A multimodal analysis of turn-taking in interpreter-mediated psychotherapy

Although interpreting in mental health care has received some scholarly attention over the past two decades, the multimodal organization of such encounters has not been investigated in detail so far. This paper highlights two types of turn-taking problems that can occur in interpreter-mediated psychotherapy from a multimodal perspective. Based on a dataset of video recorded psychotherapeutic sessions with refugees, the study aims to examine the role of nonverbal resources (especially gaze) in the negotiation of turn-transitions between the interpreter and the primary speaker in two interactional contexts: turn-transfer during extended turns and in the management of overlapping talk. The data were analyzed qualitatively by drawing on the insights from Conversation Analysis (CA). The analysis shows how interpreters use gaze direction to signal their intent to take the floor and to manage rights to the conversational floor (turn-yielding and turn-holding). The paper also demonstrates how problems in the coordination of turn-taking can result in loss of information. In sum, the present analysis points to the role of gaze in the management of speaking rights and emphasizes the multimodal nature of turn management in psychotherapeutic talk with an onsite interpreter.


Introduction
Talk is one of the building blocks of therapeutic counseling. Through the process of talking, patients gradually open up about their experiences and establish a shared ground of mutual understanding with the therapist (Peräkylä, 2013). The question is how therapeutic work is organized when the therapist and the patient have no understanding of each other's language.
Within the context of a growing number of refugees, many industrialized countries offer mental health care programs designed for refugees to help them recover from the trauma of forced migration (Miller et al., 2005; see also Bot, 2005;Tribe & Keefe, 2009;Ticca, 2018). Such mental health care programs usually rely on interpreters to enable communicative contact between the therapist and the patient. Needless to say, the interpreter's presence inevitably changes the interactional dynamics of a typically dyadic therapeutic setting into a triadic constellation (Bot, 2005). Studies have shown that, besides translating language, interpreters perform multiple tasks during the therapeutic encounter, such as turn management and meaning negotiation, and thus have a significant impact on the unfolding of the therapeutic session (Anderson, 2012;Bot, 2005;Cornes & Napier, 2005;Llewellyn-Jones & Lee, 2014: Miller et al., 2005Tribe & Keefe, 2009;Bot & Verrept, 2013;Ticca, 2018). Being a participant who actively takes part in the exchange, the interpreter's conversational needs are different from those of the patient and the therapist, or 'primary' participants. First, unlike the primary interlocutors, the interpreter has to memorize what has been said in order to be able to render it in the following turn (Flores, 2005). A common 'strategy' of the interpreters is therefore to take the turn as soon as the opportunity arises (Englund Dimitrova, 1997). Second, whereas in spontaneous conversations with more than two persons the order in which one speaks is not predetermined (Sacks et al., 1974), in interpreter-mediated encounters the interpreter typically takes every second turn to render the previous speaker's utterance in the target language. It is then important that the interpreter has enough speaking space in order to be able to provide the rendition to the other participant (Englund Dimitrova 1997). Nevertheless, the interpreter's speaking space can be impeded through simultaneous talk and interruptions by the primary participants, which can lead to omissions, loss of information and even 'communicative breakdown' (Bot, 2005; see also Flores, 2005).
Little is known about how interpreters negotiate moments of turn transfer in the context of therapeutic talk from a multimodal perspective. The aim of this paper is to examine the role of multimodality, i.e. the combined use verbal and nonverbal practices (such as gaze and gesture) in the management of problematic turn-transfers in therapeutic talk. In that way, our study aims to contribute to an ongoing 'multimodal turn' (Davitti & Pasquandrea, 2016) in dialogue interpreting research. The analysis is based on 3 video recorded therapeutic consultations, that were examined by taking the interlocutors' verbal and nonverbal behavior into account.
In what follows, we present an overview of research on the role of embodiment in the regulation of turn-taking. We then briefly discuss the data and method used for this study. The remainder of this paper examines the role of multimodality in the negotiation of turn transitions and speaking space -in therapeutic talk: first in the context of long multi-unit turns and then in the context of overlapping talk -In the concluding part of the paper, we discuss the implications of this study for our understanding of the organization of turn-taking in interpreter-mediated talk, and, at a more general level, for interpreting practice.

