Preference organization and possible -isms in institutional interaction: The case of adult second language classrooms

Abstract This study examines preference organization in adult second language classrooms in relation to possible -isms—utterances which are hearably racist, classist, (hetero)sexist, or otherwise exclusionary, although their exclusionary nature may be (re)negotiated in situ. A collection of sixty-one possible -isms from a corpus of fifty-five hours of video-recorded English second language classes was examined using conversation analysis and membership categorization analysis. The analysis shows that participants orient to solidarity by supporting -isms, progressivity by deleting -isms, and moral accountability by challenging -isms; however, participants prioritize solidarity, enacting it early, even in cases of deletion and challenges. I argue that this preference organization is rooted in the institutional roles and objectives of adult second language classrooms, where presumably competent members of diverse cultures aim to foster an environment for active participation. Findings underscore the importance of conducting microanalyses of talk-in-interaction to uncover structural constraints which facilitate the reproduction of systemic exclusion. (-isms, preference, conversation analysis, membership categorization analysis, classroom interaction, exclusion in interaction)*

It is therefore unsurprising that participants in the institutional context of adult second language classrooms struggle around -isms (Nelson 2017). Challenging -isms in this setting can help teachers maintain moral accountability and promote learner appropriacy and criticality (Canagarajah 2014); however, it can also threaten solidarity, stymieing culturally diverse learners' participation and learning (Nguyen 2007), and it can hinder the progressivity of interaction and the timely completion of instructional activities (Sercu 2006). Microanalyses of classroom interactions in situ can help uncover how various social principles and institutional obligations shape participants' treatment of -isms. Using the ethnomethodologically informed frameworks of conversation analysis and membership categorization analysis, I examine how interactants in adult second language classrooms orient to different principles around -isms-specifically to solidarity, progressivity, and moral accountability-given their institutional roles and objectives.
Focusing on -isms in naturally occurring interaction, M=CA researchers consider in fine, multimodal detail when and how -isms are produced and receipted and for what interactional purposes. The level of detail in this process is crucial, as it reveals the taken-for-granted, seen-but-unnoticed ways of (unwittingly) sustaining systemic exclusion (Rawls et al. 2020). Such microanalysis has, for instance, uncovered the often delicate, defeasible, and generally tacit (re)production of -isms, which has led Stokoe (2015) and Whitehead (2015) to propose a focus on possible (rather than 'real') -isms: utterances which could be heard as racist, classist, (hetero)sexist, or otherwise prejudiced or exclusionary, although their exclusionary nature may be (re)negotiated in situ. Prior M=CA research has also illuminated the role membership categorization and preference organization play in sustaining and challenging -isms, which I review next.

Possible -isms and membership categorization
Possible -isms entail hearably exclusionary treatment of social groups, making membership categorization analysis especially useful in their examination. Membership categorization refers to the process of constructing and invoking different social types-that is, categories-in interaction. In this process, interactants tie certain features, including attributes, activities, competencies, rights, and obligations to certain categories (Sacks 1992;Hester & Eglin 1997). Individuals that (presumably) possess features tied to specific categories may then be 'assigned' to those categories, attributed other category-tied features, and held accountable for (not) acting in category-appropriate ways. To use Sacks' (1972) classic example, a person picking up a crying infant may be categorized as the infant's parent, given that their action can be seen=heard as tied to the parent category-soothing one's child. A person who is merely sitting next to a crying infant might also be categorized as the infant's parent and then held accountable for not engaging in expected, category-tied behavior. Of course, the person and infant could be categorized in many other ways, for example, according to their activities, gender, race=ethnicity, and so on; and their categorizations will be achieved, managed, and negotiated in situ, revealing interactants' social knowledge and expectations.
