Interactional role shift as communicative project in student teachers’ oral presentations

Abstract Focusing on Swedish student teachers’ oral presentations in a rhetoric class, this article studies interactional role shift as a multimodal practice. The role shifts under scrutiny concern shifting from student teacher to teacher, thus anticipating the students’ future profession. A central feature of the article is a discussion of how role shift may be conceptualised as a communicative project, thus highlighting the different modes of communication used by the students, and consequently to examine its potential as a facilitator of students’ professional and academic development. The data was collected using an ethnographical approach, resulting in a collection of 21 video-recorded oral presentations, together with other relevant semiotic resources. The data is analysed by the employment of concepts from nexus analysis and the notion of communicative projects. Through a discourse analytical approach to social action in interaction, the analysis shows how role shifts are constructed of patterns of smaller actions that add up to three primary actions: setting the scene, changing perspective, and performing the new role. These primary actions are multimodally chained together, and the results demonstrate how social actors use instructional texts in combination with multimodal recourses in order to perform their role shifts.


Introduction
Swedish teacher education has a rather complex structure, where student teachers need to learn to cope with different institutional and professional practices (Erixon Arreman and Erixon 2017). Consequently, this may actualise different tensions between professional and academic communication in teacher education (for examples of this, see Christensson 2019; Blåsjö and Christensson 2018). In Swedish teacher education, this tension is a phenomenon possibly related to teacher education being a highly politicised matter (Råde 2014). In the midst of all of this, students also need to make sense of a constant flow of, e.g., institutional and professional discourses expected to develop their academic and professional competencies (Macken-Horarik et al. 2006). The present article highlights a nexus of practice (Scollon and Scollon 2004) where such discourses intervenea rhetoric class in teacher educationand sheds light on how student teachers manage institutional affordances and constraints in oral presentations.
More specifically, focus is directed on an interactional phenomenon that takes place during oral presentations in a teacher education setting: role shift, i.e., the multimodal process of a social actor taking the role of another character, identity or role in interaction (originally from sign language research; see Metzger 1995;Quer 2018). This is studied in a rhetoric class for high school student teachers, where they are supposed to give three obligatory oral presentations in three sessions, respectively, during which they receive feedback on presentation technique from their tutor and from fellow students. The student teachers are given two major role alternatives in which they may perform their presentations: in the role of teacher or in the role of student. Furthermore, they must define their audience as fellow students, pupils, or other teachers. Since the latter two are fictitious, the educational design presents the students with the institutional offer of engaging in role play. It is noteworthy that the students' level of commitment to a role is not considered for their final grade. The role shifts of interest in this article are the shifts from student teacher to teacher in the initial stages of a number of student teachers' oral presentations (more information about the data in section 4.1).
Thus, the object of study is an institutionally required role-play activity in a university class that offers a potential connection to a future profession as teacher, in the sense that role play is a way for a novice to practice a professional role in a controlled situation (Joyner and Young 2006). The idea that there are links between practice and social actors' identities has long been recognised by Wenger (1998), and his recognition of these links may motivate the rationale of studying actions that have the potential of facilitating outbound trajectories from the rhetoric class towards the teacher profession (see Wenger 1998). That is, role play is potentially an anticipatory activity pointing to future actions in the profession (see anticipatory discourse, De Saint-Georges 2012). Thus, this article studies a social practice with potential relevance for (1) the future profession of studentteachers, as well as for (2) the trajectory of the educational process of student-teachers towards becoming teachers.
In the present article, the negotiation of roles between social actors and institutions is seen as a communicative effort used to produce participants' social actions. Hence, mediated discourse analysis (MDA: Scollon 2001;Scollon and Scollon 2004) is used as an overarching theoretical outline. In order to analytically focus on social action, the methodological operationalization of MDAnexus analysisis employed. Furthermore, to emphasise the communicative aspects between social actors and surrounding discourses, the article strives to conceptualise role shifts as communicative projects (Linell 1998b(Linell , 2009(Linell , 2010, see section 3 for an overview). The article can be placed in a sociolinguistic tradition (see Coupland 2016;Gumperz 1999) that employs a discourse analytical approach to interactional data (cf. Scollon 1999). The aim of this article is, firstly, to account for how student teachers' role shifts from student teacher to teacher may be viewed as communicative projects and what can be gained from such a view, and secondly, to shed light on how student teachers manage institutional affordances and constraints affecting the role shifts from student teacher to teacher during academic training. In order to achieve this aim, the following research questions are asked: -To what extent does interactional role shift comply with the notion of communicative project? -How do student teachers perform the role shift from student teacher to teacher in role-played classroom interaction? -How is discourse used to cope with institutional constraints during the student teachers' oral presentations?
The theoretical framework will be presented below, in relation to relevant research, followed by descriptions of data collection and analytical procedure. After this, the introductory phases of two student teachers' oral presentations, represented through transcriptions, will be analysed. In the following discussion, the role shift from student teacher to teacher is described as a set of three main actions linked together in a potential overarching communicative project.

