The cross-lingual shaping of narrative landscapes: involvement in interpreted story telling

ABSTRACT Although consecutive interpreting of longer stretches of speech inevitably involves repetition and delays, it can also be regarded as a resource for engagement in storytelling. In an open, interpreted conversation on stage about a recently published piece of literature, the participants had to manage several tasks at once: an interview with questions of public interest, a literary reading session with captivating excerpts of adequate length, and the handling of alternating languages on the floor, all within a certain time frame. In the case explored, conversational interactions on different levels, such as repetition, gesture and gaze are analysed, to see both how cross-lingual narration is achieved and what the specific nature of an interpreter-mediated public literary conversation requires from the author, the moderator and the interpreter performing on stage. The study also touches on the role and function of the book as a talking object in the conversation.


Introduction
This article seeks to find out how narratives in a conversation are transferred across language borders, and how the teller's as well as the listeners' engagement is kept up in spite of the potentially disrupting, necessary involvement of the interpreter. This study is part of a larger research project on public literary conversations (hereafter PLCs), with special focus on the interpreter-mediated PLC. In Sweden, this kind of event with writers from abroad talking about their latest book has increasingly spread since the late 1990ies to include book fairs, international writers' stages, and literature festivals. In the conversation at hand, a writer from France, assisted by an interpreter, tells the audience about her recent book on her ancestors' life in Armenia.
The approach employed for this investigation is sociological, relying on findings in conversation analysis (CA) as founded by Sacks et al. (1974) and on topical episode analysis (TEA) as worked out by Korolija and Linell (1996). Aspects from sociolinguistics and discourse analysis by Tannen (2007) and from early theory on narratives in spoken language (Labov & Waletzky, 1997[1967) have also informed both the study and the examination of multimodal face-to-face exchange, as explored in interaction analysis (IA) by for example Goodwin (2010Goodwin ( , 2013, including gesture studies (Kendon, 2004;Müller, 2014). It should be noted that the theoretical framings mentioned so far are modelled to fit analyses of monolingual discourse, where participants interact directly. In my analysis, however, I presume that interaction in interpreter-mediated encounters is organized differently. In line with Wadensjö's approach to interpreting as interaction, this article will explore an interpreter-mediated PLC as a shared, situated, and staged activitya communicative pas de trois, to use Wadensjö's (1998, p. 12) metaphor. This implies, that bilingual discourse is seen as a joint enterprise that not only relies on the performance of the interpreter, but also demands high context-sensitivity from all three interlocutors on stage. With this in mind, section 2 explains, what specifically each framework added to the analysis.

Narrative discourse in performed speech
CA and IA methodologies are central within the rationale of my combined approach, in particular concerning phenomena such as turn-taking and sequence organization. The multimodal dimension has become of growing interest for studies of interpreting practices, as lucidly presented in a survey by Biagini et al. (2017). Moreover, dealing with literary conversations requires attention to the style of speakers such as writers and literary critics, who use language as a tool in their daily work to produce arguments as well as well-structured and complete narratives (Tannen, 2007, p. 20).

Coupled turns and episodes
From earlier single-case analyses of interpreted dialogue in PLCs (Poignant, 2018), patterns emerge as characteristic for the communicative activity type (Linell, 2009) of such conversations. Framed as semi-structured interviews on life and culture, with wide-open questions, the changing between easy, spontaneous talk and more profound ideas has formal implications for the sequence structure, as well as the shifting between languages. In CA generally, adjacency pairs are considered to be the central format of a sequence and are normally composed of two turns, such as greeting/response or question/ answer, uttered by two different speakers. In interpreted talk, one or both of these turns are, additionally, reproduced in another language. While interactionally working as a self-contained turn-construction unit (Sacks et al., 1974), interpreters' utterances are normally not self-contained but belong by definition to the turns they reproduce. This has given rise to the notion of a 'coupled turn' to designate the bilingual utterance pair in interpreted talk (Poignant & Wadensjö, 2020, forthcoming). Furthermore, the two parts of adjacency pairs in PLCs tend to be highly dissimilar: in response to merely a brief question from the moderator, the writerthe main attraction of the eventcan work out extensive thoughts and ample storylines frequently stretching throughout far more than one turn. Having this type of sequence as an extended question-answer pair format in mind, however, I here suggest to call it a turn cycle. The sequential format of a turn cycle appears as a key structure for interpreted PLCs and is prototypically transcribed below, with L s designating the writer's or source language and L t the audience's or target language; a grey font indicates actions that are seen but not heard by the audience, such as the interpreter's whispering to the writer (the professional term for this interpreting technique is chuchotage) or the note-taking, while square brackets mark the overlap of speech and/or actions (see transcript key). A prototype transcript of an interpreted PLC turn cycle looks as follows (MO = moderator, WR = writer, IN = interpreter): Lines 2 is the writer's answer and line 3 the interpreter's consecutive renderingwith or without notes, depending on length. This constitutes the coupled turn. As mentioned above, in interpreted PLCs it regularly tends to be expanded into several subsequent and long turns, in answer to the moderator's topic proposal and question. This is schematized in figure 1.
