Syrian refugees into Europe carrying COVID-19; an alarming number of migrants, exacerbates the threat which matches the narrative of the network #IStandWithGreece (Avraamidou et al., 2021).

The following tweet justifies excluding migrants because they are dissimilar and threatening (illegals, invaders) while that they may spread COVID-19 furthers their representation as threatening and unwanted: ‘One pho 1 million words. Borders of Greece  =  borders of Europe The other Europeans what are they doing to help? Thousands of illegals who may have the various #Coronavid19 want to invade.’ Class, another category of migrant exclusion, is combined with COVID-19 in the following tweet: ‘So is the future doctors and heart surgeons still battle at Greece's border? How have they all not contracted Corona.’ (March 16, 2020). The tweet is ironic particularly through two rhetorical questions. First, migrants at Evros could not be doctors. Second, the migrants are of inferior capacity and potential, making them probably coronavirus carriers; it intertwines class, race and COVID-19 to exclude migrant others.

COVID-19 was also combined with ideas that migrants were Islamic terrorists. The following is an indicative tweet of the dehumanization of migrants at Evros as violent Islamic, COVID-19 spreaders by combining Islam, Corona and hooligans into one word: ‘Coronavirus on the Borders of Greece. Islamcoronahooligans ! #IStandWithGreece <<’ (March 16, 2020). The following tweet alleges that Turkey pushes migrants to Europe who are Islamic terrorists rather than refugees (notably the word refugees is in speech marks), as they are called ‘jihad murderers’ aiming to spread the virus to Greece: ‘@TulsiGabbard @realDonaldTrump Why these “#refugees “on #Evros border yell Takbeer Allahu Akbar? Is #Turkey sending #jihadi murderers as “refugees” in #Greece? Does President #Trump2020 support this jihadi #coronovirus spread by #turkey for personal business interest?’(February 29, 2020).

Finally, in this theme, we noted the argument that media failed to objectively cover the Evros events. A tweet spoke of a ‘Media BLACKOUT’ and linked to a YouTube video (UNN) with anti-Turkish and anti-migrant messages, calling migrants and COVID-19 ‘Europe's ticking bombs’ (March 15, 2020).

This theme provided evidence of generalizations about migrants at Evros as carriers of the virus that dehumanized them. Also, COVID-19 combined with threatening words like ‘invade’ furthered the divide between us and the other, as did its combination with ideas that migrants were Islamic terrorists which reproduces pre-existing ideas of Islam as fundamentally violent (Downing and Dron, 2020). So, migrants are racialized as unwanted, dangerous others who could contaminate Europeans or who were strategically aiming to contaminate them.

Ambient threat

This theme explicates a vague association of COVID-19 and migration, as tweets did not explain how the two were linked and how the one augmented the threat of the other. A recurrent way of doing so was by randomly combining hashtags and leaving the reader to interpret them. Nevertheless, they intentionally constructed fear and threat. This is an example of a tweet using several Evros-related hashtags and a COVID-19 hashtag, mostly in English: ‘#IStandWithGreece #ΕΒΡΟΣ #COVID19 #Turkey #GreeceUnderAttack #migrants Καλό μήνα λέμɛ;’ (March 1, 2020). The tweet ends with a rhetorical, ironic question ‘Shall we say good month?’, implying that the month was not good due to the Evros events and the pandemic. A GIF picturing an explosion and heavily armed soldiers in the background and a man in the forefront saying ‘Nothing to see here. Please disperse’ completes a tweet that claims that the beginning of March brought about a catastrophe consisting of a migration and a virus threat, and that only deniers or hypocrites could not see the danger unfolding (Figure 7).

Figure 7.

Tweet GIF ambient threat.

Other tweets associated migrants at Evros to terrorism and again placed a COVID-19 hashtag randomly, although seemingly unrelated to the events. The following are gendered generalizations that migrants are threatening because they are young-male and dangerous Islamists: ‘@ErikMarquardt @ABaerbock Women and children – don't fool us twice (2015)! The majority is a crowd of young and dangerous Islamists. Propaganda channels want us to believe otherwise, to keep pushing their migration agenda even during #Coronavirus #DefendEurope #IStandWithGreece’(March 1, 2020). A conspiracy theory is also built that powerful, unnamed centres push a migration agenda (see also theme, COVID-denial).

