The Recognition and Interactional Management of Face Threats: Comparing Neurotypical Participants and Participants with Asperger's Syndrome

Erving Goffman has argued that the threat of losing one's face is an omnirelevant concern that penetrates all actions in encounters. However, studies have shown that compared with neurotypical individuals, persons diagnosed with autism spectrum disorder can be less preoccupied with how others perceive them and thus possibly less concerned of face in interaction. Drawing on a data set of Finnish quasinatural conversations, we use the means of conversation analysis to compare the practices of facework in storytelling sequences involving neurotypical (NT) participants and participants diagnosed with Asperger's syndrome (AS). We found differences in the ways in which the AS and NT participants in our data managed face threats in interaction, where they spontaneously assumed the roles of both storytellers and story recipients. We discuss our findings in relation to theories of self in interaction, with an aim to illuminate both typical and atypical interactional practices of facework.

and thus possibly less concerned with face in interaction. Interactions including participants with ASD may therefore involve more variation (cf. Maynard 2019) in the facework patterns, and the analysis of these interactions may deepen our understanding of what facework is really about.
In this paper, we compare the practices of facework in storytelling sequences involving neurotypical (NT) participants and participants diagnosed with Asperger's syndrome (AS) 1 by means of conversation analysis. We explore the possible differences in the ways in which the AS and NT participants recognize and manage face threats in interaction, both in their role as storytellers and as story recipients. We discuss our findings in relation to theories of self in interaction, with an aim to illuminate both typical and atypical interactional practices of facework.

SELF IN INTERACTION
The way in which the experience of self emerges in and through interaction has been a classical theme in sociology and social psychology (Cooley 1922;James 1891;Mead 1934). According to George Herbert Mead (1934), we are constantly reading the behavior of other people, even if they are not aware of it. There is always something in others' behavior that allows us to deduce the meaning of their actions, whether it be a glance or a bodily gesture (Mead 1934:14). We are more or less consciously seeing ourselves the way that others see us, and we are perpetually putting ourselves in the role of the other(s). This mechanism, according to Mead, is essential in the formation of (self-conscious) self (Mead 1934:68-69).
Charles Cooley, a contemporary of Mead, noted that reading the mind of the other usually generates emotions (Scheff 2003). His famous concept of ''the looking-glass self'' refers to the emotions of shame and pride: ''A self-idea of this sort seems to have three principal elements: the imagination of our appearance to the other person; the imagination of his judgment of that appearance, and some sort of self-feeling, such as pride or mortification'' (Cooley 1922:184). Goffman (1956) also pursued the idea of emotions arising out of role taking, but he formulated it less directly than Cooley and was dealing with embarrassment rather than shame (Scheff 2003:243). According to Thomas Scheff (2003), Goffman explicitly fleshed out the link between embarrassment and role taking in his everyday examples of impression management (Goffman 1956(Goffman , 1959(Goffman , 1963a(Goffman , 1963b(Goffman , 1967, making the abstract ideas of Mead and Cooley more concrete. Goffman's crucial idea of impression management made the avoidance of embarrassment a central motive of interpersonal behavior (Scheff 2003).
With his notion of face, Goffman (1955Goffman ( , 1956 referred to the participants' mutual recognition of the social value of their selves. Goffman proclaimed that the participants' reciprocal protecting of each other's face is an all-pervasive concern and that participants seem to be as sensitive to their own as the other's face, since ''in these matters, ego boundaries seem especially weak'' (Goffman 1956:265;see Peräkylä 2015). For Goffman, face means the experience of self in an evaluative interpersonal perspective, and facework, then, involves the collaborative maintenance of face in interaction (Peräkylä 2015). 1 With the publication of the Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition in 2013, Asperger's syndrome (AS) was replaced with a broader diagnostic category of autism spectrum disorder (ASD; American Psychiatric Association 2013). We use the term AS participants (instead of ASD participants) when discussing our data, since the participants in the study were diagnosed before the new diagnostic manual.