Consecutive interpretingwhich by its nature is re-iterative as a conversational move, recapitulating what already has been saidis depicted as having a specific shape of its own, in tandem symbolizing progression in time and retrospection in terms of content. The topic is picked up from behind, as it were, at the same time as talk is projected forward, preparing a relevant catching-up point for one of the primary parties, the writer or the moderator, to continue the talk. (For the interactional task and function of the interpreter's turn as the second part in a coupled turn, see Poignant & Wadensjö, 2020.) Conversation and interaction analysis often look at units of expression on a microsociological level. Yet, to embrace a PLC in its entirety, there is a need for analysing larger size sequence units, such as long turns and whole sequences. Exploring coherence and sense-making practices in multiparty encounters, Korolija and Linell (1996) (re)introduced the notions of episode and sub-episode as natural units of conversational interaction. Korolija (1998, p. 39) describes an episode as 'an action sequence internally bound together by a topical trajectory and/or a common activity' that can 'vary in size and level'. Korolija shows also that the episode concept 'can be traced to different disciplinary origins, including linguistic, psychological, and sociological theories', and that its applications are mutually compatible (ibid.). The compatibility of the episode concept makes it useful as an analytical tool in the transdisciplinary field of Interpreting Studies. It allows for analyses of larger size units within the sequential structure of interpreted PLCs.

Narrative structures
In Talking Voices, Tannen (2007, chapter 4) explores authors' conversational style in literary talk, when they (re)-tell narratives they earlier have fictionalized in their books. She describes it as a genre between orality and literatureit is spoken but based on a piece of written, literary work, which influences the choice of vocabulary and register. It often contains epic narratives, using discourse devices such as repetition, imagery, and involvement (Tannen, 2007, p. 134). This holds true also for the speakers' style in PLCs, and I therefore find Tannen's analytical use of these features to be highly applicable to my material.
With the research question in mind, namely how cross-lingual storytelling is achieved and what such a specific, staged setting requires from the interlocutors, I mapped out the episodic sequence structure (Korolija, 1998) of the interpreted PLC and identified some of its conversational style (Tannen, 2007) and characteristics as a communicative project (Linell, 2009). In order to investigate its narrative discourse functions, I chose a model, outlined by Waletzky (1997 [1967]). This model is widely known within the human and social sciences because it provides a basic typical structure of narratives. The history and metamorphosis of it within sociolinguistics, that is its uptake, critique, and later adjustments and elaborations (Labov, 1997(Labov, , p. 2013, have been comprehensively summarized and commented on by Johnstone (2016), who traced its impact on the developments of discourse and conversation analysis. The CA scholar Schegloff (1997), for example, in viewing the strengths and weaknesses of the Labovian model, criticizes Waletzky (1997 [1967]) for not taking the embeddedness of narratives in naturally occurring talk into account, but treating them rather as monologic storytelling. Nevertheless, I regard it as a fruitful tool for my analysis, as the naturally occurring narratives in PLCs, although mostly conducted by one speaker at a time, are ultimately dialogically constructed and recipient-designed for many listeners at once. The model itself has gone through various stages, from the early foundations in the 1967 article 'Oral Versions of Personal Experience' (Labov & Waletzky, 1997[1967) to the more recent book The Language of Life and Death: The Transformation of Experience in Oral Narrative (Labov, 2013). Labov, himself referring to his and Waletzky's 1967 article and to a later slightly more specified scheme (Labov, 1972), renders the quintessence of a 'fully developed narrative' as consisting of 'an abstract, an orientation with information on persons, places, times and behavior involved; a complicating action; an evaluation section, which identifies the point of the