From Narrating the Self to Posting Self(ies): A Small Stories Approach to Selfies

Abstract Selfies have by now earned a prominent place in the diverse forms of self-representation on social media. In sociolinguistic terms, they have been undergoing a process of enregisterment (Agha 2005), as attested to in moral panics in public discussions and in a developing selfie-related lexicon. A phenomenon worthy of study then, yet largely unexplored, particularly within discourse and sociolinguistic perspectives on identities (possibly due to the selfie’s visual nature). My aim in this article is to venture (and justify) a ‘claiming’ of selfies by small stories. Selfies present the semiotic hallmarks of small stories, as I will show, and so they constitute valuable ‘data’, the study of which can benefit from small stories inquiry into genres as communicative and social practices on social media. Small stories apparatus is well-suited to selfie analysis: it has been specifically developed to account for genres that challenge the assumptions and modes of analysis of conventional narrative and life writing studies. Using data of selfie postings by adolescent women on FB, I will show how within a small stories framework, far from being narcissistic expressions of ‘ideal selves’, selfies emerge as contextualized and co-constructed presentations of self, shaped by media affordances.


Introduction
alignment is designed as a response to a post that is viewed and understood as an act of performance, an artful activity that invites scrutiny and appreciation of the self as character in time and place. In addition, in both cases, alignment capitalizes on the story-making possibilities that shared interactional history with the poster or any other prior knowledge affords, so that a contributor can produce specific alignment by elaborating on, amplifying and co-authoring the selfie-post. Finally, I will discuss the implications for selfies-as-small stories for the performance of self on the one hand and alignment and affiliation with self, on the other.

Selfies through the small stories lens
In recent work, I have documented a close association of small stories with the explosion of social media and their pervasive presence in everyday life, as facilitated by media convergence (Georgakopoulou 2013a,b;2014;. I have noted a set of features that conventional narrative analysts would see as a-typical or noncanonical, being salient in different platforms and practices, from Facebook (henceforth FB) to YouTube and Twitter, from statuses to spoof videos and retweets. These features involve fragmentation and openendedness of stories, exceeding the confines of a single posting and site and resisting a neat categorization of beginning-middle-end. They also involve multiple authoring of a post, as it becomes shared and distributed. There is also a tendency for reporting mundane, ordinary and, in some cases, trivial events from the poster's everyday life, rather than big complications or disruptions. These 'textual' features have led me to recognise the role of small stories research as a paradigm that prefigured the current situation when social media affordances have made small stories more widely available and visible in public arenas of communication through circulation (see Georgakopoulou 2013a). Emerging sociolinguistic work on stories in social media confirms the validity of this view and the usefulness of small stories for analyzing mediatized narrative activities (e.g. Georgalou 2015;Page 2012;West 2013).
The aforementioned features of small stories and the ways of analyzing them can also form a point of entry into the narrative-semiotic features of selfies. Put differently, selfies could no longer be excluded from narrative analysis or questioned in terms of their narrativity, on account of these features. Small stories research has successfully made the case for the epistemological and analytical need to recognize narrativity in these features. I have also been in a position to document the migration and remediation of a specific genre of small stories, which I have called 'breaking news', from face-to-face conversational contexts to mediafacilitated conversational contexts where new technologies are present, to digital media (Georgakopoulou 2013a). Breaking news is a quintessential genre of small stories for announcing life as it happens and for capturing the moment. If we accept that selfies are about capturing the moment², along with other social media narrative activities, small stories research offers a toolkit well-suited to studying selfies, as a personal historiography of the present, in ways that acknowledge a broad definition of narrative. I have shown elsewhere (2013b) how (small) stories on social media often begin with or are confined to narrative stancetaking, a moment of position taking where the poster mobilizes conventionalized communicative means to signal that there is a narrative tale or telling in process, in the making and/or in circulation.³ This signaling, as we will see below, often applies to selfies. Media-afforded narrative stancetaking elements (for instance, check-ins, tagging, notifications) position (certain) FB friends as story (co-)tellers and recipients, thus anticipating and even proposing certain modes of audience engagement with a selfie as a story. A perspective on selfies as small stories recognizes narrative stancetaking as an important aspect of posting selfies in context and for specific viewers/users. Storying in selfies is thus viewed as a dynamic, contextually emergent process co-constructed by selfie-posters and users that engage with them, rather than as an allor-nothing, ontological and a priori definition of a selfie as being or not being a (small) story (for details on the 'definition' of small stories, see Georgakopoulou 2015). Specifically, with a small stories perspective, the subjectivity of selfies becomes part of a contextual and interactional analysis that seeks to explore the partly conventional partly contingent links amongst ways of telling (the communicative how), social media affordances and constraints and selfie posters as communicators, social actors but also individuals with biographical repertoires. Positioning analysis, as recently developed within interactional approaches to narrative and identities (for an overview, see Deppermann 2015), provides a useful heuristic for studying how with specific semiotic choices and in specific contexts, narrators present themselves (see section 3.1 below).

