Social Interaction

1. Introduction

Data collection procedures in conversation analysis, as recommended in textbooks (Heath et al., 2010; Mondada, 2013; ten Have, 2007), are often designed to reduce the salience of the researcher in the recordings in order to maximize the “naturalistic-ness” of the data. As written, such conditions strongly discourage the possibility of having a researcher also be a participant (despite well-cited work by, for example, Goodwin, 2004, 2010, 2013). Furthermore, these conditions paint all researcher participation with the same brush, despite the diversity of researcher involvement in ethnomethodology and conversation analysis (EMCA). In this paper, I argue both for a more nuanced understanding of the researcher-as-participant in data, and that being such a participant has benefits for research. Researcher-participants do not (necessarily) challenge the local “naturalness” of the data, and the situated accountability of the researcher-participant will display how the co-participants make sense of the researcher’s presence. Furthermore, researchers being co-participants can be a useful practice for fieldwork, with benefits including providing evidence for the researcher’s unique adequacy and gaining embodied access to the activity.

I will begin by discussing how the researcher has been situated as a participant in data creation in prior EMCA work. I will then briefly show how the analytic practice of grounding observations in demonstrable participant orientations guards against concerns for application of a priori, etic categories. However, this procedure also hinges on a key limitation: that the researcher must be oriented to as a member. Next, I will present some opportunities afforded by being both a researcher and participant, namely the ability to (re)enact practices. Finally, I suggest that “naturally occurring” can be a matter to demonstrate through analysis.

2. Situating the researcher in the research

EMCA research involves an array of methods in which the researcher is positioned in different ways. Drawing heavily on ethnography and participant observation, ethnomethodology specifically dictates the kind of participation that the analyst must be capable of: an analyst ought to be able to produce action that is locally adequate and sensible (Garfinkel & Wieder, 1992, p. 182). Despite suggesting that EM studies involve “Extreme immersion on the one hand and hyper-reflexivity on the other [which then] obliterate the very distinction between researcher and member, observer and observed, enquiry and object” (Pollner & Emerson, 2007, p. 131), the details of the researcher’s participation in fieldwork are not typically subject to extensive analysis, except as proofs of competency (see Garfinkel, 1986; clear exceptions include Livingston, 1986; Sudnow, 1983, 1993). The question of how the researcher reflexively is involved in research has been debated (Lynch, 2000; Pollner, 1991), but that work focused on post-fieldwork analysis. Reporting based on fieldnotes can obscure the exact details of a researcher’s participation in any given moment of data collection or information gathering. Even where it is clear from the report of methods that the researcher was a participant in the activity under study (e.g., Ayaß, 2020; Bennerstedt et al., 2012; Macbeth, 2012), by presenting only description, without supplemental recorded data of the interaction that generated the description, the researcher-participant’s situated actions are obscured.

The written guidelines in EMCA suggest that all efforts should be made to separate the researcher from the recorded activity. Several EMCA research guidelines recommend avoiding the presence of the researcher altogether (Heath et al., 2010, p. 44; Heath & Hindmarsh, 2002, p. 9; ten Have, 2007, p. 63), withdrawing the researcher if their presence is initially unavoidable (Mondada, 2013, p. 38), as well as being discreet with camera placement where possible (Heath et al., 2010, p. 48; Mondada, 2013, p. 36; ten Have, 2007, p. 89). By removing the researcher, the data is ostensibly made more “natural”, and the researcher will not be tempted to participate or overinterpret subsequently (ten Have, 2007, pp. 83–86). These guidelines suggest that naturalness is a matter of avoiding researcher co-presence and interference, so that the data is not “co-produced with or provoked by the researcher” (ten Have, 2007, p.63). Mondada (2013, p. 33) contrasts this aspect with interviews and experiments, where the researcher sets the agenda for the data to be collected.