The process of controlling how one is perceived by other people can also be called self-presentation. Interest in self-presentation emerged somewhat independently at about the same time in psychology and sociology (Leary 1996). One of the pioneers in the psychological field was Edward Jones (1964), whose methodological approach to investigating self-presentation was quite different from Goffman's: whereas Goffman engaged in anthropological observations and anecdotes, Jones tested different theoretical ideas through controlled experimentation (Leary 1996:8). Since then, selfpresentational strategies have been offered to explain many different interpersonal behaviors, such as aspects of aggression, helping, conformity, attribution, and group decision making, as well as different behavioral dysfunctions in clinical psychology (Leary 1996:10; see also Leary and Miller 1986). These psychological studies have focused mainly on the individual and their self-presentational tactics in different situations. In the current study, our perspective is more interactional, as we see impression management as reciprocal maintenance of face in interaction during actual conversational turn exchanges. We characterize interactional facework as the participants' mutual avoidance of embarrassment, and thus our approach is very much Goffmanian. However, perhaps in a more Jonesian spirit, we utilize a quasinatural setting where we can test our theoretical ideas by comparing participants who might differ in the degree to which they are concerned about face in interaction and, thus, the degree to which they engage in facework (cf. Leary 1996:13).

ASD and the Development of Self
There is an abundance of psychological literature on ASD's possible relation to a specific deficit in the so-called theory of mind, which implies difficulties in seeing the world from another person's perspective (Saxe and Baron-Cohen 2007). As the ability to read or interpret other minds is intimately connected to the ability to reflect on one's own thoughts and emotions (Bogdan 2000;Fonagy et al. 2002;Frith and Happé 1999;Mead 1934;Kilpinen 2013), there have been several theories that autism might be linked to an atypical development of self (Cohen 1980;Frith 2003;Hobson 2010;Powell and Jordan 1993;Zahavi 2010). These ideas have gained support from experimental studies (for reviews, see Huang et al. 2017;Lyons and Fitzgerald 2013). Atypicalities related to ASD have been found, for example, in body awareness (Cascio et al. 2012), autobiographical memory (Bowler, Gardiner, and Grice 2000;Crane and Goddard 2008), self-recognition of mirror images (Dawson and McKissick 1984), and engagement with others' attitudes toward themselves (Hobson 2010). The atypicalities related to self-other relations in ASD are also very much socially (re)produced, and they can form vicious cycles: ''When social affiliations are organized primarily by voluntary choice and contingent upon sustained mutual satisfaction, those who are slower to develop the kinds of competencies necessary to form and maintain such relationships are also systematically denied opportunities to develop them'' (Fein 2015:83; see also Fein 2020).
All of the earlier-mentioned characteristics of ASD can have consequences for the impression management and facework in ASD. Studies have shown that selfpresentation differs between participants with ASD and NT participants (Scheeren et al. 2010(Scheeren et al. , 2016 and that participants with ASD can have challenges in providing appropriate justifications for the emotion of embarrassment felt by others (Hillier and Allinson 2002) and by themselves (Capps, Yirmiya, and Sigman 1992). In addition, Baron-Cohen and colleagues (1999) found that children with ASD differed from neurotypical children in recognizing behavioral faux pas in an experiment that required the children to apprehend the emotional impact of an erroneous statement on the listener.
At a more general level, Ochs and colleagues (2004) have considered ''social as interpersonal'' and ''social as sociocultural'' as a relevant distinction to understand the challenges that highfunctioning individuals with ASD face. According to these authors, perspective taking in interaction encompasses more than interpersonal inferencing about another's mental states (as codified in the concept of theory of mind), but it centrally involves sociocultural ''decentering.'' This decentering draws upon members' awareness of the larger behavioral expectancies associated with socially and culturally organized situations (see also Wallace 1965:41). Ochs and colleagues (2004) thus state that some individuals with ASD can be very competent in the interpersonal level (for example, participating in conversational turn taking), but they can still encounter problems in the sociocultural domain of sociality. Arguably, the recognition and interactional management of face threats, in addition to requiring theory-of-mind skills, are deeply connected to the domains of sociocultural knowledge.
Many recent conversation analytic findings have highlighted the competencies of participants with ASD that have previously gone unnoticed (see, e.g., Dindar et al. 2016;Fasulo 2019;Korkiakangas, Rae, and Dickerson 2012;Maynard and Turowetz 2017;Muskett et al. 2010).* The sharp analytic focus of conversation analysis, however, can also be used to triangulate previously established clinical knowledge about ASD (O'Reilly, Lester, and Muskett 2016:358) and to demonstrate, for example, how certain atypicalities found in psychological experiments can realize themselves in and through specific interactional practices (cf. Dobbinson 2016; see also Hobson et al. 2012). Facework practices are very interesting to investigate in this regard, because in order for us to delve into the possible differences of facework between NT participants and participants with ASD, the participants with ASD already need to have an abundance of interactional competence. As our quasinatural data set enables a fine-grained microlevel comparison between AS participants and NT participants in the same interactional setting, our focus here is to especially investigate whether and how the practices of facework differ between these two groups and thus reveal some interactional nuances in and through which the AS participants in our data may exhibit their AS diagnosis.

