Social Interaction

1. Introduction

Embodied synchrony is a form of intersubjectivity between corporeal beings or, captured in one word, intercorporeality (Meyer, Streeck & Jordan, 2017). Our aim in this study was to examine movement synchrony as a mutual, ongoing interactional achievement among participants in the so-called "mirror game" or "mirror exercise" (referred to in the Finnish data as peiliharjoitus). The goal in the exercise is that two moving partners achieve synchronous, mirror-like movement. We analyze this activity based on video recordings of a workshop focused on dance and movement improvisation that was organized for residents in a communical living block. The participants' orientations to movement synchrony become observable in how they managed dysfluency in the mirror task. This included making fluent embodied adjustments that (re-)achieved synchrony, using other kinds of trouble-orientations by which they increasingly distanced from the joint movement, as well as moments in which the "intercorporeal connection collapsed" (Meyer & Wedelstaedt, 2017, 5) as they abandoned the task prematurely.

The aim in the mirror excercise is to produce synchronous movement as if each partner were the mirror image of the other, without separation as to who initiates the movements. However, this is not attempted right away. During the first stages of the exercise, the participants are ascribed the roles of "leader" or "follower". The leader is expected to initiate movement, which the follower tries to follow as closely as possible by producing simultaneous moves that are similarly shaped and positioned. During the next stage, the partners are instructed to freely swap roles, and finally to give them up and just engage in the flow of movement. Eventually, it should become possible to move in synchrony without perceived separation in terms of who creates the movement, and experience it as jointly created.

Given the nature of the task, orientation to synchrony is in some sense a starting point for the activity, but it still needs to be constantly accomplished and reaccomplished in the course of its doing. As reported in various multimodal studies using detailed sequential analysis, "simultaneous" actions involve stepwise building by mutually responsive conducts that are micro-sequentially organized, and rely on being in some way projected (Broth and Keevallik, 2014; Mondada, 2018, 2022; Deppermann & Schmidt, 2021; Mondada et al., 2025, yu, 2022). This leaves "simultaneous" as a descriptive term for what can in reality be diverse temporal organizations (Mondada et al. 2025, 3). The mirror game brings to light micro-sequential organization as the participants manage dysfluencies in their joint movement through subtle, mutually responsive adjustments. In addition to making adjustments – in a certain local order in relation to them – participants use other resources in problem-orientation (gaze, facial expressions, talk), which exemplify the ways in which different modalities can intertwine in embobied activities (e.g., Keevallik, 2013). The situated, activity-specific relevance of certain interactional resources and modalities over others (Mondada, 2014b, 138) is illustrated by how the participants' use of certain sets of resources for adjustment and trouble-orientation in the mirroring task, and the ordering of these resources, connects to their enacting, or shifting between, their participant role as leading vs. following mover, improviser vs. workshop attendee, and student vs. instructor.

In the analyses we explore how the participans re-accomplished synchronicity when it was at stake, how they made adjustments within the boundaries of the task and of their role in it, how they gradually distanced themselves from or abandoned the embodied task, and how they employed different modalities in these orientations. The main focus is on the "leading" movers' means of other-adjustment to the follower by decelerating or suspending movement, or returning to an earlier movement phase (section 4.1). To investigate the limits of other-adjustment, we also pinpoint moments of task abandonment (section 4.2); with regard to the role- and action-specificity of the practices we also analyze the instructor's partly similar, partly different resources during corrective demonstration of mirroring with a partner (section 4.3), and the "follower's" self-adjustments (section 4.4). Along with the embodied adjustments, the analyses show how different embodied and verbal resources come to play when momentarily distancing from or ending an improvisation task.

2. Background

2.1 Movement synchrony and other-adjustment in improvisation

Research has shown that synchronized improvisation, such as the mirror game, creates feelings of togetherness among participants (Himberg et al., 2018; Feniger-Schaal, 2021; Hove & Risen, 2009). Himberg et al. (2018) found in their experimental study of a four-person mirror game that the participants' reported experiences of togetherness were associated with external measurements of synchrony. These findings suggest that the mirror game has potential as a context within which to study attunement and emergent intersubjectivity (Himberg et al., 2018, p. 10; Yun et al., 2012).

Research on movement synchrony has been criticized, ironically, for invoking models that treat bodies as isolated units controlled by autonomous individuals. Meanwhile, the concept of intercorporeality, as developed by philosopher Merleau-Ponty, has been proposed as a means of overcoming dualistic approaches and engaging with sociality and communication as truly embodied (Meyer, Streeck & Jordan, 2017, p. xviii, Meyer & Wedelstaedt, 2017). The participants in our data were actively practicing intercorporeal attunement to each other in creating and keeping up movement synchrony. The focus in this activity is on controlling one's body to find attunement, which thereby occurs as a member's problem, and a learnable skill. The analyzed examples thus constitute a special case in relation to the mundane, constant processes by means of which our bodies are constituted by our relations and interactions with other bodies (idem.; Stevanovic & Himberg, 2021).