For analysts concerned with possible -isms, it is exclusionary categorization that becomes a key focus: when, how, and why interactants construct social categories in hearably discriminatory or offensive ways. M=CA researchers have examined how gender, racial, and ethnic categories can be invoked to perform criticisms and insults (Stokoe & Edwards 2006;Weatherall 2015), manage accusations and complaints (Hansen 2005;Whitehead 2013), and reproduce and challenge exclusion (Talmy 2008;Robles 2015;Stokoe 2015;Whitehead 2015Whitehead , 2018Shrikant 2018Shrikant , 2022. The current study builds on this work by examining how exclusionary categorization might be shaped by interactants' orientations to tacit principles of social conduct-that is, preference organization. (2023) 3

Possible -isms and preference organization
Preference organization refers to participants' differential treatment of actions which promote or threaten certain principles of social conduct (Pomerantz & Heritage 2013). One prevalent principle is that of solidarity or of supporting one's interlocutor (Heritage 1984;Clayman 2002). While what counts as support is context-dependent, in gross terms, supportive, solidarity-enhancing actions include agreement, approval, and=or appreciation, and solidarity-threatening actions include disagreement, denial, and=or disapproval. A preference for solidarity becomes observable in interactants' differential treatment of solidarity-enhancing and solidarity-threatening actions: The former tend to be delivered quickly and directly, as socially unproblematic or 'preferred'; and the latter tend to be delivered with mitigative delays and accounts, as socially problematic or 'dispreferred' (Heritage 1984). This preference organization may be complicated, however, by the concurrent relevance of multiple, even conflicting principles. For example, while recipients of assessments may orient to solidarity by treating agreement with assessments as preferred, recipients of complimenting assessments might orient to both solidarity and modesty by downgrading compliments or shifting their focus, rather than simply agreeing with them (Pomerantz 1984).
Solidarity becomes relevant around possible -isms since such utterances reveal speakers' (socially delicate) stances and make them vulnerable to criticism; and this vulnerability can be managed through solidarity-enhancing actions (e.g. agreement, approval, or appreciation). For instance, examining institutional and casual (friends-and-family) interactions, Pagliai (2009) uncovered a preference for solidarity around racist portrayals of immigrants. Recipients confirmed, co-constructed, and expanded speakers' -isms, creating a 'spiral effect' of 'increasingly racializing statements ' (2009:556). Jefferson and colleagues (1987) showed that producers of -isms (and potential improprieties in general) similarly oriented to this preference by apparently expecting and using -isms to invite laughter and enhance solidarity.
Since -isms can compromise one's moral standing, they also make moral accountability relevant (Whitehead 2015(Whitehead , 2018; and interactants can enact a preference for moral accountability through unequivocal (and solidarity-threatening) challenges to -isms. Preferences for moral accountability and solidarity therefore come into conflict: challenging -isms helps interactants maintain moral accountability but threatens their solidarity, and (co)constructing -isms helps interactants build solidarity but compromises their morality. Interactants might resolve this conflict by softening objections to -isms with delays and accounts (Whitehead 2015(Whitehead , 2018, but they might also prioritize one preference over another. Whitehead (2018), for instance, observed that hosts of call-in radio shows oriented solely to solidarity by agreeing with -isms that were restricted to callers' personal experiences. In mediations, by contrast, Stokoe (2015) found that moral accountability somewhat outweighed solidarity: mediators delicately challenged (prospective) clients' -isms despite being institutionally required to enact impartiality. Research on preference also suggests that orienting to either solidarity or moral accountability can 'complicate' sequences with -isms and halt their progressivitythe former with 'spiral effects' (Pagliai 2009) and the latter with delays and accounts (Whitehead 2015(Whitehead , 2018. These 'complications' then make relevant the principle of progressivity, that is, moving an interaction forward (Schegloff 1979). Land & Kitzinger (2005), for example, uncovered this preference in various institutional calls as callers accepted, or corrected only in passing, heterosexist references to their same-sex partners in order to promote the ongoing business of their calls. Although not originally addressed in these terms, the practices of deleting (Stokoe 2015) and disattending (Jefferson et al. 1987) -isms can also be tied to progressivity: Both practices entail disregarding the -ism and responding only to an innocuous part of the prior utterance, which helps move the sequence forward. In the adult second language classrooms examined here, interactants similarly pursued progressivity by deleting -isms, and they regularly accompanied these deletions with agreement and appreciation.

Possible -isms in classroom interaction
Research on classroom interaction has long been concerned with systemic exclusion (e.g. Cazden 2001). Studies have found that interactants may tacitly reproduce exclusion by marginalizing minoritized students' knowledge and participation (e.g.  Thomas 2013). While this research has illuminated how exclusionary practices in education can be reproduced and resisted, it has not considered how they might be shaped by underlying principles of conduct. Microanalyses of possible -isms in classrooms (cf. Talmy 2010), and specifically in adult second language classrooms (cf. Lee 2015), is also lacking within this corpus.
The issue of possible -isms in adult second language classrooms is especially pertinent considering institutional roles and objectives specific to this context. Possible -isms and the principle of moral accountability become institutionally relevant as participants work on promoting language appropriacy and criticality (Canagarajah 2014) and explore cross-cultural differences in categorization practices (Lee 2015). Solidarity and progressivity, however, become respectively relevant as participants aim to foster a supportive environment conducive to participation and learning (Nguyen 2007) and complete various instructional activities within set time constraints (Sercu 2006). Additional complications around -isms stem from participants' presumed competencies. Teachers and students in these classrooms are typically of highly diverse sociocultural backgrounds and relatively close in age; thus, despite their differing levels of second-language knowledge, Language in Society (2023) 5 as presumably competent members of different cultures, they may enact equal rights around normalizing or problematizing possible -isms. Given these various institutional roles and objectives, I explore how classroom participants orient to different principles of conduct around possible -isms.