narrative; the resolution; and a coda, which returns the listener to the present time' (2013, p. 3, italics added by me). In the current study, I relate to Labov's own visual representation ( Figure 2). The Labovian narrative model identifies two general formal functions, the referential (what the story is about), and the evaluative one (why the story is told and what the teller does with it). In the following analysis, my intention is to highlight this latter evaluative function. Evaluating the events within the narrative, as well as the co-constructed maintenance of this evaluation through sequences, has a special significance, maybe even urgency, for the practice of interpreted staged narratives and how they are conveyed to the audience across language and turn shifts. What Labov calls the evaluation section is depicted in his earlier figures as a vantage point, from which complicating actions and resolution are told and reflected on, even as he notes that 'evaluative devices are distributed throughout the narrative' and modifies the scheme accordingly yet (Labov, 1972, p. 369). Therefore, rather than a section, it now is regarded as a discursive quality that permeating the whole story.
At this point, I find that there is an interplay between the Labovian model and Tannen's discourse device of involvement. Labov and Waletzky's (1997) evaluation, showing a speaker's sensitiveness to the narrative itself, evokes involvement in the conversation in the sense that Tannen (2007, p. 27) describes, namely 'an internal, even emotional connection individuals feel which binds them to other people as well as to places, things, ideas, memories, and words.' This involvement, in turn, creates the narrative coherence between long-stretched episodes of talk characterizing interpreted PLCs. Unlike what Labov originally called personal narratives, which occur in everyday life speech and are shorter, have a simpler syntax, and often consist of a single episode, the storytelling of epic narratives in a PLC supplies the listener with backgrounds and portraits of the participants and often deals with 'the heroic individual, who is larger than life' (Labov, 2013, p. 148).
Following the above explanations, the combination of notions has relevance for my material, as the units of analysis for my investigation can be determined as narrative episodes. Their overarching topical coherence lies in the order of subsequent narrative elements, sustained over transitions between sequence boundaries of turn cycles and coupled turns.

Data and methods
The conversation was conducted and recorded at the annual literary festival in Umeå, Sweden, in 2017, and soon after accessible on the internet. To begin with, I made a rough, content-oriented transcription of it, to serve as data for an overall coding of episodes and sub-episodes. Within these units, I coded the occurring elements of narrative functions in every turn. Based on this, I was able to identify interactionally significant  (Labov, 1972, p. 369). ( = 'abstract'). moments and choose excerpts for further analysis. I also interviewed the interpreter, letting her comment on the excerpts retrospectively by playing them for her, to get an emic perspective of the participation framework as well as information on how she prepared for the task.
The excerpts were transcribed in greater detail, according to transcription conventions (based on Jefferson, 2004). The transcripts include multimodal components as movements and gestures, which are described in words within double parentheses, inserted next to the verbal expression which they are preceding or accompanying. This design has been chosen partly to facilitate readability, and partly in order not to distract focus from the emotionally evaluative quality of the narrative episode under scrutiny. The sequence analysed in detail consists of the first part of a turn cycle, or sub-episode, an extended, large-scale adjacency pair, if you wish, of the moderator's question or topic proposal, eliciting the writer's elaborated answer. It stretches over the first eight turns four provided by the author and four subsequent interpreter renditions of these in the target language. Here, these are treated as four coupled turns of bilingual storytelling. The first and the fourth are explored in-depth, whereas the intermediate two are summarized as far as content is concerned in order to support the understanding of the analysis.