It is worth noting that mirroring in this activity, or generally in social interaction (e.g. Szczepek Reed, 2020), is not about exact copying. The mirroring task rather creates a space in which to observe what the partner is doing and to respond by actions that could be locally interpreted as corresponding to it in some meaningful way.¹ Given the involvement of first-timers, the activities examined here are preparatory to more advanced creative processes (e.g., Sheets Johstone, 1981). Nevertheless, they share the core features of improvisation. It is not about doing something without rules or order but about the movers' embodied awareness, their ability to anticipate and react, modulate and engage with the flow of movement, which Sheets Johnstone discusses in terms of "thinking in movement". It is about the members' own dynamic definition of the action, and this emerging definition and redefinition creates the very improvisation: "(--) the process of creating is not the means of realizing a dance, it is the dance itself" (Sheets Johnstone, 1981, p. 399).

Our analysis of dance/movement improvisation draws on Weeks's (1996) study on musical improvisation, in particular on the notion of other-adjustment. Weeks analyzed how musicians cover co-musicians' mistakes by improvising other-adjustments, which change the musical score so that the adjustments become part of it, and thus unnoticable to an audience. In a similar vein, the leading mover's other-adjustments to the follower modify the movement to be followed, producing a "moving recontextualization" of its elements (Weeks 1996, p. 216). Weeks' analysis thus points the way to analyzing another specific activity and its time-ordering – an improvised dance – "as a members' doing". As with the musicians, the practices examined here do not necessarily enter the members' own verbal accounts, but they can also be (retrospectively) commented.

Using more recent conversation analytic terminology, the movers' adjustments can be understood in terms of micro-sequentiality. Mondada et al. (2025, p. 4) define micro-sequentiality as the organization of action responding to an ongoing action while it is in progress, to which the in-progress action correspondingly adjusts, in this sense, responding to the former. Responsivity such as this is also observable in the movers' adjustments to the partner's way of following their movement, which modify the movement in-progress. Achieving synchronous movement is therefore an interactional achievement that involves subtle responsivity between bodily conducts, and the ordering of temporally first and subsequent conduct.

The participants' other-adjustments (decelerating, suspending, and returning movement) could also be subsumed under what Lerner and Raymond (2021, p. 294) call "elementary forms of adjusting action": "accelerating an action, retarding an action, suspending an action, or reversing an action". Gradual adjustments to others in the pace and direction of movement have a role in the reflexive organization of social action (Mondada, 2014a, 2022). In the current context they are a means of accomplishing joint mirrored movement, while at the same time connecting to the enactment of specific roles in this activity.

Similarly to music or drama rehearsals (Weeks, 1985, 195; Schmidt & Deppermann, 2023), the workshop is characterized by alternation between "regular" and performance activity frames. Correspondingly, workshop participants shift between the student role (receiving instructions and chatting with others, for example) and the "improvisational role" (Keränen 2018) in the exercise. Studies on drama and dance have shown how such role shifts may be enacted bodily, involving changes in gaze direction or the speed of movement (Raevaara 2020, Keränen 2018, Savijärvi 2021) and relocating in space (Smart & Szczepek Reed, 2025). They may also involve verbally explicit role ascriptions (Smart & Szczepek Reed, 2025) or pronoun choice in person reference (Raevaara 2020).

2.2 Dance in multimodal conversation analysis

Dance instruction has proven to be a fruitful setting for examining instructional activities and multimodal interactional practices, including embodied demonstrations and corrective instructions (Keevallik, 2010; 2015). These settings render observable diverse relationships between talk and embodied action, such as hybrid bodily-syntactic units (Keevallik, 2013). Often instructed and practiced in groups, dance also includes complex participation frameworks: pairs following instructions together (Broth & Keevallik, 2014), individual dancers orienting to fellow dancers (Bassetti, 2023), the co-present facilitation of video dance instruction (Kosurko & Stevanovic, 2023), as well as rhythmical coordination between dancers and viewers (Albert, 2015). Participants thus manage layered multimodal orientations to the instruction and to the dyadic and group activity.

Coordinating synchronous movement in dance instruction may involve combinations of talk (count-ins, syntactic structures) and bodily action intertwined with specific rhythms and embodied and spatial configurations, which can serve projection of upcoming action and thereby contribute to synchronization (Broth & Keevallik, 2014). Ehmer (2021) describes two types of synchronization among teacher-student pairs in tango lessons: emergent synchronization occurs when students join the instructors' movement based on anticipation, whereas in orhchestated synchronization, joint movement is pursued by the instructor from the beginning of the movement.

There is also increasing interest in how practitioners in other embodied activities such as sports orchestrate reciprocal movement among themselves by orienting to the projectable movement trajectories of partners (Lefevbre, 2016 on Aikido; Okada, 2013 in boxing; see also Meyer & Wedelstaedt, 2017). The movement material is less pre-defined in dance/movement improvisation, however, and may therefore differ in its projectable properties, while also relying on mutual monitoring and responsivity. The present study joins the line of research focused on coordination among peers, while we also consider the presence and involvement of the instructor.

