Iterative Micro-Identity Content Analysis: Studying Identity Development within and across Real-Time Interactions

ABSTRACT

Identity development occurs in the context of real-time interactions. However, existing research on interactions has focused on identity processes and little is known about identity content development within interactions. We define real-time identity as claims about selves, formulated in the service of an interactional “social business.” The aim of this methodological paper is to introduce Iterative Micro-Content Analysis (IMICA) as an approach to studying the changes and consistencies in real-time identity content. We outline four key principles of IMICA and offer a step by step guide to its analytic stages. We provide two worked examples for illustration: a video-recorded conversation between two young women on the topic of “love and desire,” and audio-recorded speed-dating conversations between young same-sex attracted men. The worked examples demonstrate how IMICA can be used to study how identity claims change within a single interaction as well as across multiple interactions. We argue that IMICA’s empirical insights into the concrete mechanisms through which social interactions shape identities are of both theoretical and practical relevance. We discuss how IMICA may allow for a micro-level operationalization of macro-level concepts (e.g., exploration or identity centrality), outline how it may be combined with quantitative analyses, and discuss its limitations.

KEYWORDS:

• Identity development
• real-time processes
• social context
• identity content

The development of identity is the central psychosocial task of adolescence and continues into emerging adulthood (Arnett, 2015). It is in the context of everyday interactions with social others that individuals both make identity-constitutive experiences as well as integrate these experiences into a coherent sense of self (Adams & Marshall, 1996; Postmes et al., 2006). This process involves concrete actions and behaviors unfolding in real-time (Lichtwarck-Aschoff et al., 2008; Raeff, 2014; Steinberg, 1995). Importantly, within interactions others are not static developmental contexts, but are actively contributing to an individual’s identity development (Schachter & Ventura, 2008). It is through these repeated constructions and negotiations of identities within the here-and-now (i.e. micro-level) that more stable identities emerge across developmental time (i.e. macro-level; Schachter, 2015; Thorne & Shapiro, 2011). Thus, everyday interactions both provide the “content” of identities (e.g., Galliher et al., 2017), and are at the same time a pivotal site and process for the integration of these experiences (e.g., Adams & Marshall, 1996; Postmes et al., 2006). The aim of this methodological paper is to introduce Iterative Micro-Identity Content Analysis (IMICA) as an approach to the study of identity content and its change in the context of real-time interactions.

Approaches to identities in interactions

Ours is, of course, not the first approach to studying identity within the context of everyday interactions. We draw on a rich literature of social constructivist studies from discursive psychology (Edwards & Potter, 1992, 1992), positioning analysis (e.g., Harré & Langenhove, 1991; Korobov, 2015; Wilkinson & Kitzinger, 2003), and membership categorization analysis (Deppermann, 2013; Stokoe, 2009, 2012). Common to these approaches is the assumption that talk is not referential, but that it instead constitutes or performs action (e.g., Edwards & Potter, 1992). This focus might be understood as a direct antithesis to reflective or cognitive conceptualizations of identity that assume that identities become expressed or manifested as actions in the context of interactions (see e.g., Kunnen & Metz, 2015). Instead of considering identity as a primary determinant of action, social constructivist approaches consider identity itself as “a product of social interaction” (Breakwell, 1986, p. 43). While cognitive approaches to identity understand identities as something that individuals have, social constructivist approaches consider identity as something that people do.

Idiosyncratic identity content

Commonly, research on identities in talk has focused on the mechanisms through which identities are used for and established by social action. However, due to their focus on mechanisms these approaches have been chiefly concerned with identifying how identities are related to action in general and have been less concerned with studying identity development in specific speakers. In other words, micro-genetic studies of talk in interactions have been process-centered as opposed to person-centered. As a result of this focus relatively little is known about how the content of an individual’s identity changes through their social interactions. Identity content refers to the idiosyncratic issues, concerns, or topics that individuals understand as meaningfully related to who they are (McLean et al., 2016). We argue that an assessment of the idiosyncratic content of identities at the micro-level, as well as the study of how content may change within or across multiple real-time interactions, requires a person-centered approach (e.g., Magnusson, 2001). Insights into the development of micro-level identity content are not only theoretically relevant but will also allow practitioners to facilitate the development of adaptive identities in contexts such as psychotherapy.

Identities as claims about selves

In order to be able to address the idiosyncratic contents of speakers’ selves in the here and now, we follow Schachter’s (2015) suggestion that “identity is not who a person is but a claim about who a person is” (p. 3). Claims about selves provide answers to the questions “who am I” and “who are you” in terms of local attributions (Bamberg, 2011). Notably, identity claims may be explicit or implicit (Schachter, 2015). Explicitly, speakers can name a culturally relevant identity category to describe themselves or someone else (e.g., “feminist,” “Muslim,” or “gay man”), but may also formulate implicit claims in terms of attributes (e.g., Korobov, 2015; Stokoe, 2012; Wilkinson & Kitzinger, 2003). In addition, speakers may index identities by describing themselves in terms of attributes that are associated with these categories (Wilkinson & Kitzinger, 2003). These more implicit claims construct speakers in terms of actions, values and meanings that are socially recognizable as being associated with specific kinds of persons (Anderson, 2009). Importantly, the meaning of attributes is interactionally accomplished and locally negotiated, for example, through describing and evaluating what is normative in a specific context (see e.g., Deppermann, 2013).