Data and methods
I have been examining the ways in which we can extend small stories research to social media as part of a a larger project entitled 'Life writing of the moment: The sharing and updating self on social media'.⁴ My aims have been to chart the multi-semiotic forms (linguistic/textual, visual, auditory, etc.) that storying everyday life as it happens takes on a range of social media (e.g. YouTube, Facebook, Twitter), with emergent and remediated genres (e.g., selfies, retweets, spoof videos and remixes) and, where applicable, on the basis of specific (personal, political, social, etc.) incidents and issues (e.g., the Eurozone crisis). With a small stories analysis of the above, I also document the kinds of subjectivities, including ethical and political selves that life-writing of the moment engenders and how these are interactionally achieved.
Small stories research is routinely done in my study with the help of an open-ended, adaptive ethnography (Hine et al. 2009): this involves applying flexible routes to fieldwork over time to suit the mobile, ever-shifting landscape of social media. It also involves being open-minded about the use of 'remix' methods, in Markham's terms (2013), in the spirit of social media practices of bringing together unlikely modes in imaginative, reflexive and even playful ways. For instance, the researcher's own immersion and participation in social media culture with processes of catching up, sharing, and real-time tracking and life-streaming, are recognized as a major part of the development of ethnographic understandings. In addition, I have adapted digitally native methods for fieldwork: for instance, observing activities and postings systematically, as a 'lurking' participant in a specific site, so as to identify key-posters of small stories and respondents. Some of these methods involve auto-phenomenology, that is, the researcher's reflexivity about her own position, stakes, and interests in the field of social media engagement (Markham 2013). For instance, I have often examined my position as an 'outsider' to practices such as posting selfies, even using it strategically in off-the-record chats with teenage participants and their use of FB. I have also drawn upon observations and developed analytical lines as a result, on the basis of my identity as mother to a media saturated teenage daughter.
I have specifically employed the above methods for the exploration of self-presentation on FB. My initial study (2013) involved status updates and responses to them posted on FB walls. From my list of friends, I identified a female friend in her 30s, who posted the most status updates and I followed her wall for a period of 6 months, having secured consent from her and the friends. The current study of selfies began as a way of triangulating those data with users who appeared to embrace fully the increasing accessibility of video and visual facilities for status updates, including the posting of photographs and then selfies. I therefore looked into postings of a group of teenage friends who I selected from my daughter's friends. The selection was done in March 2015 on the basis of who were the top 5 selfie-posters after a systematic observation of one year. Principles of heavy disguise have been followed in both these cases and no visual material has been reproduced. For the selection and collection of data, I involved my 16 year-old daughter and some of her friends as key-informants. I asked my daughter in particular to single out 5 top selfie posters from her friends, with a preference for people from her active friends as her number of friends exceeds 1,500. It is interesting to note that none of the selected users have any privacy settings: their walls are accessible to non-friends too. My own FB 'friendship' with ca. 40 teenagers from my daughter's wide circle of friends helped me with initial orientations and subsequent intensive observations over a period of 18 months, which were akin to what FB (teenage) users often describe as 'stalking' somebody. On the basis of this scrutiny, I embarked on off-the-record chats with a few of them. In these chats, following Markham's remix, adaptive methods, I was not really following a structured list of interview questions. I was rather asking questions about practices of posting and engaging with selfies and seeking clarification on patterns that I had observed, so I was being open about questions that I had: 'why do you guys do X? what does it mean? Which selfie should I look at?' I was also interested in things going wrong, moments of gaps between what is expected and what gets done, bad experiences from posting selfies, etc. and I asked questions about those too. I also looked into cases when selfies were queried, contested, talked about, policed. The results of this analysis are beyond the scope of this article.
Finally, I involved my informants in discussions about a possible approach to selfies asking them to assist me in formulating some kind of a typology of selfies which captured their main visual arrangements in combination with what they commonly aimed to achieve. Perhaps unsurprisingly, their own descriptive language drew on existing lay characterizations of selfies: e.g. mirror selfies, beach selfies, etc. They did, however, make a very clear distinction between 'solo' selfies, which I ended up characterizing as 'me selfies' and group selfies. They also attached great significance to whether a selfie was going to be a profile picture or not. This extra-labour that goes into profile pictures has informed my analysis, as I will show below. The third type of selfies that I felt I needed to add was that of significant other selfies (see 3.1 below). This addition allowed me to tap into a common communicative act that I had observed as being associated with selfies, namely that of announcing or celebrating close relations with one person (e.g. best friend, mother, sibling) or indeed a pet. This was distinctly different, I felt, to group selfies (normally of 4-5 people).
Having singled out 5 users to be analyzed closely from a list of c. 10, I asked for consent to use the language data that accompanied their selfies in the period March 2014-March 2015, but I made it clear that no visual data would be used in any published material, not even in the form of blanking faces. I also assigned the role of each of them informing their close friends that this was happening and getting back to me with any problems or cases where people said they did not wish any of their 'comments' to be reproduced. A few users suggested that they would be happy even for visual material to be used. This openness is something that my BA students often report as part of their project for a course on social media, which I teach. It does fit in well in some respects with the phenomenon of self-branding, where ordinary individuals employ micro-celebrity techniques to achieve positive self-presentation with maximum audience reach. And these are all people who clearly bask in big numbers of friends and seek equally big numbers of Likes with their postings (400 likes was roughly the acceptable number for any profile picture posting).