3. Data and Method

The study is part of the Neighbour Dialogues project, which brought together people with different backgrounds to investigate the creation of feelings of togetherness and intersubjectivity in dialogue when doing or discussing art (Voutilainen, 2021; Rossi, 2021). The project organized dance improvisation workshops run by a professional dancer/dance pedagogist for residents of a communal block of houses. Much of the "dialogue" in the improvisation is between bodies, but as we will show, talk also has a role. The participants generally had no previous background in dance, and were of diverse ages. The group included people both with and without visible disabilities, which connects the project to the theme of inclusivity in dance (Burridge & Nielsen, 2018). All the participants gave their written informed consent.

The data gathered for this study comprise eight hours of recordings in eight workshops. From an initial set of 56 instances of dysfluency in the mirror task (identified by research assistant Iiris Joronen), we have analyzed in more detail eight segments including leaders' other-adjustment (two by the instructor), and six including a follower's self-adjustment. The instructor was either observing the pairs from the side or acting as someone's partner, occasionally giving instructions while moving. In this complex participation framework, the teacher is engaged in multiple tasks at the same time, and the participants have a certain amount of autonomy in organizing their action and assessing it by their own definition, while also laterally orienting to the teacher, as well as to other pairs. The instructor announces the start and end of the exercises, meaning that if participants abandon a task before the instructor announces its end, they can be deemed to have ended it prematurely.

As movers becoming acquainted with the mirror game, the participants observably experienced moments of dysfluency in the task. The follower might not have noticed a movement that the leader was making, and thus failed to mirror it, or followed after some delay or with the "wrong" arm or leg. Trouble also occurred in switching the roles of leader and follower. In principle, there is no right or wrong in the task, but in practice there is some normativity, to which the instructor as well as the participants orient to.

We used multimodal conversation analysis as our research method. Most of the data were transcribed using the conventions developed by Mondada (e.g., 2018). Excerpts involving (almost) no talk are, however, presented only as screen captures with time intervals indicated between each figure. This was done to anchor the unfolding of movements in visible body postures instead of chunks of time.

4. Analysis

Enacting the "leader" role in the mirror game means initiating movement. Analysis of the data reveals that participants in the leader role make other-adjustments to adapt to their partner's way of following them by means of decelerating or suspending movement, and by returning to an earlier position, which enable the partners to reachieve synchrony. In terms of social action, decelerating and suspending could be understood as waiting for, and returning as actively catching up with, the partner. In the following we examine the leaders' other-adjustments and how other resources came into play, and then illustrate the instructor's and followers' orientations to dysfluencies in the mirror task.

4.1 Leaders' other-adjustments

In the first case, in the course of the mirroring task, the leader adjusts to the follower's delayed following of her single step by suspending and delecerating the movement. Temporally extending a single step also decomposes it in terms of its phases becoming more distinguished (raising the heel, sliding the foot on the floor, and lowering it).

Extract 1. Step (Tanssi-impro – 2. Kerta_ GoPro, 32:30–32:40)

The step starts off from an asymmetrical position with TII's feet wider apart, her right foot one step ahead of RAI (Fig1.1). From this position, TII takes a step, and during its course we observe her suspending and decelerating the movement as a way of adjusting to RAI's delayed following.

Soon after starting to lift her left heel, TII monitors RAI by glancing at her (Fig1.2) (e.g., Mondada et al., 2025) and suspends the lifting for a moment. She glances only once during the stepping, from the corner of her eye and without turning her head. This illustrates the constraints built into the "leader" role: turning the whole head could constitute a movement to be followed. The glance towards RAI occurs at 2.3 seconds into the excerpt. TII started lifting her heel about half a second after the situation in Fig.1, which makes the time of non-following preceding the glance considerably long (1.8 seconds). TII's monitoring glance can thus be interpreted as an orientation to RAI's non-following. RAI, for her part, looks at TII's feet, from Fig.2 onwards, during the whole transcribed segment. RAI's gaze therefore seems to monitor the movement to follow it, not to orient to a particular trouble.

After the glancing and the brief suspension, TII brings her heel to the relatively challenging upmost position, in which she is able to take support from the floor with her bent toes (Fig1.3). When TII resumes the heel movement, RAI has about one second to follow it, but Fig1.3 shows that she keeps her heel immobile in a slightly elevated position. Only when TII straightens her toes and starts sliding the tips forward on the floor does RAI finally lift her heel higher (Fig1.4). TII continues to slide her toes at a pace that we consider decelerating to adjust to RAI's delay. RAI is wearing shoes, which prevents her from making a similar sliding movement. Instead, she advances her foot by slightly detaching it from the floor and placing the tip of the shoe on the floor a little further on. In Fig1.6, we see that TII has slid her foot almost as far as it can go, while RAI has only advanced a little. From Fig1.6 to 1.7, TII has arrived at full extension of the leg, with no more space to go further. The sliding took over three seconds (the time between Figs1.4-1.7). RAI has again advanced further, yet she lags behind in terms of the length of the step.