MATERIALS AND METHODS
The data set analyzed in this article consists of 10 video recordings of 45-to 60minute dyadic conversations, where one participant has been diagnosed with Asperger's syndrome and the other participant is neurotypical (AS-NT dyads), and 9 video recordings of control data, where both participants are NT (NT-NT dyads). The participants were adult males between 18 and 40 years old. The participants received instructions to talk about happy events and losses in their lives in a freely chosen way. The AS participants were recruited from a private neuropsychiatric clinic that specializes in the diagnostics of autism spectrum disorders. Their diagnoses were based on the International Classification of Diseases, 10th Revision (World Health Organization 1993). The NT participants were recruited to the study via student email lists, and their NT status was confirmed by using the Autism-Spectrum Quotient (Baron- Cohen et al. 2001). All participants were informed about the use of the data and signed a consent form.
The data were produced particularly for research purposes, but unlike psychological experiments or structured interviews, the discussion was carried out freely without any researcher intervention. The data can therefore be described as ''quasinatural.'' As the participants did not know each other beforehand, the conversational context for them was rather similar to that in other types of ''make talk'' situations, such as in airplanes, queues, or other places, where parties are in each other's proximity and sense a need to generate conversation (Maynard 1989:93). Similar types of settings have previously been used in conversation analysis (CA) when studying first conversations (see, e.g., Maynard 1980Maynard , 1989Maynard and Zimmerman 1984;Svennevig 2014). In our data, the participants were explicitly encouraged to talk about emotionally meaningful issues (happy things and losses), which seemingly encouraged storytelling. The topics discussed, however, also included many other genres than what the participants were instructed to talk about (such as computer games, hobbies, school, etc.).
The method utilized in this paper is CA (e.g., Schegloff 2007), which is a qualitative method for studying audio-and video-recorded interactions. CA is optimal for unraveling the recurring interactional practices through which social actions are constructed. The collection analyzed for this study comprises stories, anecdotes, and other tellings of various lengths (N = 593) that involve a display of stance by the teller to what is being told and make relevant the recipient's affiliation with that stance (Stivers 2008). In this paper, we will focus on a subcollection of ''facethreatening tellings'' (n = 60), where tellers describe moments or situations where their own selves are presented in an unfavorable light. Our case-by-case assessments of tellings as face threatening or not were based on our shared cultural membership with the participants, this members' knowledge allowing us to identify delicate or otherwise problematic topics that would require acute impression management from the tellers. These topics include, for example, losing one's job, not having friends (on the stigma of loneliness, see Lau and Gruen 1992), and failing a driving test or a subject at school. Many stories we have examined could also be considered as instances of ''troubles telling'' (Jefferson 2015) or ''complaints'' (Drew and Holt 1988), but in these particular complaints/troubles tellings, the face of the teller was under threat. We were able to gain access to the participants' own orientations to the face threats through missing recipient affiliation, as the recipients often had visible trouble endorsing the tellers' stances in these sequences. This is not surprising, as descriptions of failure with self-deprecating stances can be difficult to affiliate with (see Pomerantz 1984a). Out of the entire collection (N = 60), 28 stories were told by AS participants, 15 were told by their NT coparticipants, and 17 were told by NT participants in the control dyads. The stories of AS and NT participants dealt with very similar topics.
In the initial stage of analysis, we investigated different practices of facework in the collection as a whole, without dividing the participants into separate categories. Quite early on, however, we noticed consistent differences in the patterns that emerged between the two participant groups. The following conversationanalytic examination aims at illuminating these qualitative differences between AS and NT participants' practices of facework during storytelling sequences. From our collection of tellings, after an initial illustration from the NT-NT dyads, we examine four examples from the AS-NT dyads and one ''deviant case'' from the control NT-NT dyads. The patterns of facework we present here are representative of the whole collection.

