The ART of apologizing: Entering the black box of an intervention program

Abstract During recent decades, evidence-based treatment programs have become a given part of the youth justice system. Typically, such programs are evaluated through quantitative effect studies, in which a variety of outcome measures play a significant role. This case study offers an alternative, interactional evaluation of a treatment program. More specifically, the analysis focuses on an Aggression Replacement Training (ART) session that was held at a youth detention home in Sweden. In this session, two trainers and three detained adolescent boys perform an exercise that serves to teach the latter various apology practices. A detailed, conversation analytic examination of the interaction in the session shows that the trainers repeatedly problematize the boys’ contributions in a kind of deviant-making enterprise. Thus, rather than recognizing competencies that do become visible through closer inspection, the trainers one-sidedly highlight lack and deficiency. It is argued that the interpretative frame of ART, with its focus on pathologization, individualization, and responsibilization, amplifies the incarcerated boys’ deviancy, hence symbolically locking them up in a second, non-material or discursive, sense.


Introduction
Residential care for young people has a long history in Sweden, beginning several hundred years ago and involving a series of transformations along the way (Sallnäs 2009). One of the more notable changes during the last couple of decades has included a shift from family-emulating units that offer milieu therapy to professionalized care units offering standardized, evidence-based treatment programs (Lundström et al. 2020). This new trend within the youth justice system has been referred to as a form of "therapeutic governance" (Cox 2011: 593), a term that captures the complex mixture of care and control that often exists in contemporary psychoeducational programs. This mixture is also discernible in the present study of a specific intervention, although the primary attention is directed toward the control aspect.
In this case study, we focus on a recording of an Aggression Replacement Training (henceforth ART) session that was enacted at a special approved home in Sweden. More specifically, we analyze the interior of an apology exercise in which two trainers and three adolescent boys with a violent record participate. By applying a conversation analytic approach (Hutchby and Wooffitt 2008), including detailed examinations of transcribed recordings, we dissect the interactional machinery of the specific apology exercise as well as the problematic underpinnings of the ART program as such. Indeed, we believe it is an irony that a psychoeducational program that presupposes so many inherent shortcomings among its beneficiaries still manages to elicit rather sophisticated apology repertoires from them during training. However, rather than recognizing this, the trainers seem to be caught within an interpretative frame that continuously sustains the boys' position as deviant others.
Note that we do not want to deny the violent past of the detained residents, which made them eligible for intervention in the first place. Instead, we are interested in analyzing how the trainers and the residents, each in their own way, contribute to the making and unmaking of deviance in and through the interaction. Pollner (1978: 271) describes such an analytic project as an effort to understand "the deviant-making enterprise." Here, this includes a focus on how the trainers in various ways engage in "deviancy amplification" (Young 2004: 550). Once the notion of an underlying, inherent deviance among the boys has been firmly planted through the program, this deviance will also be reproduced in live dialogues during individual sessions. However, as will be shown, the boys also try to resist deviancy amplification by clarifying the rationale behind their actions and beliefs. The result is a rhetorical battle in which the parties, directly and indirectly, assign various identities and traits to themselves, each other, and third parties.
The article begins by providing the reader with a short background to ART. After this, we summarize some of the previous research on residential care for young people, and highlight three central features of psychoeducational programs. Finally, we analyze three episodes from the apology session that was enacted at the studied care unit.

Aggression Replacement Training (ART) -A background
A large number of Swedish studies have shown that an overwhelming majority of young offenders continue their criminal life after residential custody (e.g., Andersson Vogel 2012;Pettersson 2010;Shannon 2011). Thus, there has been a growing need to improve the youth justice system. In Sweden, like in many other countries, the leading idea in this reform work was to introduce evidence-based practices as an overarching guiding principle (Swedish Government Official Reports 2008). This paved the way for manual-based psychoeducational programs such as ART. ART was originally launched in the United States during the 1980s in order to treat antisocial juvenile criminals (Goldstein et al. 1987). Since then it has been applied within a wide variety of settings across the world. It was first introduced in Sweden in the 1990s and is now the most adopted intervention program for adolescents, operated through both open care services and residential care facilities (Swedish National Board of Institutional Care 2016).