The four tenets of IMICA

Iterative micro-identity content analysis builds on four core principles or tenets. First, as outlined above, rather than to frame the influence of social interactions on identity content in terms of static variables or factors we understand interactions as sites of negotiation in which conversation partners engage in concrete actions and behaviors (Raeff, 2014; Steinberg, 1995). As a consequence, IMICA is concerned with the microgenetic study of interactions as sites of social practices. The second tenet of IMICA is that a study of the development of idiosyncratic identity content requires a person-centered approach (Magnusson, 2001). Within the person-centered approach a person is understood as a functioning and integrated whole, and individual development is considered as a process of adaptation to a given environment. Analytically, the person-centered approach focuses on the identification of patterns with the individual as the unit of analysis.

The third tenet of IMICA is that identity claims are constructed in the course of and as tools for everyday social business (e.g., Korobov, 2015; Stokoe, 2012). As Schachter (2015) has highlighted, identity claims have pragmatic implications within a given context, and the validation of an identity claim sets “the stage for (possible) action involving the self” (p. 3). Thus, claims about selves are constructed with the aim of being recognized, “bringing off” these identities in the process (Schachter, 2015). Within one interaction a speaker may be engaged in multiple different types of social business and often these goals may be competing (Korobov, 2015). Next to “simple” social actions (e.g., such as purchasing and selling, complaining, asking for directions, and so on), speakers may also engage in relational business such as creating affiliation with their conversation partner (e.g., Korobov, 2011), or eliciting emotional support (Kerrick & Thorne, 2014). Thus, IMICA assumes that identities are constructed within specific situations, by particular speakers, for particular audiences, in the service of particular goals (e.g., McLean et al., 2007). Notably, the affirmation of a speaker’s identity by an interaction partner is also dependent on the intentions and goals of the interaction partner (Schachter, 2015).

The fourth tenet of IMICA is that a person-centered study of the development of identity content in interactions requires the consideration of the iterative nature of identity construction. Drawing on dynamic systems theory, we hold that changes in an individual’s idiosyncratic identity content have “to be understood by studying the time evolution or iterative process of the system” (Lichtwarck-Aschoff et al., 2008, p. 376; Raeff, 2014). The construction of identity claims is an iterative process in that speakers’ current utterances are building on both their own as well as their conversation partner’s previous utterances (Harré & Langenhove, 1991). Once successfully established, an identity-claim may be repeatedly invoked in future interactions resulting in its solidification, or its change should it not have been affirmed (Schachter, 2015; Thorne & Shapiro, 2011). Consequently, an analysis concerned with the iterative development of identity claims should not only consider single interactions, but study the evolution of self-claims over the course of multiple interactions.

Analytic steps

Based on the four tenets outlined above, we suggest that an analysis of real-time identity content can be broken down into different steps moving from an initial identification of claims about selves to a consideration of their effects and functions within the local context. In the following we will outline the steps in the order in which we perform them in principle, though noting that in practice an analysis often oscillates between steps. In addition, we will provide two worked examples to demonstrate how an analysis may progress both within one conversation (Worked Example 1), as well as across multiple conversations (Worked Example 2).

Prior to analysis, researchers should formulate clear research questions. Due to its consideration of the time-evolutive nature of talk, IMICA lends itself particularly well to research questions concerned with issues of stability and change in the content of identity claims (e.g., “Do self-related claims become less variable over time?“). Moreover, IMICA’s focus on individuals allows for a study of the content of speakers’ identities across different types of interaction partners (e.g., “How do maternal and peer audiences influence the constructions of educational identities?”). In addition, IMICA is also suitable for research questions concerned with the processes or mechanisms through which identity content becomes recurrent or changes over time.

Iterative micro-identity content analysis can be applied to audio and video-recordings of both prompted as well as unprompted interactions. Moreover, different types of conversational contexts are suitable for analysis, including those with equal role distribution among speakers (e.g., dating conversations or peer talk) as well as those in which the context may impose role differences on speakers (e.g., therapy sessions or class-room interactions). When considering dyads with preexisting relationships analysis may benefit from the use of play-back interviews to provide the analyst with information about the relational history of specific claims or references (see e.g., Kerrick & Thorne, 2014). While IMICA can be applied to the study of group interactions, we will discuss its application to dyadic interactions. When considering data from multiple interactions of the same dyad, we prefer to limit contact between participants outside of recorded interactions for the duration of data collection. However, this is not always possible or desired and note should be made of the frequency and type of contact outside of recordings. When data come from multiple interactions with different conversation partners, consistent conversation prompts may benefit subsequent analysis.

Once the data has been collected transcripts of the interactions need to be created. Suitable transcription notation methods should meet two requirements. First, transcription should at least be verbatim (i.e. word for word) and include all utterances by all speakers. Second, due to the time-serial nature of talk, transcripts should at least include the sequential order of individual utterances, but should ideally include time-stamps for each utterance to provide a better overview of interruptions and overlapping talk from all interaction partners. For example, while the conversation analytic transcription notation method developed by Jefferson (2004) notes intonations and pauses, it does not include information on the sequential ordering or onset of utterances. For our own research we have developed a transcription notation method that includes both notes on overlap between individual speakers as well as time-serial information (see the transcription notation method below).