ANALYSIS
Here we present examples of the interactional management of face threats in conversational storytelling. We will begin with a ''null case'' from the control data (NT-NT dyads), where the typical practices of facework by both the teller and the recipient can be observed. Then we move on to examples from the AS-NT dyads, which showcase the differences we have found in the AS participants' orientations to managing face threats, in their role both as the teller and as the recipient. We start by exploring the facework of AS tellers (first position), then we will focus on the facework of AS recipients (second position), and after that, we analyze how the AS tellers respond to the NT-recipients' potential face-saving actions (third position; see Schegloff 2007). Finally, we present a case from the control dyads, which on the surface seems to follow an atypical pattern similar to the AS-NT dyads but with a closer look provides additional evidence for the claims presented earlier.

''Null Case'': NT Teller Follows NT Recipient's Potential Face-Saving Action
This is an example from the NT-NT dyads. B has been telling how happy he was to make so many new friends when he started university, as he ''instantly clicked'' with all the other 11 freshmen. A displays affiliative understanding of B's telling by stating that it is really important to find like-minded people, as ''one does not get along with everyone that well'' and states that sometimes it can be difficult to make friends. He then launches into the story.
In the telling, A describes a trouble: it has been difficult for him to find good friends (lines 1-4) after moving to Tampere (where he lives now). There are several overt indications of the topic being a delicate matter. First, A's talk is not very fluent, and it has several pauses and hitches (lines 2, 3, 7). Second, A uses a lot of mitigations and disclaimers while describing the situation. He mentions that he did, in fact, make acquaintances, but none of them became ''such a good friend'' (line 4). He also specifies that he is now talking only about his time here in Tampere (lines 5-6) and remembers to mention that he has a girlfriend (line 7). A also brings up ''that Canadian'' (lines 4-5), who was an exchange student he had befriended (which was discussed earlier) but who unfortunately had already returned to Canada. All these practices can be described as ones of first-position facework, aiming at controlling the impression B gets from A as a person.
There seem to be some difficulties in B's reception of the troubles telling, since there are no indications of affiliation so far (in lines 1-9). There is a clear slot for recipient affiliation in line 9, but B also passes on this opportunity. In this particular example, B himself has just flaunted all his new friends (11 of them to be exact), which complicates the situation and might make it even more difficult to respond. A then continues his telling (lines 10-12) by describing a coworker with whom he gets along well, possibly doing more selfserving facework by explicating that he does have some friends after all. In addition, this elaboration modifies the stance to a more positive one, which is perhaps more easily responded to. Thus, A is ''pursuing a response'' (Pomerantz 1984b), which is very effective, as the new piece of information is responded to immediately (line 13). The most interesting part is what happens next: in his response, B offers a possible reason for why A does not have friends: ''ah yes yes yes that was it so that when you left from Oulu, you then left quite many friends there?'' (lines 13-14, 16). The quite ostentatious ''"ah . yes yes yes \ '' in the beginning of his turn can be seen to orient to the lateness of his response, and it also seems to serve as a remedy to his lack of affiliation before (the Finnish token aa indicates that the speaker now understands something relevant that conflicts with his or her previous assumptions; see Koivisto 2015). Importantly, B is topicalizing something that was implicit in A's telling and moves the focus away from A's difficulties in making friends to underlying issues that can explain the situation. B paints a picture of A as someone who has had ''quite many friends'' before he moved to another city and is thus engaging in secondposition facework by enforcing the teller's ''positive face'' (Brown and Levinson 1987). B's turn seems to be very efficient in both its affiliative import and in saving the situation (cf. Goffman 1955). After this, A accepts and follows B's line of action by confirming the assumption and continues by further emphasizing that his two best friends stayed in Oulu when he moved and that luckily they are still his best friends (lines 15-16, 18). A is thus also engaging in third-position facework by accepting the face-saving action the recipient B offered.
The beginning of the telling was characterized by missing recipient affiliation, as empathizing with the lack of friends (when one himself has an abundance of close friends) could be interpreted as showing pity. Here, as also elsewhere in our collection, recipient affiliation could even be considered as a dispreferred action (for preference for disagreement in relation to self-deprecating stances, see Pomerantz 1984a). It is possible that refraining from empathetic turns (e.g., ''Oh that is so sad'') is therefore also part of the facework recipients do in these situations.