Only when RAI has advanced for a third time and reached the distance of a full step (Fig1.8) does TII start to lower her heel to finish the step. RAI does not similarly decompose the step (e.g. by lowering her heel) but places her entire left foot directly on the floor. Therefore she cannot mirror TII's final heel movement (Figs1.8-9). Yet, TII and RAI reach movement synchrony in the final phase of the step in terms of weight shift. Whereas TII is distributing more body weight over the left foot when landing it on the floor, RAI is distributing body weight on her left foot while standing, and both sway forward (the change in RAI's position is visible from her figure against the window pane, see Figs 1.8-9). Fig 1.9 shows them having arrived with their body weight somewhat evenly distributed over both feet.

In terms of timing, RAI thereby lifts her foot off the ground for the first time (Fig1.5) only about 5.5 seconds after TII has initiated the step (by shifting her weight to the right foot and detaching her left heel from the floor). The beginning of TII's other-adjustment involves micro-sequentially organized mutual conduct in that the "leader" suspends movement and monitors the non-following mover, who continues to non-follow, and the "leader" responds by modifying their own movement (producing a very slow step, which could be quite challenging for maintaining balance). Resembling musicians' other-adjustments in covering up others' mistakes (Weeks, 1996), the ongoing improvised trajectory is itself modified by the other-adjustment. The step is transformed by the increase of time spent on physically producing it, which manifests as a decomposition of its phases. The resulting sub-phases could provide micro-initiations of movement within the larger step (detaching, sliding, lowering), to which the follower could respond by mirroring them – although in the current example clear synchrony is achieved only in the final body sway.

The subtleness of the adjustments, particularly regarding pace, makes some of them challenging to identify and analytically evidence in the data, yet they are clearly relevant to the participants. Excerpt (2) presents a more concretely observable form of other-adjustment in the data: the leader returns to the position from where a movement started, as a way of offering the partner a new opportunity to mirror it (on return practice in demonstrations during Budo sports teaching, see Råman & Haddington, 2018).

We join the participants, Tiina (TII) and Sonja (SON), when they are rotating both of their hands outwards (in anatomical terms, supination). SON has been appointed the leader. Paavo (PAA, the instructor) explains (line 1) the next phase in the task, in which the roles of leader and follower are no longer fixed. The role switch is supposed to happen spontaneously while moving. PAA is simultaneously engaged as one participants' partner in the task, and does not currently follow up on what others are doing. Dysfluency emerges when TII makes a sudden movement changing the direction of the rotation inward (in anatomical terms, pronation) (line 4). This appears to be an attempt to follow the instruction to switch roles immediately by spontaneously taking the lead, which would make it relevant for SON to follow. She does not, and we can witness how TII returns her hand to its earlier position to adjust to her non-following.

Extract 2. Rotation (Tanssi-impro – 1. Kerta_ Kamera2, 23:06–23:30)

Fig2.2 shows SON engaged in a previous movement, while TII initiates a new movement by rotating her right hand inward. When SON does not mirror it, TII rotates her hand in the opposite direction (outward), returning to the movement's starting position (Fig2.3). This was the last position in which they were still "together" (see Fig2.1). SON responds by starting to extend her left arm to mirror the arm position (Fig2.4). Once she arrives at it, TII re-initiates the inward rotating hand movement. This time, SON responds with a small 0.3 s delay (line 7, Figs.2.5-6). In other words, after the partner has missed a unit of movement – probably due to having so far been leading and therefore not anticipating the need to follow – the new leader returns to where they first departed from the joint configuration, and the movers re-achieve synchrony, thereby also establishing their new roles.

At the end of the successful joint rotation of hands, TII is holding her right hand still, when SON now takes the lead (l. 8). She initiates a rotation outward with her left hand. TII joins after 0.4 s, and the joint movement continues for more than one second, ending in stillness and in SON slightly lowering her hand (l. 8). They have now performed a full there-and-back rotation, with TII leading the inward and SON the outward moving phase, and their hands have reached the home position.

After this successful restoration of motor synchrony, TII makes the verbal comment tuntuu nii väärältä, '(it) feels so wrong' (l.9), which co-occurs with her one more initiation of a similar movement. SON now promptly joins her. Both smile, and TII lifts her gaze directly into SON's eyes. (From the camera view, it is not possible to verify whether SON gazes back.) With her verbal remark and gaze shift, TII moves away from the immediate mirroring task to establish contact, not at the level of bodies moving together but as workshop participants who can evaluate the joint ongoing activity. The turn is quiet, designed for the dyadic interaction. They do not evaluate the action as successful or not, but in terms of how it feels. At the same time the comment offers an understanding of the earlier dysfluency. Her turn is hearable, on the one hand, as an account of her insistence to try out the movement several times, and on the other hand for the lack of SON's following. The characterization of the movement as "feeling wrong" may imply both that it is extraordinary and intriguing (a motive for her to re-experience it), and that mirroring it may be less fluent.