Background
The PLC in question was conducted with the French writer (and actress) Anny Romand and moderated by a well-known Swedish literary critic. The interpreting is provided by a professional conference interpreter. The interpreter is experienced in this type of scenario and has mastered the routine of efficient note-taking and oral reproduction, which allows her in a flexible way to adjust to any speaker's pace, utterance length and communicative stylefavourable prerequisites for the subtle cooperation demanded. Furthermore, she had read the book in French and in Swedish, and she had had a short briefing with the primary parties before all three came onto the stage.
The conversation deals with Romands book from 2015, Ma grand'mère d'Arménie (My grandmother from Armenia), translated into Swedish by Madeleine Gustafsson and published in 2017. In short, the book relates an individual survivor's account of the Armenian genocide in 1915, which occurred in the aftermath of the dissolution of the Ottoman Empire. The book includes reprinted pages from a tiny old booklet, the grandmother's diary, written more than a hundred years ago in Armenia and the Middle East, interspersed with Anny Romand's own memories of spending her early childhood with her grandmother in Southern France. Thus, in the tale, locations shift across periods and between distant geographical regions: starting from a flourishing Silk Road merchant town along the south coast of the Black Sea around 1900, and educational years in Palestine, to the mountain ravines in Turkey and the deserts of Northern Syria, where the protagonist Serpouhi Hovaghian (Romand's grandmother) experienced the worst purges, and finally to Marseille, where she found another home.
Since PLCs aim to elucidate a specific work of literature and its author, an important element is the reading of excerpts from the book in question. Readings are integrated in the episodic structure of the conversation, in most cases formally either opening or closing an episode. On the stage, a printed copy of the book is usually close at hand. The topical dynamics of the conversation relate to this printed copy as to a focal point. Symbolically, the book even 'takes the floor', and its voiceand with the wording of the book in Swedish also the voice of its translatoris animated in the readings during the event. Moreover, the participants can use it as a physical device for interaction, a significant artefact, as well as a personification of the story it tells, almost becoming another 'ratified participant' (Goffman, 1981) in the conversation. For example, in the beginning, just after the presentation, Anny Romand picks up the book from the table and holds it in her hands, when she enters into the conversation, smilingly saying, 'It's a great honour for me to be in Umeå, together with my grandmother!' (Figure 3), which evokes small reactions of laughter from the audience. Romand virtually shows that she carries her grandmother with her, being the bearer and co-author of her testimony, represented by the book. Somewhat later in the conversation, the moderator also holds up the book, pointing at the front cover when introducing its protagonist and her family in Armenia: 'And here stands your grandmother, on the one side of the photograph, the young Serpouhi' (Figure 4).
During the conversation, three short readings were conducted, the first of which is cited here. They accomplished a dramaturgy in the conversation, and, not least, gave a short break to the interpreter, who was constantly busy with either whispering into French, taking notes, or rendering into Swedish. The French version of the passage from the book was read by the author herself, the Swedish one by the moderator. Here, to save space, only an unofficial English translation is provided. 1 The first reading (27 minutes from the start) quotes the early diary written in Armenia, during the persecution: And tomorrow? Everything will be over. People will sit in their comfortable chairs and read about our sufferings, which will be printed in books. But will ever such a book be able to describe what we've gone through? Never ever. They will talk about it in the salons just until other news comes up, and the prayers and voices of the poor Armenians will disappear as the smoke from a cigarette, only the ashes are left, and only the earth will save us. (Romand, 2017, p. 17) In an eerily precise way, the words Serpouhi Hovaghian wrote in 1915 anticipated the situation in which the listeners found themselves at that very moment, more than a hundred years after the lines had been scribbled down in despair, in a little notebook, sitting in comfortable chairs, talking about the cruel events of those times. The ending phrase, insinuating the impossibility for later generations to grasp the ferocious atrocities, triggers a strong sense of involvement, making Romand's continuing story still more 'tellable' (on tellability, see Herman, 2009;Hutchby & Wooffitt, 2008, p. 128).

Analysis
The whole event lasted 45 minutes. Figure 5 provides an overview of the main episodes, subepisodes, and core topics touched in the course of the conversation. The excerpts analysed below are located in the first sub-episode of episode III (underlined topic), dealing with one of the protagonist's most dramatic experiences. The turn cycle is the largest one in the conversation and contains six coupled turns in total, of which the first four are analysed.