Examples from AS-NT Dyads
Here we present examples from the AS-NT dyads that illuminate the AS participants' different orientation to face threats. The examples are ordered with regard to their sequential position (first, second, and third). In first and third positions, the face threat concerns the AS teller's own face; in second position, the face threat concerns the coparticipant's face. In each example, we discuss its similarities and differences with regard to the null case presented earlier.
AS teller's orientation to face threat (first position). Here we demonstrate the AS participants' (lack of) orientation to face threats in first position, in the role of the teller. Let us now consider examples 2 and 3, in both of which the AS teller's conduct seems to deviate from the NT teller's behavior described in example 1.
Both examples begin, similarly to the null case presented earlier, with some hitches and hesitation (''yeah so um'' [example 2, line 1]; ''so I . . . went . . . to o-on-on-one'' [example 3, line 1]). The slight hesitations are followed by quite straightforward and honest descriptions of the problematic situations (''so I did not have a friend'' [example 2, line 1]; ''I thought that I am a little bit difficult to get into this group so'' [example 3, lines 2-3]), meaning that here the tellers do not mitigate their situations by offering any explanations or disclaimers. Their descriptions of loneliness and exclusion, respectively, are responded to only minimally by the NT participants (''yes'' [example 2, line 2; example 3, line 4]). It is interesting to ponder how the formulation in example 2, ''I did not have a friend'' in singular form might differ from the plural form ''I did not have friends.'' Intuitively, by using the singular form, AS's situation comes across as even more lonely. He continues by expressing his wishes to have a friend in a direct, unpretentious, and unapologetic manner (''I needed someone'' [line 3]). This is in stark contrast to the way in which NT described his situation in example 1. In a similar vein, in example 3, the AS participant states in a deterministic and defeated tone, ''it's probably not worthwhile then'' (line 7), referring to his thoughts when he was excluded at a party.
In the null case, the NT participant did a lot of interactional work to explain his situation to the recipient. He used hedging and milder expressions, such as not having such good friends, and made efforts to describe all extenuating circumstances. Examples 2 and 3 are very different in this regard.
AS recipient's orientation to face threat (second position). Next we present an example from the AS-NT dyads where we demonstrate AS participants' orientation to a face threat (concerning the coparticipant) in second position, in the role of the recipient. Here the NT teller is complaining about having to move back to live with his parents as an adult.
When NT is talking about moving back to live with his parents (lines 1-10), he gives a lot of grounds and justifications for this. He mentions that he had lived with a roommate before, but now as he lost his job, he does not have the money to live alone anymore and is therefore forced to move back. 2 Next NT explicitly orients to the embarrassing aspect of moving back with his parents by immediately adding a comment: ''which I feel a little ashamed of but'' (lines 10, 12). This explicit mention of shame can be seen as part of the teller's facework, as he displays awareness of his situation's socially undesirable nature (cf. Goffman 1955).
In line 12, NT produces a summarizing turn, ''that's just how it is now,'' which marks the end of the telling and invites recipient uptake (Jefferson 1984). The turn has an idiomatic feel of the sort commonly found in complaint sequences where the complainant is seeking affiliation from a possibly unsympathetic hearer (cf. Drew and Holt 1988). Here, however, AS does not display affiliation or engage in any potentially face-saving actions by producing normalizing turns (e.g., ''Yes, of course'') or by refocusing the topic to some of the possible underlying, explanatory issues (as in the null case). Instead, NT's turn is followed by a 0.8-second pause, after which AS again responds only minimally (''mm-m'' [line 15]). This is followed by another pause of 0.7 seconds (line 16). What happens next is highly intriguing: NT shifts the topic and goes on to explain how in Greece, even 45year-olds have to move back with their parents (line 17 onward). In the lack of face-saving actions from the recipient, we argue, NT seems to be ''saving himself'' by normalizing the situation and comparing it to even worse cases. The new information NT supplies (about the situation in Greece) shifts the focus away from his own problematic situation, and thus it can be seen as more ''easy'' for AS to respond, and indeed, the turn receives strong alignment from AS in overlap (line 19).
