Keywords

Introduction

Digital methods and computational analysis have made great progress in recent years in the humanities and social sciences. The integration of digital data with other types of materials into heterogeneous assemblages provides intriguing grounds for further investigation. As discussed in recent digital ethnographic research (Boellstorff and Maurer 2015; Borkert et al. 2018; Blok et al. 2017; Curran 2013; Gillespie et al. 2018; Munk 2019), ethnographic materials are often treated as “thick data” due to their being generated from in situ research methods such as participant observation and interviews. In contrast, so-called big data, including API-generated social media data, is deemed “thin data” as it is generated on the basis of computational methods, which capture a broader material that does not permit in-depth investigation. In set-ups like this, ethnographic material is likely to function as adding background knowledge or context for other types of data, as discussed by Wang (2013). However, as argued by Ingold (2018, 169), placing other lives—in this case, embodied by our two different kinds of research material—within their social, cultural, and historical contexts is like “laying them to rest, putting them to bed, so that we need no longer engage with them directly. Embedding lives in context implies an already completed conversation.” Accordingly, it has been argued that ethnographic materials and “big social data” should be considered not simply as different from but also as complementary with one another and capable of being stitched or assembled into analytical compositions and insights (Blok et al. 2017). While we take these efforts as a point of departure, we also see a larger potential in the work of radically rethinking the relations between different types of materials, a potential that goes beyond either stitchingor assembling. In our work with large quantities of API-generated Facebook data (hereafter, social media data) alongside ethnographic materials, we seek to transcend conceptualising the relationship between sets of research material as either confirming, complementing, or creating context for one another. While we acknowledge the existence of other types of big social data (Manovich 2011, 460) (as well as several types of big non-social data), social media data is, as we shall see, among the most commonly used big social data that has been applied to computationally study migrants.

In this contribution, we rethink the relationship between our social media data and ethnographic material by showing them to be fundamentally interconnected. We explore the potentials of combining ethnographyand social media data by establishing relations in our material as contrapuntal. Counterpoint in music occurs when several different musical lines play simultaneously, being at once independent and related.1 Inspired by Tim Ingold (2018), we understand lines of counterpoint as translated into human movements, which carry on alongside one another, not as the summation of parts but as the correspondence of their particulars. We understand these related but not necessarily coordinated movements as contrapuntally interconnected. We further unpack some of the challenges inherent in working with social media data and using analytical data processing programmes. These involve, for example, the generation of data through logics of quantification and bias towards numbers of likes and spikes in the material, which Rogers (2018, 450–454) designates as “vanity metrics” (for further critical discussion of social media metrics, see also Tufekci 2014, 505–514; Kitchin 2014, 1–12). In order to avoid such analytical traps, in which data and materials are ordered along predefined scales of big or small, thick or thin, we pursue a strategy of non-scalability (Tsing2012). For Anna Tsingnon-scalability implies “letting scale arise from the relationships that inform particular projects, scenes, or events” (ibid., 509). As Strathern has argued, the act of scaling relies on an ontology, in which the world is composed of parts that add up to a whole, suggesting an approach of working upwards from the small details to the big picture (1991, 109f). Translated into the discussions surrounding digital methods, Latour et al. (2012, 591) have demonstrated that, by following the connections between digital traces left in available databases, it becomes possible to transcend this ontology of parts and wholes altogether. Rather than presupposing the two levels of social order—of parts and wholes, micro and macro, elements and aggregates—we follow the approach of working from the middle (Haraway 1988), in which materials are created differently and speak different languages but are nonetheless produced through engagement with the same world. With help from Ingold, we establish contrapuntal rather than summative relations in our materials. People live alongside one another; sometimes they meet, sometimes they move away from each other, yet they correspond with the same world. They are “moving on, alongside one another,” which makes them attentive, responsive, and responsible to one another (Ingold2018, 160). The task then becomes one of demonstrating the analytical potential of contrapuntal interconnectedness and how different-yet-related data and materials are answerable to the same world. By establishing contrapuntal relations in our material, we explore and qualify this affinity with the aim of identifying further potentials and questions for digital migration research when bringing social media data and ethnographic materials into conversation.

This chapter is based on ethnographic fieldwork carried out in the Danish–Swedish borderlands in 2018–2019 as well as social media data collected through API access from public Facebook pages (including posts and comments) related to irregularised migration and refugee relief in the Danish–Swedish borderlands, including German pages linking to the German–Danish borderlands, covering the period between 2011 and 2018.

In the following, we present our understanding of one world anthropologyand contrapuntal analysis. We then position our contribution in the emerging field of digital migration research and discuss how we are in dialogue with and differ from similar attempts at combining big social data with ethnographic materials. We subsequently elaborate upon our methodological approach and the ethical challenges connected to the use of large-scale social media data in the context of migration research. Finally, we test the potentials for conducting contrapuntal analysis. For the purpose of our argument here, which is to propose a contrapuntal analytical strategy for engaging ethnographic material with social media datasets, the analysis will remain illustrative. Because our research is part of the wider DIGINAUTS project (see Sandberg and Rossi in the Introduction to this book), its overall scope focused on how irregularised migrants and solidarity workers challenge the European border regime through their respective digital fields of navigation. The border thus emerged as a recurring point of tension in our ethnographic conversations, in the social media dataset, and accordingly in our analytical and theoretical discussions. The contrapuntal analysis presents three different versions of the border enacted through the material, when conversing between the different-yet-related datasets and materials. In conclusion, we discuss how the contrapuntal move can advance digital humanities and the field of digital migration studies, not only in methodological but also in analytical and theoretical ways.