Prior to excerpt 1a, after about 17 minutes of the talk had elapsed, the moderator briefly recalls the situation as set out in the book: Serpouhi was a 22-year-old mother of two children in the warlike chaos of the beginning assault on the Armenian population, with her husband just having been deported, probably murdered. The moderator then turns to the author with a question, which, according to Labov (2013, p. 7) is 'one of the most powerful tools available for an interviewer to launch into a narrative: What happened then?' In CA terms, this is a storytelling preface (Hutchby & Wooffitt, 2008, p. 126) that elicits an answer. The moderator explicitly allocates Romand as the next speaker and aligns both herself and the audience as recipients, indicating the expectation that an extensive response is underway ( Figure 6). In response, Romand, after a seven-second silence while listening to the interpreter's whispered rendition of the question, adjusts her seating position, turns to the audience and starts with Alors, ('So', line 2). She also signals that there is a longer narrative to follow (an incident from her grandmother's life) and that she is designing her telling of it specifically for the recipients and the occasion of the event.
Excerpt 1a: Armenia, April 1915. MO = moderator, AR = Anny Romand. (min. 17:40) The momentousness of the retold situation is several times reinforced by Romand's movements. Both times, when she mentions the two children (lines 3-4 and 15-16), she stages the infant as if being placed near her heart, with her left hand on her chest, and the 4-year-old son held with her right hand downward (arrows in Figure 7), as if walking along with the two of them.
In line 6, the writer interrupts herself, covers her face with her hands and makes wavering gestures, signalling an emotional charge when (in lines 7-10) she explains the Turkish soldiers' strategy to avoid making people scream and protest by pretending to put them in safety to a hospital (line 12). The remaining part of the turn is a repetition of what she said before, highlighting twice the two children and the hospital as the main subjects to keep in mind for her listeners. In her final position and gesture, Romand explicitly allocates the turn to the interpreter (after line 17). After that, she continuously directs her attention on the interpreter during the rendering.
In the following transcript of the interpreter's turn, the lines in grey note Romand's (ar)  Romand displays her intense concentration as she listens and waits for cues to recognize the emotive high notes of her own story within the interpreter's rendering, and then reinforcing them with gestures and facial expressions. She accompanies the interpreter's speech by moving her right hand, frowning, and nodding ( Figure 8). Near the end of the rendering reacting already to what Goodwin and Goodwin (1987) term the recognizable completion of a speaker's turn, that is the interpreter's onset of the phrase-ending intonation the author lifts both hands with palms up and looks straight forward, eager to continue her story as soon as the interpreter has finished. The interpreter, in ending, carefully puts the microphone back on the table, turns her face towards Romand, and gives her the turn, whereupon Romand takes it immediately and seamlessly glides into her storytelling again.
From this point on (after excerpt 1b), there is a change in the rhythm, which had been established between the writer and the interpreter since the beginning of the event (some 20 minutes previously), and the tempo accelerates. The turns become shorter and are performed at a higher speed. Romand almost thrusts the story forward: she divides it into chunks, adapting herself to the mode of telling it in passages, for the sake of the interpretation. The interpreter, passing them on to the audience in Swedish, conforms to the dramatical take-off in speed, focused on accomplishing a complete and coherent rendering. The interpreter's turns, characterized by a clear prosody and a calm tone, enable the listener  not only to catch up with understanding the narrative as such, but also to grasp the emotionally difficult aspects of Romand's unfolding, increasingly dramatical story.
In two more coupled turns (numbers 2 and 3 within the turn cycle), Romand and the interpreter go on telling what happened to Serpouhi those days in Trebizond, in April 1915, when the Turkish government started the Armenian genocide. The young mother, not knowing where to turn, obeyed and went to the hospital as she was told. A soldier announced that all children three years and older should immediately be brought to school and that her son would therefore be taken away from her. Serpouhi protested, arguing that, given the state of panic and turmoil, schools would not be open. But the soldier persuaded her by allowing her to accompany the boy to where he was going, and so she left her daughter behind in the hospital. As soon as they reached the destination, a hangar full of mothers and children, she talked to the soldier again, insisting on being permitted back to the hospital to fetch her baby girl.