D A T A A N D M E T H O D
This study is part of a larger project on diversity in adult English second language (ESL) classrooms. The data come from fifty-five hours of video-recorded ESL classes at a community language program in a major city in the Northeastern United States. The participants (all referred to by pseudonyms) were four teachers and thirty-nine students. All four teachers were US-born, with seven to fifteen years of teaching experience. The students were from seventeen countries across East Asia, the Middle East, Europe, and South America, and were in the United States for various reasons (e.g. as immigrants, tourists, professionals, etc.). All participants consented to being video-and=or audio-recorded prior to data collection. They all also had the right to withdraw from the study and to request the deletion of any recordings, although no participants did so.
The data were transcribed in minute detail according to an adapted version of Jefferson's (2004) notation system (see the appendix). This fine level of detail includes paralinguistic and extralinguistic aspects of interaction (e.g. volume, silence, gaze, etc.) and helps reveal subtle, seen-but-unnoticed features of participants' (exclusionary) actions. I analyzed the data using conversation analysis and membership categorization analysis. Through a nuanced examination of verbal and embodied actions, the former can uncover participants' orientations to implicit principles of conduct (Sidnell & Stivers 2012) and the latter participants' categorization practices (Sacks 1972;Hester & Eglin 1997), including possible -isms. Throughout my analysis, I drew on the two frameworks in concurrence: My line-by-line analysis of participants' actions focused on when, how, and why they oriented to different preference principles around possible -isms (Stokoe 2012).
In collecting possible -isms I searched for instances of racial, ethnic, gender-based, age-based categorizations which could be heard as stereotypical, discriminatory, or otherwise exclusionary and offensive, 1 although they might only equivocally (if at all) be treated as such in situ. Admittedly, this approach runs the risk of producing yet another 'weak catchall term' and an analysis grounded in my personal beliefs rather than in the participants' conduct. However, I opted for this 'generous' collection process because it allowed me to build a more robust collection, observe overarching patterns in my data, and make visible forms of exclusion which are (in the process of becoming) naturalized (Kitzinger 2000(Kitzinger , 2005a. I additionally aimed to minimize the risk of imbuing the data with my own beliefs by explicating how the design and delivery of focal utterances renders them hearable (if not observably heard) as -isms.

6
Language in Society (2023) In my dataset, I identified sixty-one possible -isms. While a preference for solidarity was observable to some extent across these cases, a preference for solidarity alone was observable in twenty-nine cases; a preference for solidarity and moral accountability in twenty-four cases; and a preference for solidarity and progressivity in eight cases. In my initial analysis, I examined possible -isms more broadly, and it was out of this broader analysis that my focus on preference organization around -isms emerged.
A N A L Y S I S This section examines participants' orientations to preference principles around possible -isms. As the analysis shows, participants oriented to solidarity by supporting possible -isms, progressivity by deleting them, and moral accountability by challenging them; and they appeared to prioritize solidarity by first supporting even -isms that they attempted to delete or challenge. While teachers and students alike enacted this preference organization, for the sake of clarity, and in light of teachers' complex institutional rights and responsibilities, I focus primarily on the teachers' actions (marking their focal turns with arrows in the transcripts). I show two examples each of teachers orienting to (i) solidarity, (ii) solidarity and progressivity, and (iii) solidarity and moral accountability. I argue that this preference organization is grounded in the institutional roles and objectives distinct to adult second language classrooms.

Solidarity
A preference for solidarity can first be observed in participants' quick and direct support-for example, approval, agreement, and=or appreciation-for possible -isms. We see this 'preferred' treatment of support in the first extract, where a teacher strongly agrees with a student's telling as soon as the student engages in potentially racist category work. The student, Catherine (CAT), is reporting a current event in Germany: an attack on refugees, tentatively identified as being from "Syria, Pakistan, Iraq, Afghanistan, those countries" (data not shown). In sharing her (ultimately supportive) stance on the refugees (from line 1), Catherine categorizes the refugees as sexually violent and receives immediate (and increasing) agreement from the teacher Greg (TG).  With contrastive stress and structure (lines 1 and 3), Catherine projects a counter to negative opinions about refugees in Germany and initially receives (mere) shows of recipiency from Greg-occasional slight nods (lines 5, 11, and 13). In building up to her counter, though, Catherine also constructs an -ism, describing the men 8 Language in Society (2023) "from those countries" as, factually (e.g. "there has been", line 9) and frequently ("many cases", line 10), sexually violent (line 14). As soon as Catherine ties the activity of "sexually assault[ing] women" (line 14) to the refugees, Greg begins to show not only recipiency but agreement with a deep nod (line 16) (see Stivers 2008), repeated confirmations (line 21), and reported familiarity (lines 18 and 23). These responses arguably allow Greg to navigate the delicate task of (a man) receipting a woman's troubles telling on sexual violence against women (cf. Jefferson 2015; Lo & Tadic 2021); however, they also overtly support Catherine's hearably racist categorization.