Theoretical background
Performing oral presentations in a rhetoric class during teacher education could be described as a kind of activity type. The notion of activity type originates from Levinson's (1979) observations that speech acts and speech activities seem to be tightly intertwined in different activities, where activities should be understood as 'larger patterns of actions and interactions' (Linell 2009: 202). Activity types, then, can be seen as settings that regulate these patterns of actions in different ways, e.g. by imposing goal-defined roles and limiting events through constraints (Levinson 1979: 69). Speaking about an activity type is a way of categorising a setting through the perspectives of its participants, i.e., it 'defines the situation for the actors' (Linell 1998b: 235).
Activity types may be intertwined and influencing each other in different aspects, thereby rendering hybrid activity types (see Linell 2009;Sarangi 2016). Adding a communicative aspect to Levinson's (1979) notion of activity type, Linell could be said to be positioning his theoretical stance, on one hand, by building on it, and, on the other hand, by deviating from different features of speech act theory (cf. Searle 1979). Moreover, Linell implements Bakhtin's (2006[1986) dialogical view on interaction to develop traditionally monological perceptions of discourse. One of the results is Linell's notion of communicative activity types (CAT: Linell 2010Linell , 2011. Examples of CATs would be seminars and meetings, i.e., settings that assign roles to their participants, thereby simultaneously enabling and limiting different types of interaction. A CAT can be recognised by its strong relation to social situations, where participants are familiar with its nature, i.e., a CAT often has a name that reveals a participant's awareness of its properties (Linell 2009: 202). In relation to this, there is a need to linger on the role of discourse in relation to context and social action in interaction. As a point of departure for such a discussion, one can view discourse as surrounded by an array of contexts, which in turn are constructed by different contextual resources, e.g., social actors and varied background knowledge (Linell 1998a: 144).
Even though he points to the importance of how written texts are used in interaction, Linell (2010: 53) often seems to focus on the mode of speech in his work. If one wants to acknowledge the importance of written text, speech and other modes in interaction, it is necessary to adopt a theoretical framework that allows such analysis. Furthermore, when interested in what discourse means to the social actors using it, directing all attention to discourse may be limiting. These are motivating factors for shifting analytical focus from discourse to social action, a main trait of MDA, where meaning is seen as a result of how discourse is used by social actors rather than embedded in discourse itself (Jones and Norris 2005: 4;Scollon 2001). Compared to Linell (1998a), which puts discourse in the centre of a group of contexts, MDA puts social action in the centre of analysis, where discourses in place is one of three main forces (the other two being interaction order and historical body) that need to be acknowledged when studying social action (Scollon and Scollon 2004). That is, instead of solely looking at a piece of discourse, one ought to ask what role discourse plays in the action taking place at that point in time. The role of written texts in interaction is a suitable example: If a written text is actualised in interaction, it is done so as a discourse used in the action taking place at that moment. In a study of communication between supervisors, student nurses, and patients, Jansson (2010: 204) shows how journal notes are used by a supervisor as a resource to reframe a professional activity as a pedagogical activity. Interpreted in the vein of MDA, the journal notes serve as an example of discourse used by the supervisor in the action of tutoring a student nurse. Texts used in this way could be described as recontextualised (cf. resemiotisised, Scollon and Scollon 2004), defined as "the dynamic transfer-and-transformation of something from one discourse/text-in-context (the context being in reality a matrix or field of context) to another" (Linell 1998a: 146). Recontextualisation of written texts, then, concerns how they are traversed and used in different ways across different practices (Ajagán-Lester et al. 2003: 230), which, in Jansson's (2010) study, becomes apparent in the altered use of the student nurses' journal notes, where the texts are used to facilitate the movement between different activities of professional nursing and tutoring/education.
In summary, the article sees social action as the main analytical unit, in combination with a dialogical view on communication in institutional settings, thus combining theoretical notions from MDA with the concept of communicative project. Although these two approaches may look differently at the role of discourse in interaction, they unite in the urge to use sociolinguistic and contextual knowledge to understand social action.