In the excerpt, the leader adjusts to non-following by returning to the last joint position, from which the movers then resume joint movement. This is coordinated through responsive conducts (initiation–non-following–return and new initiation–following), which modify the leading movement in that it becomes a repeated movement – which is later accounted for. The case also shows the movers' other resources, beyond movement itself, to deal with the lack of synchrony in the mirror task. TII smiles and gazes into the eyes of the movement partner when acknowledging and dealing with the prior hitch in the task. It is telling of the role of talk in this context that verbal comments mitigate and/or account for the trouble only retrospectively once it has been resolved in other ways. In the current excerpt, the participants remain in the task and keep moving despite also establishing contact on another parallel level through joint smiles and talk. The next section shows how such conduct may amount to prematurely ending the task.

4.2 Abandoning the task prematurely

Our focus in this section is on cases in which one or both partners end up disengaging from the joint movement task before the instructor announces its ending. Prior to excerpt 3, Terttu (TER) and Leena (LEE) verbally agreed that Leena was the leader. After Paavo gives the instruction to switch the roles (line 1), confusion gradually emerges regarding who is leading and who is following. The coordination problem is displayed (line 12 onwards) by the participants' bodily conduct, gaze, facial expressions and laughter during the task and, retrospectively, by LEE's verbal comment (l. 19, 21).

Extract 3. I do like you do (Tanssi-impro – 1. Kerta_ GoPro, 22:00–23:05)

TER and LEE are moving their arms upwards when PAA, engaged in partnering, gives a new instruction to switch the leader. TER and LEE stop moving (l. 2, 3), and LEE gazes in Paavo's direction. RAI, a participant in another pair, comments that they had already done several switches. According to the new instruction, TER should logically become the leader, as LEE has been leading so far. However, the subsequent events suggest they are not in full mutual understanding of the instruction or how to apply it. It is also not clearly verifiable to the analyst from their movements who is now leading. LEE seems to initiate small movements with her hands and fingers and TER to follow them (lines 6 and 7), but then (line 8) they continue lifting their arms with smooth and continuous movements alternating with brief moments of stillness and small, jerky movements that are difficult to recognize in terms of the ongoing action.

Next, however, TER clearly initiates the higher lifting of hands (line 9), and LEE follows slowly. However, when TER stops the movement (see Fig3.3), LEE continues past the height of TER's hands, near the face (Fig3.4). This could be LEE spontaneously taking the lead by initiating movement. TER lifts her hands a little (l. 11), but not enough to display following LEE's hand position. The lack of synchronization in their movements may be due to TER being distracted from the task: she has been gazing away slightly. LEE seems to engage in other-adjustment by stopping for 1.1 seconds, then raising her hands slightly, and again staying still (l. 11-12), but TER does not follow.

In a situation in which TER has not mirrored LEE in moving the hands close to the face (coming to l. 12), TER suddenly initiates a new movement by flexing her fingers (see her flexed fingers in Fig3.5 in contrast to Fig3.4 with straight fingers). LEE does not follow. LEE's facial expression is not visible from behind her hands, but TER seems to reciprocate it by smiling, first slightly and then more widely. LEE soon (l. 13) displays her puzzlement on the task's unfolding. She relaxes her right hand, visibly dropping out of the task, and engages in a three-part gaze trajectory. First, she shifts her gaze to TER's hands (Fig3.6), then quickly on TER's face (Fig3.7), and then glances at PAA (Fig3.8). While doing this she has a dumbstruck facial expression with round eyes and the corners of her lips pulled down (Fig3.9). TER reciprocates LEE's gaze shift to PAA by also glancing at him, and in Fig3.9 we see TER laughing. LEE now flexes her fingers, mirroring TER's hand shape, and starts lowering her arms (line 14). In doing this, she assumes the follower role.

With their gazes, facial expressions and laughter, the participants orient to confusion regarding who leads and who follows. In terms of movement, it was about whether TER should mirror the placing of the hands across the face or whether LEE should mirror TER's finger flexing (both visible in Fig3.5). LEE's glance at the instructor indexes his role as the authority. Orienting to his presence laterally (Clark & Carlson, 1982, p. 336) may either be looking for cues as to what to do, or playfully checking if the teacher saw the "mistake". From here on, they lower their arms in synchrony, TER leading. As in Excerpt 2, the trouble is resolved without talking.

After about three seconds, PAA moves on to the next phase in the task, namely to switch the roles of leader and follower 'on the fly' (l. 16). PAA's subsequent talk ('let's continue') is directed at his own partner MAI, who dropped out of the task once PAA started talking. LEE, too, treats the moment when the instructor starts to talk as an occasion to abandon the embodied activity. She releases her arms, points at TER (Fig3.10) and says elikkä mää teen nyt niinku sinä teet, 'so I do now like you do', thereby ascribing herself the follower role. By stating her understanding of the current roles, she orients to the earlier confusion surrounding them. Notably, she does this only when the issue had been resolved by embodied and some vocal means (laughter), and they had resumed the task. Although LEE withdraws from the mirrored body configuration when talking, TER maintains it by keeping her arms bent back (Fig3.10) while responding with minimal confirmation. LEE then re-engages in the task by similarly swinging her arms back (l. 23).