Catherine subsequently advances her possible -ism by positioning refugees as morally inferior to 'us' and "our culture(s)", which, being apparently free from sexual assault (lines 30-31), "should" be learned (lines [26][27]. It is at this juxtaposition of 'us' and 'them' that Greg could possibly be seen as orienting to moral accountability, that is, as tempering his support for Catherine's talk through gaze aversions (lines 23 and 29; cf. Robinson 2020; see also students' smiles and laugh particles in lines 27, 30-33). However, these gaze aversions only equivocally (if at all) suggest an orientation to moral accountability, and they are importantly surrounded by Greg's increasing agreement: nods (lines 28 and 29), repeated and intensified confirmations (note the emphasis, line 34, and "absolutely", line 35), and an elaboration (lines 38-40). With this elaborationinitiated in partial overlap-Greg also delays Catherine's projected counter ("but I think", line 37) and upgrades her -ism: He implies that refugees would do well to adopt "our culture" in general, but "especially if they're in that cul-country" (lines 38-39) and, with this straightforward formulation of the refugees' current location (i.e. "they're in that country"), glosses over the perilous circumstances surrounding their forced displacement and resettlement. Greg therefore enacts a preference for solidarity by treating agreement with Catherine's -ism as preferred-delivering it quickly, intensifying it, and temporarily halting sequence progressivity to expand it.
Greg's agreement with Catherine at her very invocation of a possible -ism might also suggest Catherine's orientation to a preference for solidarity. Namely, Catherine produced a possible -ism while receiving (mere) displays of recipiency but not (yet) agreement. It is difficult to claim with certainty though that Catherine used a possible -ism specifically "to pursue" agreement from her recipients.
However, speakers in my data did occasionally more clearly invoke possible -isms as they pursued support for their ongoing (previously resisted) courses of action, thus seemingly treating supportive responses to -isms as expected and preferred. In other words, when encountering resistance to an action (e.g. advice), speakers at times produced possible -isms, as if expecting those -isms to garner recipient support for their thus-far-resisted action. We see one such example next, when a teacher invokes a possible -ism to counter a challenge and pursue agreement with his claim. In discussing innovations, the teacher Simon (TS) has just described the light phone (a cell phone designed to be used as little as possible) as important (2023)   9 for improving people's work-life balance (data not shown). Johnny (JOH), however, disagrees and assesses this device as unnecessary, at least in countries like Colombia (Johnny's home country), where people already have a good work-life balance (e.g. lines 1-4). To further support his (resisted) claim about the significance of the light phone, Simon invokes a potentially racist categorization of Japanese people as having an unhealthy work-life imbalance.

Language in Society
(2) those people never stop working (TS0419A1_1:17:32) (JOH: Johnny; TS: the teacher, Simon; IVA: Ivana; VIN: Vincent) Simon counters Johnny's disagreement by invoking an apparent opposite to Colombian people: Japanese people, who "never stop working" (lines 7-9). Although he receives soft acceptance of this counter (lines 10 and 11), Simon does not yet move to close the sequence, but rather extends it by furthering his category work with repeated extreme case formulations ("never", lines 9 and 22) and increasingly emphatic delivery (lines 12, 22, 25-26, and 29-30). This categorization is hearably (and designedly) unfavorable, although Simon only equivocally orients to it as morally accountable in his reference terms: He shifts from a generic referent ("those people", line 8) to specific category incumbents (lines 15-16 and 18-19; cf. Whitehead 2018) and initially uses a non-ethnic term ("those students", line 15), as if treating explicitly ethnic references as delicate (Hansen 2005).