Communicative projects
A CAT encapsulates different communicative projects (hereafter: CP) that, in turn, are embedded in each other, e.g., the CAT of a meeting being constructed of the CPs of greeting, opening the meeting, arguing for one's idea, summarising, etc. CPs have previously been studied in various settings, both in institutional and everyday interaction. The concept of CP has been used in research of institutional settings to describe local activities, such as admitting or denying guilt in legal settings (Linell et al. 1993). Furthermore, on a larger scale, Rönn (2009) shows how the CP of a joint story is constructed by students at an acting school. This leads to students depending on other participants, which in turn may be an obstacle for them. The notion of CP has also been used to study and understand interactional features in everyday interaction, e.g., getting children prepared for bedtime (Goodwin and Cekaite 2013) or writing a wishlist for Christmas (Ottesjö 2007). Furthermore, Norén (2007) is interested in the communicative function of a specific rhetoric figureapokoinouand how it can be used as a tool to close a CP or connect to pending CPs. A unifying feature of previous takes on CPs is the focus on interaction between several speakers, which is logical, given that Linell (1998b) points to the dialogical nature of a CP. In the data used in this study, the students are both sole speakers and are responsible for managing interaction in their presentations. However, I will argue that the role shifts analysed in this article are still examples of dialogical communication.
A CP is rather ambiguous in its nature (Linell 1998b), and there are several characteristics that may constitute a CP. I have discerned three general criteria based on what could be described as the main traits of a CP, which are specifically based on Linell's (1998bLinell's ( : 218-219, 2009Linell's ( : 188-197, 2010 research: -A CP is partly planned or anticipated, and its main goal is to resolve a communicative task. It emerges from participants' actions and is completed through interaction. -A CP can be both the acting (process) and the act (accomplishment); thus, it is complex and multifunctional. Adding to the complexity is the tendency to be embedded in other CPs, meaning that a whole encounter can be described as a CP. -A CP is dialogical. It involves at least two participants and is, to its core, in opposition to monological ideas, such as an autonomous actor producing speech acts in isolation.
Looking at the example of a CP used in the literature in light of these criteria, Ottesjö's (2007: 210) analysis of the CP of completing a wishlist for Christmas is a good vantage point. It is obviously an anticipated activity, due to the yearly occurrence of Christmas. It is also constructed of several actions linked through dialogue, e.g., where two children tell their mother about what they wish for Christmas (act), which she then writes down (accomplishment). The criteria above will furthermore be used as a basis for discussion in section 6.

Methodology
In this section, the data construction is initially described, followed by some notes on transcription. In order to highlight how the student teachers use discourse and multimodal resources to produce their actions in the process of role shift, the analytical approach employed in this article is a combination of Scollon's (2004, 2007) nexus analysis, and the concept of CP (Linell 1998b(Linell , 2010(Linell , 2011. Thus, nexus analysis will be described in further detail.