Transcription conventions

Table

-	shorter than expected silence/no silence between words
,	micropause
(.),(.),(…)	pauses of less than .5s, 1s, 1.5s
(2.)	measured pause in seconds
wo(h)rd	laugh particles within words
?	pitch rising to high at end of phrase
wo:rd	segmental lengthening, according to duration
[word]	uncertain transcription
[0:00:00]	time-stamp noting the onset of an utterance
word <	interrupted talk, immediate floor change
word< word	interrupted talk, no floor change
word<word>	interrupted talk, floor change with overlapping talk
<word	interrupting talk, floor change
<word>	interrupting talk, no floor change
[…]	excerpt was shortened

Step one: familiarization with the data

Once all interactions have been transcribed, researchers should familiarize themselves with the data through repeated reading. Special notice should be made of explicit references to identity categories or self-descriptions (“I am a gay”), as well as of explicit evaluations and comparisons (“Wow, we’re so different”), and the repetition of phrases both within and across interactions. While the analysis is focused on individual speakers, at this step it is useful to compare the type and frequency of explicit constructions across different interactions from the full sample to gain an understanding of how typical characteristics within one interaction are of all interactions within the study.

Step two: identification of claims about selves

After all transcripts have been read, all subsequent steps are carried out at the level of the dyad. First transcripts are segmented into codeable units. Codeable units describe the level at which claims will be coded. In our work we code claims at a turn level, defining turns as changes in conversational floor. After codeable units have been determined, researchers should identify identity-related claims. In our coding we are considering two different types of claims. First, claims about selves can be made in terms of category references (Stokoe, 2012). Importantly, category references do not need to refer to socially recognizable categories (e.g., “woman,” “man,” “cheater,” etc.), but may instead refer to categories that are interactionally constructed within the context of the specific interaction. For example, Deppermann (2013) has documented how a group of young male friends locally constructed the identity category “alcohol sloven.” Second, claims about selves can also be formulated in terms of speaker’s attributes and general tendencies (Wilkinson & Kitzinger, 2003). In coding general tendencies and attributes the analysis should consider whether a claim is formulated or constructed as extending beyond the specific context of its utterance. For example, in the fictitious exchange below Adam’s utterance should not be considered as extending beyond the current context, while the attribute in Bob’s claim would be considered persistent. One indicator for whether a claim contains a general tendency is the presence of extreme case formulations, which invoke maximal or minimal properties of a characteristic (e.g., “always,” “never,” “completely”; Whitehead, 2015).

Adam:

I like your shirt.

Bob:

Thank you. I never used to like wearing shirts, but now I love it.

Notably, claims about the self can be formulated to be about the speaker, the interaction partner, the dyad as a unit, but also about a third party. It should be noted that making a claim about someone else may also have reflexive identity implications for the speaker (Korobov, 2015). Prior to analysis the researcher should thus consider which targets will be included in an analysis. In our own work we consider speakers’ direct self-related claims, including those made about the dyad, but do not consider the reflexive implications of claims made about others. In addition, claims made about a speaker by their conversation partner are only considered if the speaker orientated toward them in their own claim. In the example below, our analysis would consider Bob’s response but not Adam’s question.

Adam:

You’ve never been on a blind date, right?

Bob:

No, never.

After all relevant claims have been identified, they should be extracted from the transcript preserving information on the dyad, speaker, and the temporal location of the claim within the conversation. In the extraction of the claim researchers should consider whether to extract multiple claims per codeable unit, or whether to extract one integrated claim per unit. In the example below, Adam’s claim could be segmented into two separate claims (i.e. “I love basketball” and “I never go to any matches anymore”) or understood as one integrated claim. Importantly, for the extraction of some claims additional information about the context of its utterance should be provided. For example, an extract from Bob’s claim below would include a reference to Adam’s previous claim (“Yeah, me [I never go to any basketball matches anymore] neither.”).

Adam:

I love basketball. Unfortunately, I never go to any matches anymore.

Bob:

Yeah, me neither.

Step three: themes and domains

After claims about selves have been extracted from the transcripts, all claims of the same dyad are sorted according to their topic or domain. Coding in this step could either be inductive or deductive. Inductive coding may employ a form of thematic analysis (see e.g., Braun & Clarke, 2006). While deductive coding could make use of existing taxonomies of identity content domains (McLean et al., 2016) to code claims in terms of ideological domains (eg. education or values) or interpersonal domains (e.g., family or friends). The choice of either inductive or deductive coding may in part be determined by the conversational context. Deductive approaches will allow for the coding of a larger set of conversations, and are useful in interactions in which speakers construct identity-related claims across a variety of topics and issues. In contrast, an inductive approach aimed at assessing the idiosyncratic content of speakers’ identity claims might be better suited for the study of interactions in which speakers construct nuanced claims across few topics or domains. To be coded as present a topic should be a central, unexchangeable component of the claim, rather than its context. For example, the claim “I’ve been with my partner since high school” is concerned with issues around romance and dating, not education. This is evidenced by the fact that “since high-school” could be exchanged for a formulation such as “when I was a teenager” without changing the meaning of the claim. Individual claims may be about multiple themes or topics, or construct a self across multiple domains, for analysis it is useful to determine a “central” theme or main topic for each claim.

Step four: content and formulations of claims

Subsequently, the contents and formulations of all claims within a theme or domain are analyzed at the dyad-level. An analysis of content considers whether information provided within a claim is consistent with information contained in previous claims within the same domain or topic cluster. To this end it is useful to note for each claim whether it restates, adds to, or contrasts previous content, or whether it introduces new content within a topic or domain. For domains that contain many different identity claims an inductive grouping of claims into clusters with similar content may benefit analysis (see e.g., Braun & Clarke, 2006). In addition, claims within the same topic should be considered in terms of their formulation (see Deppermann, 2011). For example, some claims may repeat key words or phrases, or may even recycle previous claims verbatim. Similarly, some words or phrases may be introduced by one speaker and then later taken on by their interaction partner. While the focus in this step is on individual speakers, analysis oscillates between all present speakers. An analysis of content should adhere to the sequential order of claims and work retrospectively, considering how every additional claim relates to all previous claims within the same topic.