However, there is a different conceptualization of naturalness, found in Potter (2002): the dead social scientist test. Under this rubric, the recorded interaction is “natural” if it would have occurred anyway, in the way that it does on the recording, with or without the researcher having recorded it. This definition of “natural” more neatly captures the fact that the camera is not an omnirelevant concern for participants (Mondada, 2013, p. 34); the camera is made relevant as necessary for situated action (Hazel, 2016; Laurier & Philo, 2006; Tuncer, 2016). Every recording will inevitably be a produced, situated object, molded by the researcher to preserve the details of the recorded activity, make them available for analysis, and make sense of the action (Mondada, 2006). The researcher’s molding of the recording—and, critically, their necessary co-participation in the unfolding interaction—is particularly clear in recordings of mobile interaction, where the researcher must follow the participants. Even though the researcher is not visible on camera, their active participation is clear through the camera angles, which show their ongoing sense-making about where the participants will go, how they can pass each other, and what they are engaged with (Broth & Lundström, 2013; Broth & Mondada, 2019). Such local sense-making is thus analytically useful, providing insight into ongoing spatial relevancies.

A small number of studies have taken researcher involvement further, acting as a researcher and participant, as one who participates in the main activities being recorded, in front of the camera with the other participants. Charles Goodwin’s work (see below) is the earliest to mention his own role in the videos, referring to himself first by a pseudonym and later by his name in the third person, analyzing himself as any other member. A similar technique is used by Reynolds (2017), in analyzing power lifting training sessions. Reynolds also mentions the unique adequacy his membership granted. Finally, more recently, Schindler (2017) has provided an analysis of her own learning of a martial art, and its coaching from co-participants, though this focuses more on a first-person account of developing skill, Sudnow-like. None of these studies address whether their co-participation has relevance for the “naturalness” of the data, but take it as a matter of course that the researchers are valid co-participants.

Much as researcher sense-making may be encoded in camera angles (see above), being an active co-participant has advantages, including access to relevant ethnographic details (see Pehkonen et al., 2021/this issue). The ethnomethodological studies of work (Garfinkel, 1986) particularly drew on the researcher as a participant in order to reveal the situated orders involved in understanding a mathematical proof (Livingston, 1986), or learning jazz piano (Sudnow, 1993) or a video game (Sudnow, 1983). Robillard’s (1996) detailed recounting of the anger-inducing non-compliance of his own body provides evocative evidence of the researcher’s position as part of the documentation of the phenomenon. Similarly, Garfinkel’s breaching experiments¹ require the presence of a researcher to accomplish the breach of typical action, and have been used as a means to collect rare instances (Baranova, 2020). Although breaching experiments can hardly be said to be natural, they would not be accountable if they were not, at least at first, made sense of as everyday behavior; indeed, Garfinkel reported that the instances that were treated as jokes from the outset were “failed” cases (Garfinkel, 1967, p. 47). At what point do such involvements cross a line between a naturally occurring instance of the phenomenon and artifice? Despite textbook stipulations, not all studies with the researcher as a participant are considered the same kind of unnatural as a breaching experiment. Rather than arguing for a binary, the question should be on what basis we may raise concerns about naturalness, and moreover, what we may gain, or concede, in different forms of researcher co-presence. As I argue below, video analysis is both augmented and complemented through researcher participation.

3. The researcher-participant under analysis

Any EMCA analysis must adhere to the principles central to doing any rigorous analysis of members’ sense-making: ground observations in participant orientations (Schegloff, 1987, 1996), demonstrate procedural consequentiality and relevance (Schegloff, 1992), and make the data on which observations are based as available to readers as possible (Sidnell, 2010). The use of recorded data “serves as a control on the limitations and fallibilities of intuition and recollection” (Heritage & Atkinson, 1984, p. 4; see also Lynch, 2002). Adherence to these principles prevent speculation, intuition, memory bias, and mind reading.