One World Anthropology

In acknowledging the interconnectedness of our ethnographic material and social media data and bringing them into conversation without positioning them as each other’s stand-ins or contextual backgrounds, we find inspiration in Tim Ingold’s (2018) oneworld anthropology. Ingold challenges the idea of life-as-a-whole as a sum of its parts and proposes the idea of correspondence, which entails that “parts are not components that are added to one another but movements that carry on alongside one another, so too, in the human family, lives lived in counterpoint are not ‘and … and … and’ but ‘with … with … with.’ And in answering—or responding—to one another, they co-respond” (160, emphasis in original). He contends that “life itself, then, is not the summation but the correspondence of its particulars” (158). Ingold argues that this calls for “a ‘turn’ that is not ontological but ontogenetic” (169), that is, a turn which focuses on the ongoing generation of being rather than its essence (167). And this “leads us to conceive of the one world as neither a universe nor a fractiverse but as a pluriverse” (169). As mentioned, Ingold proposes the analogy of music, where the “relation between parts and whole is not summative – neither additive nor multiplicative – but contrapuntal” (160, emphasis in original). Ingold thus thinks of “the life of every particular soul (…) as a line of counterpoint that, even as it issues forth, is continually attentive and responsive to each and every other” (160).

The idea of contrapuntality was introduced to cultural studies by Edward Said (1994, 2000). It is especially pertinent for us to (re)turn to Said since the state of exile, which is at the heart of our endeavour, was crucial to his life and scholarship. Said’s idea of contrapuntality is based on the same premise of connectedness as that of Ingold. Relations between coloniser and colonised, former coloniser and formerly colonised take centre stage in Said’sreflections. Said contends, “we must be able to think through and interpret together experiences that are discrepant, each with its particular agenda and pace of development, its own internal formations, its internal coherence and system of external relationships, all of them co-existing and interacting with others” (1994, 32). With the musical metaphor, “various themes play off one another, with only a provisional privilege being given to any particular one; yet in the resulting polyphony there is concert and order, an organized interplay that derives from the themes, not from a rigorous melodic or formal principle outside the work” (Said1994, 51). As Said suggests, we must not foreground any composer or mastermind behind “the music”; it is necessary to remind ourselves that there is always a limit to the metaphor. For Said, contrapuntality becomes a method for simultaneous awareness of different voices, one that acknowledges their connectedness and answerability to each other and that allows alternative or new narratives to emerge (1994, 51).

Life as Experienced, Traced Life

While Ingold, working from the perspective of anthropology, speaks of lives and souls, Said, grounded in the tradition of literary criticism, speaks of texts and voices. We do not regard these contrapuntal understandings as contradictory but instead find it useful that the two foci explicitly address our two types of material, namely the lived lives that take centre stage in our ethnography and the texts that we choose to foreground in our social media data. That is, while we maintain that our materials are inherently interconnected—corresponding with the same world—we remain aware that the different modes of data production have facilitated different paths to knowledge: one foregrounds life as experienced, and the other foregrounds textual traces.

Before elaborating upon why we find contrapuntal analysis particularly relevant for our analysis of disparate data and material, it is useful to expand upon how we perceive these data and materials. We cautiously designate the content we collected through Facebook APIs as data. This is because the data is actually produced by the API on the basis of activities that take place on the platform (Lomborg and Bechman 2014, 256) and because this term is commonly used in computational sciences. When we refer to our ethnographic observations, reflections, and transcripts as materials, we do so in recognition of Ingold’s (2013, 5) understanding of ethnography as a way of “knowing from the inside.” Ingold contends that:

to convert what we owe to the world into ‘data’ that we have extracted from it is to expunge knowing from being. It is to stipulate that knowledge is to be reconstructed on the outside, as an edifice built up ‘after the fact’, rather than as inhering in skills of perception and capacities of judgement that develop in the course of direct, practical and sensuous engagements with our surroundings. (Ingold,2013, 5)

Approaching our social media data with an ontological commitment to knowing from the inside entails recognition of these as actively produced in correspondence between our research team’s decisions, social media platforms and programme logics, irregularised migrants, solidarians and other Facebook users, writings and movements, and more. That is, despite their diverse ontological heritages and the frictions these differences produce, our social media data and ethnographic materials have a fundamental affinity that we wish to bring into focus through a contrapuntal analysis. Our point is not to seek out particular connections between our data and material but to instead acknowledge their interconnectedness. The texts of our social media data are fragmented and partial, but they are produced relationally with the ethnographic settings we have explored, in Ingold’s words, “with…with…with.” As we bring different elements together in a contrapuntal analysis, we allow new narratives—non-singular and situated—to emerge. The contrapuntal approach highlights a tension between on the one hand recognising materials and data as fundamentally interconnected and on the other hand recognising that the researcher actively composes connections. Pursuing a contrapuntal analysis, we consciously position ourselves within this tension (cf. Haraway 1988).

Contrapuntal analysis is particularly relevant to our material because it not only acknowledges the interconnectedness of lives but also speaks to the particular historical moment of 2015, when the stream of irregularised migrants to and through Europe brought people together and strengthened connections between people who had never before had direct engagements. Rather than being detached fragments, disconnected from people’s experiences, our social media data speaks directly to how many of our research participants experienced this time, a time when precisely these types of digital messages frequently proved crucial to the course of their lives. It was a moment that few of the people who were directly involved experienced coherently or cogently. This shared moment illustrates the affinity between our material and data, and it is exactly this co-existence between incoherence and instability—and connectedness—that we wish to capture with the contrapuntal approach.

On Non-Scalability

In her article “On non-scalability,” Tsing (2012, 507) defines scalability as the ability to expand without rethinking the basic elements, features, or designs of a project: “To ‘scale up’, indeed is to rely on scalability – to change the scale without changing the framework or knowledge of action.” Tsing also presents the notion of precision nesting of scales, in which “the small is encompassed neatly by the large” without any revisions to the design or nature of the project (ibid.). Whereas scalability is a tenet of capitalism’s growth ideology, which promotes the idea of endless growth of businesses along the same scale, non-scalability presents the idea of “letting scale arise from the relationships that inform particular projects, scenes, or events” (ibid., 509). “In that work, there are big stories as well as small ones to tell. There is no requirement that the scales nest or that one performs wizardry of conversion from one to the other without distortion” (ibid., 509f). While tracing contrapuntal connectedness, the question arose how to build an understanding of the phenomenon that “scales smoothly from minute details to aggregate patterns and back” (Munk 2019, 169). In other words, through our contrapuntal move, we find it crucial to go beyond the particular case, albeit in non-scalability mode.