The 5 people who I ended up choosing comprised 3 young women and 2 young men: I could have opted for 5 young women, but having the 2 young men in the list can provide a point of entry into studying gendered aspects of selfies in subsequent work. Preliminary analyses suggest that selfies by men often attract banter from male friends and humorously offered criticism about their looks (e.g. 'you are sooooo ugly'). This notably appears to be as highly conventionalized as the 'ritual appreciation' which abounds in the comments on selfies posted by the female participants (see discussion in section 4 below). The choice of zooming into the 3 top female selfie-posters for this article is justified by the fact that young (in particular, adolescent) female women are the primary selfie-posters. If we accept that selfies have developed some kind of enregisterment in their short social-mediatized history, to explore normativity by empirically studying demographically archetypal users, can be a first step. Table 1 provides information about the total number of selfies in each case, singling out me selfies posted as profile pictures and average number of Likes and comments that such selfies tend to attract. In terms of other information about the chosen users that is notable, as there were around 30 nationalities involved in the list of observed users and their friends, which hardly mapped with the people's actual locations, I decided to note where users were located. Of the list of chosen individuals, Kate is London-based with American parents. Maria is half-English, half-Greek, also London-based. Saachi has got Indian parents and she too is based in London. Petros is half Greek, half Romanian, based in Greece. Aris is Greece-based Greek. It is also notable that 4 languages are used systematically in the analyzed data, in the following order: English, Greek, Spanish and French, perhaps reflecting the languages that my daughter and I have knowledge of. I did not exclude any of these languages from my analysis of comments. In sum, this chapter is based on the close analysis of 189 selfies from 3 female selfie-posters (aged 16-17) and of a total of 1,713 comments that these selfies received.

Coding
All selfies were characterized as one of three types: me selfie, significant other selfie, group selfie (see 3.1 below). The number of likes and comments were also added along with names of people who liked, commented on and/or were tagged in posting selfies: this allowed me to see who emerged in each case as those who liked and/or commented most on somebody's selfie postings. Building on my previous analyses of FB data, I used the technique of what I call 'zooming-in': this involves going back to data already analyzed to further check and add coding for any new analytical lines of investigation. In this case, based on previous analyses of FB and YouTube data (Georgakopoulou 2013a), I added coding to check for narrative stancetaking elements, i.e. any language-based or visual references to time, place, characters, and emotions and any proposed connections amongst them. I also added coding to check for the frequency of certain knowing participation patterns that had previously emerged as salient. In particular, I coded explicit and implied references to knowledge of specific events, activities and/or characters and any information about the provenance of this knowledge (e.g., shared participation in an activity offline). I also included references to and reaction to the state of non-knowing. Finally, I conducted more off-the-record chats with some of the studied participants on FB to check for patterns of reciprocation to comments after the posting of a selfie. This confirmed my finding about the normativity of reciprocation in these cases, as it revealed what we can call 'hidden' responses too: for instance, the participants who were not thanking after a positive comment and reciprocating, claimed that they still reciprocated in other ways, e.g. by going back and liking a picture or post that a friend, who commented positively on their selfie, had posted. Another methodological and analytical angle which is beyond the scope of this article involved tracking the migration of selfies in the course of my study from FB to Instagram. This process is by now near-complete and most of my informants, particularly the London-based ones, see Instagram as the main platform for posting selfies while the frequent posting of selfies on FB tends to be associated with vanity and narcissism.

Analyzing selfies as small stories genres
In previous work, I have discussed how I move beyond a conversation-analytic perspective to document stories not just as interactionally achieved activities with sequentially organized implications for what is to follow, but also as social practices embedded in specific contexts (Georgakopoulou 2007). To do so, I employ heuristic methods which guide the analyst to forging links amongst ways of telling, the sites of the stories' tellings and taleworlds as well as the 'tellers' in the broad sense of communicators (e.g. producers and active consumers of stories). In this heuristic, drawing on identities-in-interaction perspectives, I assume that people's identities are partly manifested in participation roles, contingent on local interactions and shaped by a given interactional activity. In particular, narrative arrangements project specific roles and types of action for recipients and these are consequential for the participants' relations. For instance, it has been found that casting an interlocutor as a story recipient increases the requirement for showing alignment (i.e. understanding of) with the type of activity and affiliation with the speaker's stance on the reported experience and feelings (see Jefferson 1978;Lerner 1992;Mandelbaum 1987). This heuristic dictates a dual focus on online postings and responses to them without however predetermining what from each of the multi-layered ways of telling, sites and tellers will be of analytical importance and how their relations will be configured. In line with nexus analysis (Scollon & Scollon 2007), the aim is not to provide an inventory of all semiotic resources or participation frameworks in specific tales and tellings but, instead, to identify which ones are routinely associated with them and are facilitated or constrained by the social actions operative in them. A broad perspective on interaction online also needs to take into account, when applicable, the sharing of a posting across multiple events and spaces with multiple and unforeseen recipients and the multiple related stories that this may generate through media enabled processes of linking, replicability and remixing.

