1.1Previous research on media representations of refugee

Previous research on representations of RASIMs has largely focused on the study of discourse representations in the news, particularly print news, given that these are more easily amenable to the compilation of corpora and analysis. The significant body of work produced in this area is based on the recognition that media play a key role in establishing specific frames for the interpretation and assessment of current realities relating to refugees and, consequently, in affecting public opinion and sentiment towards people in suffering. A recurrent finding in these critical approaches to the study of media representations of RASIMs is the reproduction of negative perceptions, which results in the ‘othering’ of these social groups and their dehumanization in public discourse (Gabrielatos and Baker 2008Gabrielatos, Costas, and Paul Baker 2008 “Fleeing, Sneaking, Flooding: A Corpus Analysis of Discursive Constructions of Refugees and Asylum Seekers in the UK Press 1996–2005)”. Journal of English Linguistics 36 (1): 5–38. ).

Dehumanizing discourse further sediments existing negative representations of RASIMs and creates feelings of contempt for them and their lives (Esses et al. 2013Esses, Victoria, Stelian Medianu, and Andrea S. Lawson 2013 “Uncertainty, Threat, and the Role of the Media in Promoting the Dehumanization of Immigrants and Refugees”. Journal of Social Issues 69 (3): 518–536. ). Some of these pervasive negative representations include their discourse construction as a “threat” or an “invader” to “national borders”, neglecting the realities that motivate migration in the first place. Ruth Wodak (2015)Wodak, Ruth 2015 “Protecting Fortress Europe?” The Negotiation of ‘Borders’ and ‘Benchmarks’in National and EU Arenas.” Seminar presented at Forksning Centre., more specifically, points to representations of refugees as a threat to “the national body” in both right-wing and mainstream discourse about refugees in Europe. Male refugees, in particular, have been the target of bias and negative categorizations as “rapists” or “terrorists”. These are often fronted as justifications for populist movements that get disseminated via slogans, such as #RefugeesNOTwelcome (Walker et al. 2016Walker Rettberg, Jill, and Radhika Gajjala 2016 Terrorists or Cowards: Negative Portrayals of Male Syrian Refugees in Social Media, Feminist Media Studies, 16 (1): 178–181. ).

From 2015 onwards, however, the increased negativity surrounding the discourse about refugees and the dehumanizing images in circulation has been countered to a significant extent by more positive and sympathetic portrayals. For example, shifts from the representation of migrants as “fortune seekers” or “adventurers” to “innocent victims” or “vulnerable”, “tragic asylum seekers” in need of support and assistance have been noted by a host of researchers (Blommaert 2015Blommaert, Jan 2015 “One Crisis, Three Photos: How Europe Started Caring for Refugees”. Alternative Democracy, Ctrl+Alt+Dem. Research on Alternative Democratic Life in Europe [online]. Available at: https://​alternative​-democracy​-research​.org/. Accessed: 29/09/2018.; Parker 2015Parker, Samuel 2015 “ ‘Unwanted Invaders’. The Representation of Refugees and Asylum Seekers in the UK and Australian Print Media”. eShapr23 (Myth and Nation). Available at: http://​www​.gla​.ac​.uk​/media​/media​_404384​_en​.pdf.; Gualda and Rebollo 2016Gualda, Estrella, and Carolina Rebollo 2016 “The Refugee Crisis on Twitter: A Diversity of Discourses at a European Crossroads”. Journal of Spatial and Organizational Dynamics 6 (3): 199–212.; Smith Dahmen et al. 2018Smith Dahmen, Nicole, Jesse Abdenour, Karen McIntyre, and Krystal E. Noga-Styron 2018 “Covering Mass Shootings: Journalists’ Perceptions of Coverage and Factors Influencing Attitudes”. Journalism 12 (4): 456–476.). This shift to more positive representations is largely achieved through the use of human stories (Steimel 2010Steimel, Sarah 2010 “Refugees as People: The Portrayal of Refugees in American Human Interest Stories”. Journal of Refugee Stories 23(2): 219–237. ; Hickerson and Dunsmore 2015Hickerson, Andrea, and Kate Dunsmore 2015 “Locating Refugees: A Media Analysis of Refugees in the United States in “World Refugee Day” Coverage”. Journalism Practice 10 (3): 424–438. ; Cooper et al. 2016Cooper, Samantha, Erin Olejniczak, and Caroline Lenette 2016 “Media Coverage of Refugees and Asylum Seekers in Regional Australia: A Critical Discourse Analysis”. Media International Australia, 16 (1): 78–89. ). For example, Smith Dahmen et al. (2018)Smith Dahmen, Nicole, Jesse Abdenour, Karen McIntyre, and Krystal E. Noga-Styron 2018 “Covering Mass Shootings: Journalists’ Perceptions of Coverage and Factors Influencing Attitudes”. Journalism 12 (4): 456–476. report high emotional and sympathetic reactions to images of refugee children, including Alan Kurdi. Other scholars have pointed to the expression of strong political emotional reactions, directing blame to national targets, governments and humanity (Mortensen 2017Mortensen, Mette 2017 “Constructing, Confirming and Contesting Icons: the Alan Kurdi Imagery Appropriated by #humanitywashedashore, Ai Weiwei, and Charlie Hebdo”. Media, Culture and Society 39 (8): 1142–1161. ; Olesen 2017Olesen, Thomas 2017 “Memetic Protest and the Dramatic Diffusion of Alan Kurdi”. Media, Culture and Society 40 (5): 656–672. ).