With close examination, one can find subtle displays of affiliation and acknowledgment of the face threat in the AS recipient's minimal responses. The nii tokens (line 11) can be seen as agreeing and supporting (on the affiliative use of nii, see Sorjonen 2001) NT's description of his circumstances and the reasons why he had to move back with his parents. In line 13, AS changes his agreement token nii to a more 2 In Finnish, line 4 actually translates ''the job left from under me,'' which is very different from ''I lost the job'' or ''I got fired,'' indicating even less agency and blame for the NT teller. neutral mm right after NT has produced the word ha¨vettäa¨, ''ashamed.'' This can be seen as a subtle orientation to the fact that one should not agree with the other person feeling ashamed (cf. Pomerantz 1984a). Ultimately, however, it is the NT teller who engages in explicit facework by normalizing his own situation.

AS teller does not follow NT recipient's potential face-saving action (third position).
Next we present a relatively long example where the AS participant describes his difficulties in socializing at school parties. This is an example of what we previously called third-position facework: the original teller accepting (or in this case, rejecting) the face-saving action the recipient offers. The beginning of the following extract was already presented in example 3. AS tells NT about how it was difficult for him to get into the freshman group at the university: after attending one party he decided that it was not worthwhile (lines 1-7). There are clear slots for recipient uptake in lines 4 and 8, but NT passes on these options. AS then makes a summary assessment that his telling probably belongs to the category of unpleasant things (lines 9-10). NT agrees with AS's assessment (lines 11, 13) and then asks a question about the admission rate of his major (line 14). At first sight, NT's question seems to shift the topic quite a bit. However, its main function here, we argue, is to move the focus away from AS's personal difficulties to some underlying issues that can explain the situation. With his question, NT offers AS a possibility to explain his exclusion from the group on objective grounds. On one hand, because AS has previously described the freshman group as a small, inside group (not shown in the extract), NT's question could orient to the small number of accepted students overall, thus making the tight-knit group a result of this. On the other hand, if the admission rate were high but only few students participated in the events, they could talk about the small percentage of students who attend the parties. Support for this interpretation is also found in the question's design, as it ends with sitte, ''then,'' which indicates that the question has somehow arisen from what has just been said (cf. Selting 1992:326).
This strategy, however, does not work here, as NT seems to have forgotten that AS transferred to the current study track from another major and therefore does not even know the admission rate. AS explains this (line 15 onward) and eventually answers how many students got accepted to the original study track to which he applied, which is not relevant here and thus removed from the transcript. NT then returns to the business of AS not getting into the tight-knit freshman group (possibly because his original attempt to deal with it from another angle failed) and displays affiliation with ''those are indeed unpleasant such cliques emerge'' (lines 32-33). Again, NT shifts the potential blame to the others and then generalizes (and in some way even normalizes) the experience. NT's turn is topicalizing the underlying issue of forming inside groups, which he offers as an explanation for AS's situation and, importantly, as one that is not related to AS's persona per se. However, AS does not display agreement with NT or elaborate on what he has said. Instead, he acknowledges what NT said (''mm'' [line 34]) and then disagrees with the implication (''I don't blame them'' [line 37 onward]). AS seems to stay focused on his own perspective instead of looking at the situation from a larger sociocultural angle, as NT suggests. It is also possible that AS favors a more honest and veridical description of the situation instead of one that would depict his own character in better light. The example shows the difference in the AS and NT participants' orientations to Example 5. I don't blame them (A13; 5:11) face threats also in third position, when the recipient's face-saving action is not accepted and followed by the AS teller.

''Deviant Case'': NT Teller Does Not Follow NT Recipient's Potential Face-Saving Action
This final example is a deviant case from the control (NT-NT) dyads that shows us that NT tellers do not always necessarily follow the NT recipients' potential facesaving actions, either. Early in the discussion, A told a story about his girlfriend cheating on him. The following telling occurs in an environment where B has described a breakup with his girlfriend, and A responds with a ''second story'' (Sacks 1992) about another breakup of his. In example 6, A describes how his girlfriend left him after a prom rehearsal, which came by surprise (lines 1-6). After a pause (line 7), B displays minimal agreement (''nii'' [line 8]) and then agrees with A's evaluation that one cannot really do anything if one's partner does not want to continue the relationship (line 10). After a relatively long pause (line 12), B makes laughingly a generalization, ''women have mistreated you a little bit'' (lines 13-14), and A reciprocates the laughter (line 15). After this, B then continues, ''like more than one'' (line 16), seemingly referring to A's earlier story about A's girlfriend cheating on him. B's response here has similar face-saving implications as example 5 (the forming of cliques), as he shifts the blame for the situation on the women. In response to this, however, A does not seem to be going along with B's line of action. A no longer smiles or laughs, and he then produces a turn filled with hitches and pauses (line 17). The inhale, the pause, the ''yes well,'' and the ''ee'' (possibly going for ei, ''no'') quite clearly project disagreement with B's previous turn. B orients to the upcoming dispreferred action by A and quickly alters his formulation ''or dumped at least'' (line 18). This does not seem to make matters any better, however, and the turn is followed by a 0.5second pause. After this, A produces an admission: ''well I do know that I was a little bit obnoxious towards her now and then'' (lines 20-21). The turn starts with the Finnish particle nii, which is this context could be translated as either ''well'' or ''yeah,'' and may carry the meaning of weak agreement or hesitation. As A then continues his utterance, he is clearly changing the depiction of the situation: he portrays himself not as the one who was mistreated but rather as the one who was mistreating the girlfriend (cf. example 5 where AS did not follow the action of blaming the others but referred to his own shortcomings). Thus A does not accept the potential face-saving action B is offering.