Excerpt 2a: ' … it's over '. (min. 20:29) Approaching its peak, Romand's storytelling evolves into reports of direct speech in the present tense, with short accounts of who is speaking ('she says', 'he says', lines 56, 61, 62, 63), making use of 'direct quotations as an essential feature of dramatized experience' (Labov, 2013, p. 34). From line 64 onwards, her telling proceeds with restrained gestural movements, with a suppressed and, at the same time, emotionally charged tension, mirroring an agitated state in her facial expression. She ends with quick, sweeping sideward gestures, what Müller (2014, p. 145) identified as semantically indicating a negative evaluation, or rejection and exclusion of a topic from the conversation and distancing oneself from the just pronounced fact, in this case the killing of the infants in the hospital.
In excerpt 2b, the interpreter subsequently renders Romand's direct speech, though slightly abridges the four quoted utterances into three. As before, she holds the microphone in her right hand and, while keeping track of her notes, she also takes rapid glances at the audience. This time, her gestures, in spite of being naturally constrained by the handling of her notepad and pen, are characterized by the repeated left-hand beating strokes. She marks the direct speech onsets and emphasizes the turning point of the storyline (lines 78, 79) when Serpouhi goes back to the hospital but can't get inside. Meanwhile, Romand's gestural movements, which in the earlier turns had accompanied the interpreter's speech, are now noticeable absent, except for two tiny onsets, as reflexes, suddenly rising and falling (in lines 74 and 79), yet neither time leading to an interruption. Her gaze is motionless throughout, fixed slightly downward in front of the interpreter (Figure 9).

Narrative elements in the episode
The turn cycle covers the whole story about what happened to Serpouhi during those first days of the genocide. The moderator's topic proposal sets it off, and then follows a sequence of four coupled turns, each containing a directional (forward pointing) turn produced in French, followed by a (confirming and prospective, pointing backward as well as forward) rendition of that turn in Swedish, all evolving in rhythmically progressing sub-episodes. In terms of narrative elements, the first coupled turn picks up the moderator's introductory orientation section. Romand describes how Serpouhi holds her children as the Turkish military directs her to the hospital. She then immediately proceeds to a complicating action section (the soldier and his commands), which is elaborated and continues throughout the second and third coupled turns in the sequence, reaching a dramatical climax in the fourth. When the storyline is cut off abruptly by the deictic expression 'it's over' (line 66), it immediately changes into the resolution part, the result of what had happened in the hospital. What follows with 'she never saw her little girl again' (line 67) is a sudden coda, catapulting the storyteller and the recipients away from the heart-breaking moment and letting them watch it from a later waypoint of what was said about that 'hospital'. This passage is significantly marked by no movement in the interaction at all.

Repetition and imagery
The storytelling author creates imageries, both vocally and in her gestures and poses, when talking about the children and the soldier, the town in chaos, and the hangar full of people. whether in dialogue or direct speech, and using repetitions and exclamations, 'vividly evoking a scene one can hear and see' (Tannen, 2007, p. 21). In words and gestures, Anny Romand repeats the image of a mother holding her two children and being taken to the hospital, reinforcing twice the fatal negation ne … jamais ('never') about the loss that is forever. In reporting facts live in front of an audience, Romand emotionally identifies herself with Serpouhi at the time of the traumatic event, taking her stance in the vulnerable situation: by believing she was saving her one child by not letting him go away with the soldier alone and trusting her other child to be safe in the hospital, she cruelly loses first the baby girl and then the boy (who luckily survives and eventually reunites with her much later, as the listeners learn from other parts of the conversation).
There is still another kind of repetition active in the conversation, namely the one which is inherent in the practice of interpreting. On the interactional level, interpreted talk accomplishes the goal of including one or more persons who did not understand the preceding utterance. Thus, the fact, that every sub-episode of the story is interpreted has the repetitive quality and effect of a repair and functions as a topical retake in the course of the conversational flow, attended to by the author. In that meaning, 'repetition not only ties parts of discourse to other parts, but it bonds participants to the discourse and to each other, linking individual speakers in a conversation and in relationships' (Tannen, 2007, p. 61). The deeper effect of the interpreter's rendition as a repetition within the narrative process will be discussed in the result section.