Despite Simon's clearly unfavorable category construction, Johnny formulates a candidate positive result of incumbents' work-life imbalance (line 37), again disagreeing with Simon (i.e. if "they are happy", they might not need to improve their (im)balance with the light phone). While it could be argued that Johnny's turn here is ironic, neither Johnny nor Simon treats it as such: Johnny seemingly Language in Society (2023) 11 expects a confirming response (rather than laugher or further irony) given his declarative turn-design (line 37; Heritage 2010) and repair initiation (line 39), and Simon strongly and repeatedly disconfirms Johnny's formulation as unironic (lines 38, 40, and 41). Simon finally receives agreement from Ivana (lines 42-46), and the two engage in a brief expansion, upgrading each other's claims about the dangers of (Japanese category incumbents') work-life imbalance (lines 47-54). Only after he has received support from Ivana does Simon initiate a sequence closing with a summative "so" (line 56), and, after reasserting the importance of the light phone, he also finally receives a concession from Johnny (line 60). In this case then, Simon seems to orient to a preference for solidarity by using an -ism to elicit recipient support: He produces an -ism upon receiving disagreement with his claim and maintains it until he receives agreement instead.
In the above extracts, interactants oriented to a preference for solidarity by quickly and directly supporting possible -isms and invoking possible -isms to garner support. As they delivered and pursued supportive responses, participants only equivocally (if at all) oriented to moral accountability through gaze aversions, smiles, and 'safe' reference terms. Overall, however, they increasingly upgraded and collaboratively extended -isms, thus engaging in a 'spiral effect' (Pagliai 2009) and halting progressivity. However, even when they oriented to a preference for progressivity, participants seemed to enact solidarity, as we see next.

Solidarity and progressivity
Participants enacted a preference for solidarity and progressivity by tacitly supporting an -ism that they otherwise attempted to disregard or 'delete' while moving the interaction forward. Participants deleted an -ism by explicitly commenting only on an innocuous part of the utterance containing it; but they simultaneously supported the -ism by subtly smiling and=or laughing at it. We see this in the following extract: In response to a student's possibly sexist contribution, the teacher moves the sequence forward by explicitly commenting only on the complexity of the student's turn; however, he implicitly also shows prompt appreciation for the -ism with smiles and light laughter. The students are taking turns to speculate about how people woke up on time before alarm clocks, and Johnny (JOH) has just proffered that "they should have been married [sic]" (data not shown). We join the class as the teacher Simon initiates a repair of Johnny's answer, which eventually leads to Johnny invoking the hearably sexist 'nagging wife' category (line 6; cf. Weatherall 2015).  Drawing on an idiomatic expression-"make them crazy" (line 6)-Johnny characterizes the "wife" category as irritating its presumed standardized-relational pair 'husband' ("them", line 06) (see Kitzinger 2005a,b on implicit heteronormative reference). This potentially sexist characterization could be heard as an impropriety inviting laughter (Jefferson et al. 1987), and it is treated as such by Ivana, who immediately laughs (lines 8 and 13) and offers a candidate upshot in a smiley voice (line 11, "you don't sleep at all"). Simon, by contrast, offers continuers, nods, and smiles (lines 5-12), which are somewhat supportive, although they arguably fall short of the laughter Johnny's -ism may be inviting.
Once Johnny reaches a transition relevance place (line 14), a response to his impropriety again becomes potentially relevant. At this point, Simon seems to delete Johnny's -ism: He explicitly comments only on an innocuous aspect of Johnny's Language in Society (2023) 13 response-its complexity (lines 16-19)-then (re)initiates a grammar-oriented sequence ("put it into a modal of… speculation", lines 24-26; cf. Waring, Reddington, & Tadic 2016 on invoking learning orientation), and finally progresses the overarching activity ("what do you think", line 33). Notice, however, that Simon still surrounds his apparent deletion of Johnny's -ism with subtle signs of appreciation: smiles, first occurring in line 5 precisely after Johnny has introduced his possible -ism (line 4), and a laugh particle (line 16), occurring after Johnny's embodied turn completion (line 14) (see also line 15). We thus see Simon orienting to both solidarity and progressivity: By promptly smiling and laughing at Johnny's possible -ism, he subtly supports it; but by explicitly commenting only on the complexity of Johnny's contribution, he preempts any expansions of the -ism and smoothly moves the activity forward.