Step five: effect and function of claims

For an analysis of effect and function claims need to be considered within the context of their utterance (i.e. the transcript). Initially, analysis should identify which discursive action a claim is accomplishing within the interaction. For example, a claim may provide an account to a previous challenge or may itself challenge the previous utterance of a conversation partner (Potter, 1996). Subsequently, analysis should attend to the response of the conversation partner to the claim in question. While researchers can draw on specific concepts from conversation analysis in this step, we suggest that all analyses should at least consider a partner’s evaluations, challenges, or affirmation of the speaker’s claim (Schachter, 2015). If a claim referenced an earlier claim in the interaction, both moments should be considered jointly.

Subsequently, researchers can explore the social business that claims may discursively accomplish (e.g., Locke & Budds, 2020). This exploration can employ either an inductive or deductive analysis. An inductive analysis of function departs from a consideration of (in)consistencies in the claims of speakers. For example, a series of consistent claims could be coded as functioning to construct the speaker as positively distinct, if a conversation partner’s responses repeatedly expressed surprise. In contrast, a deductive strategy would focus on a specific social business as a lens to understand the unfolding interaction (e.g., “How do speakers construct self-related claims in the service of positive distinctiveness?”). Here analysts could focus on single functions and code whether a claim does or does not serve them. Both a deductive as well as an inductive analysis should consider multiple social businesses simultaneously to reach a more nuanced understanding of the interaction.

Evaluating quality

In order to assure the quality of analysis, we suggest multiple strategies. First, while IMICA is concerned with the identification of patterns and regularities within the coding, researchers should pay special attention to the identification of deviant cases (Peräkylä, 2011). Deviant cases are instances in the data that do not conform to previously established patterns and can allow for the further specification of mechanisms and their boundaries. In addition, for the validation of an analysis of´function (Step 5) we suggest the application of the next turn proof procedure (e.g., Sacks et al., 1974). In this procedure evidence for an interpretation of a claim’s function is derived from the way that it is taken up in subsequent turns. To further increase the confirmability of their analysis researchers should openly discuss emerging findings with members of their research team to reach consensus (Noble & Smith, 2015). Together, these strategies will help to convince a reader of the transparent adequacy of the findings, strengthening their apparent validity (Kirk & Miller, 1986). Finally, researchers should also reflexively consider how their own identities may have influenced data collection and analysis (Kuper et al., 2008), for example, through the keeping of a research journal and frequent peer debriefing (Noble & Smith, 2015).

Some procedures within IMICA can also be subjected to a quantitative assessment of reliability (see also Peräkylä, 2011). Specifically, the coding of content domains (Step 3) and the analysis of effect and function (Step 5) can also follow a deductive approach, which would be suited for the calculation of inter coder reliability (O’Connor & Joffe, 2020). For example, a subset of identified claims could be coded for domain content by two independent raters and their agreement assessed. Similarly, in a research project investigating the way that claims about selves may function to accomplish a sense of positive distinctiveness independent raters could assess the presence or absence of a function for a subset of claims.

Worked Examples

Worked example 1: change in claims within a single interaction

Data for the first example came from a conversation between two young women who are referred to here as Caro and Jenny both aged 19. The participants took part in a study in which they were asked to freely discuss the topic “Love, romance, sex and desire” in three different session of each 20 minutes; in the following the first of these conversations is analyzed. After being initially greeted by the first author, the participants were brought to a room, where two chairs were positioned at a right angle to each other and a video-camera was visibly positioned for recording. The participants were asked to engage in an “ice-breaker” conversation for 5 minutes, after which they were asked to provide consent for the video-recordings. The participants were instructed to freely talk about the topic in terms of their own experiences, thoughts and questions. The study received approval from the Ethical Review Board of the first author’s university. For illustration we selected all claims by Caro that were inductively coded with the theme “stance on intimate relationships”; note that time-stamps in Tables 1 and 2 were removed for presentation.

Table 1. First exchange between Jenny and Caro

Jenny: yeah (3.) What was your longest relationshi(h)p (chuckles)?

Caro: I don’t do relationsh(h)ips (laughs)

Jenny: (laughs) O(h)kay(h) (shrugs) Tha(h)t’s interesting – why not?

Caro: I’m not really interested in relationships and romance and all that. It’s not my <big thing>

Jenny: <I don’t like romance, but I like relationships (.) I only had one (chuckles)

Caro: That’s enough

Jenny: Yeah

Caro: I mean. I don’t – I don’t really need a boyfriend (2.) I–I feel like I don’t

Jenny: Yeah, if you are happy with that – that’s pretty<

Caro: <[I am]>

Jenny: >cool. Cuz I would like be totally lost (chuckles)

Caro: Really?

Jenny: Yeah (nodds)

Caro: But does he live here?

Table 2. Second exchange between Jenny and Caro

Jenny: But do you often like have guys over, or?

Caro: No:, I don’t?

Jenny: < No? >

Caro: < I had like (.) in the time I was living here I had two guys <

Jenny: < okay >

Caro: > over.

Jenny: And in total?

Caro: In total?

Jenny: Yeah, like how many?

Caro: (.) U::hm (.) Three? in total?<

Jenny: <hm>

Caro: >But I had longer thi::ngs::

Jenny: Hm. Okay.

Caro: So friends with benefits, I don’t know

Jenny: (nodds)a h Okay.