On the role of multimodality in the regulation of turn taking
One of the basic principles of conversation is that speaking rights are restricted to 'one party at a time' (Schegloff, 2000). While taking turns at talk, interlocutors generally orient to minimizing gaps (no one talking) and overlaps (several people talking) between their turns (Schegloff, 2000;Mondada, 2007;Oloff, 2012). Thus, once the speaker has got the turn, (s)he generally has exclusive rights to it until the first transition relevance place (TRP), i.e. the moment in the talk where the transition to a next speaker becomes possible (Sacks et al., 1974).
Such moments of possible completion are usually projected in advance through various resources, such as syntax, prosody, the type of action in progress (e.g. question, elaborate tellings) and embodied cues, which enable the next speaker to prepare their turn (Clayman, 2013). The actual transfer of speakership is interactionally negotiated between the current speaker and the listener in one of the following ways; (a) the current speaker may select the next speaker ('current-selects-next'), or (b) another speaker may self-select to produce the next turn (Sacks et al., 1974). One particularly important resource for selecting the next speaker is speaker's gaze (Kendon, 1967;Goodwin, 1981;Heath, 1986;Stivers & Rossano, 2010).
Previous research has shown that speakers tend to gaze away at the beginning of their turn and gaze back to the recipient toward the end of their turn, which indicates that they are ready to hand over the floor (Kendon, 1967;Duncan, 1972;Bavelas et al., 2002;Auer, 2017). In multiperson interactions gaze at an interlocutor appears to be an explicit way of selecting that person as addressee (Goodwin, 1981;Lerner, 2003). At the same time, recipients who are being addressed are usually also expected by the speaker to display their availability and orientation to the ongoing turn by gazing at the speaker (Goodwin, 1981;Oloff, 2012).
As for self-selection, the next speaker can claim incipient speakership through both verbal and nonverbal resources such as appositional beginnings, audible inbreaths, gaze orientation and gestures (Hayashi, 2013). Recent years have witnessed a growing interest in nonverbal resources for self-selection in conversational interaction. Studies have shown that incipient speakers tend to gaze away just before starting to speak, which appears to signal to the interlocutors that they are about to take the turn (Kendon, 1967;Duncan, 1972;Brône et al., 2017). Even by using gestures speakers can make a claim for speakership publicly visible. For instance, in the context of meetings, listeners can establish themselves as next speaker before the end of the current speaker's turn by using pointing gestures towards relevant objects in the interactional space (Mondada, 2007).
In dialogue interpreting research, there has been an increasing interest in the role of multimodality (and especially gaze) in the coordination of interpreter-mediated interactions (Bot, 2005;Pasquandrea, 2011;Mason, 2012;Davitti, 2013;Krystallidou, 2014;Authors 2018). Recent studies have shown that interpreters use their gaze to organize turn transitions (Mason, 2012;Davitti, 2013;Authors 2018) and to select next speakers in certain sequential contexts (Authors, 2018). Furthermore, it is through their gaze that participants display their mutual involvement and recipiency when interacting with the aid of an interpreter (Davitti, 2013;Krystallidou, 2014;Authors, 2018, Theys et al., 2019. Altogether, these studies suggest that there is much more to be learned about the role of multimodality in interpreter-mediated interaction, A multimodal approach "can give us insights into how and, most importantly, to what extent interpreters can intervene in the ongoing encounter without substituting any of the primary parties" (Davitti & Pasquandrea, 2016, p. 19).
Although interpreting in mental health care has received some scholarly attention over the past two decades, the multimodal dimension of such encounters has not been investigated in detail so far. This is partly due to the sensitive nature of therapeutic encounters, which makes it an extremely difficult task to get permission to video record the sessions. By exploring the role of embodied resources in the regulation of turn transfers in interpreter-mediated therapeutic sessions, this study aims to make a contribution to the growing body of knowledge on the multimodal coordination of talk in dialogue interpreting.

Dataset
The analysis is based on three naturally occurring interpreter-mediated therapeutic sessions, videotaped in two mental health facilities in the Netherlands, (see Authors, 2005; 2018 for further description of the data) 1 . The use of video recordings is important, as it brings the details of the "original" event to the attention of the researchers (Jones & Lebaron, 2002).
The patients spoke either Russian or Dari, whereas the therapists were Dutch-speaking (see Table 1 for further information about the sessions). Each consultation was interpreted consecutively by a professional interpreter 2 . All participants agreed to be recorded by signing a written informed consent, which ensured their anonymity and stated how the data were going to be used and presented. We focused on the moments of turn transition between the patient 1 This study relates to spoken-language interpreting. Whether these findings are applicable to sign-language interpreting is an empirical question that could be addressed in future research. 2 By 'professional' we mean that the interpreter practices interpreting as a professional occupation with an hourly fee and a code of conduct to adhere to (see Bot 2005). and the interpreter. The analysis is based on the insights from Conversation Analysis (CA), which studies interaction in its emerging, co-constructed context (Gardner, 2001). female ambulatory patient, suffers from post-traumatic stress disorder. She has some understanding of Dutch.
-female therapist -male therapist -female therapist