Positioning analysis and selfies
To forge links between the communicative how, the FB spaces that selfies are part of, and the posters, I drew on positioning analysis in its connections with small stories (Bamberg & Georgakopoulou 2008;Georgakopoulou 2013c), which seeks to identify and establish any iterativity in the following interrelated yet analytically separable levels of self-presentation: Level 1: This comprises the characters & narrators in the taleworld, their relations, evaluative attributions, activities and overall placement in time and place. In this case, Level 1 included the 'photographed' people, the selfie-takers and posters, the tagged people, any visual arrangements and narrative stancetaking elements.
Level 2: This comprises the ways in which a story is locally occasioned and told in interaction with other people. Who participates and how? Who ratifies, legitimates or contests which part of the story? Who co-authors, what and how? In this case, level 2 identifies modes of uptake & engagement with the selfie (e.g. with Like, Comment).
Level 3: This responds to the question of who the teller 'is' above and beyond the here and now of a storytelling activity. I have argued in earlier work (Georgakopoulou 2013c) that this question should not be viewed as a fossilization or essentialization of fleeting and interactionally constructed processes. It should however be a contextualized question that seeks to establish which 'sites' of storytelling may encourage or engender a more 'stable' presentation of self and statements about self that seem to be constructed as having some durability. In the case of FB, profile and cover picture spaces emerge as such spaces and it is fair to assume that the interaction of ways of telling, sites and tellers in their case is aimed at producing a more 'rehearsed', edited or 'stable' self-presentation.
Although it is difficult to suggest that the above levels of analysis were kept separate at any stage of the research, it is still fair to add that Level 1 was mostly analysed using a narrative-semiotic approach for the identification of the above emplotted elements; Level 2 was mostly identified using the small stories interactional approach to online data and Level 3 resulted from combined insights from level 1 and level 2 but it also included any ethnographic insights as explained in Section 2 above. It also took into account two notable affordances of FB that can be viewed as 'directional' (van Dijk 2013) to specific ways of presenting self: a) the status accruing affordances (Marwick 2013) which consist in the metrics of FB as indicators of popularity (e.g. number of Likes, comments, friends) and b) the tracking, monitoring and editing facilities (e.g. notifications, news feeds, being able to view who has liked, viewed your pages, etc. but also being able to view other friends' pages and postings). Both facilities are aimed at creating popular selves and can be conducive to what has been described by social media research as 'self-branding' activities, in which the self is presented as a brand, through marketing strategies applied to the individual (Marwick 2013). While the role of self-branding cannot be automatically assumed, it usefully shaped the starting points of a positioning analysis of selfies⁵, seeking to scrutinize their performative elements: how is 'displaying self and putting self up for scrutiny and appraisal engaged with by the audience' (Bauman 1986)? Positioning analysis showed that talking about selfies as an undifferentiated activity was too crude to capture the subtle yet important configurations between Level 1, Level 2 and Level 3. The placement of selfies, the visual arrangements of them and the commonly associated communicative purposes and ensuing engagements called for the following typology⁶: 'Me' selfies: display self Profile picture/(can be included in an) Album Significant other selfies: signal relationships Cover picture/Album Group selfies: capture the moment Timeline/Album In each of these types, the self is presented in specific visual perspective patterns. Me selfies tend to be retrica-filtered, designed and edited portraits. Full-length mirror selfies are much less common in my data. Particularly as profile pictures, me selfies present a left placement which, according to Kress & Van Leeuwen (2001), has the effect of proposing the 'photographed self' (the character) as a given. This works well with FB affordances of constructing a more or less recogniseable identity over time and profile pictures being a major part of that. Significant other selfies on the other hand normally present visual and verbal 'tie-signs' (Goffman 1971: 226pp) which inform and contain evidence about a relationship, its name, stage and terms. For instance, two best friends or partners may mirror one another as they are placed next to one another by doing a duck face or sticking their tongue out. The caption may also provide 'tie-signs": e.g. 'my bae', 'me and my gorgeous girl', 'besties', commonly followed by emojis, in particular heart signs. Finally, in group selfies, fun 'moments' and 'scenes' are captured (e.g. eating or drinking out with friends), usually with landscape techniques. The selfie-taker in these cases, who tends to be the selfie poster too, clearly emerges as the main stage-director or narrator of the moment, as it is down to them what is captured and how. As we will see below, the above features of Level 1 positioning were found to relate to how a selfie was engaged with. The more designed and edited a selfie is, the more it is engaged with as a performative display attracting positive assessments along with Like. This was certainly the case in me selfies.