ART is successful in the sense that it is both well-known and widespread, but its status as an evidence-based program has been questioned. Originally, those involved in developing the program reported positive results (e.g., Goldstein and Glick 1994). However, Brännström et al. (2016: 40) conclude their recent review by stating that it is impossible to say "whether ART helps, whether it has no effect, or even whether it is a harmful intervention." The main reason for this conclusion is that most of the effect studies that have been performed suffer from extensive flaws, for instance, selection biases, limited follow-ups, and involvement of researchers with vested interests in the interventions.
Against this background, we believe it is time to study the interactional dynamics of the program itself. By entering the black box of a single session of ART, we hope to provide an alternative type of critical examination of this intervention. It is our argument that the raison d'être of this program should not be judged solely on quantitative effect studies, but also on close analyses of the interaction occurring within individual sessions. Indeed, as pointed out by Sankofa et al. (2017), qualitative studies may be more effective than quantitative studies in elucidating the internal workings of evidence-based practices.

Previous research
Previous qualitative studies of youth detention homes show that residents tend to describe the institutional environment in negative terms, depicting it as boring The ART of apologizing (Bengtsson 2012), unforgiving (Cesaroni and Alvi 2010), insensitive (Phoenix and Kelly 2013), autonomy depriving (Inderbitzin 2006), irrelevant to life beyond confinement (Cox 2011), and rife with conflict, anger, and resentment (Abrams and Hyun 2009). However, staff members are considerably more inclined to make positive judgements, claiming that it offers individualized treatment (Andersson and Johansson 2008), helps offenders to create new identities (Stokholm 2010), and refrains from using escalating moves (Kivett and Warren 2002). Thus, the two parties involved seem to emphasize very different aspects of the complex mixture of punitive and treatment-oriented practices that previous research has often identified (cf. Cox 2018;Enell 2017;Gradin Franzén 2014;Kaunitz 2017;Phoenix 2009;Silow Kallenberg 2016).
In reviewing the more critically oriented research literature, we also identified three features of psychoeducational programs that were repeatedly presented as its core: The widespread pathologization, individualization, and responsibilization of the clientele. It is our argument that these features represent an important interpretative frame of the ART program, which also comes to life in the microinteractions at individual facilities.
The pathologization of residents with an aggressive record means that they are not judged solely against the backdrop of their crimes, but that everything they say and do in retrospect of their dubious past can be used against them as evidence of inherent flaws (Cox 2011;Crewe 2011;Fox 1999a). This pathologization is also a central feature of the ART literature. According to Goldstein et al. (1998), the founder of ART, aggressive juveniles possess a series of interlocking and compounding deficiencies: They lack prosocial skills, exhibit primitive levels of moral reasoning, have weak anger control, etcetera. We argue that once the residents are viewed in this way, that is, as deficient in behavioral, cognitive, and emotional terms, this interpretative frame will also affect how the trainers approach them during sessions. As Goffman (1961) points out, the adoption of such a frame may cause even the most innocuous actions to appear highly suspicious. All conduct can, under such premises, be read as evidence of a flawed self, morality, thinking, etcetera.
The individualization of incarcerated aggressors means that the justice system holds them, and them only, accountable for what they have done in the past. In close alignment with the pathologizing ascription, psychoeducational programs assume that cognition alone causes behavior, and therefore dismiss situational explanations. Thus, when residents try to explain why they acted in a certain manner, and use archetypal sociological explanations, like "bad home environments," this becomes yet another proof of distorted thinking patterns (Abrams and Hyun 2009). The result is a decontextualized version of criminals' past behaviors, which promotes a view of them as "intrinsically dangerous" (Foucault 1978: 17). Finally, responsibilization, described by Phoenix and Kelly (2013: 424) as a transformative "ethical reconstruction" that seeks to produce rational individuals who actively strive for law-abiding social participation, represents a belief in the possibility of moral improvement of human beings. It is the logical third step after the pathologizing and individualizing practices, and it is necessary in order to justify psychoeducational programs. Within such programs, the participants are expected to constitute themselves as objects of knowledge and objects of transformation both for the trainers and for themselves (Cox 2011). In essence, this corresponds to a subtle integration of coercion-technologies and self-technologies (Foucault 2007(Foucault [1980), which together are presumed to result in improved selves. Responsibilization is thus seen as being productive of particular identities, which are more desirable for society at large (Phoenix and Kelly 2013).