At the same time, however, emotions of fear, hostility and the belief that borders ought to be closed continue to occupy a prevalent place in public discourse (Chouliaraki and Stolić 2017Chouliaraki, Lilie, and Tijana Stolić 2017 “Rethinking Media Responsibility in the Refugee ‘Crisis’: a Visual Typology of European News”. Media, Culture and Society 39(8): 1162–1177. ). These different portrayals underline the continuous shifts in the framing of the so-called ‘refugee crisis’ in media narratives. As Georgiou and Zaborowski (2017)Georgiou, Myria, and Rafal, Zaborowski 2017 “Media Coverage of the “Refugee Crisis”: A Cross- European Perspective”. Council of Europe Report (DG1 (2017) 03). Council of Europe. note in their study of print media across eight European countries, the emphasis on tolerance, solidarity, and humanitarianism in September 2015 quickly gave way to discourses of securitization and fear in November 2015.

The large bulk of media studies and critical discourse analyses of representations of RASIMs has tended to foreground aspects of these representations as discourse products. In other words, the focus of these studies is placed on the recurrent lexical, grammatical and discourse choices – mostly verbal – which create specific ways of talking about refugees and of interpreting events and realities relating to them, i.e. producing different types of frames.

More recent work in critical discourse studies has considered aspects of multimodality (Machin and Mayr 2012Machin, David, and Andrea, Mayr 2012 How to do Critical Discourse Analysis. A Multimodal Introduction. London: Sage.; van Leeuwen 2008van Leeuwen, Theo 2008 Discourse and Practice. New Tools for Critical Discourse Analysis. Oxford: Oxford University Press. ) in relation to photojournalism (Chouliaraki and Stolić 2019Chouliaraki, Lilie, and Tijana Stolić 2019 “Photojournalism as Political Encounter: Western News Photography in the 2015 Migration ‘Crisis’”. Visual Communication 18 (3): 311–331. ; Caple 2013Caple, Helen 2013 Photojournalism: A Semiotic Approach. London: Palgrave Macmillan. ), highlighting how news photos also carry meaning and how they engage and position readers attitudinally (Economou 2014Economou, Dorothy 2014 “Telling a Different Story: Stance in Verbal-Visual Displays in the News”. In Critical Multimodal Studies of Popular Discourse, ed. by Emilia Djonov and Sumin Zhao, 181–201. New York: Routledge.; Bouko this issueBouko, Catherine this issue). “Emotions through Texts and Images: A Multimodal Analysis of Reactions to the Brexit vote on Flickr”. Pragmatics (Special Issue: Networked Emotion and Stancetaking ed. by Korina Giaxoglou and Marjut Johansson X x xx xx ). There is still ample scope for investigations of how such multimodal aspects combine with narrative practices in online contexts, especially given the fact that in media and critical discourse studies, the concept of narrative is often used metaphorically, broadly referring to ideologies (De Fina and Georgakopoulou 2012De Fina, Anna, and Alexandra Georgakopoulou 2012 Analyzing Narrative. Discourse and Sociolinguistic Perspectives. Cambridge: Cambridge University Press.).

In this article, we are interested in how representations of RASIMs are reworked into visual representations producing and reproducing specific kinds of frames, which guide the further circulation of stories and sedimentation of particular stances to them in connective media. The next section discusses in more detail the concept of frames.