The above case illustrates the complexity and fragility of the mirroring task in terms of the sometimes vague recognizability of whether a movement is to be interpreted as an act of leading or following, which again affects the projectability of subsequent actions. Making the spontaneous switching of roles smooth requires not to rely too much on what the current trajectory in leading the movement projects, as the movers need to be ready to change their responsive conduct instantly to act as a follower.

The participants in excerpt 4 abandon the task without the presence of any external distractions. This case exemplifies a characteristic situation in which participants end up making a silly movement and start laughing, sometimes also verbally commenting on the incident, which leads to at least momentary abandonment of the task. The laughable trouble-source in this case again reveals the constraints of the leader role: anything they do is potentially follow-able. The series of still images below illustrates how TER (on the right) innocently adjusts her clothing, first the hem and then the neck. LEE mirrors this action (Figs.4.2-3), which TER apparently did not mean to be followed. Noticing LEE's mirroring of it, she bursts into laughter, and LEE joins in (Fig4.4). After laughing for some time they resume the task by taking one more step towards the window, but then again abandon it and turn to face the center of the room, with broad smiles on their faces (Fig4.5).

Extract 4. Laughter (Tanssi-impro toinen kerta GoPro 28.38-28.42)

These incidents were not part of the workshop agenda, as evidenced by PAA later in commenting on the performance of all the pairs, saying that this time the task fell apart as a result of the participants losing their focus. Nevertheless, given the aim of the Neighbour Dialogues project to foster encounters between residents, these moments of shared laughter may be as valuable as any feeling of attunement during the dance. In these shared affective moments the participants connected in their role as workshop participants who were engaging together in a new and unfamiliar activity, and who were able to take a playful stance towards the activity and the possible mishaps that came with it. Their embodied conduct was synchronized on a different level, in laughter and smiling, which encompassed their entire bodies. Their co-creativity grew somewhere between compliance and creativity in performing an instructed task (Kosurko & Stevanovic, 2023).

Abandoning the task after such incidents may initially be momentary. However, it seems that after fully dropping out it is hard to restore the focus, and after one or two attempts the participants characteristically end up abandoning it altogether.

4.3 The instructor's other-adjustment and correction

The role of the instructor (PAA) in the excerpts so far has been to partner with a participant while guiding the exercise, or quietly observing it from the outside. In excerpt 5, we examine his demonstration of the mirroring task with a partner as a separate activity. His resources for orienting to dysfluency in joint movement during the demonstration are partly similar to, yet differently ordered from the participants' resources when engaged in the task.

In this variant of the mirroring task the participants are positioned side by side with the pair, moving around in the space. This is more advanced and demanding than in the earlier face-to-face version. Mirroring side-by-side requires laterally paying attention to the partner's whole body, instead of focusing mainly on arms and hands when face-to-face. Before the segment to be analyzed, the participants have had difficulties in keeping on the side of their pair while moving. PAA engages in a corrective demonstration with RAI. He demonstrates how to lead simply and slowly enough for the pair to follow it.

Extract 5. Easy (Tanssi-impro – 2. Kerta_ GoPro 30:06–30:29)

At the start of PAA's demonstration of how to move in sync, both he and RAI are gazing forward. Shortly, after PAA's turn and the lifting of his right heel, RAI turns her head slightly to the left. She gazes laterally up at PAA's shoulder level (Fig5.2). She does not visibly focus on PAA's feet but appears to monitor the movements of his upper body. After almost a second, when RAI has re-directed her gaze forward, PAA continues the step by sliding his foot on the floor for 1.6 seconds, but decelerates and finally stops the movement (l. 4, Fig5.3). Keeping his foot still, he now gazes and points at RAI's feet (Fig5.4) and gives the verbal corrective instruction sun pitäs liikuttaa eteen(päin), 'you should be moving ((it)) forward'. In other words, when decelerating and stopping to wait for the follower does not succeed, he resorts to talk, and thereby shifts from enabling the emergence of synchronization to actively pursuing it. This order within the overall corrective demonstration reflects the distinctions in Ehmer (2021), in that according to their study, emergent synchronization is more typically related to corrective instruction than actively pursuing it. PAA holds his body position while talking, and resumes sliding his foot before the end of the utterance.

RAI orients to PAA's pointing by looking at their feet, and responds to the corrective instruction by bringing her foot forward (l. 6). Note that she is wearing shoes: here she lightly rubs her shoe against the floor to make the sole slide on it (detail not shown in the transcript). When RAI advances her foot, PAA is already lowering his right heel to finish the step, and when RAI similarly comes to lowering her right heel PAA is already initiating a new step with his left foot, by raising the heel off the floor (l. 7, see Fig5.5). Their movements are thereby severely out of synchrony. RAI does not mirror the new left-foot step, instead still gazing at PAA's right foot. During a trajectory lasting 3.6 seconds, PAA first slowly lifts his left heel, stops the movement, again slowly advances, then stops (l. 7-8). Finally, he gives another corrective instruction (l. 9-> kato, jos sä… 'look, if you…') pointing at their feet. After this, they do not return to continue the demonstration.