Digital Migration Studies: An Emerging Field

Digital migration studies has emerged as a relevant field for providing new and meaningful insight into the phenomenon of human migration and migrants’ practices (Kok and Rogers 2017, 23–46; Leurs and Smets 2018). Following the rapid adoption of digital material and methods in the humanities and social sciences, digital migration studies has developed in numerous, often only loosely connected directions. While there is fundamental agreement that digital media and digital technologies repurpose—and frequently facilitate—the process of migration (Diminescu 2008, 565–579) and affect migrant populations’ processes of social integration and political participation (Komito 2011, 1075–1086), there is considerable diversity in research foci, empirical foundations, and methods (Leurs and Smets 2018).

Calls have been made recently to systematise existing approaches by further reflecting upon methodological implications of studying migrants’ digital practices. Leurs and Prabhakar (2018, 247–266) map the field of digital migration studies by identifying three distinct paradigms: (I) migrants in cyberspace, (II) everyday digital migrant life, (III) migrants as data. These three existing paradigms include main representative scholars, alternative theoretical discourses, and often non-overlapping methodological preferences. While the first paradigm is rooted in the hermeneutical approach to digital humanities, the second builds on social science theories and methods that precede digital data (e.g., ethnography, interviews, and participatory observation). Finally, the third paradigm is organised around the idea of digital methods as research methods and practices uniquely tailored to handling contemporary digital traces.

Following the call to establish bridges between these different paradigms, we situate our research in conversation with the non-digital-media-centric ethnographic approach, which focuses on everyday digital migrant life (paradigm II), and the digital-media-centric digital approach, which appreciates migrants as data (paradigm III). We do so by developing a research design that is from the start equally rooted in social media dataand ethnography. In contrast to common practice within mixed methods perspectives (Creswell 2014, 1), we do not define a convergent or fully sequential research design (Snelson 2016) in which one type of data complements or drives the analysis undertaken upon the other type of data. As will be detailed in the methods section, ethnography, data production, and analysis were carried out in relative independence following an initial phase of alignment. For instance, previous knowledge and preliminary field studies informed the initial digital seed for the data production, but data production subsequently followed its own internal process for selecting relevant sources.

Stitching, Complementing, Remixing…

Moving beyond the migration context presented here, several existing studies have approached social media data alongside qualitative or ethnographic materials (Gillespie et al. 2018; Borkert et al. 2018; Curran 2013). Boellstorff and Maurer (2015) bring together anthropological and comparative insights with “big data,” highlighting the complexity of enabling conversation between highly divergent methodological approaches and epistemological perspectives. Among recent attempts to combine different data formats and materials, the focus has been on “assembling” or “stitching” heterogeneous data worlds (Blok et al. 2017; Blok and Pedersen 2014). Blok and Pedersen outline a research collaboration between anthropologists and sociologists experimenting with combining big transactional data (based on GPS and Smartphone Bluetooth signals) and ethnographic fieldwork (fieldnotes) into a joint social network analysis. The core metaphor used to build together the data formats is stitching “as a possibility of mutual fertilization across apparently incommensurable fields” (ibid., 2), which recognises the relationship between the two as complementary, flagging an ontological ground of assemblage (cf. Carter 2018). The aim is to create an experiment in “‘big data-ethnography’ in which the hyphen suggests an initial, inherent, and deliberate uncertainty as to which of the two terms (big data, ethnography) is here the context for the other; which is figure and which is ground” (ibid., 4). A further example is provided by (Markham et al. 2013), who, though not focusing specifically on social big dataand ethnography, suggests a method of remixing data materials as “a powerful tool for thinking about qualitative, interpretive research practice” in order to “better grapple with the complexity of social contexts characterized by ubiquitous internet, always-connected mobile devices, dense global communication networks, fragments of information flow, and temporal and ad hoc community formations” (65). Markham focuses on the complex manner in which different types of material can be brought together to create particular frames of interpretation. However, the remix metaphor also encompasses a practice of cutting, copying, and pasting (cf. Markham 2017). The contrapuntal approach differs in that it does not embark on “cutting” elements and pasting them into new contexts for new purposes, seeks instead to shift the gaze in order to allow new meanings to emerge from already-connected material. It also differs in highlighting the material types as already connected, rather than focusing on the researcher’s remixingor stitching practices. That is, in a contrapuntal analysis, the researcher establishes ways of working analytically with connectedness, as detailed below.

In the specific context of conversing with big data and ethnographic materials, we contend that we are not simply dealing with data of different scales in which big data indicates a sum of the ethnographic material but a more meticulously described part. Anders Munk (2019) argues that, across the “great divide” (cf. Latour2005) between qualitative and quantitative analysis, a shared disconnect exists among ethnographers and digital researchers. This is because both fields struggle to chart the “native’s point of view” into more quantifiable formats while maintaining the intended meaning and original context of those “on-life traces” made in the digital world: “These questions are not simply about making sense of data but about making sense in a way that claims to reflect a native point of view” (Munk 2019, 163). Apart from the complementary model pursued by (Blok and Pedersen 2014), Munk suggests three additional modes for further overcoming the quali-quantitative divide: single-level analysis, curation, and algorithmic sense-making: Whereas single-level analysis sees onlife traces as embodying both qualitative richness and quantifiability, curation presents itself as a critical practice scrutinizing different media environments. Algorithmic sensemaking tends to emulate qualitative sense-making to be replaced by quantitative pattern recognition. (Munk 2019, 164f).

While we understand digital and ethnographic material to originate from the same world, we move beyond the complementarity model. We thus position our approach on a possible overlap between single-level analysisand curation, as proposed by Munk (ibid.). We manually curated a digital data collection of pages—bearing in mind that this was ultimately mediated through Facebook’s API and involved repurposing the API for our research goals—to avoid fully automated data crawling and to build a corpus with a purpose (cf. Munk 2019, 172). We simultaneously approached the observed digital data and the quantitative trends not as something to be complemented with our sense-making obtained from ethnographic material but as a means of acknowledging a connectedness that transcends the particular case. We can perhaps see the contrapuntal approach as a fifth mode for overcoming the quali-quantitative divide, a mode in which digital data and ethnographic material can produce insights that are not necessarily complementary but nevertheless refer to the same world.