1.2Approaches to frames

The notion of frames is used in different ways across disciplines, with no consensus over its definition or the methods for its study, except perhaps for its recognition as an important concept in the study of representation, meaning, and the organization of social experience (cf. Goffman 1974Goffman, Erving 1974 Frame Analysis: An Essay on the Organization of Experience. Cambridge, MA: Harvard University Press.). Among these different approaches, specific types of frames, including news, discourse (or cultural), and interactional frames have emerged as the object of study for media and cultural studies scholars, critical discourse analysts, and interactional sociolinguists, respectively.

The study of discourse representations in the media often draws – directly or indirectly – on news frames, defined in terms of specific configurations of verbal and visual information that propose particular angles on what the issue is about, how it can be addressed, and who is responsible for creating and for doing something about it (Tewksbury 2015Tewksbury, David 2015 News Framing. Oxford: Oxford University Press.). The concept of discourse frames, which has also been applied to the study of media representations emphasizes cognitive aspects of frames as “knowledge representations about the world” (van Dijk 1977van Dijk, A. Teun 1977 “Semantic Macro-Structures and Knowledge Frames in Discourse Comprehension.” In Cognitive Processes in Comprehension, ed. by M. A. Just, P. A. Carpenter, 3–32 New York, London: Taylor & Francis., 19). These approaches to frames tend to restrict analysis to the investigation of topics and language choices and to pay little attention to the way these discourse representations are negotiated in everyday interaction. As Fisher (1997Fisher, Kimberly 1997 “Locating Frames in the Discursive Universe”. Sociological Research Online 2 (3). Available at: http://​www​.socresonline​.org​.uk​/2​/3​/4​.html. Accessed: 15/09/2018. , n.p.) notes, “scholars cannot identify frames by counting the appearance of key words and phrases, or by specific argumentative structures. Instead, one must look for storylines about what is to be comprehended.” In addition to storylines, it is also important to look at interaction and story participation patterns.

From an interactional perspective, frames are conceived of as situated orientations of participants to a discursive context, which makes meaningful interaction possible or explains communication breakdowns; recognizing, for example, whether something is said as a joke or as a serious comment is key for successful communication. In other words, frames are understood as “part of the interpretive means by which participants understand or disambiguate utterances and other forms of communicative behaviour”, including aspects of metacommunication (Jaworski and Coupland 2006Jaworski, Adam, and Nicolas Coupland 2006 The Discourse Reader. 2nd edition. Oxon: Routledge, 41–48., 28–29). Frames also refer to participants’ stancetaking, i.e. the way participants in conversation shift their stance in response to new or changing situations and to “the alignment participants take up to themselves and others in the situation” (Tannen and Wallat 2006Tannen, Deborah, and Cynthia Wallat 2006 “Interactive Frames and Knowledge Schemas in Interaction: Examples from a Medical Examination/Interview”. In The Discourse Reader, ed. by Adam Jaworski, and Nicolas Coupland, 2nd edition. 332–349. Oxon: Routledge., 363). As Coupland and Ylänne (2006)Coupland, Nikolas, and Virpi Ylanne 2006 “Relational Frames in Weather Talk”. In The Discourse Reader, 2nd edition, ed. by Adam Jaworski and Nicolas Coupland, 349–362. Oxon: Routledge. suggest, frames can be also investigated as a process through which societies reproduce meaning (cf. Goffman 1974Goffman, Erving 1974 Frame Analysis: An Essay on the Organization of Experience. Cambridge, MA: Harvard University Press.). This understanding of frames points to their importance as acts of stancetaking, which shape – and are shaped by – participants’ political, social, or cultural positioning in specific discourse domains.

Acts of stancetaking in interaction are not solely speaker-centred language phenomena associated with an individual user’s linguistic choices. Rather, they are: (1) sequential in that the act of taking a stance often becomes a target for the next speaker’s stance, (2) dialogical, as stances emerge in and through interactional co-participation and (3) discursive, in that acts of stancetaking are embedded in other communicative acts, such as conversing, arguing, or narrating (see Du Bois 2007Du Bois, John 2007 “The Stance Triangle.” In Stancetaking in Discourse. Subjectivity, Evaluation, Interaction, ed. by Robert Englebertson, 139–182. Amsterdam, Philadelphia: John Benjamins. , 140).