Let us consider a model example of how a sequence involving a researcher-participant may be analyzed. Goodwin recorded himself in interaction with a man with aphasia (he later specified that this was his father, Goodwin, 2018, pp. 62–63). The first paper from this corpus (1995) did not include Goodwin in the data, but among the many published since then, several include extracts where Goodwin is an interlocutor and in which he analyzes himself in the third person (in Goodwin, 2004 he is called Scott but also identified as Chil’s son, 2010, 2013). These studies are not consistent with the texts quoted above, in that the researcher is absolutely co-producing the interaction (though Goodwin is treated as a participant, see below), but Goodwin’s analysis is completely in line with EMCA methodology. For example (2010, p. 385):

It seems clear that Chil is displaying objection…to Chuck’s inability to provide an appropriate frame for interpreting what Chil is trying to say. Chuck hears it this way. He begins his response with a classic change of state token…and thus displays that what Chil has just said has led him to abandon the assumptions that led to the gloss he produced.

Goodwin grounds the observations in what is demonstrably relevant for the participants in situ. For example, in claiming that “Chuck hears [Chil’s utterance] this way”, he provides evidence from what Chuck does that suggests such a hearing—namely, that he produces a change-of-state token, and displays that he has changed his assumptions about Chil’s talk. Goodwin does not draw on remembered information in order to accomplish this analysis (such as characterizing his own confusion about Chil’s referent, stating possible motivations for making incorrect assumptions, etc.). He focuses on the sense that Chil and Chuck produce together, as is demonstrable in the data. Additionally, by making the data available to the reader, Goodwin demonstrates precisely what his participation consisted of, and allows readers to evaluate what kind of participation the researcher had.

There is no evidence that Goodwin’s interactions with Chil are organized in a way unlike everyday, naturally occurring activities, nor that Chil oriented to Goodwin as pursuing a research agenda during the interactions. Goodwin is oriented to as a situated member, specifically as a family member interacting with another family member who has aphasia. The analysis in these studies can speak to the question of how speakers with aphasia interact with relatives precisely because the participants treat each other as being engaged in that activity.

Thus, as long as the researcher’s actions are treated as adequate for the situated interaction, and is oriented to as such in the data, they may be analyzed as “natural”. Fortunately, evidence as to whether the researcher is treated as a member will be available in the data itself. The data stand as a demonstration (or not) of the researcher’s membership and unique adequacy, and the basis on which co-participants understand any given action. Here I will turn to my own data.

My own research has involved multiple scenarios where I was a co-participant in some recordings in the corpus (Hofstetter, 2019, 2020a, 2020b; Hofstetter & Robles, 2019). At times this has been to make up a necessary number of players at a game, or because I was the most skilled or trusted to safely belay a co-climber (what Edmonds, 2021/this issue, usefully calls “satisfying community expectations”), and at times because friends agreed to me opportunistically recording a get-together. In all cases, the interaction would have occurred anyway: Where I was invited in, the game or workout was already underway, and where I was recording from the outset, the meeting would have occurred regardless. In both settings, I have been a member for a decade or more. I have been a hobbyist board gamer, playing with friends and at cafes, since 2010, and I have been rock climbing since 2008.

For example, below is an example of my participation in rock climbing fieldwork. Several of my climbing friends had been planning to take a trip to a particularly nice climbing area for a long time and invited me to join. I asked if I could record on days as opportunities arose, and they agreed. A variety of skill levels came on the trip. At times, due to lack of enough competent belayers, I was the only available person to belay (that is, manage the safety rope) a climber for a climb they wanted to film (some participants wanted recordings they could look at later, for their own purposes). My participation at this point was necessary in order to do “being a good team member” on the climbing trip, but it also happens to stand as a good demonstration of how the climbers oriented to me as a regular, uniquely adequate participant.