Caro: (nodds) Yeah (3.) Yeah (.) one was a (.) kind of an one-night-stand. bu::t we are still in contact and we met and and stuff. (1.)

Jenny: (.) Is he nice?

Caro: (nodds, smiles) Yeah. And he is very good looki(h)ng (chuckles)

Jenny:(chuckles) And you don’t want a relationship?

Caro: I do, but he is not from here

Content and formulation

In our analysis of content (Step 4) we identified five claims constructed by Caro regarding the topic “stance on relationship.” The first claim occurred within an exchange taking place within the first minute of the conversation (Table 1). This claim constructed Caro as the kind of person who generally does not engage in relationships (“I don’t do relationships”). Subsequently, different attributes become linked to this initial kind in three different claims. These attributes are being “[interested in] romance and all that,” “not needing a boyfriend” and “being happy not needing a boyfriend.” The fifth claim occurred around 14 minutes into the interaction (see Table 2) and referenced Jenny’s question regarding Caro’s frequency of casual sexual contact in which Caro constructed herself as the kind of person who does not frequently engage in casual sexual contact (“No, I don’t [often have guys over]”). While the subsequent turns contained information about Caro’s stance on relationships they did not construct a general tendency and were thus not coded as identity-related claims. Notably, the content of the final claim in this series contrasted the initial claim and constructed Caro as a person who does “do” relationships (“I do [want a relationship], but he is not from here,” Table 2), highlighting a change in Caro’s claim of self.

Effect and function

To contextualize Caro’s identity claims we considered their effects within the interaction. First, we noted that Caro’s initial self-related claim occurred as a response to a question by Jenny that implicitly constructed the assumption of having had relationships as normative (Wilkinson & Kitzinger, 2003). Caro’s response (“I don’t do relationships”) challenged the identity ascriptions contained in Jenny’s question. Jenny’s subsequent evaluation of the claim as “interesting” and her request for an explanation further highlighted the normative assumption that was resisted by Caro’s response. The defensive minimization of expertise in Jenny’s self-related claim (“I only had one”) suggested that Caro’s second claim was taken up as a challenge. Notably, in her response Jenny challenged the conflation of “relationships” and “romance,” resisting the attribute of “liking romance” that was marked as dispreferred by Caro. The turn-initial discourse marker “I mean” in Caro’s third claim functioned to side-step Jenny’s challenge (Fox Tree & Schrock, 2002) and allowed her to invoke an additional negative attribute of the kind of person she was resisting (“Needing a boyfriend”). Arguably, Jenny’s opening question positioned Caro as aligned with some unnamed normative kind characterized by compulsory romance (Hammack et al., 2019). Caro’s subsequent four identity claims functioned to demonstrate how she did not possess the attributes that she ascribed to this kind. In doing so, Caro distanced herself from the normative kind, establishing an identity style that marked her as unique (Korobov, 2011).

To understand the change in Caro’s final claim we turned to the turns preceding it. Caro’s fifth claim was a response to a question by Jenny which offered “frequent casual sex” as a new attribute to the kind of person who does not “do” romance. We noted the careful design of Jenny’s question that was illustrated by her use of the turn-final discourse marker “or” which invited further elaboration (Drake, 2015). Arguably, Jenny constructed “frequent casual sex” as an attribute that was implied by Caro’s earlier constructions of self, as was suggested by her subsequent clarification probes (e.g., “No?” and “And in total?”). Together, this suggested that Jenny’s question functioned to establish distinctiveness between both speakers by constructing Caro as “sexually experienced,” while constructing herself as relatively less experienced (“I only had one”). In her account Caro hearably orientated toward the challenges contained in Jenny’s probes establishing herself as experienced (“But I had longer things”), while rejecting the attribute ascribed to her (“So friends with benefits”). The characteristics and behaviors Caro described in her account of “an one-night-stand” with a “very good looking” guy prompted an additional question by Jenny (“And you don’t want a relationship?”) that was formulated as a challenge to Caro’s earlier self-description (“I don’t do relationships,” Table 1).

The analysis highlighted that Caro’s initial claim functioned to differentiate her from Jenny. This initial construction of herself as an “unconventional” kind of person implied attributes (e.g., frequent casual sex) that Caro either was not able to or chose not to align with. After challenges to her alignment with this non-normative kind, Caro adjusted her claim to align with Jenny. The adjustment occurred after Caro had successfully differentiated from Jenny in terms of their relative sexual expertise. Thus, while the content of Caro’s identity claim changed throughout the interaction as a response to Jenny’s challenges, the initial differentiation was maintained.

Worked example 2: change in claims across interactions

While the first example focused on changes in claims about the self within one interaction, the second example assessed changes in the content of one speaker’s identity claims over the course of multiple interactions. Data come from audio-recordings of one speed-dating event including eight young men who were interested in dating other men. Each of the eight participants met all other participants in a round-robin design in a total of seven rounds. Each individual round lasted approximately six minutes. At the end of each round participants anonymously indicated whether they were interested in meeting their partner again, only if both partners agreed both received the other’s contact information. The study received approval from the Ethical Review Board of the first author’s university. The example followed one target speaker (John, age 20) whose claims across seven different conversations with different partners (Mean Age = 25.65, sd = 3.46) were at the center of the analysis. For illustration we selected all claims that were coded with the theme “Where are you from?” as the remarkable consistency in John’s claims allowed us to demonstrate an analysis of identity claims over multiple interactions.