In order to apply this more interactional concept of frames and stancetaking to the study of story participation patterns in social media communication, we start from the widely established distinction in narrative analysis between the taleworld, that is the world where characters move and live, and the storyrealm, that is the storytelling event at the centre of which the taleworld lies, which is, in turn, embedded in a conversation (or in other kinds of communicative practices). According to Young (1987)Young Galloway, Katharine 1987 Taleworlds and Storyrealms: The Phenomenology of Narrative. Dordrecht: Martinus Nijhoff. the storyrealm and taleworld echo Goffman’s (1974Goffman, Erving 1974 Frame Analysis: An Essay on the Organization of Experience. Cambridge, MA: Harvard University Press., 21) notion of frames in that they work as “schemata of interpretation” used by people “to locate, perceive, identify, and label events and occurrences to understand participants’ perspectives on stories” (De Fina 2016De Fina, Anna 2016 “Storytelling and Audience Reactions in Social Media”. Language in Society 45, 473–498. , 479). Story frames are, in essence, the sites where participants’ stancetaking and positioning are articulated, and thus constitute key sites for the study of stancetaking as embedded in narrative practice.

In the present study of Twitter reactions to the shared story of the life and death of Alan Kurdi, we are interested in identifying the salient elements of the story and storytelling event that draw participants’ attention and evaluative comments and create specific kinds of stancetaking and positioning for sharers. We argue that the narrative practice lens is necessary for moving beyond the identification of ‘static’ discourse representations, be they negative or positive, and consider instead the ongoing (re)constitution of discourse representations in (inter)action as an integral part of meaning-making and positioning practices. The next section presents the data for this study and the methods we have used to collect and handle them.

2.1Coding scheme and research questions

In this article we look at acts of stancetaking embedded in story frames in the context of online participation to the shared hashtag story #JeSuisAylan. Our research questions are the following:

1. how do networked users participate to the hashtag story #JeSuisAylan?

2. how do these participation patterns relate to existing media representations of refugees?

To address these questions, we look for the different story frames produced and reproduced through acts of stancetaking taken in the process of participating to the shared story. To identify these frames, we draw on Anna De Fina’s analysis of YouTube users’ comments (De Fina 2016De Fina, Anna 2016 “Storytelling and Audience Reactions in Social Media”. Language in Society 45, 473–498. ). In her study De Fina examined participation roles in digital storytelling as they relate to the range of frames available in an interaction.

The comments analyzed by De Fina are somewhat different to the material examined here given that they revolve around a monologic, ‘big’ story that had been shared in video format by Charlamagne Tha God, a radio host and TV personality in the US, and then reposted on Gawker titled “This is an amazing story about Rihanna partying”. We focus, instead, on a ‘small’ story, in the sense of a transportable, seemingly fragmented, distributed and condensed story of an iconic death shared via images and their accompanying comments. Despite this difference in the story format of our data, the patterns of participation in which these comments are produced and shared are similar in that they are intricately linked to platform-specific affordances of commenting in polylogal (Page 2018Page, Ruth 2018 Narratives Online: Shared Stories in Social Media. Cambridge: Cambridge University Press. ) or poly-storying (Georgakopoulou and Giaxoglou 2018Georgakopoulou, Alexandra, and Korina Giaxoglou 2018 Emplotment in the Social Mediatization of the Economy: The Poly-storying of Economist Yanis Varoufakis. Language@Internet 16: 1–51.) contexts, where comments may be physically adjacent but not necessarily referentially and explicitly connected to each other. For the analysis of tweets, we use Young’s (1987)Young Galloway, Katharine 1987 Taleworlds and Storyrealms: The Phenomenology of Narrative. Dordrecht: Martinus Nijhoff. types of story frames with a focus on “the taleworld, the storyrealm, or the external world outside the interaction” (De Fina 2016De Fina, Anna 2016 “Storytelling and Audience Reactions in Social Media”. Language in Society 45, 473–498. , 479); this analytical framework has been selected for pinpointing how comments are discursively connected through the production and reproduction of particular types of frames.

For the coding of our data, we marked individual posts for three different categories noting (1) details about each post (date and time stamp, username, content, interactivity metrics), (2) social media activity type and (3) story frame type. With respect to the details we noted about each post (coding 1), in order to ensure the anonymization of posts usernames recorded in the initial coding were deleted at a later stage.

The content of the post refers to the tweet’s content including message and hashtags, and any hyperlinks that may have been part of the post. In the case of hyperlinked content, we sought to retrieve the image, cross-linked tweet or news article, where possible (i.e. unless hyperlinks were broken links when we accessed them).