Level 2: Narrative stancetaking and selfies
In addition to the visual narrative stancetaking elements of the selfie, language semiosis and emojis play a key-role in framing and (further) explicating selfies as small stories. They thus create an imagetext. In general, as I have argued elsewhere (Georgakopoulou forthcoming), the projection of narrative arrangements is enhanced by FB platform facilities such as: As we can see, these FB facilities encourage the inclusion of time, place, events, characters and/or condensed or indexical associations amongst them. Furthermore, posters can 'select' certain friends as ratified recipient: they can for instance tag friends in the photograph (for which they get a notification), include their name or post the selfie as part of a shared check in with them. We will see below what preferred responses this inclusion generates.

Selfies as small story performances
Posting selfies, particularly as profile and cover pictures, are high stakes activities both for the poster and for the audience. The heavily discussed anguish over how many likes a selfie will get and how many selfies young women in particular take before uploading one was easily attestable to in the case of my data. There was a lot of backstage editing and discussion about which selfie to post. It is therefore fair to assume from the outset that the posting is proposed as a performance act that raises the task of positive appraisal by the audience as an appropriate response. According to Bauman's (1986) influential definition of performance, a speaker assumes the responsibility to display communicative artistry and skill so as to be appraised and scrutinized by an audience. In this way, every performance calls forth special attention to and heightened awareness to the act of expression, in this case, the language, visual and other choices in the posting, and it sets up and an 'interpretative frame within which the messages being communicated are to be understood' (168). Performance is thus intimately linked with the deployment of conventionalized resources which come to be recognized as part of it. In the case of the data at hand, displaying a selfie or a picture comes with a conventionalized visual prominence of an edited self, who is placed at front-stage, the window dressing of FB, namely as a profile picture or cover picture. These photographs change less frequently than others that are uploaded sometimes almost on a daily basis. In my data, profile selfies tend to change once every 5-6 weeks, even by the most popular selfie-takers. There is also a clear 'labour' (Marwick 2013) that goes into uploading this type of photograph which can be easily seen as part of a project of positive presentation of self and self-branding: profile pictures become part of FB's timeline too, so they are instrumental in proposing a self-identity over time. Overall then, the extra-editing that tends to be applied to profile pictures combined with their viewability and relative permanence renders me selfies as the perfomative selfies par excellence. Below, we will see how this shapes 'friends'' engagement with them. Performative aspects combined with narrative stancetaking (see 3.3 above), position the viewers/audience as story recipients of the post and so it projects some sort of a narrative participation as the appropriate one. As I will show below, (more or less) implied associations between events, characters, emotive states and reactions in initial posts create conditions of story recipiency, more specifically, what I call positive story recipiency. This places certain members of the audience in a position to align with the stance in the original posting and to elaborate on, amplify and co-author it, on the basis of (shared) knowledge.

Me selfies: Ritual appreciation
As already suggested, Level 2 positioning analysis brought to the fore two systematic patterns of engagement with selfies, namely ritual appreciation and knowing participation. In ritual appreciation, highly positive assessments on the photographed person(s) in the selfie are employed while in knowing participation, the friends display some kind of knowledge about the selfie and the photographed person(s) that goes beyond the here-and-now. As Table 2 shows, these two modes of engagement occur in all three types of selfies, but with distinctly different frequencies. Me selfies are the par excellence type of selfies for ritual appreciation, while significant other selfies tend to attract knowing participation more. Positive assessment which is done in a ritualized way, as we will see below, is at the heart of processes of alignment, a keyfeature in Goffman's (1981) study of self-presentation in communication, consisting in the participants' signaling of their positions and recognizing the positions of others. Alignment is thus an interactionally constructed position that signals that the activity underway and the speaker's previous contribution have been understood. Alignment is very often signaled with supportive embodied and paralinguistic contributions. In addition, a typical feature of alignment, as documented within conversation analysis, consists in creating contributions that are lexically, syntactically, grammatically and sequentially similar to previous contributions (Guardiola & Bertrand 2013;Stivers 2008). Such convergent sequences are viewed as a linguistic manifestation of convergent relations and behaviours, a sign within psychological research of affiliation with another. Although the exact relations between doing alignment vis-à-vis the communicative action of a speaker's prior contribution and expressing affiliation with them and their stances are a matter of some dispute (see Guardiola & Bertrand 2013), a position that seems to be more prevalent in CA is that alignment is a prerequisite for affiliation (Stivers 2008). Aligning with what and how a speaker communicates displays support and endorsement of his/her conveyed stance. This is a keyfeature in the intimate association between ritual appreciation and me selfies. In me selfies, the 'self' is put up for scrutiny and appraisal and so alignment in these cases as understanding the purpose of the activity underway requires displaying affiliation with the portrayed self. Hence, alignment by ritual appreciation emerges as more common (and normative) in the case of me selfies. Alignment has often been documented as expressed with congruent contributions that mirror one another. In this case, my contention is that the congruence-creation goes one step further, enabled by the visual and spatial architecture of commenting facilities on media platforms and the diachronicity of contributions they allow for. This results in a pattern of several, atomized contributions by commenters, which tend to be concentrated in a specific time-frame (close after the posting) but diachronic contributions, separated in time, are also possible.