Managing turn-taking during extended turns
In our data set, the patients often produce long, multi-unit turns or extended tellings, that involve elaborate actions such as extended descriptions, explanations, accounts of events and the like (Houtkoop & Mazeland, 1985;Selting, 2000). Not only do extended turns challenge the interpreter's memory capacity 3 , but they may also pose a challenge for the interpreter as far as the turn-taking is concerned. The primary speaker may decide to facilitate the interpreting process by producing shorter utterances (or 'chunking' their turns), allowing the interpreter to interpret as closely as possible (Flores, 2005). However, this is not always the case. Below, we illustrate an interpreter's failed turn-taking attempt during the patient's extended turn.
In the following excerpt (Excerpt 1), the interpreter displays readiness to take the turn during the patient's extended turn, but eventually fails to do so. Prior to the excerpt, the therapist had asked the patient whether she knows why the doctors decided to amputate her leg above the knee instead of below the knee. The transcript is presented in two lines: the original utterance 4 appears in a first, numbered line; the translation into English is written in italics just below the corresponding original turn. Relevant gaze information and the screenshots are presented under the corresponding lines in the transcript. Dari is provided in the transliterated original and in English glosses, whereas for Dutch only the English glosses are given due to space limitations 5 .
Excerpt 1 5 Also, some parts of the excerpt are not presented in the transcription due to space limitations. This was done after careful consideration of its content to ensure that its omission does not contain information that would affect the presented analysis.
The interpreter maintains his gaze at the patient while listening to her extended turn.
Research has shown that activities such as extended narratives "require more sustained gaze by the recipient toward the speaker" as a display of continuing attention and engagement (Rossano, 2013, p. 313). Around the point when the patient's turn reaches its pragmatic completion (line 5), the interpreter opens his mouth but does not take the turn as the patient continues talking and gazing away from the interlocutors. By gazing away, the patient displays her wish to maintain the turn (see also Lerner 2003). However, as the patient continues with the story ('Later, when they understood that…' line 7) a behavioral change begins; the interpreter starts displaying his readiness to take the floor by shifting his gaze to the therapist (line 7) and by inhaling audibly ('.hh') in line 8. The interpreter's display of self-selection through gaze shift and audible inhalation does not have an 'interruptive' effect (Mondada, 2007) on the patient's ongoing turn at that moment, as the patient continues with an elaborate account of the circumstances surrounding that traumatic event while orienting her gaze at the therapist. Thus, the patient's gaze aversion from the interpreter at the moment of the interpreter's turn-taking attempt appears to function as a turn-holding cue. At the same time, she orients her gaze to the therapist to secure her attention as recipient, which can be seen as a strategy to maintain her turn (Zima et al. 2018). The interpreter then quickly abandons his claim for speakership and reverts his gaze back to the patient. By directing his gaze at the patient he displays his ongoing availability and attentiveness as a recipient of the patient's utterance. The actual turn-transfer occurs only in line 13, where we see that the patient establishes mutual gaze with the interpreter, thus signaling her readiness to yield te floor.
In the subsequent interpreter's translation of the patient's turn (from line 14), we see that a large portion of what the patient had been tellingin particular the portion before the interpreter's turn-taking attemptis not rendered by the interpreter. Most importantly, the interpreter does not convey the reason for the high amputation of the leg (i.e. the doctor's fear for gangrene, line 2) to the therapist. Consequently, the therapist continues to think that such high amputation of the leg was unnecessary, as becomes evident from her reaction in lines 23-25 ('because the doctors did something which in fact was not necessary at all').
This extract illustrates a communicative breakdown 6 (Bot, 2005 p.209) which may be the result of the interpreter's lack of initiative to take the turn. Although the interpreter's turn-taking strategy may be motivated by the specific context of psychotherapy and his orientation towards the patient, it appears to conflict with his own need for speaking space at that moment. Davitti (2018) notes in this respect that it is the interpreter's responsibility "to identify appropriate times to intervene, deliver the rendition and give the floor back, in the least disruptive possible manner" (p. 18) in order to be able to render the patient's telling completely.