Acknowledging that we communicate through “culturally recognized activities” with identifiable goals among community members (Levinson 1979Levinson, Stephen 1979 Activity Types and Language. Linguistics 17 (5–6): 66–100., 69), the coding for social media activity type was aimed at providing an insight into the different types of sharing users select from when participating in an unfolding Twitter ‘conversation’ and it is summarized in Table 1. Commenting refers to posts that feature user-created content in its own right or added to reshared content. Posts that feature one or more hashtags or hyperlinks to an image, a news article, or other social media profiles without any added comment to them are coded separately as Resharing. Replying includes any public replies to a previous message or contributions added to a thread. In the first round of coding, we had also included the category of cross-posting, i.e. posting automatically from another platform but given that there were very few instances of this type of activity the category was left out of the scheme. As Adami (2014Adami, Elisabetta 2014 “Retwitting, Reposting, Repining, Reshaping Identities Online: Towards a Social Semiotic Multimodal Analysis of Digital Remediation”. LEA – Lingue e Letterature d’Oriente e d’Occidente 3: 223–243., 226) notes automatic crossposting is usually discouraged and users instead prefer to personalize a post each time it is re-posted in a different social media profile involving minimal intervention to reshape it to suit the site- specific audience.

Following De Fina’s categorization, our coding for story frames initially included four different foci: the storyrealm, the taleworld, the outside world, and a second story, but ended up focusing on the first three, given that we did not identify any second stories in our corpus. The categories used either bracket a discourse activity as a story (Taleworld) or reveal attitudes towards the story or the storytelling event (Storyrealm) (De Fina 2016De Fina, Anna 2016 “Storytelling and Audience Reactions in Social Media”. Language in Society 45, 473–498. , 479). In our data we followed a similar path to De Fina, which involved looking first for the main topic of the message and then recognizing the following foci and sub-foci. This process resulted in the coding categories summarized in Table 2:

Table 2.Coding scheme for story frames

Category	Content type and examples from the corpus
Storyrealm “the storytelling event at the center of which the taleworld lies” (Young 1987Young Galloway, Katharine 1987 Taleworlds and Storyrealms: The Phenomenology of Narrative. Dordrecht: Martinus Nijhoff. )	Participants: direct address to networked publics, often calling to (some form of) action, e.g. “#JeSuisAylan; We went to the streets for Charlie. But for #Aylan?”
	Tellability: reference to why the story is worth telling, e.g. “#jesuisaylan Cette photo qui fait la une de tous les journaux remue toutes les consciences.”
	Shareability: exclusive use of the hashtag or a series of hashtags, e.g. #JeSuisAylan
	Metacomment: explicit reference to the hashtag #JeSuisAylan, e.g. “On est à deux doigts du #JeSuisAylan ou #JeSuisUnMigrant”
Taleworld “the world where characters move and live” (Young 1987Young Galloway, Katharine 1987 Taleworlds and Storyrealms: The Phenomenology of Narrative. Dordrecht: Martinus Nijhoff. )	Character: comment on the character of the story, e.g. “#JeSuisAylan cet enfant est le mien…; Il s’appelait Aylan Kurdi, et il avait 3 ans. #jesuisaylan [hyperlink to image].”
	Plot elements: comment on other aspect of the story, e.g. “Aylan et Ghalib fuyaient la guerre pour un avenir meilleur [hyperlink to news article] #JeSuisAylan #JeSuisGhalib.”
Outside World	News story: explicit comment on coverage of the story in the media e.g. “Le choc des photos. L’arrogance européenne résumée en une et en page 5 du Monde. #AylanKurdi #JeSuisAylan #réfugiés.”
	Europe or the refugee crisis: explicit orientation to more general issues relating to Europe or the refugee crisis e.g. “The European Value is Humanity. The European Valuta is Euro Don’t mix Values and Valuta. #jesuisAylan #refugeecrisis”)

Recognizing the multimodal nature of the posts in our corpus, we also coded all the images that were shared alongside the posts, applying similar categories to the ones described for the verbal story frames (see Table 3).

In the next section we present and discuss the findings from our analysis.

3.1Social media activity types

As Table 4 shows, in terms of social media activity types, the analysis of our corpus yielded an almost equal number of posts categorised as ‘commenting’ (87 instances) and ‘resharing’ (85 instances). In terms of resharing, more specifically, users seemed to opt for resharing just the keyword hashtag #JeSuisAylan (54 instances) and to a lesser extent a series of connected hashtags (25 instances). There were also few instances of resharing news articles about the event (8 instances).