Content and formulation

Table 3 shows excerpts from all interactions, in the following we will focus on John’s initial claims in each of the individual interactions. There were two main-components of John’s self-description that initially emerged in the first conversation. The phrase “This is the tenth country I’ve lived in now” (1.2) first emerged as a response to the question “where are you from,” while the phrase “I moved around a lot growing up” (1.1) was introduced in response to a question about the process through which John arrived in the University town. While initially presented in separate claims, the two phrases became integrated into one claim in the second conversation (see Table 4). Over the course of the next five interactions the claims became more specified at first (e.g., “everywhere” (2.1) compared to “ten different countries at this point”) and then less specified through the addition of modulators such as “lot’s” (5.1) or “kinda” (6.1).

Table 3. Excerpts from all of John’s 7 speed dating conversations

Round		Question	Response
1	1	Partner 1: So, how did you get here?	John: Uhm:- Hmm. What do you mean? Like- Partner 1: Just- Just in [university town]. John: […] Uhm:: (.) Alright, so I moved around a lot growing up.
1	2	Partner 1: And where originally, did you come from?	John: Uh(hh)m, I::’ve- This is the tenth country I’ve lived in now. This stage of my life. […] Partner 1: Da:::mn John: Yeah. I moved around a lot Partner 1: Oh, wow! We’re so different. < John: <(chuckles)> Partner 1: > I just- I was just born here, and I just stayed here. John: Yeah, I couldn’t- I:: couldn’t imagine that. It seems (.) Like, was it nice? Do you enjoy living here? […]
1	3	Partner 1: So, where were you born?	John: Uh, I was born in South Africa. And <
1	4	Partner 1: Ok, quickly! List all the countries! I’m very curious.	John: (chuckles) Uh, so South Africa, then we moved to Ireland for a little bit. Then we moved to Honduras, then to Wales, then Scotland. Then Uganda, then Malawi. Then back to Ireland. Then London, then Switzerland (hh) now here (chuckles).
2	1	Partner 2: Where [are] you from?	John: Uh (hh). I moved around a lot when I was growin’ up, so I, like, traveled everywhere. Partner 2: Okay. John: (Ind.) Uhm, so I was born in South Africa < John: < Okay. > John: > and then I was there for a little while, then I moved to Ireland. And then from Ireland we moved to Honduras (English pronunciation) < […] John: > and then, uhm, we were there for a few years and we would move to Wales. And then Scotland, then Uganda, then Malawi (chuckles) < Partner 2: < Oh my god. > John: > Yeah. And I moved back to Ireland, then I moved to London, then I moved back to Ireland again. Then I moved to Switzerland < Partner 2: < Okay, that’s a handful … (chuckles) > John: > Yeah. (chuckles) Seriously. And then I moved he(hh)re. And<
2	2	Partner 2: < O::h. Where did you spend most of your life?	John: U:h, most [of] my life was < Partner 2: < Most years, or months, or I don’t know > John: > between Malawi and Uganda. Spent 4 years each there.
2	3	Partner 2: But, do- (.) Is it true that you do not, like, have a place that you call home? Like, you have a difficult identity? You know what I mean?	John: No, really! It didn’t really affect me as much as it affected my sister < Partner 2: < Okay. > John: > and, uhm, no. Like, I just kinda- I see home as Ireland. < Partner 2: < Okay! > John: Like, slash where my granny lives, ‘cus most summers we’d spend a lot of time there and my (Ind.) and I got a good group of friends, so (.) yeah, I just kinda see Ireland as my home and I love being there <
3	1	Partner 3: How about you?	John: I’m- (chuckles) I’m from- I moved around a lot when I was growing up, so I’ve lived in, like, ten countries. Partner 3: Cool. John: Yeah.
3	2	Partner 3: Which one would you say- So, if someone asked you “What nationality are you”?	John: Uhm, hmm. It depends. If I’m outside of Ireland I’m Irish. < Partner 3: < Yeah. > John: > But if I’m outside of Ireland, I kinda- Partner 3:You’re like “I’m an expat” John: Yeah. (chuckles) Partner 3:Cool! John: Yeah. (Ind.)
3	3	Partner 3: So where have you lived?	John: Uhm, so I lived- I was born in South Africa, then I moved to Ireland for a little bit. < Partner 3:: < Cool. > John: > Then I moved to Honduras and then Wales, Scotland, Uganda, Malawi. Then Ireland again, then London, then back to I(hh)reland < Partner 3:< (chuckles) > John: > then Switzerland, now here Partner 3: Cool. John: Yeah.
4	1	Partner 4: Yeah. Where are you from?	John: Uhm, uh:, I moved around a lot when I was growing up. So I’ve- I lived in ten different countries, now at this point. Partner 4: Ok. John: Yeah.
4	2	Partner 4: And- But do you feel represent with something or.?	John: Uhm, yeah, like, do I- Yeah, in Ire- like, Ireland would be the one I most, like, identify with and < Partner 4: < Mhm. > John: > just the place I enjoy being most. (.)
5	1	Partner 5: Nice to meet you. (chuckles) (.) Where you from?	John: I moved around a lot when I was growing up. So I lived < Partner 5: <Ah:::> John: > in lot’s of different countries.
5	2	Partner 5: So you don’t have, like, a- What- Where were you born?	John: Uh, in South Africa. Partner 5: In South Africa? Ah::: John: Yeah. Partner 5: Interesting.
5	3	Partner 5: < And which parts of- I think you were born in Af[rica]- in South Africa, but then? Where did you go?	John: Uh, then I moved to Ireland for a little while. Then < Partner 5: <Yeah.> John: > I moved to Honduras. Then, from Honduras we moved to Whales, then Scotland, then Uganda, Malawi,< Partner 5: <Ah:::> John: > then back to Ireland, then England and London < Partner 5: <Oh my god.> John: > then Ireland, then Switzerland, < Partner 5: < You have been < John: < then here.> Partner 5: > all around the world. (chuckles) John: Yeah. Just every(hh)where.
6	1	Partner 6: So. (.) Where do you come from?	John: Uhm- (hh) Uh, I moved around a lot when I was growing up, so I kinda- I lived in a lot of places. Partner 6: Ah::, o
7	1	Partner 7: Where are you from?	John: Uhh, from Ireland. Partner 7: Ok. John: Yeah. Partner 7: Interesting