In sum, it is our argument that the ART program, much like other psychoeducational programs, provides a totalizing interpretative frame that affects the trainers' actions within sessions. As Goodwin (1994) argues, the ability to evaluate specific phenomena is not a given talent, but always rests on certain unpronounced practices to which one has been sensitized. Hence, all vision is perspectival and may vary between people. In the present case, the power to speak as a professional belongs to the trainers, and everything the residents say is evaluated against the backdrop of their aggressive history and against the ART ideology. In contrast, the residents speak from their horizon and try to defend themselves in various ways. The result is a bidirectional power struggle between the parties (cf. Kivett and Warren 2002) to which we will soon direct our attention.

Data and method
This study draws on video recordings from a larger corpus of data collected at a youth detention home in Sweden . The institution was a special approved home, run by the Swedish National Board of Institutional Care (cf. Shannon 2011). Young people at such care facilities may have been involved in crimes, drugs, or other socially destructive behavior, and can be placed by means of compulsory care, with parental consent, or in substitution for a prison sentence.
The institution prescribed the ART program to all residents deemed eligible to attend it. The studied session, in which the residents practiced what is referred to as the advanced social skill of apologizing (Goldstein et al. 1998: 225), took place in the morning and was about two hours long. Three of the twelve residents, and two of the five staff members that were engaged as ART trainers, took part in the session. The former were given the pseudonyms Emil, Felix, and Roger, and were between 16 and 18 years old. The latter were given the pseudonyms TOM and TONY, and are marked with capital letters in the transcriptions.
The session was transcribed in its entirety using Jefferson's (2004) symbols (see Appendix A). However, in order to increase the readability of the transcripts, we also used slashes (/) to indicate "breath units" or "tone units." According to Juzwik and Sherry (2007), transcribing in such units foregrounds the poetic dimensions of speech and captures how it unfolds as a verbal performance.
In line with conversation analysis, the primary focus of the analytical work was the actions that the participants accomplished in and through their speech. Thus, the aim was to show the detailed intricacies of the intervention program in naturalistic interaction rather than to provide a full ethnographic account of the institution. This type of "single case analysis" (Hutchby and Wooffitt 2008: 113-114) is aimed less at producing generalizable findings and more at analyzing "the various conversational strategies and devices which inform and drive its production." The fine level of granularity of the analysis does not allow more than a few sequences, in this case three, to be chosen. All data were initially analyzed in their original Swedish form and only later translated into English (see Appendix B).
The study was conducted in concordance with the ethical principles of research in Sweden, including informed consent, confidentiality, and usage rules (Swedish Research Council 2011). In line with the "The ethical review act" (Swedish Code of Statutes 2003), ethical approval was obtained before the research project started.

Findings
Before we turn to the analysis of the apology exercise, it is important to clarify that previous research on apologies presents a picture of a very complex human activity (for an early review, see Meier 1998). Rather than trying to account for this extensive literature, we here only point to three types of apology that the participants spontaneously oriented toward in the session, and which proved to be important in our analytical work. We refer to these as perfunctory, precarious and nonsensical apologies. Perfunctory apologies occur when the transgression is rather small, when the guilty party did not commit it on purpose, and when the apology is easy to perform, sometimes even being executed automatically (Fraser 1981). Conversely, precarious apologies occur when the transgression is substantial, when the perpetrator presumably performed it intentionally, and when the apology, as a result, is often difficult to perform, demanding more elaborated strategies (Goffman 1971). Finally, nonsensical apologies ought not to occur at all, but if they do, they are inappropriate because the transgression is normally huge, and the guilty party is also assumed to have committed it in a highly premeditated manner, for instance in relation to a perceived guilty enemy (Exline et al. 2007). It needs to be pointed out that this distinction is not made in the ART manual.