Managing overlapping talk
In the context of interpreter-mediated interaction, overlapping or simultaneous talk poses another challenge for interpreters. Given that the basic feature of conversational interaction is "one speaker at a time", overlapping talk is seen as one of the major departures from it (Schegloff, 2000, p. 2). When overlapping talk occurs, the interpreter will need to make certain choices on how to resolve it, deciding who will get the turn (see also Roy, 1992). In our analysis, we distinguished between non-problematic overlap (short listener responses such as 'yeah' and 'mh hm' and 'that's right', terminal overlap and choral speaking) and more competitive forms of overlap (e.g. in which simultaneous speakers appear to be contesting for a turn space) (Schegloff, 2000). Competitive overlap can occur when two speakers simultaneously co-start a new turn, or turn-finally, when a new speaker tries to take over the turn. Such overlaps require some sort of overlap resolution, such as dropping out of the turn by one of the speakers, in order to return to "one speaker at the time". According to Schegloff (2000), speakers employ a set of 6 Bot (2005) defines 'communicative breakdown' as a situation in which the communication comes to a halt not because a topic has been dealt with sufficiently for the time being, but because of a marked misunderstanding (p. 209). In the presented case, the communication does continue, but without correcting this and reaching a mutual understanding on this topic. The therapist continues to believe during the remainder of the session that the amputation was unnecessary.
devices for the management of overlapping talk. Examples include hitches (that is, momentary interruptions in the progressivity of the talk's production), prolonging or stretching of some next sound, or repeating a just prior element (Schegloff, 2000, p. 12). In addition, recent research has argued that gaze also plays a role in the management of overlap, as prevailing speakers (i.e. those who triumph in the competition for the floor) tend to avert their gaze away from the competing speaker as a turn-holding strategy (Zima et al., 2018).
Overlap has received scant attention in dialogue interpreting research, especially from a multimodal perspective. In the following, we examine how competitive overlap between the patient and the interpreter is resolved in the context of therapeutic talk.

Floor-yielding to the overlapping speaker
In the following excerpt, a patient is talking about his nightmares and how they cause his blood pressure to rise. Towards the end of his turn (in line 5), the patient turns his head towards the interpreter, which can be seen as a turn-yielding cue. The interpreter takes the turn after a slight pause (line 6) by uttering the acknowledgment 'mh hm' and shifting her gaze away from the patient. Her gaze shift to the therapist marks the transition from the activity of listening towards the activity of translating (Merlino & Mondada, 2014), which is initiated in line 8. Shortly after the interpreter starts rendering the talk, the patient redirects his gaze at the interpreter and suddenly takes the floor with slightly raised volume in overlap with the interpreter's turn (line 9). The overlap seems to result from their different treatment of the patient's turn: whereas the interpreter treats it as complete, the patient appears to treat it as still in progress (Ford et al., 2002).

Excerpt 2
The overlap is immediately resolved as the patient and the interpreter abruptly cut off their turns. We can see that both the interpreter and the patient display orientation towards the provision 'one speaker at the time' (Sacks et al., 1974). The patient seems to treat his turntaking attempt as face-threatening, as he utters an apology ('excuse me') and yields the floor to the interpreter (see also Schegloff, 2000). The interpreter quickly finishes her initiated turn unit (line 11, 'high blood pressure'), while shifting her gaze from the therapist to the patient. In this setting, the interpreter is the only one who can assess the importance of the patient's overlapping talk and decide whether she should maintain or yield the turn. The interpreter's gaze shift towards the patient, accompanied by nodding and smiling, functions as a turn-yielding cue. The patient's understanding and acceptance of this transfer of speakership is made evident through a prolonged "a:h" (line 12), after which he continues with his turn. We also see that the therapist acknowledges this course of action by producing a series of expansive nods (line 12). This example thus shows how the patient and the interpreter collaborate in resolving overlap by employing both verbal and nonverbal resources (gaze and head nods). In this process, the patient orients toward the interpreter as a real participant in the talk by withdrawing from his floor-taking attempt and acknowledging her rights to the conversational floor.
In excerpt 3 below, the overlap also results from the patient's and the interpreter's different treatment of the preceding unit. Here, the patient is explaining why she does not want anyone to know that her leg was amputated.