  Open in a separate window

Several practices make available that the co-participants treat me as a typical climber. First, most straightforwardly, Pat trusts me to belay him (i.e., to use the equipment in an adequate way to prevent his falling to the ground). He does not warn me with an explicit cry of “falling” or “prepare to catch” or the kind of coaching that would accompany a novice, nor does he wait for my confirmation that I am prepared to catch his fall; he simply falls, and trusts my competency. Second, Pat’s “noh” (L3) accounts for his own falling—it rejects the possibility of being able to continue the climb. Thus, he treats me and the other climbers as relevant recipients for an account, a group to whom he is accountable for not continuing to climb. Third, when Pat begins to haul himself up the wall (he has to rearrange some equipment at “the bail” point), he designs his notification to me as a reminder rather than a detailed instruction of his next move. This is evidenced in both the elliptical dropping of the pronoun, and in the use of technical terminology (“jug up” and “the bail”, L15. “Jug up” refers to this specific rope-assisted haul up the climb, and “the bail” refers to a type of carabiner he will install at a specific location). The process of getting back up the wall, using the rope, requires my coordinated involvement. Pat does not wait for confirmation before initiating the haul, projecting that I will respond in time, as a competent climber would do. Overall, Pat is treating me as a co-participant, specifically a uniquely adequate partner in the activity at hand. My membership and adequacy as climber are made relevant here, not my potentially available role as a researcher, suggesting this data can be collected together and analyzed with any other instance of climbers doing a climb.

Sometimes, the dual role of the researcher will be made available, making both the researcher-as-researcher and the researcher-as-co-participant relevant. Furthermore, sometimes the participants’ own research role (i.e., as participants, rather than ordinary interlocutors, or gamers, or climbers, etc.) will be made relevant. The following instance shows both of these complexities.

Punjabi is the name of a snack mix that the players are eating. John is making a slightly naughty pun about the residue of the snack’s spice mixture on his fingers after eating (L3). My initial reaction appears to excuse the problem of John’s dirty hands (L4), but after a beat I respond additionally to the joking component of John’s turn. I do not align with the joke as humorous, making John accountable for the comment (L7). John uses the camera as a resource for (playfully) accounting for his comment, highlighting its humorous and risqué nature by making the recording relevant (L12). In orienting to the camera, rather than to me as the researcher, John treats me as a gamer and a friend participating with him in a research-observed situation. Thus, making the ongoing video recording relevant does not necessarily invoke the researcher’s dual role. Furthermore, John’s statement shows that it is not only the researcher who has a relevant research role to invoke. By treating the camera as situated, researching observer, John is making his own membership category as a research participant relevant. John is also doing “being a good participant”, providing material “for the camera” (see also Chen, 2021/this issue), which makes his own research participation relevant, but does not explicitly index my own.

As long as there is evidence that the participants make sense of the researcher’s actions and the activity at hand as naturally occurring, then the analyst should treat them likewise. The reader of any such study should also be able to assess such orientations through the presented material. As mentioned above, the adequacy of the researcher-participant is conveniently demonstrable through the same materials, at minimum for the situation in the extract. In this sense, researcher participation augments video analysis, providing additional cases and displaying adequacy. I will next discuss some other opportunities that being a researcher-participant affords.

4. Opportunities for a dual role

So far, I have argued that the researcher’s co-participation can be treated as member co-participation, as long as their contributions are treated as adequate on the basis of local activities, rather than research agendas. I have also suggested that an inherent benefit of having recordings of the researcher is the situated demonstration of their unique adequacy. There are two further benefits I will highlight: first, the opportunity to (re)enact member actions, and second, the possibility to conduct breaching experiments.

A researcher who is simultaneously a participant is able to experience first-hand the contingencies, constraints, affordances, and stakes involved in the activities under study. The additional benefit of having such experiences recorded is the opportunity to re-review them in slow motion and in detail (see Schindler, 2017). A researcher can, in this way, experiment with different actions, for later review, while still responding and behaving according to local accountability during recording. Sormani (2016) demonstrates such a possibility by reenacting the use of a squeeze bottle in a microscope laboratory exercise. In filming his own attempted use of the squeeze bottle, Sormani can highlight the skills (and learning hurdles) of the teacher and pupil in the original exercise.