Data overview and transcription
The data used in this article consist of video-recorded oral presentations, performed by seven student teachers over three sessions. Additional data of observational field notes and instructional texts were gathered as well.
The oral presentations took place over the course of three sessions during a rhetoric class for student teachers, where the participants were supposed to perform oral presentations in order to develop their presentation techniques. Each session was approximately 3 h long, and attendance at all of them was required for the students. Students attended the class during the second year of a teacher education programme oriented towards high school teaching, at a Swedish university. Prior to each seminar, a lecture was given that focussed on different aspects of classical rhetoric, aiming to supply the student teachers with relevant tools for their subsequent presentations. As a final assignment, the students wrote a short self-reflection based on their performances and the feedback they received.
At the beginning of the rhetoric class, the students were provided instructions; two main instructional texts were of particular importance in the study: (1) a template text providing information on how an instructional presentation should be structured, and (2) a study guide (see Appendix A) with information on the written assignments included in the course, as well as information on potential role combinations for the presentations. The student teachers could choose from performing their presentations as 'themselves' or as a teacher, and they could address the audience as fellow student teachers, teacher colleagues or pupils. An overview of the distribution of roles regarding presenter and audience for all students participating in the rhetoric class, across all three sessions, is shown in Figure 2 (Section 5.1).
When performing the oral presentations, student teachers were expected to use a manuscript (as the study guide instructs them to) and to write a situation analysis, where the role of the performer and the audience was explicitly described, together with a general aim of the presentation. The situation analysis was to be handed to the tutor prior to each presentation. Besides performing oral presentations, the student teachers were expected to produce written texts for themselves (manuscript) as well as for their tutor (manuscript + situation analysis). The audience, consisting of the tutor and the other students, had the task of observing different rhetorical aspects of the presentations in order to give feedback when the presentations were finished. In order to highlight the complexity of instructions affecting the student teachers, Figure 1 strives to show relevant instructions, together with expected oral and written production of an individual student teacher during one oral presentation.
As Figure 1 highlights, the student teachers were given two main texts consisting of different instructions. They were then expected to produce an oral presentation, together with two new written texts: a manuscript and a situation analysis, which they were supposed to provide to their tutor. Besides this, the manuscript also served as a physical text that the student teachers could use during their presentations.
The transcriptions of spoken discourse were made in accordance with the Jeffersonian transcription system (Jefferson 2004), whereas the more detailed transcriptions of multimodal actions were influenced by Mondada's (2016) and Norris' (2004Norris' ( , 2011 conventions for multimodal transcription. Although the transcriptions in the article pay attention to rather detailed interactional features at times, the analysis does not aim to be in the field of conversation analysis. However, a basic influence of conversation analysis is almost ineluctable when descriptively showing what happens in interaction. The translations from Swedish to English were non-literal, with the main purpose being to facilitate understanding of the content. Each transcription consists of two partsexcerpt a and excerpt bwhere excerpt b is a more detailed multimodal transcription of a specific part in excerpt a. For further information on transcription conventions, see appendix B.

Nexus analysis
A nexus of practice could be regarded as a reoccurring action in space and time where different practices are intertwined (Scollon and Scollon 2004). Implementing nexus analysis means approaching a nexus of practice through three general phases: engaging the nexus of practice, navigating the nexus of practice, and changing the nexus of practice. These three phases act as a guide for a researcher to approach data construction, analysis, and the relevance of results. There are no clear-cut boundaries between the phases; however, I will discuss them one by one in order to make the methodological choices of the study as clear as possible. Interactional role shift as communicative project My way of engaging the object of study was to employ an ethnographic approach. First, I found out about the rhetoric class through discussions with the tutor. I met with the student teachers, who all seemed interested in participating in the study. All participants in the study were anonymised and signed informed consent. I also participated in some of the student teachers' activities during the course, such as voice warm-ups. I recorded and charted all seven student teachers' presentations (a total of 21 presentations), searching for instances where role shift might occur. As I had an interest in role shifts with potential relevance for the future profession of student teachers, an explicit role distribution in which student teachers perform as teachers served as a sample principle. That is, I observed presentations more closely when student teachers explicitly stated that they were performing their presentations in the role of teacher, since such presentations made the occurrence of role shift between student teacher and teacher highly probable.
While navigating the nexus of practice, I found potential in the introductory parts of the oral presentations and made a collection of transcriptions of each student teacher's introduction. I noticed the students' use of different semiotic resources, mainly texts, and employed the notion of recontextualization as a tool to trace the student's actions to larger discourses. Finally, two student teachers were selected as representatives for this article. The two student teachers -Anna and Joelperformed their presentations in the role of teacher during session two of the rhetoric class, and I made a more detailed analysis of their presentations. The reason for this sample was (1) the explicit role distribution, which signalled a shift from student teacher to teacher, and (2), that the student teachers were in the midst of an ongoing rhetoric class and had become relatively familiar with the presentation form.
Regarding the phase of changing the nexus of practice, the video-recorded presentations were almost immediately integrated in the educational process in collaboration with the students' tutor, through their status as feedback resources for the students. Video-recordings are often used as a pedagogical tool in rhetoric education, and here they served as a chance for the participants to develop features that were otherwise difficult to capture through feedback, such as minor hand movements. Hence, because of the data construction, the students were able to watch the video-recordings of their own presentations. Furthermore, the research question concerned with how student teachers cope with institutional constraints points to possible results suggesting changes in educational methods that may facilitate the students' development as future teachers.