Searching for the magic middle
Before Excerpt 1 starts, the trainers, as part of the first step of apologizing, encourage the residents to suggest various situations in which an apology might be appropriate. The assumption is that aggressive youngsters have limited interpersonal abilities, and that social situations resulting in violence could have been avoided with adequate skills (Goldstein et al. 1998). Thus, being able to envision such situations, and knowing how to deal with them, is an important part of the responsibilization training included in many psychoeducational programs (Phoenix and Kelly 2013). In responding to the trainers' quest for concrete situations, several suggestions are offered by the residents. For instance, Roger argues that you should apologize if you accidently break something. To help the residents, trainer Tony gives another example: You have to apologize if you say something mean, like "bloody monkey," to your mom. In the first line of the excerpt, Felix problematizes these suggestions by presenting two vastly more dramatic examples. Our argument is that these examples, and the objections Felix raises, initiate a pedagogical criticism which is not recognized by the trainers, but which is instead read as resistance.
Excerpt 1 (Time: 20:29) Present: TOM, TONY, Emil, Felix, Roger, and researcher 01 Felix: =what the hell / you can't just / ok, we say like 02 thi:s then / (1.0) I shot someone / then I come 03 after ten minutes / "yeah, sorry that I shot you" 04 heh heh / or something / you have to thi:nk also 05 TONY: that's right ( (nods) In the ART manual, the trainers are presented with a few examples of situations in which one is expected to offer an apology, for instance, when one has interrupted someone, made an error, or hurt someone's feelings (Goldstein et al. 1998). However, the manual does not mention that there are situations in which it is not appropriate to apologize. By pointing to two such examples -(1) when you have shot someone (lines 1-4) and (2) when you are responsible for genocide (lines 6-9) -Felix manages to show that the task at hand, inventing apology scenarios, is not just a matter of judging whether you have committed a regrettable action, but also whether an apology is appropriate at all (Petrucci 2002). In short, Felix raises the question of the function of apologies, emphasizing that apologies must be understood in relation to the social context in which they operate.
Earlier, we made a distinction between three kinds of apologies. Applying this to the case above, it is possible to categorize Roger's initial example of breaking something as making a simple perfunctory apology relevant. Tony's example of saying "bloody monkey" to one's mom is more complex and demands a precarious apology, clearly admitting guilt and taking responsibility for repairing the relationship. Finally, Felix's more extreme examples fall into the realm of nonsensical apologies, as the deeds are beyond repair. All three types of apologies suggested can be seen as valuable and competent contributions to the session, indicating knowledge of different interactional contexts.
Looking more specifically at Felix's contribution, he does not really use the extreme cases in order to discuss them any further, but to point to the futility of the suggestions made by the rest of the group, saying they are not challenging enough.
Thus, after having presented them, he suggests that they should discuss "something else" (line 11), such as when one has "slept with someone's sister" (line 13). This suggestion from Felix can be read as a realistic scenario that potentially would necessitate a precarious apology, hence being highly relevant for teenage boys to discuss and practice.
Whatever Felix's proposal might signal, it is not picked up by any of the trainers. Their first responses suggest that they read it as a ridiculous proposal or a joke (lines 14-15), and a kind of resistance that is to be expected according to the ART manual. After this, Tom makes another attempt to come up with a more mundane case: Slamming a door in front of someone (line 18). Depending on the circumstances, this transgression could demand either a perfunctory or a precarious apology, but even if it required the latter, it would obviously still not match Felix's delicate case, as judged by his responses: "it's SUCH SIMPLE stuff you raise" (line 19) and "think a bit more difficult" (line 21). However, having had sex with someone's sister, presumably a friend's sister, would be a very delicate matter that could jeopardize the relationship, hence requiring a truly precarious apology.
This quest for a challenging case, in terms of both difficulty and relevance, could be seen as a search for the so-called "magic middle" (Berger 2012), that is, the area between trivial and unattainable learning, where the task at hand is exciting and demanding rather than boring or exhausting. Moreover, it also points to the lack of relevance of the program for Felix, a common criticism put forward by young offenders in retrospective interviews (cf. Cox 2011; Phoenix and Kelly 2013; Silow Kallenberg 2016). Here, this criticism is conveyed in situ, but to no avail. Instead of paying attention to Felix's criticism, Tony gets defensive. He objects to Felix's extreme Hitler case (line 22) and argues that the group cannot discuss "all the examples" (lines 25-27). However, he does not pick up on Felix's alternative example of having had sex with someone's sister. Indeed, not even when Felix agrees that his own extreme examples are "too bi:g" to discuss (lines 23-24), that is, outside of the desirable magic middle, does Tony return to the "sex case" as a suitable one. As gatekeepers, both Tony and Tom have the privilege of having the last word, which Tony now exercises by returning to the question of shooting someone, asking Felix if he has ever shot someone (line 28), thereby undermining the relevance of discussing the scenarios suggested by him.