⁷ This happens, however, when the initial post for some reason becomes topical again. For instance, it could be that a newly added friend likes 'this' picture and this may bring it back to the consciousness of other friends (especially followers), through notifications. In addition, although the commenters as a rule do not engage with one another, as their contributions look highly similar and are often format-tied (Goodwin 2007), it can be assumed that a contributor has read at least some other comments before posting theirs. The conventionalization of expression, particularly of positive evaluative language, in such cases is readily detectable, as we will see in the example below. Besides the recurrence of specific evaluative lexis, there is a systematic coupling of language with emojis, in particular hearts. Drawing on Zhao's concept of coupling as a systematic combination of semiotic resources, levels of expression or meta-functional variables (2010), in her study of Twitter, Zappavigna (2012) documents certain stable coupling patterns between interpersonal and ideational meanings, i.e. of a positive attitude with the ideational topic in hashtags. Her claim is that such coupled relations, which hold in a specific point of time and for specific activities, are conducive to the creation of communities on Twitter, based on shared values. In the case of coupling of evaluative language with emojis in the comments on selfies, the conventionalization of this relationship affords a sense of collectivity as part of the act of displaying alignment: numerous highly similar positive appreciations of an initial post, serve as the visual counterpart of lots of people clapping and cheering at the same time. This collective response is arguably less about shared values and more about many people acting together as spectators and as 'fans', responding appropriately to an act of performance, as suggested above. This shared appreciation and enjoyment is what, in my view, qualifies such aligned responses as instances of 'ritual appreciation', in Collins' terms: "a mechanism of mutually focused emotion and attention producing a momentarily shared reality, which thereby generates solidarity and symbols of group membership" (2004: 7).
At the heart of alignment by means of ritual appreciation is a performance frame of 'make believe'. To display yourself as a 16-year old young woman 'in view' of potentially hundreds of people certainly evokes occasions of catwalk and modelling and tallies with a micro-celebrity presentation of self, a set of practices which position the audience as a fan base (Senft 2013).⁸ At the same time, to view, scrutinize and 'admire' your 'friend', often a close friend offline, as a micro-celebrity on display, requires joining in (potentially many) others in a frame of spectatorship that is not part of your ordinary relationship with your friend. The result is alignment based on shared appreciation and expressed in similar terms across contributions. We can see this in the example below, from a selection of comments on Maria's updated profile selfie. As I have suggested, it is commonly the case that selfies posted as profile pictures by the selected participants receive hundreds of Likes (the number of friends that the participants have is c. 1500; see Table 1). 35 comments were posted in this case, and below, I have selected the 13 comments that were posted within a time-frame of 2 hours after the selfie was uploaded.
Like · Reply · December 29, 2015 at 7:19pm Maria beth baby Ellen chicaaaa!! Like · Reply · December 29, 2015 at 7:22pm Leah Stunning <333 Like · Reply · December 29, 2015 at 7.31pm Lizzy whaat you're soooo hot Like · Reply · December 29, 2015 at 7.43pm Maria You are xxx Like · Reply · December 29, 2015 at 7.45pm Kate Beyond beaut Maria!! Like · Reply · December 29, 2015 at 8.13pm Maria Like you can talk miss u :(( see me soon bb xxxxxx Like · Reply · December 29, 2015 at 8.18pm Selena Fit af Like · Reply · December 29, 2015 at 8.24pm Maria Aw thanku <33 We can see in the example above the use and recurrence of a closed set of evaluative lexical choices (e.g. beautiful/beauty/beaut, gorgeous, hot) that relate to physical appearance and are coupled with emojis, mostly heart symbols and heart-shaped faces. In fact, the typicality and recurrence of the same evaluative language with the same emojis across posts of (me) selfies in the data and across English, French, Spanish and Greek is almost unexceptional. Even comments that are set up as more 'individual' repeat formulaic and widely circulating assessments: e.g. 'It must have hurt when you fell from heaven' (referring to the photographed person as an angel).
This coupling of language with other semiotic modes, e.g. punctuation (exclamation marks), typography (capitalization, repetition of vowels that simulate elongation in speaking: e.g. sooooo) has the effect of amplification and upscaling of the meaning. This upscaling was reported by Zappavigna (2013) too in her study of the expression of appraisal (e.g. appreciation, evaluation) on Twitter. The effect of the addition of these contributions in a short space of time is, as already suggested, visual and linguistic congruence, which is spatially clustered and aggregated. The other notable conventionalized choice in the data is the preference -if not normativity-for reciprocity on the part of the poster that these responses create. We can see above how Maria routinely reciprocates to the comments by thanking ('thanks so much babe'), paying back a compliment (e.g. 'you can talk') and/or replying with affective terms of address ('my luv'). These replies too are concentrated in a short space of time and there seems to be a normativity for a minimal timelag (2-3 minutes) between a comment and a reply from the poster. My ethnographic observations suggest that after posting a selfie, posters are online for the first couple of hours at least, waiting for the comments and likes, and responding to the former. Maria's responses also present a coupling of affective language with emojis and they too are very similar to one another, even though they are tied to a specific comment thus forming dyadic exchanges between poster and commenter.