The excerpt shows how PAA first uses other-adjustment in waiting for his partner, but when this does not succeed, resorts to verbal corrective instruction, stepping out of the task mode and putting the movement trajectory on hold. As revealed in earlier excerpts, students tend to resolve such situations by means of embodied adjustments, and if they fail (or in retrospection), incorporating other embodied and vocal resources. Talk was their last resource, and it was never corrective.² Unlike the instructor, the participants were embracing a symmetrical position, which did not warrant verbally correcting the other. The set of cases thereby shows how resources for achieving synchrony at moments of dysfluency are specific to the movers' role in the activity (see, e.g., Ehmer, 2021, p. 2).

This ordering of embodied and verbal modality is also evidenced in the fact that in the data, the turn to verbal "debriefing" seemed to be occasioned by the instructor starting to talk. During the exercises, PAA announces transitions to next phases without pausing the embodied activity. The participants have difficulty remembering not to let go of the task, and in several instances they abandon it immediately when PAA starts talking (see excerpt 3). For them, the verbal modality is not parallel to the embodied task in those instances, but alternates with it (Mondada, 2018).

To elaborate further on the resources for adjustment in different participant roles, in the next section we show the types of adjustments in which followers engage.

4.4 Follower's self-adjustment

According to PAA's instructions, all leading movements should be slow to make them easy enough for partners to follow. In most cases the leaders adhere to this. For example, the return movements (Excerpt 2) are done slowly, and could in principle be taken as part of the trajectory to be followed. In other words, other-adjustments are designed to be interpretable within the "moving recontextualization" (Weeks, 1996) of the movement trajectory. The follower's adjustments, in turn, are characterized by abruptness, thereby not incorporated in the ongoing trajectory.

In Excerpt 6, TII (on the left) appears to notice that she has been using the wrong arm to follow RAI, as she quickly changes from raising her right arm to raising the left.

Extract 6. Wrong arm (2. kerta GoPro, tii ja rai 28.49-28.50)

Prior to switching the arm that follows, TII verbally displayed trouble in a single word, ei, 'no'. Notably, this occurred after the parties had abandoned the task for about a minute and then re-continued it for 20 seconds by taking a couple of steps. In contrast to the leader's slow-paced adjustments shown earlier, TII self-adjusts by switching the arm quickly, in only 0.7 seconds. This movement was not designed to be part of the mirroring trajectory, but to "cancel" the current position, and reposition the body to better correspond to that of the leader.

Another type of adjustment made by followers is to change the spatial configuration between themselves and the leader to secure space for performing the movement without the bodies colliding. In Excerpt 7, the follower takes a step to the side to allow stretching out arms in a face-to-face mirror task. TER and LEE raise their arms, and LEE takes one step to her right. The increase in distance between herself and TER is visible in relation to the dotted line.

Extract 7. Making space (kerta GoPro 24.10)

The followers swiftly adjust their own movement or positioning in space to allow the mirror task to continue fluently. They thereby adjust to what the leader is proposing with their ongoing movement by ameliorating their own conditions for mirroring it. In both examples of self-adjustment it takes less than one second, which is relatively fast compared to the extended time that other-adjustments take in earlier excerpts. In cases of other-adjustment the leader ameliorates the other's conditions for rejoining the movement. In terms of micro-sequential organization, self-adjustment does not similarly invite a change in the partner's ongoing action, but allows its maintenance. However, its influence is to some extent reciprocal: making space for oneself to mirror a raising arm also enables the other to continue the current trajectory, which they might otherwise need to transform.

In sum, resources for managing dysfluencies in the mirror task are differently distributed between the "leading" and the "following" mover: temporally extending and slowly returning movements is the leader's resource, whereas abrupt fixes are mainly the follower's. This is also evident elsewhere in the data, when the instructor sanctions leaders for making abrupt and therefore "too difficult" movements (prior to corrective demonstration in Excerpt 5, not shown). The leader's abrupt fix in Excerpt 4, which was not meant to be followed, became a source of humor. The type and pacing of movement is therefore intimately connected to its recognizability as to-be-followed or not. Consequently, it is also connected to the participants' enactment of their complementary roles as leader and follower, which became a source of confusion in Excerpt 2, and even more so in Excerpt 3.

5. Discussion

There is a saying in dance that a good leader is a good follower. This means that the leader also needs to follow in the sense of being attentive and adjusting to the partner. The other-adjustments examined in this study are an example of how this can be practically achieved during the mirror game. The analyses show how the mover in the leader role decelerates, suspends, or returns to an earlier movement phase to adjust to the partner's following (sometimes related to a confusion regarding the current roles), creating opportunities to resume joint movement. The characteristics of these practices, e.g., being produced in a way that makes them part of the follow-able trajectory, were made more salient by a comparison to the followers' quick self-adjustments.

The improvisation task has no predefined movement at which the participants should perform and that would thereby constitute "correct" action, but there is an ongoing negotiation in terms of what is "similar" enough in time and shape. This manifests "moving recontextualization" (Weeks, 1996) in the sense that the ongoing action is modified by the way the leading mover attunes to the follower. This becomes particularly visible at moments of dysfluency. Consequently, the presence of synchrony, and its disappreance, comes across as a fluid and gradual phenomenon and a mutual, practical achievement of the dance partners. In negotiating staying in the task vs. stepping away from it at moments of dysfluency the participants display what kinds of temporal and embodied conducts fit in their ongoing definition of joint synchronious movement. This negotation thereby exemplifies the "time-ordering" of a specific activity "as a member's doing" (Weeks, 1996).