Excerpt 3
The patient finishes her turn by directing gaze to the interpreter who immediately takes the floor (in line 3) to render the patient's utterance. Towards the end of the interpreter's turn (line 6) the patient appears to treat the interpreter's turn as complete, as she turns her gaze to the interpreter and starts speaking again by adding a specification ('my morale dwindles') followed by a directive ('tell her'). This is produced in overlap with the interpreter's turn-unit "I don't like that" (line 6). We see that the interpreter, who was looking at the therapist during his rendition, quickly shifts his gaze to the patient (figure 5) and then back to the therapist (figure 6) to start rendering "and that has bad-" in overlap with the patient. The interpreter's shifting gaze from the one participant to the other reflects his double orientation at that patient therapist moment; on the one hand, he tries to maintain his speakership, while at the same time displaying attention to the patient and trying to comprehend the import of the patient's overlapping talk.
The interpreter appears to treat the patient's overlapping talk as a replacement at first, as he immediately starts rendering it ('and that has bad-') to the therapist. However, as the patient persists in her claim for conversational floor, the interpreter cuts off his rendition and redirects his gaze to the patient (figure 7), thus signalling that he is handing over the floor. As in the previous example, the interpreter orients to the conversational rule of only 'one party at a time'.
We cannot discern from the video recording whom the therapist was looking at at the moment of overlap. However, it is clear that in the competition for the floor, the patient orients primarily to the interpreter and seeks to secure his gaze. As soon as patient and interpreter establish gaze contact again, the patient averts her gaze, which indexes her intent to hold the floor, and continues with her turn.
In this section, we have seen how gaze towards the overlapping speaker functions as a floor-yielding cue. In the following section, we show how gaze aversion from the overlapping speaker is employed to maintain the floor.

Resisting floor-taking attempts from overlapping speaker
The interpreter may also choose not to yield the turn to the overlapping speaker. This is illustrated in the example below. Here, we find a lot of competition for the speaking space between the interpreter and the patient. The patient is telling about the injuries she suffered during the war. She produces overlapping talk several times during the interpreter's turn, making corrections and providing specifications about her injuries.
As he takes the turn (in line 4), the interpreter directs his gaze at the therapist. In line 6, the patient interrupts the interpreter with the specification 'shoulder-launched rocket', after hearing the interpreter's translation in Dutch. The interpreter merely acknowledges the patient's intervention with the token 'yeah' without looking in her direction nor adding this in his rendition. By gazing away from the intervening speaker, the interpreter signals a wish to maintain the floor. He does not seem to treat the patient's intervention as a relevant contribution to his ongoing turn as he does not render it to the therapist. At that moment, the interpreter is focused on retrieving information, as he appears to struggle to remember which leg was injured ("tha:t it's the right (.) the right leg was completely uh damaged", line 7). Towards the end of the interpreter's rendition (line 9), the patient self-selects again by speaking in overlap with the interpreter. In fact, she seems to detect a trouble source in the interpreter's preceding utterance "the left lower leg was not damaged" (line 8) and provides an unsolicited correction in the following turn (lines 11-12, "it has a dent") 7 . Throughout this excerpt, the patient appears to monitor the interpreter's output and correct it (see also Kredens 2017). Given that the patient acts as the 'principal' (Goffman, 1981) of the interpreter's talk, it is in her own interest that her words are rendered correctly.

Excerpt 4
During the interpreter's subsequent rendition of the patient's utterance (lines 13-15), the patient does not wait for possible completion of the rendition, but launches her turn again in overlap by providing further specifications about her injuries (line 16). While doing so, she points at her left leg, possibly in an attempt to attract the interpreter's attention. Interestingly, the interpreter does not yield the turn at that moment (line 15), but pauses for 0.8 seconds while maintaining gaze at the therapist. He then, still in overlap with the patient, finishes his turn ("is damaged"). By maintaining his gaze at the therapist, the interpreter indicates that he is not yet willing to yield the floor and that he intends to finish his turn. Only at the end of line 16, after patient interpreter therapist the therapist directs her gaze to the patient, does the interpreter turn towards the patient and look at the place on her leg that she is pointing at (figure 9).
After a considerable silence of 1.4 seconds (line 17), which could be a signal of some interactional trouble (Jefferson, 1986), the patient initiates overlap resolution by repeating her preceding turn "on this part still no flesh" (line 19) with a slightly rising intonation contour.
This repetition seems to be informed by an inference that the interpreter did not understand her.
Also, by repeating her previous turn when there is no danger of interruption, she displays herself as the "surviving claimant" (Schegloff 2000, p. 34) for turn space.
To summarize, by overlapping with the interpreter, who is engaged in rendering the talk to the other participant, the patient inhibits the progressivity of the interpreter's utterance.
She displays no awareness that such overlaps could be problematic for the interpreter. In this excerpt, the interpreter clearly displays resistance to the patient's overlapping talk in his effort to maintain speakership, which is made evident through his gaze aversion from the patient. He eventually yields the turn to the patient by pausing and orienting his gaze at her. We can assume that, during the patient's repeated overlaps with the interpreter, it becomes not only difficult for the interpreter to keep track of his thoughts, but it also becomes impossible for the therapist to