I have attempted re-enactment while in situ in climbing fieldwork, namely by attempting the same routes as other climbers in the corpus (including ones far above or below my skill level). Interestingly, the climbing community recognizes that each body will have to solve a climbing route according to its abilities (Dutkiewicz, 2015). As a result, situated reenactments of climbing routes are not only fresh achievements from an analyst’s perspective, but also from a participant’s perspective, with respect to climbing culture expectations. The responses from the other climbers can make sense of the researcher-participant’s climb as “just another climb”; that is what climbers do in the course of regular activities.

Additionally, a field reenactment allows the researcher-participant to gain access to difficult to film areas. In this climb, for instance, there was no way to get a camera view of the portion of the route that was giving the climber in Extract 1 difficulty: it was forty feet in the air with no nearby vantage points. Even though the climb was above my skill level, making my own attempt was one way both to demonstrate the skills that would be needed to manage this climb, as well as to see how others responded to another climber’s attempt of the challenge (again, it is common for climbers to attempt moves on climbs that stretch their skill set).

  Open in a separate window

When I hesitate, placing and replacing my feet (L3), Pat shouts an encouragement (L4), treating my bodily reorganization as a sign of trouble. Similar to Pat’s “Noh” (Extract 1), I give accounts that project upcoming failure (L7, L10) before falling. In contrast to Extract 1, however, I do not then announce being “done”—leaving the possibility open that I will try again. Pat treats this as an opportunity to encourage me again, challenging the logic of my account (L16-17). Here, again, we see Pat treat me as a member, albeit a less skilled or physically weaker one. By repositioning Pat as a situated belayer and observer on the very route he previous climbed, his re-situated involvement becomes a matter for public consideration. This makes available alternate perspectives on the affordances that this route provides. Such alternates are analytically useful, as they provide opportunities for the occurrence of deviant cases, breaches, and even slightly different contingencies that may better demonstrate the practice to the analyst (similar to how Schegloff, 1968, found his "500th call" to be critical). Pat’s own reflection on the route he himself had just attempted now becomes analyzable, by virtue of his responses to my climb.

Furthermore, my own experience of the climb becomes part of the research data. The camera could not capture a view of the climb that showed the nature of the rock wall well, but furthermore, no camera angle can capture the affordances of the holds for the body, the embodied sensation of the subtle (yet challenging) backwards slope of the wall, or the care of placing my hands to avoid camouflaged wolf spiders.² These elements guide climbers’ ascents of the wall, but are only accessible through participation (see also Pehkonen et al., 2021/this issue).

The opportunities to take a more active role in recorded activities again test the boundary of what is considered “natural” data, as the researcher is more actively co-producing the video recording. Experimenting with participation—that is, taking advantage of opportunities to question or pursue actions—can provide insight into what responses they give rise to, and how those actions are made accountable. One useful application of this is to attempt rare or unusual actions that would be difficult to capture or find, as Baranova (2020) did in breaching experiments on the phone, wherein she called participants and, without explaining until after the sequence, began a specific script designed to elicit rare “why” inquiries. Below is an example in a board game where I spontaneously decided to try an unusual move, in order to have to it recorded: I bluffed (lied) about a game move. I would not have done this action were I not interested in it as a research object; however, I had never planned to do it, let alone planned it for this moment in the game. In the game below—a combative game, wherein players score by forming specific shapes out of tiles and wrecking other players’ shapes—Adam is choosing where to place a tile. The placement is meant to be bad for the other players—it should give Adam an advantage, and disadvantage John and I. However, I state “Thank you” (L5), implying that I am happy about Adam’s action. This causes Adam to reconsider his move, and eventually change it (data not shown that far).