The migrant other in the following tweet is an animalistic racial other called ‘islamomonkey’: ‘we also witnessed this with the islamomonkeys (sic), ><♂ #StayAtHome #covid_19GR #EuropeUnderAttack #κορονοιου #CoronaOutbreak #isis #terrorist’ (March 16, 2020). It includes a link to a news story about ISIS advising ‘terrorists’ to avoid Europe because of the virus; equating migrants to animals furthers their otherness as inferior non-human.

One tweet wondered ‘Does the Kalergi plan relates to what we live? #covid_19GR #IStandWithGreece’ (March 13, 2020). ‘Kalergi plan’ is a conspiracy theory of population replacement in the West, and the tweet is an example of how a vague association of COVID with migration opened up discussions beyond the specific events, to exacerbate migration as a threat. Similarly, in the following, COVID-19 hashtags are used with #BuildTheWall, which associates with ultra-restrictive migration policies in the Global North: ‘Fake Refugees and economic migrants teargassed trying to enter Greece after our ‘greatest friend’ and Nato ally Turkey opened borders… #CoronavirusOutbreak #CloseTheBorders #BuildTheWall via @MailOnline.’ (February 29, 2020).

Hashtags, as affordances of Twitter, were mobilized in this theme to randomly exacerbate threat but perhaps also to gain more visibility as COVID-19 hashtags were trending. The end result is the combination of two threats to spread fear and draw boundaries of us/other.

COVID-19 denial and migration threat

In this theme, two inter-related arguments are put forward: COVID-19 was not an actual threat in contrast to migration, which was a real imminent threat, or that migration is a bigger threat than COVID-19. Two examples follow:

#IStandWithGreece this is the real problem #european_union under siege that spreads faster than #coronavirus <<. (March 1, 2020)

Right now Greece is legit struggling to keep a tsunami of refugees out of its land (basically defending the entirety of Europe) because Turkey just opened the borders for them to freely invade our homelands but ja, keep on tweeting coronavirus memes y’all ==’ (March 1, 2020)

The following, has the word refugees in speech marks, denoting that the author does not think that they are actual refugees: ‘Greek women and Greek men, you are NOT in danger by the #coronavirus, or by other viruses. You are in danger by all these “refugees” who are openly threatening that they will destroy us. Wake up, don't be sheep. #IStandWithGreece #Εβρος.’(March 1, 2020)

Another tweet denied COVID-19 even more explicitly and used sexist language to exacerbate the threat at Evros (March 16, 2020). It compares Greece to a woman who claims to be a ‘virgin’ but who in fact has affairs. The sexist reference probably means that while Greece was shutting down, Evros remained a possible entry point for multiple migrants. Simultaneously, the tweet called for disobedience to lockdowns and social distancing. The following mingles COVID-19 denial, anti-migration and anti-media stances: ‘Watching German fakenews today and ups no refugee crisis at the greece border is found! ‘Suddenly death horse’. Fakenews drive CoVid-19 foreward. Media cartell don't report only brainwashing people with deepstate propaganda.’ (March 16, 2020). Reference to ‘cartel’, leads to associating media with some form of criminality that aims to profit from and dominate over people's ideas.

In this theme, tweets downplayed COVID-19 to shift attention from the pandemic back to the Evros crisis, which was still unfolding. The end result was to downplay the threat of COVID-19 and exacerbate migration as a threat.

Discussion

This study explored the ways COVID-19 and migration intertwined on Twitter by focusing on the outbreak of the pandemic in Europe, which coincided with a border crisis on Europe's periphery. Through a combination of digital and traditional research methods, we provided the broader picture of the COVID-19 and migration nexus on Twitter and an analysis of boundary making. Specifically, we showed that COVID-19 related with migration debates on Twitter across a spectrum of positive (to include migrants and refugees) and negative ways (to exclude them), but each related to different migration events and topics.

From early on, we noted COVID-19-related hashtags which appeared positive about migrants, such as #LeaveNoOneBehind and #COVID19Solidarity, and negative, racializing hashtags, like #AyatollahsSpreadCOVID19, #CoronavirusJihad and #StopCOVID19infectedMigrantsInvaders. The co-hashtag analysis confirmed that the positive side was clearly organized around #leavenoonebehnid, and the negative side was less clearly organized around #IStandWithGreece. Notably, we found no positive hashtag towards migrants of the centrality of #IStandWithGreece for the Evros events. While the co-hashtag network analysis showed clearly the main positive narrative, to protect migrants stranded at the Moria camp from COVID-19, it was not so enlightening about the negative stances. The frequency of negative/positive hashtags and their relation to Evros, Moria and other migration events confirmed the prevalence of negative stances towards migrants when a tweet was about Evros and the prevalence of positive stances when the tweet referred mainly to Moria or other refugee camps.