In sum, Felix's objections in the session are interpreted as evidence of resistance against the training, rather than as specific (apology) competencies. Goldstein et al. (1998) emphasize that participants in ART often seek to circumvent the training, and Felix's actions can be seen to fit several types of "trainee resistance" (e.g., digression, monopolizing, and interruption) that are enumerated in the ART manual. For instance, when Felix brings up the hypothetical case of Hitler as a nonsensical apologizer, he can be seen to digress from the directive to suggest everyday examples of transgressions, and to disrupt the normal procedure of ART. At the same time, he can also be seen to monopolize the episode by talking more than all the others together, while persistently interrupting the trainers (lines 13, 23, 27). Hence, the specific competencies that Felix can also be seen to exhibit (e.g., identifying situations in which an apology would be inappropriate, inventing relevant scenarios, and putting forward justified pedagogical criticism) are left unrecognized. As the trainers easily become the "judges of truth" (Fox 1999b: 98), "resistance" is a very handy part of the interpretative frame provided by the program. It allows them to subject the residents to the "worst possible readings" (Goffman 1971: 108), and it protects them from having to engage in pedagogical self-criticism.

A place for apologies
The next example occurs a few minutes after the previous one. The participants have been discussing the second apology step, the one focusing on different ways of apologizing, and have brought up a series of suggestions (e.g., calling, texting, sending flowers). Immediately before Excerpt 2, the group has turned to the third step, choosing the best time and place to apologize. Our argument is that the trainers in this sequence operate within the same form of narrow interpretative frame as in the last example, which makes them read Roger in a bad light. It all begins when Roger, immediately before the first line in the excerpt, suggests that the pub could be a good place to apologize in. In lines 1-3, we can see how Tom's rhetorical question turns Roger's proposal into a "virtual offence" (Goffman 1971: 108-109) by maximizing its disagreeable implications. Note that Roger only suggested that a pub could be a suitable place where to apologize, but not that he should be "sloshed" while doing it, as Tom now proposes. Of course, there is nothing in the ART manual that invites Tom to make this kind of reading. However, the interpretative frame of the session, for instance, the assumption of "cognitive distortions" among the residents (Fox 1999b: 89), contributes to this. If you are a juvenile with a criminal background, you are not only bound to choose the wrong place for an apology, but also to be "sloshed" rather than "sober" while performing it. An important part of the creation of the virtual offence is the associative force of commonsense. Most people would agree that an apology delivered in a pub would be less appropriate than in, for instance, a café, in particular with regard to precarious apologies. The pub environment, associated with alcohol, must always be a suspicious place to beg for forgiveness as there is an increased risk that you are drunk in that context. Yet, when Roger justifies his choice in front of the trainers, he creates a scenario in which it is indeed possible to defend the pub environment: If you have had exactly the right amount of alcohol, and "everything kinda flows" (line 8), then it could indeed be "perfect" (line 8). Through this bickering with the trainer, Roger shows himself to be an independent thinker, who is reluctant to go along with the either/or (sloshed/sober) quality of the trainer's question.
At this point, Tom uses Roger's defense as a resource for instigating a new critical appraisal: "is it because you have the courage to… apologize then?" (lines 11-12), hence converting Roger into a coward. Goffman (1961: 36) refers to this rhetorical practice as "looping" and argues that it is more common in incarcerating institutions than in society at large. Through this practice, an offender's attempt to offset fault-finding comments, or to normalize one's own behavior, always risks paving the way for further fault-finding. Whatever you say, it can always be turned against you, just like in the case above. By trading on commonsense knowledge of The ART of apologizing what alcohol presumably does to you, for instance, making you more confident, Tom suggests that Roger does not have the guts to apologize unless he is drunk.