These findings suggest the importance of Twitter in this case as a space for users’ comment- and hashtag-based reactions to the shared story of Alan Kurdi with little need to refer to the details of the events – as evident in the low number in the corpus of resharing news articles or other more extended accounts of the related events.

There were few ‘conversations’ prompted by different types of posting activity, as evident in the small number of replies to tweets. This suggests that reactions to the story are not shared as part of an extensive engagement with user comments, but instead, they take the form of one- off individual contributions to an unfolding ‘conversation’ online made up of accumulating posts and aggregating hashtags in indirect interaction with one another. This finding prompts us to propose the term inter-reactions to capture this indirect relationship created between sequentially-ordered and spatially contiguous posting formats on social media comment spaces, which foster modes of interaction as participation to a shared story. This mode of participation will be further clarified in the analysis of verbal and visual story frames and acts of stancetaking.

3.2Story frames

The analysis of story participation shows that users’ emphasis is largely placed on the contribution of meta-comments to the shared story of the death of Alan Kurdi (114), as shown in Table 5. Fewer posts related to the Outside World (48) or focussed on aspects of the Taleworld (20).

The majority of the posts categorized as relating to the Storyrealm involved the resharing of the hashtag #JeSuisAylan or a series of related hashtags (79 instances). The use of the hashtags highlights a concern with the shareability of the story and the further promotion of its circulation and hence its visibility online. Fewer posts were oriented to networked publics calling for action (18 instances; see Example 1) or explicitly commented on the tellability of the story (12 instances; see Example 2).

Lastly, a limited number of posts called attention to the use of the hashtag, either negotiating the meta-message expressed by the hashtag (Example 3) or ironically nodding to the quick succession of hashtags in reactions to very different kinds of events, turning ‘JeSuis…’ statements into a cliché, i.e. an empty expression (Example 4).

Posts categorised as invoking the Taleworld included posts, which explicitly referred to Alan as the main character of the story (Example 5) or which invoked some of the details of the life of the boy and his brother, Ghalib, who also died when the boat they were on capsized.

Finally, comments categorized as ‘Outside World’ mainly involved comments on the broader issues the story raised about Europe, the world and humanity, attributing blame and responsibility to different social and institutional actors (38 instances; Example 7).

Comments in this category also included a number of posts (7), which highlighted the ambivalent coverage of the story as encapsulated in two images featured in the French newspaper Le Monde: the image of the dead child in the first page and the image of a model advertising Chanel in a similar pose on a shore (Example 8).

The above findings suggest that Twitter affords a space for inter-reaction – rather than interaction – taking the form of individual contributions of meta-comments, assessments, and evaluations to a story-in-circulation, i.e. a story that is shared among networked audiences who contribute brief assessments and evaluations of it. This is evident in the predominance of comments pertaining to the Storyrealm and highlighting a concern with further increasing the visibility of the story by sharing it, and comments relating to the Outside World, which underline its wider importance and relevance. The identification of considerably fewer comments pertaining to the Taleworld in the corpus suggests that users’ affective engagement with the main characters of the story is limited. Instead, affective engagement with the story appears to be driven by associations of the boy’s story to other stories and reactions to them, e.g. the Charlie Hebdo attacks or the refugee crisis and its implications. Users’ participation to the story is found to be centred around the shareability of the story in the recognizable and portable format of the hashtag story #JeSuisAylan, which extends the mediatization of the story online. The next section looks more closely at the story frames created by the images that are shared as an integral part of the tweets.

3.3Visual story frames

The majority of the images in the corpus focused on the Storyrealm, i.e. images involved some form of visual rescripting (in 33 instances). To a lesser extent did images concern the Taleworld, in other words fewer images involved the visual representation of the boy’s death as portrayed in Demir’s widely shared images (in 15 instances) and even less so, did they invoke aspects of the Outside World (i.e. images of users in solidarity rallies; in 6 instances). More specifically, images relating to the Storyrealm were found to involve predominantly rescriptings of the image of the boy’s dead body on the shore, which projected positive after- life scenarios (in 20 instances). Through these rescriptings, often elaborated and made more explicit in the verbal component of the message, users projected stances of solidarity and positions of alignment with Alan and refugees, more generally, as shown in Figure 1.