Our understanding of contrapuntal analysis specifically goes beyond an idea of assembling. We approach our materials neither as (interchangeable) complementary add-ons to one another, as confirmations of one another, nor as contextualisations of one another, regarding them instead as answerable to one another, as fundamentally interconnected. We contend that the complementarity approach, alongside the confirmatory approach and contextualisation, fundamentally assume the ability of one type of data to “fit together” with alternative types of data within a larger representation of the world, either by providing an alternative perspective, supporting evidence, or background information. This assumes known or expected relations between the data (how the data types speak to each other) that cannot always be taken for granted.

Methodological Design and Ethical Considerations

The general methodological design of this research builds upon two parallel data sources: social media data generated through API access to a curated list of public Facebook pages and ethnographic material produced through participant observation and interviews. The production of digital data was initially informed by early pilot studies and pre-existing knowledge, while the ethnographic study was at times inspired by insights from our digital data production. However, the actual definition of the digital data sources was not made to mirror or complement the ethnographic observations (see also Makrygianni et al. Chapter 2 in this book). Our ethnographic material was produced through ethnographic fieldwork carried out with Syrian refugees and solidarians in and around the Danish–Swedish borderlands, the Øresund Region, in 2018 and 2019. The research participants, who were refugees, had arrived in Denmarkand Sweden as irregularised migrants between 2014 and 2016 and had subsequently obtained juridical refugee status. We therefore maintain the designation irregularised migrants since our conversations with research participants concerned the time when they fled, travelling to the Øresund Region without being able to depend upon regular modes of travel. Taking our cue from Rozakou (2018), we use the term solidarians to highlight that the various refugee support initiatives included in this study were not necessarily part of established NGOs; they were instead constituted by informal networks of people in the vicinity of the locations where the irregularised migrants arrived, with a common aim of “standing up in solidarity.”2 Key topics in our study were how these irregularised migrants had navigated on their journeys to and within Europe, the role of solidarians in this navigation, and the significance of digital practices for both groups. We therefore used mainly in-depth, retrospective interviews. In our interviews with refugees, we looked back on their journeys, and in our interviews with solidarians, we revisited their work to help irregularised migrants. We interviewed several members of the same family networks in order to hear different accounts of shared journeys and establish trust (Mollerup and Sandberg, forthcoming). This also provided us with detailed insight into the significance of family members who had fled previously.

We interviewed some people on multiple occasions, and in some conversations, we included what we term device tours (Mollerup 2020), in which interviewees showed us old conversations, pictures, etc. on phones and computers, allowing these to become focal points in the interviews. These device tours at times played a role in bringing up memories and emotions. For ethical reasons, the contents of our research participants’ devices were never systematically recorded, archived, or documented beyond our fieldnotes. Some interviews lasted half an hour, while others lasted several hours and were interwoven with meetings and socialising with family members and others. We carried out interviews in Danish, English, and Arabic. In total, we undertook 16 interviews with 12 refugees, and 16 interviews with 16 solidarians situated on both sides of the Danish–Swedish border. After transcribing all interviews,3 we manually coded the transcripts and field notes (526 pages in total) for different themes that had emerged through the fieldwork and through reading and re-reading material. We ended up with 97 codes. One of these codes was “border.” The notes and transcripts that were coded “border” deal with stories of irregularised migrants planning to cross, crossing, or failing to cross borders and of solidarians helping irregularised migrants. Many of these stories describe extreme danger and difficulty. They also at times describe remarkable creativity and unexpected success in being able to move.

The digital data production followed a relatively well-established procedure in digital methods: data selection, data enrichment, and data analysis (Kok and Rogers 2017; Rogers 2013). The starting point was constituted by a set of public Facebook pages focused on refugees and irregularised migrants, identified through a combination of methods, including manual snowballing from relevant pages previously known to the research team as well as through Facebook’s search function, using topic-specific keywords in English, Arabic, German, Greek, Swedish, and Danish. This produced 200 Facebook pages, which were then manually coded with additional information, such as the type of actor behind the Facebook page, physical location of the actor, date of page creation, and language (or languages) used. All the publicly available content (posts, comments, and reactions) on the pages was downloaded using Facebook’s API. This produced a final dataset comprising 200 pages, 84,359 posts and 2,254,923 comments, produced between 20/12/2010 and 24/09/2018. A detailed description of these elements and of the methodological consequences can be read in Chapter 2 (Makrygianni et al.) of this volume. Given the specific focus on the Danish–Swedish border region, we filtered the data so that it contained only pages created by actors from Swedenor Denmark or German pages that connected with the German–Danish borderland. This narrower dataset comprises content posted on 80 public Facebook pages (49,162 posts and 1,238,794 comments). The language distribution is illustrated in Fig. 3.1.

Fig. 3.1
A bar graph of activity. Arabic, roughly 1,150,000. Domestic, 20,000. English, 100. Multilingual, 10,000.

(Source This image is used with permission of the authors of this chapter [Rightsholders])

Distribution of languages in social media data, based on number of comments. “Domestic” includes Swedish, Danish, and German

Full size image

On the Ethics of Digital Data for Migration Studies

Use of large-scale social media data in the context of migration research raises important ethical questions. When using digital traces produced by irregularised migrants, we are dealing with a type of material that has been collected without explicit authorisation from the subjects (who at times are unaware of the possibility of data collection). At the same time, gathering and storing large amounts of such data creates additional risks when the practices documented in the data—e.g., undocumented border crossing—are illegal in several countries. While a full analysis of these problems is beyond the scope of this chapter and touches upon several methodological (Lomborg and Bechmann 2014), legal (Kotsios et al. 2019), and policy issues (Bruns et al. 2018), it is possible to adopt specific research practices to mitigate such risks. In the context of the present research, digital data has been collected only from publicly accessible Facebook pages (private and public groups and user profiles were deliberately excluded). In this sense, it is important to stress that our selected social media data from publicly available groups represent only the tip of the iceberg, given that non-public information (including private messages on Messenger, WhatsApp, and similar platforms as well as posts and comments in non-public groups) is not included in the research material for ethical reasons. Identifiable information about the authors of the messages has not been collected at any stage and is not available to the researchers. Furthermore, the data is securely stored and is not accessible to anyone outside the research team. While this somewhat conservative approach is important to secure the anonymity of and prevent harm to people who are active on social media and who never agreed to participate in our research, there is also an important ethical argument for researching this particular historical moment in order to document, analyse, theorise, and contribute to discussions about the EU border regime and its consequences. Our research design, combining computational methods with ethnography, ensures that we can attain a deeper understanding of this moment, acknowledging both the ethical need to research this moment and the ethical need to do so without compromising privacy.