Significant other & group selfies: Alignment through knowing participation
I have argued elsewhere that FB posts present systematicity in terms of what types of responses/action they project and how these are taken up by commenters (Georgakopoulou 2013a). In particular, posts that report disruptive events in the poster's life are more likely to receive comments from their friends than a simple Like. ('Like' only is the case for routine everyday events). The report of disruptions is also more likely to lead to a further post from the poster where she/he thanks for any wishes and interest and updates on the situation. In all these cases, I have argued, close friends with whom the initial poster has had frequent interactions with offline (too), tend to bring in their knowledge when participating in posts with comments (Georgakopoulou 2016). Knowing participation can take many forms and it can serve many functions but what is notable for our discussion is how it becomes the main type of contribution at the Level 2 of selfies' positioning for significant other selfies and to a lesser extent 'group selfies'. These selfies seem to serve different purposes for different FB 'friends': they may be an announcement for non-knowing recipients and an opportunity for display of alignment through offline selfie-taking and other knowledge for knowing friends, as we can see in the example below from a significant other selfie: Example 2 Kate: [next to selfie of her and her best friend Anna, which is not reproduced here].
Waaaay up I feel blessed. With Hannah Bates.
Hannah: Awh luv u. xx May 7 at 9 pm Kate: Luv u too . We're gonna have so many more great times esp. now that we've got Mike .
May 7 at 10.47 pm Hannah: Ha ha very tru two hearts let's hope we don't run into bryan again tho … May 7 at 10.58pm The private chat which develops between the two friends above elaborates on the caption of their selfie, a line from a song in fact ('Way up I feel blessed'), in ways which allude to their closeness. In particular, with the reference to 'Mike' and 'Brian', a back story of shared interactional history is referred to cryptically. This is an example of specific alignment that signals understanding of the stance on the caption of the selfie and ability to amplify it.
We can see another example of alignment through knowing participation as well as the juxtaposition of contributions from knowing vs. non-knowing friends in the comments to another significant other selfie (of Helena, the selfie-poster, with a young man). The selfie suggests a developing romance, as it is accompanied by hearts.

Example 3
Knowing friends such as Kate L. propose a positive assessment of the selfie which is based on offline/or any other insider's knowledge in relation to the selfie posted. Such knowing allusive references are the main way in which specific alignment is provided and in particular affiliation of the type that we find documented in 'with' type of relationships. Mandelbaum (1987), for instance, has shown how couples signal their withrelationship interactionally, through collaboratively undertaken actions when telling a story. These range from collaborative completions to monitoring the story for accuracy. In the data at hand, this is done in different ways, shaped by media affordances. In particular, narrative stancetaking in the original self or some other selection of knowing participants (e.g. tagging) is 'read' by recipients as an invitation and even a requirement for participation that displays knowing status. This participation separates friends in the know from other friends, however aligned the latter may be with the post or poster. This is seen in Helena's response to Kate's comment that makes a typically cryptic reference to a backstory of Kate and Helena having discussed this relationship and Helena 'having taken her advice on board'. The back-story is normally part of the events surrounding a selfie and any allusions to it arguably elaborate on the selfie as part of a narrative and thus propose how it should be understood. Alignment through knowing participation may thus narrow down the interpretative options around a selfie. Overall, the task of showing knowledge takes the following forms: commenters can expand on the narrative stancetaking of the initial post by constituting it as a story or providing more of the story; they can also refer more or less allusively to pre-posting shared activities ('the back-story').
Knowing participation emerges in the data as part of a status accruing and validating position which is compatible with the well attested phenomenon in social relations online of FOMO (fear of missing out; see Marwick 2013). This is evident in the interactional arrangements that have to do with non-knowing recipients: not knowing is often expressed as missing out and as explicitly asking for the back-story. It is also notable that such requests lead to some kind of (more) storying in relation to the selfie from the poster. We can see this in the example above with Megan's comment which seeks to find out 'what she has missed'. A narrative explanation form the selfie-poster begins in the publicly available comments ('went out for bubble tea with a mate') but the fuller story, we can speculate, is provided in the private chat area to which the friends claim they will switch. (I did not have access to private messages). While ritual appreciation validates a self-branded self by means of generic alignment in comments, hence the striking resemblance of comments, knowing participation displays a specific type of alignment, as it brings in something extra about the selfie and/or the selfie-poster which not every FB friend has access too. In both cases, however, the identities projected by the selfie and validated by the comments are overwhelmingly positive. We will discuss the implications of this below.