The example above shows a reworked image of the dead boy’s body as it was found on the shore. The lifeless body of the boy is rescripted as a body with wings, i.e. as an angel, while the earthly place of the shore is rescripted as an imagined space where the boy is flying, instead of just lying on it, lifeless. The image also features the slogan ‘Je suis Aylan’, which is repeated in the space of the message and hashtagged alongside additional hashtags that link this story to the crisis in Syria and refugees. This visual rescripting accompanied by the verbal component of the tweet offers a meta-comment on the shared story and elevates the image of the dead boy’s body and his story into an icon of affective solidarity with refugees.

This example shows how the visual and verbal components of the message create a coupling of the visual tellability and shareability of the story to project forms of affective solidarity typical of online participation modes (e.g. ‘JeSuisX’). As De Cock and Pizzaro Pedraza (2018De Cock, Barbara, and Andrea Pizarro Pedraza 2018 “From Expressing Solidarity to Mocking on Twitter: Pragmatic Functions of Hashtags Starting with #jesuis across Languages. Language in Society 47 (2): 197–217. , 214) have shown in their corpus-based study of the broadening of meaning in the case of the hashtag #JeSuisCharlie, such uses attest to the gradual evolution of the phrase ‘JeSuis’ into an independent element which becomes emblematic of Twitter users participating to public displays of mourning and solidarity via the use of hashtags.

Other images pertaining to the Storyrealm include slogan memes, which are reworkings of the popular ‘Je Suis Charlie’ meme, as shown in Figure 2.

In this case, the image becomes the vehicle of visual meta-comment on the story and projects alignment with the story character, Alan. The slogan’s presentation re- uses the format of the slogan ‘Je Suis Charlie’ and invokes statements of solidarity that had become the centre of media and social attention across large parts of the world earlier that year. The rescripting of the slogan super-imposes the story of a refugee child to the story of the attack at the offices of Charlie Hebdo as a way of claiming attention to it and calling for affective engagement with this story in an equally large scale of sharing.

In the corpus, there were also a number of instances of shared images of the boy’s dead body, as it had appeared in the print news, creating frames focused on the Taleworld (21 instances), i.e. as comments on the main character and the event of death. These were often accompanied by posts featuring the hashtag #JeSuisAylan or posts including more extensive comments that picked up on aspects of the taleworld rendered in an affective tone (see Example 9) or that commented on the wider importance of the story for European institutions and values (Example 10).22.In the examples we have refrained from reproducing the image of the dead boy for ethical reasons. In these cases, we only reproduce the text that accompanies the image.

In Examples 9 and 10, the visual component of the message shows the story, zooming in on the boy’s lifeless body, while its verbal component adds an evaluative assessment of why it’s worth telling. These messages combine textual, visual, and hashtag components into multimodal ensembles, which are amenable to further sharing as visual small stories. The circulation of such stories contributes to the creation of what Ruth Page (2018)Page, Ruth 2018 Narratives Online: Shared Stories in Social Media. Cambridge: Cambridge University Press. calls a shared story, which emerges from the engagement of multiple tellers in the production, consumption, and reproduction of particular representations of events and characters and assessments about these. These visual small stories select specific stances on the death of Alan Kurdi and the wider issues this death relates to, such as migration, attitudes to RASIMs, and the grievability of lives. The next section discusses in more detail the identified story frames in relation to the acts of stancetaking that they encode and make available for sharing.

3.4Acts of stancetaking

Story frames which focus on the Taleworld are based on the selection of salient visual and verbal aspects of the story as worthy of being shared. More specifically, visual taleworld frames involve the selection of the image of the boy’s body in focus. These frames invoke the event of the boy’s drowning as a shared story worthy of public attention and political action. By showing, rather than just telling the story, the image registers an affective stance about the death of Alan elevated to a symbolic level of interpretation. These visual frames are, thus, instrumental in emblematizing the image.

Sharing the emblematic image prompts the production of more explicitly articulated stances to the event as part of the Storyrealm, i.e. as meta-comments to this shared story. These meta- comments highlight the tellability of the story and the wider importance and relevance of the events for networked users. The use of verbal taleworld frames, on the other hand, points to the selection and foregrounding of basic story elements, for instance the tragic journey of the boy expressed in the first person (see Example 9 above) or the report of story orientation elements, such as the boy’s name and age expressed in the third-person (see Example 11).