Creating a Data Environment Relevant for Contrapuntal Analysis

In order to be able to work contrapuntally with the two different-yet-interconnected types of research material, we first needed to establish a sustainable data environment relevant for our purpose (cf. the Introduction to this book). Our efforts to approach these two sets of material contrapuntally was challenged by the inbuilt logic in the social media material, which prioritised an approach focused on comments and shares, which encouraged us to search for spikes in activity (Rogers 2013). Looking at intensities in activity enabled us to identify certain spikes that probably related to events we knew to be ethnographically significant (such as the spread of the images of Alan Kurdi, which strongly influenced one of our interviewees’ decision to pack his bags and travel to Lesvos, and the advertisements published in Lebanese newspapers by the Danish Minister of Integration at the time, Inger Støjberg, which made one irregularised migrant with whom we spoke reconsider Denmark as a destination). However, these spikes told us little about how these events mattered to the people in question. In addition, we could not assume a relationship between spikes and significant events. For instance, one large spike we looked into in the Swedish material for October 2015 was produced by comments mainly on two particular posts, in the same Arabic language group (we saw a similar spike in the sharing of these two posts). One was a rather humorous video of a Lebanese man in Germany trying to learn Syrian dialect. The other was a lively video of Syrians teaching Germans how to dance the Syrian folk dance dabke (see Fig. 3.2). These types of videos were posted on pages that also carried very serious messages and images detailing the devastation of war, fleeing conflict, and quests for information about how to reach safety. Spikes at times corresponded with and contained posts about major events, but the occurrence of a spike did not necessarily provide information about irregularised border crossings.

Fig. 3.2
A line graph of activity. It fluctuates and peaks after October.

(Source This image is used with permission of the authors of this chapter [Rightsholders])

The lure of spikes: Activity in comments in the Swedish material. While at times relationships between events stood out in our ethnographyand spikes in activity, we often found attention to spikes to be counterproductive

Full size image

Seeking instead to move beyond a quantified logic of relevance, in which a relevant period of time would be identified through a peak in the data, and to create a sustainable data environment relevant to our purpose, we filtered the data so that it contained only the period from August 2015 to October 2015 and contained only posts. This led to a smaller dataset containing 57 pages and 3456 posts. The period was chosen because it coincided with the arrival of unprecedented numbers of irregularised migrants in Denmark, Sweden, and Europe as a whole, including many of the irregularised migrants with whom we spoke. Correspondingly, a vast array of solidarity initiatives emerged during this period. Many of the solidarians with whom we spoke became active during or immediately in advance of this period. We focused only on posts because comments often strayed far from the topic of the posts, and it proved unfeasible to reconstruct the conversational dynamic between often apparently disconnected comments. Moreover, a focus on posts alone made the quantity of material more manageable. A negative consequence of this decision is, however, a reduction in the diversity of voices, prioritising those individuals who were actually able to and wished to post on the pages.

Having delineated a subset of our digital data, we approached it with the keyword “border.” As mentioned above, this recurring keyword was chosen because of our overall research focus on irregularised migrants’ and solidarians’ strategies for navigating the European border regime through their respective crossings of borders and refugee aid initiatives. Using the Tableau data analysis software, we searched for any occurrence of the word border in Danish, Swedish, German, Arabic, and English in our subset of data. Overall, the data contained a reasonably small number of posts containing explicit reference to border in the various languages (10 in Danish, 23 in Swedish, 13 in German, 86 in Arabic, 15 in English). Using our ethnographic material as a backdrop, we read and manually coded these posts, paying attention to the nuances in the various references to border, with the aim of conceptualising them as different versions of the border (Mol1999; Law and Urry 2005; Sandberg 2016). We then approached our ethnographic material, which we had manually coded for occurrence of the border concept (not, as with the social media data, for occurrence of the word itself). We extracted those parts of the ethnographic material that were coded with border and then manually coded them for different versions of borders, as we had with the social media data.

We thereby created a sustainable data environment relevant for carrying out contrapuntal analysis, allowing us to simultaneously acknowledge the interconnectedness between our data and material while also allowing us to juxtapose these, enabling productive tension between the ways in which our materials were already connected and the connections we established. Our goal here was not to establish direct relations between our different materials. Instead, we continually approached the two types of materials, seeking to understand how they corresponded with one another (see Fig. 3.3).

Fig. 3.3
A whiteboard with written information and drawn graphs with a peak or a trough.

(Source Photo by Nina Grønlykke Mollerup)

Contrapuntal correspondence. Rather than identifying direct relations between our different types of materials, we aimed to establish correspondence. In the analytical process, we tried to visually illustrate this relationship of “moving alongside one another” through arrows and waves, which react with each other without clashing, meeting, or otherwise making direct contact, and also without necessarily being unidirectional in their movements. The peaks in the top three waves and the corresponding nadir in the bottom wave illustrate when irregularised migrants were told to get on a designated train in Germany, which many did, but which prompted suspicion in one couple with whom we spoke and caused them to choose a different path

Full size image

During the process, three different but related versions of the border emerged as significant for this chapter. We designate these: politico-legal border, solidary border, and border navigated. Unsurprisingly, these three versions of the border did not emerge equally from the two sets of material and from the different subgroups of our material. For instance, our Arabic-language social media data, irrespective of country, particularly spoke to border navigated and politico-legal border, while English and local languages in the respective countries particularly spoke to solidary border and politico-legal border. Our ethnographic material from conversations with solidarians spoke to all three border versions, while our ethnographic material from conversations with refugees rarely spoke to the solidarian border. While each subgroup of our research material tells stories about the border independently from the others, our analysis emerges from both the social media and the ethnographic material. The ways in which the different border versions formed through and within the different subgroups is thus crucial to our ability to engage in contrapuntal analysis.