We may now examine B's reception turn (lines 13-14, 16) in more detail, as it actually has important distinctions to the potential face-saving actions presented in previous examples. First, B's response does concern A as a person. It is not formulated in a generalized manner (e.g., ''Women always misbehave like that''), but B's turn implies that several women have specifically mistreated A. Second, being mistreated by women (or to be dumped by many) is culturally very different from the forming of cliques, for example-the former being more discrediting for the person being dumped. Here, we argue, A's misalignment with B's line of action (blaming the women) and taking the blame by owning up to his own bad behavior, amounts to the protection of his face. A rather claims more agency over his own (obnoxious) actions than receives the stigmatizing label of ''being mistreated by women.'' It is also important to note that A remedies this temporary injury on his self-image by emphasizing that he has now changed his ways (lines 29, 33). This deviant case brings to light the NT participants' global orientation to the management of face threats, even in circumstances that require them to temporarily portray themselves in a bad light.

RESULTS AND DISCUSSION
In line with earlier studies (see, e.g., Huang et al. 2017) concerning autism spectrum disorder's atypical relation to self in interaction, we found that the orientations to face threats of participants with Asperger's syndrome differed from those of the neurotypical participants. Whereas the NT tellers utilized several facework practices in describing their own difficult situations, such as mitigations, disclaimers, justifications, and explanations, the AS tellers did not utilize any of these practices when talking about the same topics. This is in line with previous experimental studies where participants with ASD were found to be less preoccupied with impression management than were NT participants (e.g., Scheeren et al. 2010). As story recipients, the NT participants in our data remedied their initial lack of empathy by producing affiliative face-saving actions later in the tellings. Even though we found some subtle indications that AS recipients acknowledged the face threats (see example 4), the AS recipients did not engage in similar facework as NT recipients, which sometimes led to the NT tellers producing such turns by themselves. This finding might be related to the challenges found in ASD in giving explanations to others' emotions of embarrassment (e.g., Hillier and Allinson 2002).
The results concerning AS participants' atypical facework both in first position, as tellers, and in second position, as recipients, are very much in line with what earlier experimental research has already suggested. What is most intriguing, however, is that the AS participants did not ''accept'' or follow the recipients' face-saving actions in third position, either. There can be several, possibly cooccurring, explanations for this new finding. First, the face-saving actions by NT recipients involve a perspective change that requires cognitive flexibility that can be more characteristic to NT participants than to AS participants. As seen in example 5, participants with ASD can sometimes stay in their own egocentric perspective (Frith and de Vignemont 2005), instead of looking at the situation from a larger sociocultural perspective (see also Ochs et al. 2004). When recipients then shift the focus of talk, tellers have to renegotiate the coherence and intelligibility of the events as relevant circumstances change (see Goodwin 1995:118), and this type of sociocognitive operation might not be characteristic for AS participants Solomon 2005, 2010). Second, the facesaving implications of recipients' utterances are often quite implicit. Participants with ASD are known to have challenges in interpreting nonliteral meanings (Kalandadze et al. 2018), so responses that for the NT participants involve face-saving implications might not be such for the AS participants.