When Roger again tries to defend himself (lines 13-14), he is exposed to yet another looping practice from Tony. This time, he implies that Roger remains unaware of what the recipient of the apology might think of him if he is drunk while performing his act (lines 15-16). Roger thus has to defend himself once again, this time by clarifying that some people, like his girlfriend, actually would accept such an apology, whereas others, like his parents, would find it less tolerable (lines 18-21).
In sum, the analyses above show that the participants continuously add a variety of concrete components to the hypothetical case at hand, apologizing at the pub. And they do so according to a predictable pattern: The trainers try to add details to the virtual offence, making it more and more problematic, whereas the resident, Roger, tries to add alternative details that cleanse the shabby picture. While the trainers' othering practices contribute to the shaping of a young offender in need of professional help and responsibilizing reformation (cf. Fox 1999a), Roger's defense creates a young man who is already making responsible decisions, that is, whether or not to stay sober in relation to different recipients of the apology. "Lack" is the logic of the former, "competence" the logic of the latter.

Talking like a lion
Excerpt 3 deals with the fourth apology step, making the actual apology, and takes place a few minutes after the last sequence. We here examine how Felix follows up on Roger's argument that apology practices are contingent on the addressee. According to Felix, you cannot offer apologies to just anyone without distinction. Our argument is that we encounter descriptions of two forms of lifeworlds in the excerpt: A consensus-oriented lifeworld and a conflict-oriented lifeworld, each with their own specific rationales, mentalities, and language games (cf. Wittgenstein 1953). When we enter the scene, Felix argues that it is impossible to make apologies to enemies. At the beginning of the excerpt, Felix presents two main arguments as to why an apology to an enemy is deeply problematic. First, the enemy "takes it as a sign of weakness" (lines 1-2) and, second, the enemy then "wants to go against you even more" (lines 2-3). Observe how Felix presents these arguments with a very strong "epistemic stance" (Heritage 2012: 6), that is, in a manner that conveys definite knowledge of what would follow from such an apology. For instance, he does not say "he might take" or "he might want to," but claims to know the outcome beforehand. In this way, he portrays himself as having a deep knowledge of the violent world he normally inhabits, and positions the trainers as the "not knowing." He is the master; they are his apprenctices. Note also that before Felix even presents the arguments, he says "if you really start to apologize" (line 1), which not only suggests that it is a highly unlikely scenario, but also makes it clear that he is talking about nonsensical apologies in a conflict-oriented world filled with enemies. Given this context, only some actions are thinkable, reasonable, or even intelligibleoffering apologies not being one of them.
When Tony ironically responds to Felix (lines 4-6), he focuses only on the sign-of-weakness maxim (the first argument). This accomplishes an important individualizing and pathologizing shift in the interaction. In a consensus-oriented lifeworld, the display of "weakness" does not constitute a problem for people, so consequently Felix's understanding of the importance of frightening, imagecreating practices is clearly abnormal. Thus, by making fun of a vivid display of this phenomenon, Tony transforms Felix's unwillingness to offer apologies into a personal problem, a mental disposition that is based on an irrational fear of being seen as weak. If Tony had instead been prepared to acknowledge various environmental justifications for Felix's non-apologetic stance, it would not have been possible to reach this reductionist conclusion (cf. Fox 1999a).
Tony enacts the looping appraisal of Felix's answer in a clearly ironic manner. Yet, when Felix responds, he does so in a solemn, "po-faced" manner (cf. Drew 1987: 220), "no of course not" (line 7), thereby holding on to his argument, and opposing the implicit ascription of deviance. In his world, it is vital to appear fearsome in order to deter attackers, and this is not a game to be taken lightly. Thus, while Tony pathologizes Felix and his answer, Felix normalizes his own reasoning and safety measures. Felix also explicitly explains that it would not work to apologize (lines 7-8), but then quickly enacts a hypothetical situation in which the magic words indeed are uttered to an imagined enemy, "'I'm sorry' heh 'I'm sorry'!" (line 8). The distancing laughing particle in the middle of his solo role playing reveals that he imagines it to be nonsensical to apologize in the conflictoriented context in which it would be uttered. He is visualizing his own submissiveness in the situation and the negative consequences that would most definitely follow.