Border Enactments

The following analysis serves to illustrate how a contrapuntal lens can help in understanding social media data and ethnographic material as different-yet-interrelated materials. This lens enables us to identify the three different versions of the border enacted through social media communication as well as in situ practices. We present them here first through some of their general characteristics, followed by an exemplary analysis, which simultaneously deploys a contrapuntal approach to our research material and uses the border versions as an analytical lens.

Politico-legal border is concerned with the regulation, politicisation, and control of borders. It includes contestations of both physical and legal borders, recognises borders as changeable, and includes visions for new legal models for a borderless world and longer term plans for managing and controlling borders. Across our research materials, the border emerges as a particular site of struggle (cf. Hess and Kasparek 2017) through calls to action at particular borders and through questioning the moral implications of borders. However, the politico-legal border focuses on borders as a matter of political negotiation and is also addressed in places well removed from physical border crossings, including parliaments and other sites of political debate. Solidary border focuses on border-spanning activities that aim to ease irregularised migrants’ crossing of borders through refugee relief, everyday humanitarian aid, and transit assistance. Solidary border also includes negotiations of particular borders, for instance, through demonstrations and advocacy for opening borders and organising border crossings. Solidary border works in opposition to several “significant others” such as against racists and related neo-Nazi movements. Border navigated deals with practices of illegalised border crossing and thus takes into account the irregularised migrants’ perspectives. It attends to different types of border crossings and includes attention to political, social, and material circumstances (weather conditions, landscape, border fences, etc.). Unlike politico-legal border, it does not deal with longer term perspectives or moral implications but is instead invested in the here and now of the border and how it might facilitate or hinder movement. Together, the three versions of the border advance an argument concerning the border multiple (Andersen and Sandberg 2012), which highlights borders as practice and thus as constituted through a multiplicity of actors while reaching beyond the individual action, site, or event. When we say that borders are enacted, we do so to refer to the different ways in which borders are negotiated in experienced life (as depicted in ethnographic conversations) and in traced life (as documented in social media data).

Example #1: The Border as an Event

Our social media dataand ethnographic fieldwork focused on the Danish–Swedish borderlands, but it was clear through both sets of material that the border extended well beyond this and was entangled with other borders and solidarity practices spanning the wider European border regime. Thus, the Serbian–Hungarian border emerged as significant in both our social media data and our ethnographic material in the time leading up to Hungary completing the construction of a border fence on 16 October 2015. In the context of this emergence, the different sections of our two sets of material come to have particular significance, as the different data and interviews in different languages with former irregularised migrants and solidarians presenting us with different paths towards understanding the events. A contrapuntal approach allows us to show how the two sets of material present different-yet-related paths to knowledge about irregularised border practices and solidarity actions at this border and beyond, while ensuring a non-scalable mode of working. We see scale as arising from the relationships informing the events or stories that have been told or the words that have been posted or said.

The following information posted in an Arabic-language Facebook group speaks to the politico-legal border but can simultaneously affect navigational decisions about where and how to cross borders.

[Social media data] As of tomorrow, Tuesday, Hungary will tighten border crossing procedures and punish illegal crossings with imprisonment. They have sent more than 900 additional police officers to protect the border and close it completely. Posted on Arabs in Denmark, 14 September 2015, translated from Arabic.4

A related call for human presence at the border, posted in an English-language German group, shows the solidary border as a particular site of negotiation that spans well beyond the actual site:

[Social media data] Please inform everybody who can come should come NOW to the serbian - hungarian border (Horgoš border crossing)! Everything, everything is needed but human presence the most! Thousands of people are demanding without a break the opening of the border! Stand up for solidarity against borders and repression!Posted on Refugees Welcome (Germany), 1 October 2015.

Such calls, public as well as private, have contributed to prompting solidarians with whom we have spoken to radically change their lives by dedicating themselves entirely to helping irregularised migrants for many months. This emerges through our interviews with solidarians. Some could name the precise post, at times including an image, which caused them to get involved. Jens was one of the solidarians who placed his life on hold for months to respond to the situation, including at the Serbian–Hungarian border. He told us:

[Ethnographic material] So, after that, then I was in Presevo [Serbia], driving some emergency aid down there and trying to help out. I had some Danish contacts, who had been down there, when it was complete chaos with 10,000 people [arriving] a day, and there was just no infrastructure or anything. People were soaked in the rain and with two degrees cold [Celsius]. So, I went down there with the car packed with tools and - we just took blankets and children’s clothes and everything we could get our hands on and threw them into the back of the car and drove down there. And smuggled it in because we weren’t allowed by the authorities in Serbia. Jens,5 Danish solidarian, 21 March 2019, translated from Danish.

The situation at the Serbian–Hungarian border emerges as chaotic and overwhelming through our social media data and interviews with solidarity workers. Thousands of people are at the border, in bad weather conditions, and with little organisation. Shifting the focus to the border navigated provides us with a different view of the situation. When approaching the border through our ethnographic interviews with irregularised migrants, the chaos is still visible, but it is backgrounded, whereas the danger and uncertainty of the situation are highlighted, along with the actual decisions that were made. That is, the chaos is a temporary obstacle that recedes as soon as it is overcome. Most irregularised migrants with whom we spoke singled out Hungary as the most dangerous place or one of the most dangerous places they had encountered in Europe. One irregularised migrant with whom we spoke had been imprisoned and had his belongings “confiscated,” and others had, themselves, heard similar stories. Ghada and Ziad, a young couple travelling together, in separate interviews told us of the situation at the Serbian–Hungarian border, which they crossed in mid-late September 2015. Ziad said:

[Ethnographic material] The weather was so bad we couldn’t continue, like, raining so much. So, then we took a bus again, to a city between Hungary and Serbia. I forgot the name. Then everything was hard, because on the way, I mean, the army, they were on the way, and they took people and you have to sign so that means you cannot go to Sweden if you sign in Hungary. Ziad, Syrian refugee, 3 March 2019.