Next, we see a teacher enacting the two preferences more collaboratively with a student. The teacher lightly smiles at the student's possible -ism only after the student seemingly pursues recipient support with her own smile; and once the teacher does not expand on her support (e.g. with explicit agreement), the two jointly delete the -ism and initiate a sequence closing. In answering the teacher Casey's (TC) question about whether the class would dissuade their children from pursuing any careers, Hanna (HAN), a female musician, has just said that she would dissuade her children from becoming musicians. After first accounting for her response by assessing the music career as "competitive" (lines 2-4), Hanna offers an alternative, possibly sexist account which positions women as financially dependent on men.  We see Casey (merely) showing recipiency (lines 7 and 9) as Hanna partitions (Sacks 1992) the children category in terms of gender and projects a contrast between female categories ("girls", line 5, "womans [sic]", line 6), for whom a music career would be "fine", and other gender categories, for whom it would not. As Hanna produces a tacit -ism, categorizing women as lacking or not requiring financial independence ("they don't have to earn money… regularly", lines 8 and 12), she reaches the first possible completion points of her multi-unit turn; and although Casey could shift to more clearly supporting Hanna's categorization at these points through, for instance, agreement (e.g. 'right'), she continues to only show recipiency with nods (lines 10 and 13). It is only after Hanna furthers her possible -ism, insinuating that men, unlike women, do need to earn money regularly (and should therefore not be musicians) that Casey offers a more supportive smile (line 15; see also lines 17 and 18). Importantly, however, this smile comes in return to Hanna's own laugh particles and smiley voice (lines 14 and 16), which ambiguously invite laughter (Jefferson et al. 1987) and=or mitigate her potential sexism (cf. Potter & Hepburn 2010). What's more, Casey's smile appears to be the full extent of her support. When Hanna creates space for additional responses to her -ism with a trail-off "yeah" (line 19), Casey expresses no further (more explicit) appreciation or agreement.
In light of this unexpanded support, Hanna and Casey both orient to progressivity by deleting Hanna's -ism and terminating the sequence. Hanna deletes the -ism by shifting to a non-gendered account for dissuading her children from becoming musicians ("it's really hard", line 20), and Casey deletes it by omitting gender from her reformulation of Hanna's response ("it's maybe not stable", line 24), but notably not omitting Hanna's earlier reference to financial stability (from lines 8 and 12). With the -ism collaboratively deleted, the sequence comes to a close (lines 27-29) and the class return to Casey's initial question ("how about the rest of you", lines 30-31). Interestingly, as they delete the -ism and progress the sequence, Casey and Hanna also more clearly enact solidarity-strongly Language in Society (2023) 15 ("really", line 22) and quickly (at the first point of possible completion and in partial overlap) agreeing with each other's non-gender-based contributions (lines 22-25). Throughout this segment, Hanna and Casey jointly orient to solidarity and progressivity by pursuing and offering minimal, unexpanded support-a subtle smilebefore collaboratively deleting the -ism to move the exchange forward.
As the above cases show, participants displayed an orientation to solidarity and progressivity by showing implicit support (e.g. smiles and laugh particles) before deleting possible -isms. By offering (and pursuing) implicit rather than explicit support for -isms, participants managed to protect their solidarity without triggering a 'spiral effect' and could thus promote smooth sequence closings; and by showing support before, as well as while, deleting -isms, they appeared to slightly prioritize solidarity over progressivity. Smiles and light laughter further allowed participants to enact solidarity without explicitly endorsing each other's -isms, but also without clearly holding each other morally accountable for them. As we see next though, participants enacted (and prioritized) solidarity even when they clearly oriented to moral accountability.

Solidarity and moral accountability
A preference for moral accountability and solidarity emerged as participants delayed and mitigated challenges to possible -isms with supportive smiles, laughter, and pro-forma agreements. We see this in the following example, where a teacher softens her challenge of a student's potentially ageist remark by repeatedly foregrounding her agreement with the student. Prior to the extract, Veronica (VER) noted that, unlike the "younger generation" (including all present participants), the "older generation" struggles with technology. After sharing a story about her older uncle's persistent tech difficulties (data not shown), Veronica repeats her claim about people struggling to keep up with (i.e. "track") technological advances as they age (lines 1-2 and 4). In response, the teacher Erin (TE) offers a turn-initial agreement token (line 5) and launches a second story (Sacks 1992) which eventually, and delicately, challenges Veronica's ageist category work.  Erin initially focuses her second story on her ninety-year-old grandfather (lines 5-6), who, as it later turns out, uses new technology with ease (lines 45-46). However, she delays this storyline at its inception (line 7) and instead launches a telling about her 100-year-old grandmother (line 9), which reaffirms Veronica's claims about older people struggling with technology ("but she doesn't know", lines 13-14; "she's always surprised to see me", lines 20-21). With a second saying (line 23; Wong 2000), Erin then initiates a return to the story about her ninety-year-old grandfather, but once again delays it with claims that the elderly can and should learn to use new technology (lines 26-34 and omitted lines). These inserted claims hearably challenge Veronica's possibly ageist categorization, but notice that Erin mitigates them by: (i) repeatedly ascribing the claims to an article, rather than to herself (e.g. lines 33-34), and (ii) maintaining some of Veronica's ageist presuppositions (e.g. implying that older people might not be successful in their learning by inserting "at least like attempt", line 30).