The qualitative analysis complemented the co-hashtag analysis on negative stances, showing the strategic use of COVID-19 to reinforce racialized representations of migrants as a cultural and security threat, as well as a health threat. This supports extant literature about migrant racialization and exclusion during the pandemic (Pelizza, 2020) and before the pandemic, as Islam as a religion of extremism and terrorism was already a common theme in social media (Hashmi et al., 2020). Our findings supplement existing studies about how traditional and social media spread anti-Muslim racism (Poole and Williamson, 2021) accusing migrants in general (Boberg et al., 2020) and Muslim migrants in particular of spreading COVID-16 (Nizaruddin and Islamia, 2021). The study further highlights the use of COVID-19 in combination with other hashtags to create an ambient threat from migrants. It also showed migrant attacks on social media from a COVID-denialist position: migrants were represented as the real threat, whereas the virus was considered fake news or propaganda. Our findings that migrants are contagious also match pre-existing media content associating migrants with disease. This suggests that, during the reporting period, antimigrant and disease arguments were combined in tweets with two other prominent far-right and populist right-wing arguments, that of media hostility (Farhall et al., 2019) and corona-denialism (Falkenbach and Greer, 2020). Simultaneously, the study brought new insights to the existing literature about migrant solidarity on social media during the pandemic, and which was also found during earlier crises, as demonstrated in studies about the use of hashtags to mobilize pro-refugee action in 2015 (Nerghes and Lee, 2018).

The combination of the methods of analysis allowed the study to not only answer the research questions but to address its overarching aim to investigate how COVID-19 and migration content on Twitter relate with pre-existing representations of migrants in the media and social media of various reception contexts in the Global north. The two poles of the wider migration debate (Siapera et al., 2018) were found in this study. The threatening representation appeared to be prominent in relation to the Evros crisis and the humanitarian representation was prominent in relation to the refugee camps, that is when migrants were already in the Greek reception context but not when they were en route to Greece. This fits the problematization of both positive and negative migrant representations as inherently Eurocentric (Avraamidou, 2020), because in the case of the camps the migrants fit well within the victims/voiceless/helpless frame, justifying calls for humanitarianism, this time on health grounds. In the case of Evros, the other fits the Eurocentric frame of violent/male/inferior (Gutiérrez, 2018) and thus unwanted other, who may also be a virus spreader. The humanitarian side of the debate on Twitter appeared weak about the Evros events but stronger in relation to the camps because the migrant-situation therein made it easier to mobilize humanitarian action. We argue that colonial logics may leave humanitarians occasionally powerless to stand for migrants in more complex contexts like Evros, which entailed geopolitical antagonisms and state sovereignty, as migrants were still at the border demanding entry. The threat/negative side of the debate appeared stronger on Twitter during the Evros events rather than Moria, because the migrant-situation at Evros again made it easier to mobilize anti-migrant action.

Therefore, the mobilization of COVID-19 by both sides of the debate is also influenced by colonial logics. On the one hand, the COVID-19 fear is both medical and political (McFarlane, 2008), reproducing colonial fears of the unhealthy other (Flint and Hewitt, 2015). It therefore justifies a reproduction of longstanding politics of exclusion in two inter-related directions: migrants should be excluded so that they do not spread the virus to us or/and they should be excluded because they will burden the health system if they get sick as they will need medical attention at taxpayers’ expense (Abel, 2007). On the other hand, the existence of a new virus alerted that vulnerable others became even more vulnerable, needing protection. All in all, Twitter debates did not break away from the negative and the positive poles of the migration debate during the reporting period. By contrast, the COVID-19 pandemic was recurrently used to reinforce the social media role in maintaining the racialized, Eurocentric representational field of migrants from the Global South.

One limitation of this study, however, is that we focused on the Global north and more on the negative spectrum of the migration debate in social media. If we provided a more in-depth analysis of the solidarity expressed towards refugees in the COVID-19 pandemic, we believe it would have allowed us to show the wider spectrum of the debate, opening up our knowledge to social media participation from a progressive perspective and looking beyond the Global north. This is an area for future research in the field, which we intend to undertake.