Ghada also told us of their difficulties at the Serbian–Hungarian border and the challenges with the weather:

[Ethnographic material] I remember, Hungary borders, it was really difficult. This I won’t forget. It was really bad. And even the police there were really mean and they almost hit people, like this, ‘Just go back! Go back!’ They weren’t nice. And then we had to walk at night, and we had to wait two nights, I think, two nights. The first night it was raining so [the smugglers] said, ‘No, we can’t go like this, it’s a bit difficult, and we can’t find the way.’ Ghada, Syrian refugee, 20 March 2019.

Ghada explained how they had finally managed to cross with the help of a smuggler, escaping a volatile situation at the border, where people were being pushed, and they had to hide from the police to avoid getting caught. She elaborated:

[Ethnographic material] We knew that something was happening there in front of us. We didn’t know what it was exactly. And we didn’t want to just go there. So, we just followed the people who were going on the side. In the bushes. And then we contacted some guy, I think Ziad contacted him. (…) It was a lot of people and I was, like, afraid, I would say. I was really afraid. Ghada, Syrian refugee, March 20, 2019

This ethnographic material highlights the danger and uncertainty experienced at the border; the border navigated. For solidarians, this uncertainty was, of course, experienced very differently and instead played into their planning of relief work, allowing what were often much-needed rests and breaks. As a Swedish solidarian, working in a solidarity space that housed irregularised migrants upon immediate arrival, said:

[Ethnographic material] A border would be closed here and there. In Germany, and somewhere else, Serbia for example. We could be told: ‘Now the border is closed; it is expected to be opened again in two days. Make sure to rest and sleep now, fewer will come, and later everyone will come again.’ And later, we could then be told: ‘Now they have reopened. The men have gone first. In the second wave, women and children will come. If you need to collect prams and such, you will need to do it within three days’ Like that. So, it was extremely effective. Åsa, Swedish solidarian, 21 March 2019, translated from Swedish.

Approaching the events surrounding the Serbian–Hungarian border around September 2015 contrapuntally through our different materials and their subsets allows us to show how chaos, danger, and uncertainty are differently foregrounded and configured. Our point here is not to seek out direct relations but instead to show contrapuntal correspondence between the different aspects of irregularised border crossings, spanning well beyond individual people and places.

Example #2: The Instability of the Border

Irregularised border crossings across the built border occurred during the summer and early autumn of 2015, before EU member states had introduced temporary border controls. This allowed large groups of people to enter and exit territories without legal documentation once they had entered the EU. Similarly, the temporary lifting of borders could occur when border control was enforced only as spot checks, leading irregularised migrants to cross borders, for instance, on trains, or as a result of “fissures” in the border enforcement. An instability of the border emerges concurrently through our different materials. For border crossers, the border environment could change very quickly, for instance, due to weather conditions, border closures ahead, or border police approaching at certain times (cf. Mollerup 2020). This border instability requires attentiveness to the here and now of the border characteristic for border navigated as it emerges from our social media data through posts, such as the update on the Hungarian border situation presented above, and through efforts to obtain information, such as:

[Social media data] Who has more information about the Danish border today? Posted on Arab Hamburg, 10 September 2015, translated from Arabic.

There is often a future orientation to these updates, which aim to facilitate or inform navigation. The border-crossing practices comprise navigation both through and around the built border. Turning to our ethnographic material, the affinity between traced lives and experienced lives in navigating the instability of the border becomes visible:

[Ethnographic material] After arriving in Northern Europe in September 2015, Ghada and Ziad took a regular bus, trying to blend in among regularised travellers. As they reached the border, a police officer entered the bus and asked everyone to take out their passports. Ghada silently told her husband to remain quiet as the police officer approached them. She then explained to her that they were from Lebanon but living in Germany, and that they were on their way to visit her uncle who lived in the country they were trying to enter, a story which only bore a vague connection to the truth. The police officer asked if they had their passports with them. Ghada confirmed, hinting at her bag. Without asking them to show their passports, the police officer let them stay on the bus. Ghada and Ziad saw other irregularised migrants being led off other buses at the border, but they were themselves allowed to cross the border despite never showing their – non-existent – Lebanese passports. Ziad and Ghada, Syrian refugees, based on interviews on 3 and 20 March 2019.

The instability of the border is also followed closely by solidarians, who would assist irregularised migrants in crossing the border or negotiate with border authorities concerning irregularised migrants’ rights and how to best manage the border in specific situations. This is characteristic for the solidary border. In the following excerpt from our social media data, we learn how the solidary border can simultaneously cooperate with and work against authorities. In the German–Danish borderlands, the Flensburg-based network Refugee Welcome Flensburg announced through a press release that they had decided to follow new measures after the Danish border police had forced migrants to register upon arriving in Denmarkfrom Germany. Refugee Welcome Flensburg had previously assisted irregularised migrants by providing donation-funded tickets for the transit buses (“Bahn und Schienenersatzverkehr”) provided by Deutsche Bahn due to overfilled trains and deliberately with the purpose of avoiding “Schleusserei,” i.e., illegal border crossings:

[Social media data] Press Release: The organisational team of the initiative ‘Refugees Welcome Flensburg’ reacts to the current conduct of the Danish (border) police. This happens after two public transit buses [Bahn und Schienenersatzverkehr] within the last 24 hours, shortly after arrival at the Danish border, have been stopped in order to forcibly register all the refugees onboard [which is why] the volunteers feel incapable of assisting with further transit help. For the past 7 weeks, the volunteers at the Bahnhof provided humanitarian aid and thereby worked cooperatively and transparently together with all the relevant authorities on both sides of the German-Danish border. Posted on Wir Sagen Moin (repost from Refugees Welcome Flensburg), 29 October 2015, translated from German.