This bluff, uniquely among my entire corpus of board games, can be confirmed as not just an attributed bluff, but an “actual” bluff, in the sense that at least one person—me—can claim to have intended it as a deception. From an EMCA perspective, one can only ever have “possible” bluffs; claims to confirm any such unavailable motivations are unvalidatable. However, for recipients, this distinction is irrelevant, as there is never a way to have certainty, only ascription. It is, as with any move in this kind of play scenario, a “possible” deception (Schegloff, 2006). There is even evidence from Adam’s receipt of the “bluff” that the very “possibleness” of the bluff is at issue for him: He displays difficulty deciding what “possible” action my utterance is doing (L9, 22). With respect to what can be analyzed with sequential analysis and members’ sense-making, this experiment does not provide any better access to manipulation than standard recording, because all that matters is manipulation attribution. What this practice achieves is, first, a means for attempting to see rare events in recording (see also Baranova, 2020), and second, an opportunity to experience what such an action is like, beyond the limitations of third-person viewing of participants that video affords.

Adam could also draw on the omnirelevant issue of my membership category “researcher”, and/or the presence of the camera, to make sense of the bluff as a (possible) research-motivated action, though he does not. My experiment was not made accountable, neither as an accountable game action (generally speaking, bluffs are not a sanctionable event anyway, although they may be made accountable through teasing, Hofstetter & Robles, 2019), nor by being oriented to as a research-motivated action inappositely occurring in a game. If I did not remember the brief decision to “try out” doing a bluff for the sake of recording it, this could have been any other game action. The experiment seems to sit at the threshold that Robinson (2016, p. 15) described concerning accountability—ambiguous as to whether it is accountable, whether it is a breach. Adam makes sense of my utterance as deviation from typical action (typical for gamers, as most gamers suffering the loss of a piece do not thank the aggressor), rather than as a breach of relevance rules. In other words, Adam could make sense of my actions without recourse to treating me as being “up to something”, at least not “up to research”.

The experiment above was much less systematic and planned than other examples of breaching experiments (Baranova, 2020; Garfinkel, 1967), as evidenced both in my recollection of the event and in Adam’s sense-making. However, I still exerted a kind of control over the data that is rarely seen in CA. Overall, this extract shows an example of how breaches may be accomplished without necessarily implicating the participant-researcher’s research agenda or role. The benefit of doing so is capturing at least some version of rare events, and the experience of triggering them.

5. Discussion

As discussed above, for the researcher to provoke or co-create the data is for the data to be seen, by EMCA textbooks, as not-natural, and thus deviant. Speer (2002a) has argued that the distinction between natural and contrived does not hold beyond methodology. Actual research practices within EMCA are far more diverse than either definition of natural (“do not interfere” and “would have occurred anyway”) permits. A range of data collection techniques have arisen: some studies invite people to a lab in order to get higher quality data (Ruusuvuori & Peräkylä, 2009; Umair et al., forthcoming), others require participants wear special equipment (Holler & Kendrick, 2015; Stevanovic et al., 2017). Some studies involve even further intervention, such as breaching experiments (Baranova, 2020), or staging meetings between non-acquainted people (Maynard & Zimmerman, 1984) in order to view events that are uncommon in natural occurrence. As EMCA expands its investigations into topics that include rare (or even taboo)³ events, or often inaccessible features such as sensory perception, we reach the outer boundaries of what video recordings can offer. Perspicuous settings, where the participants are inclined to make the phenomenon relevant or accountable, can mitigate some difficulties (as with most multisensoriality studies so far, Mondada, 2019), but not all. Furthermore, the development of unique adequacy and ethnographic understanding of a setting relies on some degree of participation (Garfinkel, 1996)—filming those experiences can supplement that process and provide analytic opportunities.

I suggest a better question than natural or non-natural is: to what other data are these data comparable? This question can only be answered by looking at how the participants make sense of the situation. If they demonstrably make sense of it as an experiment, an interview, a chat after work, or a board game, then it is under those auspices we are called to analyze it. Everything is a natural something. This restores Garfinkel’s sense of natural as spontaneously and locally organized (Lynch, 2002). Using this rubric, we rely on locally organized evidence to assess the impact of researcher-participants and avoid excluding them a priori. The adequacy of the researcher’s participation is respecified as a members’ concern. This allows us as researchers and readers to examine the data for the ever-emerging membership(s) and activities being made relevant (so that we need not worry about intervention without basis, Edmonds, 2021/this issue). We can use the sense-making as a heuristic for evaluating the comparability of instances.