Such worries about submissiveness and negative consequences are not unreasonable. Research shows that apologies are a lot more likely to occur in close, positive relationships (Exline et al. 2007), and when they are likely to be accepted (Chiles and Roloff 2014). In contrast, expectations of rejection dampen the willingness to offer apologies. Certainly, it is possible that Felix commits a "forecasting error" (Leunissen et al. 2014: 322) in that he overestimates the potential negative effects of apologizing (e.g., the loss of respect, status, and power). However, his experiences of a conflict-oriented lifeworld justify a quite skeptical attitude toward apologies. Whereas the ART program and the trainers envision a situation in which a minor offence occurs within the parameters of a well-built relationship, Felix pictures a situation of mutual conflicts and unsafe relationships. Research shows that in the latter situation, in which standing up for oneself is a virtue, people often regret their apologiesif they are ever offered (Cesaroni and Alvi 2010). No one wants to be on bended knees, to take a stance of submission, under such conditions (Exline et al. 2007).
When Felix continues to defend his non-apologetic position, it is exactly the stance of submission he refuses to adopt: "he will keep at it until you fight with him / and then he will calm down" (lines 9-11). Felix here refers to the kind of "dog eat dog mentality" that one of Cox's (2011: 598) young offenders construed as essential to being "street." From the outside, this Machiavellian approach could be characterized as abnormal, but Felix himself describes his own aggression as a necessary response to another man's aggression. In his eyes, the violence is warranted and represents a refusal to be victimized rather than a deviancy. In comparison, several of Fox's (1999b: 96) incarcerated interviewees seriously doubted that they could ever learn to act differently, with one of them describing the debilitating effect of the intervention program: "It makes me feel weak." As Fox (1999a) makes clear, the type of situational explanations for aggression that Felix here brings into play have to be dismissed by cognitively sensitized trainers. One way of accomplishing this is to invoke the workings of a welladapted, consensus-oriented lifeworld as a contrastive image. Tony does exactly this when he suggests that he would certainly appreciate an apology from "someone" (lines 12-13). Like previous looping practices by the trainers, he presents this objection from a fairly decontextualized horizon. For instance, Felix is not informed about the blameworthy actions that Tony was supposedly exposed to in order for the apology to be relevant in the first place, and nor is he told about the relationship between Tony and the culprit. Thus, Tony replaces a nonsensical apology situation between enemies with a perfunctory/precarious apology situation between friends/unacquainted people. Whereas Felix tends to reason from the perspective of a specific hostile situation, Tony adopts a universalized, commonsensical, and conflict-free position. From the latter position, Felix's skewed reasoning appears indisputable (cf. Fox 1999a).
Following this, Felix paints an alternative picture of accidently spilling beer on a stranger in a bar (lines 15-16 and 18-19), hence making a perfunctory apology relevant. This creates a scenario which is more in line with what Tony probably had in mind in lines 13-14, although this latter example is contextually richer. At any rate, this scenario allows Felix to demonstrate to Tony that he is indeed capable of offering apologies. Moreover, he here exhibits a kind of "moral autonomy" (Fox 1999a: 442), that is, of being governed by internal rather than external forces, which is the desired outcome of the ART program.
Tony initially greets Felix's rapid "change" with surprise (line 16), but then praises him: "exactly (.) there you go!" (line 20). This latter phrase construes Felix as someone who exhibits a sudden improvement, hence showing progress within the ART program. Unlike when Roger paints a potential scene in a bar (see Excerpt 2), Felix's suggestion is not questioned. Yet, keeping the argument along the previously established conflict-oriented logic, Felix then issues a reservation that he would not apologize to "those you live next door to / or (.) same area / that are like thi:s-" (lines 24-25). Felix does not finish the utterance before he is interrupted by Tony, but we can project that he is about to describe enemies with hostile dispositions. This, in turn, occasions a new looping practice from Tony, who detects an apparent inconsistency: "you can't apologize to people you're going to see again? / I don't really get that" (lines 26-27). Given Tony's consensus-oriented worldview, Felix's reasoning is irrational.
In this situation, where two distinct rationales once again stand in opposition to each other, we are reminded of Wittgenstein's (1953: 223) famous words in his Philosophical Investigations: "If a lion could talk, we could not understand him." Adopting this metaphor to the present case, it is clearly applicable in two different senses. First, Felix appears like an incomprehensible lion in the eyes of Tony, who is unable or unwilling to understand his conflict-oriented rationality. Second, if Felix indeed began to offer apologies to his enemies in the neighborhood, he would appear like a speaking lion in that context as well. He would be inventing an entirely new, unintelligible language game, foreign to the indigenous morality that currently prevails there.