Evaluation and involvement
The element of evaluation, which according to Labov is not bound to an order of chronological logic and is thus able to occur anywhere in a narrative, is tangible throughout the whole storytelling. Waletzky (1997 [1967]:32) define it as 'that part of the narrative that reveals the attitude of the narrator towards the narrative', simply put, that which makes it worth telling, that makes it a narrative. In the analysed interpretermediated PLC, evaluative loadedness and involvement activity occur not only in the interlocutors' speech but also in their concentrated, concomitant listening activity (Romand, moderator) and in the note-taking for rendering (interpreter). In a contextsensitive way, this kindles involvement (Tannen, 2007) in all parties. Perhaps because she is also an actress and thus used to timing, Romand utilizes the interpreter's turns as a device for emotionally 'filled' pauses and artistic caesuras. Her body is then motionless, except for a few, restrained hand movements, covertly accompanying the interpreter's speech. She is slightly directed toward the interpreter, her head lowered, intensely concentrated in her facial expression as if 'doing being within the tale' (cf. Sacks, 1984), that is, mentally following the narrative while listening to its Swedish version. Meanwhile, the interpreter delivers her renditions, keeping up the tension of their common performance. In turn, when the writer is speaking again, the interpreter concentrates on listening, analysing, gripping the storyline and putting down her notes, preparing to re-tell it conscientiously in Swedish. In the post-event interview I conducted with the interpreter, she said that although she had read the story twice beforehand, she was again strongly touched by it, when hearing and noting down Romand's oral version.
Romand's repeated passing of the turn over to the interpreter, and her following up on what is happening to her own story when retold in another language, shows her impulse to include the interpreter's turn as a repetition, a variation, or a prolongation of her own utterance, and that both of them have a shared focus on keeping up the tense and engaging story. At one point in the later part of the conversation, when the moderator briefly inserts a short, personal compliment to the interpreter, Romand adds a quick remark, asking the audience, 'How do you find my Swedish?' This shows her awareness of how her voice is being embodied in this context by the interpreter. Figure 10 features a scheme depicting the storytelling's progress by tying the subsequent narrative elements of orientation, complicating action, and resolution to the writer's and the interpreter's two adjacent coupled turns. The entire sequence is pervaded by the quality of evaluation itself and its variously generated involvement activities. Here, the evaluation in the Labovian model and Tannen's notion of involvement cover activities with similar functions, integrated within the sequence structure of a turn cycle with its coupled turns.
Obviously, the reality is far more complicated than what can possibly be captured schematically, but the figure can indicate, how the non-speaking party maintains the narrative tension by conducting and displaying involvement activities: the interpreter by taking notes for rendering, and the writer by listening to the interpretation with multimodal expressions of evaluation. Not accounted for in the figure, yet present throughout the fulfilment of the narrative, is its embeddeness in the attention of the moderator and the audience, sustaining continuous engagement in the entire speech event.

Results and discussion
This article explores how cross-lingual storytelling is achieved. The analysis of the interpreted turn cycle shows that the participants move along the narrative phases of the story and perform them in relative orderliness (Labov & Waletzky, 1997[1967). They progress in episodic arcs, from orientation to complicating action, pursuing the complicating action more and more dramatically in several coupled turns, and end up in the crucial resolution and a short coda. Regularly shifting languages in their management of turntaking on a collaborative floor, the author and the interpreter effectively use the discursive elements of repetition and imagery, interactionally displayed in verbal and other modal expressions, to accomplish functional narratives and conversational dynamics. Their use of dialogue, and direct speech, renders situations in a dramatized, 'evocative way' (Tannen, 2007, p. 21), as for example the scene when Serpouhi walks to the hospital, falsely designated as a safe haven by the Turkish military, and her ensuing dispute with the soldier. These pictures and scenes, or imagery, show the speakers' emotional stances in moments of loaded evaluation (in the sense of Labov & Waletzky, 1997[1967) and affective involvement (in the sense of Tannen, 2007).