I am essentially reformulating Speer’s closing argument (2002b, 2002a) that procedural consequentiality is the key on which the naturalism debate can hinge, at least for a discipline interested in empiricist claims. Member orientations answer the question of how to make sense of the data: They have done this work as part of their situated activity. From an ethnomethodological perspective, to treat researchers as invalid participants a priori is to deny all participants of their analyst capabilities, to treat them as cultural dopes, because all participants are analysts. The question instead should be in what way they treat their actions as “investigatable”. For members doing everyday social action, accounts are investigatable for “why that now” (Schegloff & Sacks, 1973, p. 299), but not “interesting” (i.e., worthy of research). As Lynch (2000, p. 47) writes, “the ‘essential’ reflexivity of accounts is ‘uninteresting’ and ordinary”; were participants to treat practices as “interesting” (in the sense of Garfinkel, 1967, p. 9), it would undermine their ability to achieve action. Accomplishing action requires that it be investigatable for sense, but not intensely scrutinized for all possible actions. It is the analyst who treats practices as “interesting”. Were an analyst to do so in situ in a recording, it would be difficult to treat them as a member for long. So, on the one hand, researcher-participation makes for a demonstrative version of Pollner’s (1991) radical referential reflexivity, wherein the fact that analysis is itself a sense-making and situated practice is incorporated into the analysis. Researcher participation achieves this in that it literally performs the fact that all participants are analysts already. On the other hand, reflexivity in situ can only be done in Lynch’s sense, i.e. as a member, or one risks undermining the activity at hand with research interests.

What this paper contributes is an expansion of what Speer hinted at: That a dual role as researcher and participant is possible, where one is oriented to as a member. This affords us needed opportunities to use ourselves as tools to uncover and display members’ uniquely adequate practices, especially in increasingly specialized settings, and as concerns difficult to access phenomena, including the rock climbing angles and embodied experiences in this paper, as well as multisensoriality. Researcher participation both augments (through additional participant perspectives and demonstration of unique adequacy) and complements (through access to otherwise unavailable experiences and development of skill) video-based analysis. If nothing else, we can move forward from here to systematically investigate how the researcher-participant is treated by other participants in interaction. But, in order to do this, we will need more researchers willing to step in front of the camera.

Acknowledgements

This work was supported by the Vetenskapsrådet [2016-0087]. Thank you very much to Yumei Gan for suggesting I write this paper and getting me to properly formulate some of these ideas. Thank you very much to the editors Yumei Gan, Sara Goico, Candy Goodwin, and Julia Katila for very helpful comments and superb organization and steering. Thank you to Leelo Keevallik for wonderful discussions about this work. Thank you finally to two anonymous reviewers for their excellent insights, to which I have done my best to do justice.

^{1 Garfinkel often had his students engage in ‘breaching experiments’ (or later, ‘tutorial exercises’) wherein the students would act as if they didn’t know or refused to uphold various expectably knowable ‘rules’. For instance, the students might constantly clarify words (‘what do you mean, you’re ‘fine’?), pretend they are a houseguest rather than the offspring of their parents, or place tic-tac-toe/noughts and crosses marks on the wrong part of the paper or when it was not their turn. The outrage and confusion that these exercises provoked made visible the taken-for-granted nature of social ‘rules’ that organize everyday life, and demonstrated how we are accountable for upholding those taken-for-granted expectations.↩}

^{2 They feel like marshmallows, squashy, until they bite.↩}

^{3 Taboo events are less likely to occur in general, or to be attributed as occurring, due to their accountability. Participants will avoid exposing taboo occurrences, even providing opportunities for their revision or withdrawal (Robles, 2015). Depending on the situation, swearing, innuendo, racism, sexism, lying or bluffing, intimate talk, and so on, may be avoided or done covertly in ways that are defeasible.↩}