Such individual reactions to the story shared on Twitter prompt specific modes of story participation and stancetaking online. In this type of stories, the image serves as the target of the evaluation of the story and the post as the space for expressing assessments on the event and characters, as discussed in the previous section.

The hashtag #JeSuisAylan is the most common act of stancetaking in the visual small stories in our corpus. This stance is expressed in the first-person point of view, anchoring the evaluative angle on the story to the individual user and creating a position for oneself in inter- reactions to this shared story. In some cases, the first-person singular focus is extended to the first-person plural signalling an inclusive ‘we’ as taking up this stance (see Example 14).

The expression #jesuisaylan can be taken to illustrate Du Bois’ definition of stancetaking as follows: “I evaluate something, and thereby position myself, and thereby align with you” (Du Bois 2007Du Bois, John 2007 “The Stance Triangle.” In Stancetaking in Discourse. Subjectivity, Evaluation, Interaction, ed. by Robert Englebertson, 139–182. Amsterdam, Philadelphia: John Benjamins. , 163). At the same time, this act of stance-taking is communicated as a dialogic act through the use of hashtag, which points to an intended connection to networked audiences. Importantly, this act of stancetaking is iterable and transportable in that it is produced for further circulation.

The (inter)textual and visual combinations that make up the visual small stories invoke sociocultural value by expressing evaluations at different levels. These evaluations involve picking up the image of the dead Alan as the most salient element of the storytelling event, freezing it as a storytelling moment and showing it for further sharing. This mode of showing rather than telling mobilizes the expression of affective stances relating to sympathy and solidarity with the plight of children like Alan. In some cases, these stances extend to expressions of alignment to migrants (e.g. We are all migrants. #JeSuisAylan) and refugees more broadly (#Réfugiés pour moi l’accueil c’est OUI! #AylanKurdi #JesuisAylan #Jesuisréfugié #refugeeswelcome #PS; Trans.: #Refugees for me the welcoming is YES! #AylanKurdi #JesuisAylan #Iamarefugee #refugeeswelcome #PS).

Despite the mimetic character of acts of stancetaking in reaction to the tragic story of Alan, it is important to note that such reactions can be said to be motivated by viewers’ need to restore order in the face of an event that causes high levels of discomfort and distress, namely a child’s death, which is the most marked type of death. Participation to the shared story via visual and verbal meta-comments seems to attest to the phenomenon of what is known in psychology as parasocial mourning (De Groot et al. 2015De Groot, M. Jocelyn, and Alex P. Leith 2015 “R.I.P. Kutner: Parasocial Grief”. OMEGA – Journal of Death and Dying 77 (3): 199–216. ), whereby distant viewers who are emotionally moved and affected by a death of someone they didn’t know before. Media scholars have pointed how social media have afforded increased opportunities for constructing such parasocial relationships with celebrities or TV characters (e.g. Lawerence Kutner, a character on the series House, M.D.) through which fans claim entitlement to public displays of mourning (Klastrup 2015Klastrup, Lisbeth 2015 “I didn’t know her, but...”: Parasocial Mourning of Mediated Deaths on Facebook RIP Pages”. New Review of Hypermedia and Multimedia 1 (1–2): 146–164. , 2018Klastrup, Lisbeth 2018 “Death and Communal Mass-mourning: vin Diesel and the Remembrance of Paul Walker”. Social Media + Society 4 (1): 1–11. ).

The case of this emblematic death, in particular, makes available positions of distant witnessing to the death of the ‘Other’. These positions resonate with stances to the Other’s suffering mediated by emblematic photographs in earlier historical moments. As Sontag (2004Sontag, Susan 2004 Regarding the Pain of Others. London: Penguin., 107) notes “certain photographs – emblems of suffering, such as the snapshot of the little boy in the Warsaw Ghetto in 1943, his hands raised, being herded to the transport to a death camp – can be used like memento mori, as objects of contemplation to deepen one’s sense of reality; as secular icons, if you will”. The amenability of the media image of Alan’s death to artistic rescriptings and visual small stories extends its meaning and upscales it to a symbol of (distant) death. Similarly to the ‘about-to-die’ images whose presentation draws attention to the generalizability of the event, rather than its specificity (Zelizer 2010Zelizer, Barbie 2010 About to Die: How News Images Move the Public. Oxford: Oxford University Press., 25), Alan’s image produces an emotional and contingent audience involvement which sidesteps the complex details of the life narrative of the boy, his family, his country or other refugees.