In sum, as Crewe (2011) points out, the unwillingness of staff members to understand the life of incarcerated people transfers the responsibility for their situation entirely to them alone. Supposedly, the real enemy of the detainee is not the figures of the threatening street environment, but the enemy within. However, for Felix, "responsibilization" does not mean that he will cool down and stage a series of apologies in his crude environment. It would be devastating for him. Instead, responsibilization means standing up for himself, as no one else would do it. Phoenix and Kelly (2013) argue that if there is anything adolescents have learnt in rough environments and institutional care settings, it is that "only they can do it for themselves" (p. 435). Hence, they are not easily transformed into "prudential citizens" (p. 420), the implicit aim of many psychoeducational programs, but adhere to an alternative responsibilization rationale that has been firmly established in less refined environments.

Discussion and conclusion
After dissecting the contents of the recorded apology exercise, it is not unreasonable to reach the conclusion that ART, or at least this social skills component of the program, can be described as a fault-finding enterprise, governed by a very narrow interpretative frame. Indeed, in line with Ponnert and Svensson's (2016: 586) argument, the specific manual-driven exercise does not leave much room for professional discretion, but rather encourages the trainers to enact "a strictly mechanical form of work," or a very narrow professional vision (Goodwin 1994), that is bound to produce deviancy. In this way, the young criminals become trapped in vice and negativity under the disguise of a treatment program (cf. Cox 2018).
Certainly, the residents have not been paragons of virtue in their past, but our analyses show that the residents actually display a series of quite normal competencies, for instance, they prefer challenging over simplified cases, can distinguish between various types of apologies (perfunctory, precarious, nonsensical), and do understand that apologies must be designed to suit the recipient. Yet, the trainers respond to and discount the residents' expositions as symptoms of inherent flaws. Whatever the residents say, it can always be construed as "thinking errors," "limit testing," or "negation and manipulation" (cf. Abrams and Hyun 2009;Fox 1999b;Kivett and Warren 2002), which makes the assumed deviance "impervious to contrary evidence" (Fox 1999a: 450-451). Thus, rather than reading the residents' utterances as situated actions, and as displaying specific cultural knowledge of apology practices, the trainers use them as diagnostic evidence of flawed minds that risk producing violence.
Our conclusion is that the interpretative frame of the ART program amplifies the detained boys' deviancy, hence symbolically locking them up in a second, nonmaterial or discursive, sense. As Fox (1999b: 91), points out, this constitutes a form of "language as coercion," which can be characterized more adequately as "punishment" rather than "care." Thus, while psychoeducational programs are often depicted as care-oriented, they can also be described as sanctioning of a "normative imperialism" (Crewe 2011: 516), which consistently disregards the perspectives of those who are subjected to them. If the trainers instead imagined themselves in the context of conflict-filled street life, they could begin to grasp the rationality of the various practices and language games that the residents express in the analyzed session. Once the residents are understood from their own horizon, and "once their nefarious activities are put in proper context" (Matza 1969: 40), they do regain some of their human value.
So what is the alternative to psychoeducational programs of the analyzed type? One answer to this question is to abandon methods that focus exclusively on emphasizing various shortcomings among young criminals. Such othering techniques risk leading to the very instantiation of selves that are fundamentally different from the "normal" population, without ever recognizing these selves as artefacts of the very same technologies (cf. Pollner 1978). If the production of new identities is truly the purpose, why not fashion such a production in ways that are less influenced by reductionist, psychologistic discourses, and more oriented toward approaches with a focus on the outer rather than the inner world? Interestingly, when the incarcerated boys in Wiberg's (1976) early Swedish study were asked to evaluate various ways to accomplish change, "outer-world alterations"finding a partner, engaging in creative activities, and participating in politicswere rated highly. Today this list may look different, but institutional care geared more toward resident participation would certainly be more consistent with both the Convention on the Rights of the Child (United Nations 1989) and the welfare and rights model proposed by Smith (2010