With "so" (line 34), Erin next projects a logical connection between her ninety-year-old grandfather and (the article's) claims about the elderly benefiting from learning new technology. She quickly abandons this trajectory, however, in favor of a but-prefaced assessment ("what's interesting with him", lines 34-35), which, particularly with its added emphasis on him, marks the story of her grandfather as special and out of the ordinary-in effect, an "interesting" exception rather than a logical, representative example of an elderly tech-user. As the story unfolds, we see Erin finally more clearly challenging Veronica's agism by positioning herself-a member of the younger, presumably tech-savvy generation-as technologically unskilled (lines 39, 49, and omitted lines) and her grandfather as skilled (lines 40-46). Interestingly, Erin describes her challenge as a story "similar" to Veronica's (lines 50-51) and then terminates the sequence with a non-age-related moral about "willingness" (line 61), which she frames as co-constructed by her 18 Language in Society (2023) and Veronica (note the agreement tokens, "yeah. that's true", line 60, and the tag, "right", line 62). Therefore, while Erin orients to moral accountability by ultimately challenging Veronica's possible -ism, she also orients to solidarity by foregrounding her agreement with Veronica and framing her delayed challenge as a collaborative extension of Veronica's contribution rather than an explicit counter to it.
Participants also oriented to solidarity and moral accountability by using smiles and laughter to ambiguously show appreciation for -isms and=or mitigate (incipient) challenges to them, as we see in the final extract. Marianna (MAR), an au-pair, has just attributed her host-family's wealth to one "key word: own [ership]" of a textile business. As we join the class, the teacher Simon (TS) laughingly suggests that he could be equally successful in the textile industry (lines 1-2)-a suggestion which Johnny (JOH) rejects (from line 3) with a potentially racist account for the host-family's success (lines 5, 7, and 9).   (2023) 19 challenge of Johnny's -ism to an increasingly direct one. Importantly, however, Simon not only delays this clear challenge but also continuously sprinkles it with solidarity-enhancing smiles and laughter. As this section shows, even when participants oriented to moral accountability by challenging possible -isms, they still also enacted a preference for solidarity by surrounding their challenges with shows of support (i.e. subtle appreciation and agreement). Arguably, participants could have oriented to these two preferences concurrently by, for instance, delivering softened challenges without delays. However, that they regularly foregrounded (partial) support and delayed (and otherwise mitigated) challenges to -isms suggests that they prioritized a preference for solidarity over that for moral accountability. Through a microanalysis of possible -isms in adult second language classrooms, this study has shown interactants orienting to principles of solidarity, progressivity, and moral accountability, and prioritizing solidarity. Participants oriented to moral accountability and progressivity by delicately challenging and deleting possible -isms, and to solidarity by quickly and directly showing (and apparently expecting) support for -isms; and they prioritized solidarity by offering early and regular support even for -isms that they otherwise deleted or challenged. In managing these preferences, interactants used various resources: Delays and repair initiations allowed speakers to withdraw -isms and pre-empt (further) threats to underlying principles; softened challenges helped interactants protect solidarity and morality; and agreement helped enhance solidarity. Systematically ambiguous smiles and laugh particles also often accompanied these resources, simultaneously allowing for the distinct interactional possibilities of (innocently) appreciating and=or delicately deleting and challenging -isms.
These findings align with prior M=CA research on preference around -isms in institutional settings, which similarly showed interactants orienting to solidarity (Jefferson et al. 1987;Pagliai 2009;Whitehead 2018), solidarity and moral accountability (Stokoe 2015;Whitehead 2015Whitehead , 2018, and progressivity (Jefferson et al. 1987;Land & Kitzinger 2005;Stokoe 2015). However, the current study diverges from prior research in finding that these classroom participants prioritized solidarity by offering regular and early (even if only ambiguous) agreement and appreciation for ultimately challenged and deleted -isms. This observation suggests not only that solidarity is of crucial concern in adult second language classrooms, but that classroom participants might treat possible -isms as calls for building solidarity, which they may then orient to even when pursuing progressivity or moral accountability. The dispreferred design of teachers' challenges, despite their institutional authority on appropriate language use, also suggests that challenges to -isms may (still) be treated as outside teachers' professional purview. This raises the questions of how interactants design and respond to -isms in Language in Society (2023) 21 environments where challenges may not be (as clearly) institutionally relevant (e.g. doctors' offices) and how these practices might change over time. Additionally, while this study brought together a wide range of possible -isms, interactants might treat different preference principles as more or less relevant for different types of -isms. Future research could examine preference structures around specific -ism types and consider how those structures may tie to various institutional and sociopolitical circumstances.