Conclusions: Selfies as small stories -Insights & rewards
Using selfie postings by 3 English-speaking adolescent women on FB, I showed how, within a small stories framework, far from being mono-semiotic expressions of 'ideal selves', selfies emerge as contextualized, multi-semiotic and co-constructed presentations of self, shaped by media affordances. Drawing on positioning analysis and small stories, I singled out three salient types of selfies, namely me selfies, significant other selfies and group selfies. I showed their systematic positioning configurations in terms of the characters portrayed, visual arrangements, placement (Level 1) and the audience patterns of engagement (Level 2). In particular, I showed that both the posting of selfies and the interactional engagement with them are closely linked with story-making and story-recipiency roles. I documented two systematic interactional patterns of engaging with selfies which display alignment with the activity of posting a selfie and affiliation with the 'character(s)' in the selfie: ritual appreciation and knowing participation. Ritual appreciation involves positive assessments of the selfied person(s), expressed in highly conventionalized language coupled with emojis. These semiotic choices result in congruent sequences of atomized contributions, which despite not directly engaging with one another, are strikingly similar, visually and linguistically. Doing alignment through knowing participation on the other hand creates more specific and individual alignment responses, by bringing in and displaying knowledge from offline, pre-posting activities or any other knowledge specific to the selfie or selfied person(s), as a means of validating the post. It also results more often than not in some kind of a dialogue between the commenter and either the selfie-poster or other commenters.
In both cases of alignment by ritual appreciation and alignment by knowing participation, narrative arrangements are important for explaining the patterns of engagement: a selfie is shared as a performance and/or responded to as a story. In particular, alignment through ritual appreciation is premised on the selfie projecting commenters as viewers and spectators through some display of positive presentation of self. There is a tendency for the most viewed/viewable and edited selfies, which in turn are the me selfies, normally put up as profile pictures, to create this focus of projected alignment on the 'character', the self that is put up for scrutiny. In knowing participation, on the other hand, audiences seem to respond on the basis of displaying knowledge by elaborating on and even producing a story through their comments. Overall, seflies come with media-shaped narrative potentiality, e.g. FB designed narrative stancetaking elements, which can lead to multi-semiotic, co-authored and expandable small stories through audience engagement.
I began this article by suggesting that selfies seem to have become enregistered as a social media activity, particularly in terms of who the typical selfie-takers are. Although the principles and methods of my analysis do not aim at generaliseability, the findings about the ways in which selfies are engaged with can prove a useful starting point for exploring further the conventionalization of selfie-posting and commenting activities. The relation in my data between performance and alignment through ritual appreciation suggests the need for further studies of the conventionalization of participation modes that become part of validating self-branding activities on social media. At the same time, it also points to a close link between alignment and affiliation which merits further investigation. As suggested, CA studies have shown how alignment, which is primarily a way of signaling understanding of the action performed by the previous turn or speaker contribution, more often than not leads to affiliation with the speaker's stance as well. In this case, the two seem to be connected from the outset. In other words, by aligning with a posted selfie, commenters show affiliation with the 'character(s)' of the selfie. Such character-based assessments and affiliations, as opposed to affiliation being premised on events, emotions and other story plot components support a performative, (micro)celebrity-inclined presentation of self and render the narrated self as the most 'assessable' one.
At the same time, the results of my analysis caution against narrowly associating one posting practice with one type of self-presentation. Positioning at Level 3 in the three types of selfies, and the modes of engagement with them, can be posed as a cline from microcelebrity selves, mostly applicable to me selfies, to selves that are presented as more casual, fun-loving and seizing the moment with friends. We can talk about a continuum of more individualistic to more relational selfies and selves. This is the result of semiotic choices and placement at Level 1, which also show a continuum of possibilities, from the edited self of profile me selfies to the ordinary self of group selfies, from the most to the least viewable and commented upon selfies. The modes of engagement with selfies, as we saw, also range from appreciation and validation of 'me' to knowing participation that brings in the back-story that only few people know about 'me'. We can thus revisit Walker Rettberg's claim (2013) that selfies, along with other technologies of self, allow us to see ourselves as an object and through this to know ourselves and add the irreducibly intersubjective element of this self-recognition: from the moment we post selfies and make them part of social and communicative practices and contexts, we see and know ourselves through others' engagement with them. We tell our stories with and for others.
Alignment is at the heart of such intersubjective understandings of selfies and exploring further forms and practices of alignment-based participation on social media, by employing micro-analytical modes, as above, can safeguard against normative conceptualizations of sharing practices. To be specific, the distinct alignment responses in different types of selfies bring to the fore their interactional and contextual aspects thus calling into question any descriptions of selfies in the media as narcissistic self-presentations.¹⁰ I started by suggesting that small stories modes and methods of analysis are well-suited to capturing the narrativity of selfies. It is worth considering at this point what rewards are to be gained for small stories research from the analysis of selfies. In my view, these are mostly to be found in the ways in which selfies navigate competing requirements which themselves emanate from competing narratives of relationality in the designing of social media platforms. On FB in particular, the directional drive to share the moment as soon as it has happened is in a trade off with constructing edited and polished selves. Marwick (2013) similarly suggests that there is tension between self-branding and authenticity, performing and rehearsing self at the same time as coming across as genuine and spontaneous. I would add to this the double bind and paradox between em-plotting the ongoing present and sharing mundane incidents of everyday life at the same time as making it look spectacular it at the same time, making the familiar look unique. Uncovering the nuanced ways in which social media users discursively and interactionally manage these tensions is worthwhile and a small stories perspective on selfies could be at the heart of this inquiry.