We have interpreted our findings in the light of assumed cognitive differences between AS and NT participants, but another general interpretation is also possible. It can be that the AS participants were less preoccupied with maintaining face and more concerned with being genuine or ''staying true'' (cf. Cage, Bird, and Pellicano 2016)-that for them, face was not, after all, such an all-pervasive concern as it seems to be for NT interactants. This interpretation would be in line with the study by Scheeren and colleagues (2010), who found that some participants with ASD know very well what kind of self-presentation is expected from them, but they prefer to be veridical rather than adhering to audience preferences (see also Cage et al. 2013). Some situations might exist, such as psychotherapy contexts, where NT individuals are also expected to drop their defenses and share their genuine feelings without resorting to selfserving facework. Future research could ask whether and how the atypical facework patterns we have observed in this study differ from the facework of NT participants in therapy and related contexts.
Our study has one key limitation regarding the categorization of participants (as AS/NT). Schegloff (2003; see also Wong and Olsher 2000) has warned against invoking external categories and letting certain kinds of material dictate the terms of the analysis. We very much agree that not everything the participants say is related to their differing neurological statuses, and it should remain an open question what relates to AS diagnosis and what does not (cf. Schegloff 2003:45). Here, however, we have based our analysis on the distinctive, observable differences between AS and NT participants' practices of facework that are supported by earlier psychological research and theory. Although the autism spectrum is known to be very heterogeneous (Masi et al. 2017), the qualitative findings in our limited data set can work as a basis for future comparative work using larger samples, including samples of naturally occurring data, and also quantitative investigations. In a similar vein, we also acknowledge the contradictory views regarding the concept of ''face'' in the field of conversation analysis: while some studies found this concept helpful in analyzing conversational practices (Lerner 1996;Heritage and Raymond 2005;Peräkylä 2015), others claim it distracts the analyst from the study of structures of action (Schegloff 1988). We believe that as such, most of the observations that we made in this paper could have been made also using more conventional conversation-analytical concepts (such as complaints, troubles telling, and preference). Yet, we found the notion of face to be extremely helpful as we sought to account for the patterns identified in our particular data set. The notion of face integrated the conversation-analytical findings and also served as a bridge between these findings and the psychological theories concerning autism.
On a more general note, our study shed light on the social construction of self as it happens through the minor details of the turn-by-turn unfolding interaction. The self-images of (neurotypical) individuals arise from, and are offered for, acknowledgment in the interactional process of facework (Goffman 1955(Goffman , 1956; see also Perä kylä 2015). These practices of facework, however, are usually so taken for granted by the participants themselves (as well as the analyst) that they can sometimes be hard to pinpoint in typical interactional data. It is therefore important to also examine different participant groups in order to bring to surface how face threats are (or are not) oriented to in interaction and how the reciprocal maintenance of face is actually implemented in and through conversational practices (for narcissistic participants' orientation to face threats, see Janusz et al. forthcoming). Furthermore, our study suggests that perhaps not all individuals rely on the same unspoken assumptions about the mutual avoidance of embarrassment (see Scheff 2003). Even though the neurotypical ''looking-glass self '' (Cooley 1922) might be very much in the hands of others, we agree with Fasulo (2019:627) that it is possible that individuals with ASD ''may develop a self more autonomous from the social environment than is generally the case for neurotypical individuals.'' The NT participants in our data, however, were visibly engaging in facework throughout the examined storytelling sequences, both as tellers and as recipients. For NT individuals, then, the interactional management of face threats seems to be a real concern.
BIOS Emmi Koskinen is a doctoral candidate in sociology at the Faculty of Social Sciences, University of Helsinki, Finland. Her research focuses on the phenomena of affiliation and facework in storytelling interactions. Utilizing quasinatural settings, she investigates how certain interactional deviances, such as autism spectrum disorder, narcissism, or depression, potentially show in the participants' behavior and automatic physiological responses.
Melisa Stevanovic is a senior university lecturer at the Faculty of Social Sciences, Tampere University, Finland. She has conducted conversation-analytic research on the interactional management of power relations, collaborative decision making, and affectivity both in naturally occurring interactions at the workplace and mental health rehabilitation and in quasinatural experimental settings. She has also published on the topic of social interaction deficits and experiences of interaction.
Anssi Perä kylä is Academy Professor at the Faculty of Social Sciences, University of Helsinki, Finland. His current research interests include self in social interaction, narcissism and personality disorders, and social interaction in psychotherapeutic and psychiatric encounters. He is involved in both conversation-analytic and experimental studies. In recent years, he has published in outlets such as American Journal of Sociology, Social Psychology Quarterly, and Research in Language and Social Interaction.