Given that time constraints are often seen as a hinderance to addressing systemic exclusion in classrooms (Sercu 2006), we might have expected a preference for progressivity to outweigh solidarity and moral accountability. Yet these teachers (delicately) let -isms pass for the sake of progressivity in only eight out of sixty-one cases. This suggests that solidarity and moral accountability outweighed progressivity when it came to -isms, possibly because of the importance ascribed to exploring cross-cultural differences in categorization practices in second language instruction (Lee 2015). Given language teachers' obligations to promote appropriacy and criticality (Canagarajah 2014), we might have also expected a preference for solidarity to be 'relaxed' and for moral accountability to be prioritized (cf. Stokoe 2015;Whitehead 2015Whitehead , 2018. However, other institutional roles and objectives in these classrooms may have contributed to the overall preference for solidarity uncovered here. First, interactants might have treated support for possible -isms as a means of fostering participation through rapport (Nguyen 2007). By delicately challenging possible -isms and allowing them to unfold as misunderstandings (rather than intentional offenses), interactants might have been orienting to each other's differing language competencies. Finally, participants' (subtle) support, even for deleted or challenged -isms, might have been tied to their symmetrical relationships as competent members of diverse cultures. A 'tolerance' for possible -isms may therefore be grounded not only in speakers' presumed primacies over their personal experiences (Whitehead 2018) but also, more broadly, over their distinct cultural knowledge and values.
By unraveling how institutional roles and objectives may 'sway' participants toward supporting possible -isms, this study underscores the importance of closely analyzing classroom interactions in situ. Examining preference organization in this process can be particularly enlightening, as it can uncover underlying principles of conduct which not only complicate 'idealized' practices for confronting systemic exclusion but might even tacitly reproduce it. An important next step in this line of research would be identifying and helping teachers implement practices that effectively challenge exclusionary talk without undermining solidarity and stifling participation. Additionally, while support for possible -isms in these classrooms might have been treated as a path toward building solidarity and promoting participation, longitudinal microanalyses could help clarify how different treatments of possible -isms in fact shape participation, as well as appropriacy and criticality, over time. Future research could also help identify other implicit 22 Language in Society (2023) N A D J A TA D I C principles or institutional features that shape interactants' engagement with possible -isms in classrooms-for example, a principle of neutrality (cf. Stokoe 2015), orientations to 'overhearing' audiences (cf. Whitehead 2015Whitehead , 2018 or hidden identities (Vandrick 1997), and so on.
On a larger scale, this study helps further illuminate the tacit ways in which -isms may become reinforced in talk-in-interaction. Possible -isms seemed to slip into these classrooms under the radar: They emerged during speakers' mundane, otherwise inoffensive courses of action (Hansen 2005); and apart from (at times only equivocally) being treated as potential improprieties, they were generally delivered as recognizable and commonsensical, and receipted with (light) appreciation and (partial) agreement. Since interactants' approaches to possible -isms in interaction can reflect and shape broader sociopolitical discourses (Pagliai 2009), more microanalytic research is necessary to reveal how possible -isms are tacitly reinforced as well as effectively challenged across contexts. Without such research, we risk overlooking the structural constraints that facilitate the seamless reproduction of exclusion and the complex interactional work necessary to disrupt it.  length of a silence in 10ths of a second (.) micro-pause, less than 0.2 seconds (word) uncertain hearing (x) non-transcribable segment of talk; each 'x' stands for a syllable of talk () unidentified speaker Ss multiple student speakers nods non-speech activity or transcriber comment {nods-words} dash to indicate co-occurrence of nonverbal behavior and verbal elements; curly brackets to mark the beginning and ending of such co-occurrence when necessary N O T E S *My deepest gratitude to the teachers and students who so generously welcomed me into their classrooms. I would also like to thank Hansun Waring for her invaluable guidance and feedback, as well as the editors and anonymous reviewers for their thoughtful suggestions, which helped sharpen the arguments and contributions of this article.
1 Although it might be argued that 'stereotypes' are, for instance, distinct from -isms, I found that a potentially offensive categorization may be (and in my data at times was) formulated as either an '-ism' and=or a 'stereotype' (or a 'mere description', as in Stokoe 2015), with the latter being treated as a more mitigated alternative to the former (see extract (6), lines 13 and 37). I therefore examine these various potentially offensive categorizations together.