Discussion
This study has sought to investigate the role of multimodal features in the management of turntaking in the context of interpreter-mediated psychotherapy. While previous research has examined discursive features in interpreter-mediated therapeutic talk, few have focused on the negotiation of turn-taking, especially from a multimodal perspective. We have analyzed the role of multimodality in the management of turn-taking in two specific interactional contexts: turn transfer during an extended turn and in the context of overlapping talk. This has led to some preliminary observations. First, the interpreter's gaze aversion from the current speaker towards the recipient functions as a floor-taking cue. In the first extract, we have shown how the interpreter abandons his turn-taking attempt during the patient's extended telling and how his failed turn-taking attempt appears to be linked to the primary interlocutors' visual behavior: the patient did not display readiness to yield the turn, which is evident from her gaze aversion from the interpreter at the moment of the interpreter's self-selection. The interpreter was thus not able to secure mutual gaze with the patient nor with his recipient (the therapist) at the moment of self-selection (see also Oloff 2012). We have also seen how the interpreter's failed turntaking attempt eventually resulted in loss of important information. In such cases, Bot (2005, p. 245) suggests that the therapist may aid by stepping in and taking control as chair of the session and 'preventing' such long turns.
Second, if the interpreter gazes at the overlapping speaker, s/he is likely to withdraw from the turn. This can occur close to the beginning of te interpreter's turn (as shown in excerpt 2) or towards the end (excerpt 3). The overlapping speaker who gets the turn will, by averting his/her gaze, signal the intent to maintain the floor. Third, the interpreter may resist the turn- Our study also suggests that interpreters are in a constant field of tension between their role as communication facilitators and their own conversational needs as participants in the exchange. While the interpreters' main role is to enable communicative contact between the therapist and the patient, they also need to safeguard their own speaking space. Our analysis confirms that turn-taking in therapeutic talk with an onsite interpreter is a collaborative achievement between the primary participants and the interpreter. Acknowledging the interpreter as a co-participant with a certain (professional) role and speaking rights within the exchange supports the interpreter's interpreting activity, allows the interpreter to focus on the task of translating instead of competing for speaking rights and allows the interpreter to decide when to take the turn in order to optimize his/her rendition. However, delegating the organization of turn taking entirely to the interpreter may eventually overburden the interpreter which may point to a role of the therapist in this matter (Bot, 2005). Bot (2005) suggested that therapists should therefore monitor the interaction between the patient and the interpreter for potential problems (e.g. the interpreter not being able to take the turn or patient interrupting the interpreter) and intervene if necessary. This is important, as the quality of the therapy in part depends on the smooth organization of turn taking between the interpreter and the primary speaker (see also Miller et al., 2005). Problems in turn-taking canas we have shownlead to loss of information, which may have an impact on the quality of the therapy session.
Finally, this study should be understood as an invitation for further investigation of the multimodal dynamics of interaction management not only in mental healthcare interpreting, but also in other contexts. For instance, it remains to be examined in detail how interpreters deal with overlapsboth verbally and nonverballyin other conversational settings, which will undoubtedly increase our understanding of the interactional choices that interpreters make during the interaction, the way they try and/or succeed to implement those choices and their motivations behind those moves. Consequently, important insights may be drawn not only for dialogue interpreting theory, but also for interpreter training.

Acknowledgments
The authors wish to thank the anonymous reviewers for very insightful comments and suggestions for an earlier version of this paper.

Conflict of Interest
We confirm that all personal identifiers have been removed or disguised so the person(s) described are not identifiable and cannot be identified through the details of the story.
a period indicated a falling intonation contour , a comma indicates rising intonation contour ? a question mark indicates a rise stronger than the comma ((comment )) information in double parentheses provides details about the nonverbal behavior of the participants. fig. the exact point where a screen shot (figures) has been taken is indicated # with a specific sign showing its position within turns-at-talk ---gaze continues across subsequent lines --->> until the >> symbol is reached → analyst's signal of a significant line