Legend: Ireland is highlighted bold

Table 4. Initial claims by John across the different rounds

Round	Claim
2	I moved around a lot when I was growin’ up, so I, like, traveled everywhere.
3	I’m- (chuckles) I’m from- I moved around a lot when I was growing up, so I’ve lived in, like, ten countries.
4	Uhm, uh:, I moved around a lot when I was growing up. So I’ve- I lived in ten different countries, now at this point.
5	I moved around a lot when I was growing up. So I lived in lot’s of different countries.
6	Uh, I moved around a lot when I was growing up, so I kinda- I lived in a lot of places
7	Uhh, from Ireland.

Notably, in his response to the question “where are you from” in the final interaction John abandoned the previous core-phrases and changed his response to “[I’m] from Ireland.” Prior to this interaction John mentioned Ireland in seven different segments across five different conversations as a response to his partners’ follow-up questions. Ireland was introduced as part of a description of John’s history of “moving around a lot” (1.4; 2.1; 3.3; 5.3). While John minimized the duration of his first stay in Ireland in these narratives using the phrases “a little bit” (1.4; 3.3) and “a little while” (5.3), other references to Ireland constructed it as a more central place. In the remaining segments Ireland was brought up as a response to follow-up questions about his home (“I just kinda see Ireland as my home and I love being there,” 2.3), nationality (3.2), and identification (“Ireland would be the one I most, like, identify with,” 4.2).

Effect and function

In our analysis of effect we first noted the striking consistency both in the formulation of the initiating question by conversation partners as well John’s initial response, which foregrounded a complex history of movement in the first six interactions. This response violated the single-location response preference of the question “Where are you from.” Response preferences are the socially expected formulations and responses to questions (Clayman & Loeb, 2018). This non-adherence to response preferences was further illustrated by the fact that the majority of partners requested single location answers in follow up prompts (Table 3). In addition, the formulation of John’s multi-location response was followed by partner evaluations that were both implicit (e.g., “Damn::,” 1.2; “Cool,” 3.1; “Interesting,” 5.2), as well as explicit (e.g., “We’re so different,” 1.2).

Arguably, the construction of himself as someone whose origins may not easily be pin-pointed, functioned to positively contrast John from his interaction partners (e.g., “You’ve been all around the world,” 5.3). It is important to note that rather than to represent an internalized moment of identity development, the change in John’s initial claim may suggest the existence of a repertoire of identity claims fulfilling different functions. While the initial claim allowed for differentiation, the change in the initial claim to a single location response adhered to the response preference of the question and afforded fewer follow-up questions. Thus, this example illustrated how an analysis of identity claims over multiple interactions may indicate the influence of conversational conventions on identity exploration on issues such as the meaning of “home.”

Discussion

There is theoretical agreement that identity development is a socially embedded process and that every day interactions are both the site of as well as the mechanism for this process (Adams & Marshall, 1996; Postmes et al., 2006). While a rich field of research has studied the mechanisms through which identities are used for and established through social action (e.g., Deppermann, 2013; Kerrick & Thorne, 2014; Korobov, 2015; Stokoe, 2012), little is known about the real-time processes through which identity content develops within and across everyday interactions (e.g., Kerrick & Thorne, 2014; Schachter, 2015; Thorne & Shapiro, 2011). However, empirical insights into the concrete mechanisms through which social interactions shape identities are not only important for a theoretical understanding of their development but are also useful for clinical practice. In this paper we introduced IMICA as an approach to the study of identity content and its change in the context of everyday interactions.

Research on identity development is centrally interested in how individuals maintain a sense of self-sameness and consistency in the face of developmental changes across the life-span (Hammack, 2015). To this end IMICA offers a micro-analytic approach to studying the mechanisms through which social contexts influence identity development. This fine-grained analysis is necessary as “any contextual influence must be mediated through some sort of interpersonal process” (Steinberg, 1995, p. 252). Concretely, IMICA allows for the investigation of processes that are commonly only operationalized at the level of developmental time (i.e. months, years or decades). For example, analyses of the recurrence and variability in identity claims can translate concepts such as identity centrality to the context of micro-level interactions (e.g., McLean et al., 2016). Similarly, a consideration of the speaker’s actions within conversations may outline how central macro-level processes such as exploration unfold in real-time. Finally, as the second worked example highlighted, one important benefit of IMICA is that it links an analysis of micro-identity content to macro-social contexts, and can illustrate how cultural practices (e.g., what questions are asked, Raeff, 2014) and conversational conventions (De Fina, 2013) may shape identity development.