In line with the view of Sheets Johnstone (1981, p. 399) on creativity in dance improvisation, embodied coordination during the mirroring task is not only about producing a given movement, but about embodied awareness and mutual responsivity constantly redefining the ongoing actions – this dynamic, in itself, is the improvised dance. Success in the activity of mirroring, such as when the movers are in sufficient synchrony, is thereby flexible. There is space for leader adjustment in terms of slowing down, suspending or returning movements, which change the timing and details of the current movement to be followed, and indeed become it (Weeks, 1996).

The study has suggested that improvisation includes micro-sequential ordering of embodied action (Mondada, 2022, p. 40–42) in the sense of the partners' responsivity to each other, which reflexively shapes the ongoing trajectory. A question that emerges regarding micro-sequentiality in the mirror game concerns what are the "sequential" units that host this micro-organization. In responding to each others' movement (or lack of it), the participants establish what the relevant, follow-able units are. Raising a heel or rotating a hand could become mutually recognizable units of action, and accordingly project further movement trajectories: note that participants may also become accountable for (not) producing them. Achieving movement synchrony, and members' ongoing definition of what exactly is being synchronized, thus involve the type of micro-sequential organization that has been shown for other "simultaneous" actions in interaction (e.g., Mondada et al. 2025). The subtle initiations and responses in relation to the partner enable their staying in the task without engaging in a separate repair sequence. In contrast to this, repair was evidenced in the instructor's corrective demonstration. Future research could explore further the relationship between gradual adjustments orienting to dysfluencies, and repair organization (Lerner & Robinson, 2021).

The participants handled synchrony by embodied means. Additional resources came into play in trajectories of noticing and displaying dysfluency and orienting to it on diverse levels: gaze at a body part -> gaze into the partner's eyes -> gaze at authority; smile -> laughter -> facial expressions -> retrospective verbal comments. Verbal comments accounted for dysfluency in a no-fault (Heritage, 1988, p. 136) and humorous quality. By prioritizing embodied resources, the participants maintained the embodied task frame, while by using other resources they coordinated their attention and stances towards the embodied details. In the latter case, they distanced themselves from the task in a stepwise manner, enacting shifts from the improviser role to the workshop attendee role (Keränen, 2018). The shifts could be momentary or, in the case of abandoning the task, more permanent.

The instructor used these resources somewhat differently, easily turning from other-adjustment to the authoritative action of correcting the other's embodied conduct. Talk has a more central role in this action in singling out the embodied details (Keevallik, 2010). The instructor's treatment of talk and embodiment as parallel or tightly successive (Mondada, 2018) is also shown through the action of (non-correctively) instructing during ongoing movement. By comparison, participants characteristically oriented to the instructor starting to speak by dropping the embodied task. Their orientation to talk and embodiment as alternate (instead of parallel) modalities may relate to the ordinary feature of embodied responses as deliverable once the initiating embodied action becomes recognizable, and verbal responses rather orienting to the completion of prior actions (Lerner & Raymond 2021, p. 280). The analyses suggest that the instructor-style parallelism of modalities (e.g. talking or listening without abandoning the embodied activity – with their micro-temporalities), is a learnable skill that participants could appropriate during the workshop. The study has thereby explored the mirror game as an interactional ecology that hosts diverse orders of multimodal resources (Mondada, 2016, p. 341; Mondada, 2014b; 2018; Keevallik, 2015, cf. Stevanovic & Monzoni, 2016). It has also illustrated how the movers' resources in (re)achieving movement synchrony are role- and activity-specific and connected to the "regular" and performance frames in the workshop.

Finally, the key to synchrony in the mirror exercise is not the perfect copying of movement, but becoming fluent in mutual adjustment. This seems to be the pathway to a state in which nobody knows who initiates the movement, which may thereby be experienced as jointly created.

^{1 For this description, as well as much insightful discussion on the data, we thank the participants of the "Creativity data session" on March 24th, 2023: Saul Albert, Anna Heino and others, and organizers An Kosurko, Chris Leyland, and Joseph Webb.↩}

^{2 Other-adjustments do not necessarily disrupt progression of the action – on the contrary, they enable it through modification. Thus we do not primarily conceptualize them as repair. Moreover, as Lerner & Raymond 2021 argue in cases of 'body trouble', because embodied conducts are not organized one-at-a-time, as talk is, their remediation may not be best described in terms of repair organization, but rather considered on its own terms. Fluent other-adjustments nevertheless orient to dysfluency in the task. They could be placed along a continuum with trouble-orientations that do break away from the ongoing embodied trajectory (gazes, laughter, retrospective comments in other-adjustment, and abrupt movements in self-adjustment). At the end of the continuum is the instructor's explicit correction, which, as we have discussed, is differently organized.↩}