Realising that the Danish border police would stop transit buses in order to register the irregularised migrants upon arrival, Refugees Welcome Flensburg ceased providing this kind of transit assistance:

[Social media data] So far, the refugees have been able to travel largely unhindered. The form of the random spot checks was typical of counter-terrorism measures. The current controls can only be interpreted as deterrent measures. People’s fates are thereby at risk. The current situation lacks any clarity, as the transports are stopped randomly, making it impossible for the organisational team to inform refugees about their onward journey. From the team’s point of view, the assistance at the station can currently be limited to supplying only food and clothing. (ibid.)

Similar kinds of stories are told in our ethnographic material. These materials move alongside one another without being directly linked to each other. In the following case, we highlight an example characteristic of the politico-legal border, which also focuses on borders as a matter of political negotiation and is thus addressed both at particular borders and in places well removed from these. For a volunteer based in Denmark, this border contestation played out when he learned from a secret Facebook solidarity group that a Rudolf Hess march was planned in Hamburg at the same time as the German–Danish border was being closed, leaving thousands of irregularised migrants stranded in Hamburg with thousands of neo-Nazis gathering. Realising the severity of the situation, he called the Danish police and explained, as he recalled us:

[Ethnographic material] If you close the border, there is going to be a large accumulation of refugees at the train station in Hamburg. Then there is no one who can protect these people against the Nazis (…). So, in the worst case, it will cost human lives, and there are certainly a lot of refugees who will get some unusually unpleasant experiences. Jens, Danish solidarian, 21 March 2019, translated from Danish.

The Danish border did indeed remain open at this time. While our social media data would typically not reveal such detailed insights into conversations between solidarians and the police, this data shows various traces of similar negotiations with the authorities to the one mentioned here.

The contrapuntal move of bringing our different research materials into conversation does more than just show how experienced and traced lives are part of the same world. When brought together, they also document the instability of the border in ways that transcend the particular case. The unstable border relates to practices of border navigated as well as solidary border and politico-legal border. Finally, the contrapuntal move allows for new narratives of the border: the border’s instability is indicative of its fragility can be contested, crossed, and conquered (cf. Hess and Kasparek 2017).

Contrapuntal Moves: Conclusion

In this contribution, we have rethought the relationship between social media data and ethnographic materials by establishing them as contrapuntally interconnected. With this contrapuntal move, we have actively sought to advance beyond logics of quantification and conventional scales of big and small, thick and thin, foreground and background that are offered in the computational data processing programs. For this purpose, we created a data environment suitable for transcending the criteria of relevance built into the software’s logic (such as spikes based on number of entries). Pursuing Tsing’snon-scalability approach, we refrained from nesting the “smaller” ethnographic material into the “larger” social media data.

Positioning our contribution in relation to similar, recent attempts at working with digital methods and ethnographic materials, we moved beyond stitching, complementing, assembling, or remixing strategies. Alongside Munk’s four models for overcoming the qualitative–quantitative divide (complementarity, single-level, curating, and algorithmic sense-making), we positioned our research at the overlap between single-level and curating. We thus suggest that contrapuntal analysis could be thought of as a fifth model for overcoming the qualitative–quantitative divide.

By establishing relations in our material as contrapuntal, we have explored affinities between seemingly disparate materials: the social media data and our ethnographic material. We have done so with the overarching aim of identifying potentials and further questions for digital migration research: On the one hand, we aimed to create a conceptual framework, beyond the existing practices regarding online ethnography, in which ethnographic material and social media data can co-exist without being forced into a relational structure. On the other hand, we wished to show how scrutinising digital fields of navigation among vulnerable groups, such as irregularised migrants, poses specific ethical challenges and methodological issues that, if anything, become more prominent when different types of materials co-exist. We hope that this contribution can stimulate further critical reflection in the field of digital migration studies, particularly regarding the conceptual understanding of how big social data and ethnographic material can be brought into conversation.

Developing a contrapuntal lens enables us to show not only the multiplicity of the border, enacted through these three different-yet-related versions of the border, but also how these border versions are related beyond the particular case. We thus argue that the border multiple (Andersen and Sandberg 2012) is likewise enacted in social media communications, which can be counted, measured, and visualised into peaks while retaining their qualitative affordances. The opposite is true as well: recognition of how the border is enacted through social media data has more far-reaching potential when brought into conversation with the ethnographic material. The different materials thus present different-yet-related paths to knowledge of the multiplicity of borders. The contrapuntal approach, we argue, thereby establishes ground upon which we can expand our ethnographic insights without compromising them. As we have demonstrated, the insights are expanded because of their connectedness to the social media realm established and highlighted through our contrapuntal lens.

Our contrapuntal analysis has thus allowed us, following Said’s call, to bring together discrete and internally coherent yet simultaneously co-existent and interacting experiences. In so doing, we have enabled new narratives of the border. The border multiple is at once fragile and robust: fragile because the border can be destabilised through the actions of solidarians and irregularised migrants, and robust because of governments’ increasing militarisation and fortification of the border. We suggest that this conversation on materials foregrounds the interconnectedness between practices of irregularised border crossings—an interconnectedness that cannot remove or diminish the chaos, changeability, and insecurity connected to these practices but that enables new narratives of the border’s fragility.

Notes

  1. 1.

    Thanks to our colleague Tine Damsholt for reminding us about the metaphor of the contrapuntal.

  2. 2.

    According to Rozakou (2018, 200) the term solidarian draws on a specific, emic notion (allilegyoi) of standing up in solidarity (2018, 190). The term has occurred in different, specific historical conjunctures and has been still more widely used, also among solidarity initiatives for refugee and asylum-seeker relief in Greece. However, we find that while acknowledging its localised legacies, several of the general traits of this notion can be applied to other local contexts as well. For instance, its anti-authoritarian legacy (in Greece related to the post-dictatorship anarchist movements) as well as the absence of, and in some cases deliberate distancing from humanitarianism discourse, applies to several of the solidarity initiatives included in our study.

  3. 3.

    We are indebted to Alaa Almeiza for transcribing our Arabic interviews into English translations.

  4. 4.

    All translations have been done by the authors. Thanks to Anna Sandberg, Annika Lindberg, and Mahmoud Alsayed for checking, respectively, German, Swedish, and Arabic translations.

  5. 5.

    All names are made up for the protection of research participants.