There are some notable similarities between IMICA and approaches such as thematic analysis (e.g., Braun & Clarke, 2006), positioning analysis (e.g., Korobov, 2015) and membership categorization analysis (Stokoe, 2012) due to their shared socio-constructivist underpinning. For example, both thematic analysis and IMICA can be used to identify content themes underlying speakers’ everyday talk. Similarly, IMICA shares the micro-analytic approach to studying identity construction in everyday interactions that characterizes membership categorization and discursive positioning analysis. However, IMICA critically differs from these approaches through its focus on the individual as the unit of analysis. Due to this focus on idiosyncrasies within specific speakers, the most salient difference between IMICA and other approaches to identities-in-talk is its focus on identifying change and consistency in identity content over time. Thus, while approaches such as conversation analysis or membership categorization analysis are aiming to identify stable conversational patterns across multiple speakers and conversations, IMICA’s aim is to study the dynamic development of identity claims within a specific speaker over time.

The results of IMICA can be presented qualitatively, but can also be used as input for further quantitative analysis. Similarly to approaches such as thematic analysis, researchers may find it useful to present results in terms of frequencies or group differences (e.g., by gender or other social categories of speakers). For example, researchers could report on the range of topics or domains made relevant within the interaction, or assess how many different types of claims are contained within a specific domain. Due to the fact that IMICA preserves the time-serial structure of claims, it is especially well-suited for further time-serial and sequential analyses. For example, analysis could usefully assess whether claims occur in temporal bursts, or are more evenly distributed throughout the interaction (e.g., Xu et al., 2020). In addition, the sequential organization and patterning of the domains of identity claims could be studied using approaches such as T-pattern analysis (e.g., Casarrubea et al., 2015). Similarly, analytical methodologies such as state-space grids can allow researchers to study how speakers’ identity claims affect each other and highlight the presence or absence of frequently shared identity claims over time (e.g., Hollenstein, 2013).

Limitations

Importantly, there are some theoretical and methodological issues that researchers wishing to apply IMICA should consider. The first theoretical issue concerns the ambiguity of everyday talk. Notably, the social business of an interaction is rarely explicitly stated and similarly the precise implication of an identity claim is not only context dependent but also negotiated by the speakers (Deppermann, 2013; Kerrick & Thorne, 2014; Korobov, 2015). As a consequence, both interaction partners as well as analysts may need to rely on interpretation in making meaning of the actions of a speaker (Harré & Langenhove, 1991). We suggest that these ambiguities in the meaning of identity claims are not a limitation of research on social interactions but features of everyday talk itself. Arguably, the task of a micro-analysis of interaction is not to uncover what social business or intention is “really” influencing the formulation of an identity claim but to use different types of social businesses as lenses through which to make sense of changes and consistencies in identity claims (i.e. taking an intentional stance, see Dennett, 1989). We thus attempt to learn more about the often inconsistent and complex nature of identity construction by analyzing speakers’ utterances as if they functioned to create a sense of interpersonal sameness or difference, or as if they were aimed at maintaining intrapersonal consistency. Importantly, the choice of specific analytic lenses inevitably limits the richness of social interactions to a small number of concepts (see Benwell & Stokoe, 2006). However, we argue that such a reduction is necessary to systematically make sense of the complexity of everyday life.

Importantly, as Korobov (2015) has highlighted, one limitation of micro-analytic studies of social interactions is that, due their focus on local constructions, identity development beyond the recorded conversations cannot be tracked. As a consequence, we do not know whether a momentary identity claim enters a speaker’s repertoire and will become relevant across developmental time (Thorne & Shapiro, 2011). Consequently, approaches to social interactions such as IMICA should be combined with research on long-term developmental outcomes such as identity status interviews (Kunnen & Metz, 2015). Despite its limitations in prediction, it is important to note that IMICA is able to capture local change and development as was highlighted in both worked examples. Moreover, we suggest that IMICA’s insights about processes and mechanisms may transfer more easily beyond the conversations under study, as it highlights the possibility of a specific mechanism to occur across a variety of settings (Peräkylä, 2011). Thus, an application of IMICA needs to carefully consider who is speaking to whom, in what context, and in the service of which social business (see e.g., McLean et al., 2007).

Methodologically, the study of identities in interaction is a long and labor-intensive process. First, the collection of conversational data can be especially time-consuming when interactions between the same speakers are recorded repeatedly over long intervals of time. Second, the creation of verbatim transcripts including time-stamps or sequential information is an intensive but necessary step for high-quality analyses, and less detailed transcripts may result in lower-quality outcomes. In contrast, the analysis may be sped up by the use of coding schemes for the identification of identity content domains especially where a large number of interactions are studied. Researchers should carefully consider the trade-off between the choice of sample-size and the depth of the resulting analysis. Importantly, in its current form IMICA is restricted to an analysis of identity claims within a given domain and does not allow for a consideration of identity content across domains.

Conclusion

Despite these issues, iterative micro-identity content analysis can provide researchers with the tools to study the influence of everyday interactions on identity development. We understand IMICA as a pragmatic approach to the study of interactional identity construction. While our own work is embedded in a social constructivist understanding of identities as interactionally accomplished phenomena, we hope that those who conceive of identities as “expressions” of “relatively firm choices about identity elements” (Lichtwarck-Aschoff et al., 2008, p. 384) will find our guide instructive. Interested readers may discover that conversation and discourse analytic work can offer useful insights and concepts toward the systematic study of the interactional effects and functions of identity claims. However, the approach outlined in this paper does not require any familiarity with either of these literatures, because at a basic level IMICA is concerned with understanding how individuals speak about themselves and others over time.

Data availability statement

The data that support the findings of this study are available upon reasonable request from the corresponding author, Jan-Ole Harald Gmelin. The data are not publicly available due to their containing information that could compromise the privacy of research participants.

Disclosure statement

No potential conflict of interest was reported by the author(s).