General results and tendencies
In order to discuss results more generally, I will provide an overview of all of the student teachers' explicit role distributions in the rhetoric class, as well as some information on tendencies found in the different stages of analysis. Figure 2 shows the different roles explicitly distributed by the student teachers in all the data in three categories and is useful for highlighting the student teachers' trajectories (here, more local than in the case of Wenger 1998) of role distribution in the rhetoric class. The theme of the first session is personal presentations. During the second session, student teachers perform instructional presentations. Finally, the third session concerns argumentative presentations. Students with two oral presentations during the same session had been absent during previous sessions; thus, they needed to compensate with an extra presentation. The 'no explicit roles' tag means that the participants did not explicitly reveal the role distribution.

Student
Session one Session two Session three The overview of role distribution shown in Figure 2 provides information on how the student teachers chose to perform their presentations. It is interesting to observe that none of the student teachers chose to perform in the role of teacher during all three sessions, and that performing in the role of teacher was most common during instructional presentations. It is also noteworthy that several presentations during session three were performed without explicit mention of the roles of presenter and audience, thus revealing a tendency among most students to not explicitly state role distributions. This highlights a potential development in which students used previous presentations as social resources in the contextualisation process and can be interpreted as students gradually appropriating the presentation form as the course proceeds. Closely related to the phenomenon of appropriation is embodied experience (e.g., historical body), and the effect of experience can be observed in the special case of Eric, who was in the final stages of his teacher education. He performed all three presentations without an explicit role definition. However, his case is not subjected to detailed analysis in this article (instead, see Author forthcoming).
As for the general tendencies, the analysis shows that student teachers' role shifts may be seen as three main actions linked together: setting the scene, changing perspective and performing the new role. This is illustrated in Figure 3.
A majority of the other student teachers who chose to perform their presentation in the role of teacher shares this pattern in their introduction. To provide a more detailed analysis, this article focusses on the cases of Anna and Joel, which will be presented in detail in the following section.

Anna
From this point, the article provides a more detailed look at two student teachers who explicitly distribute the roles of teacher (presenter) and pupils (audience) during session two -Anna and Joel. Excerpt 1a shows the introductory sequence of Anna's instructional presentation. Anna has chosen to give a presentation in the role of teacher, instructing a class of eighth-grade pupils on the practicalities surrounding a school trip to a museum. She is one of the few student teachers in the data who engage rather vividly in the role-play aspect of the presentation.  Interactional role shift as communicative project In Excerpt 1a, Anna engages in the three main actions: setting the scene (lines 01-02), changing perspective (lines 02-03, described in further detail in Excerpt 1b) and performing the new role (lines 03-04). First, the main action of setting the scene appears to have the function of framing (Goffman 1974) the situation. This is done in interaction by (1) distributing the anticipated roles to presenter and to the audience (lines 01-02), and (2) defining the situation in which presenter and audience are going to be involved ('we're going away on a trip', line 02). Setting the scene this way makes it clear that at least two activities can be expected to blend during the presentation: the activity of giving an oral presentation in a rhetoric class, and the activity of informing pupils about a class trip. Thus, hybridity is prefigured at this point. The expression 'we're going' (line 02) points forward in time, i.e., the action of setting the scene encloses anticipatory discourse (see De Saint-Georges 2012) in the mode of speech, since it anticipates future actions. Simultaneously, this limits the future line of preferred actions in the presentation, and participants in the room are now aware of the framing of the presentation.
To explain why Anna frames her presentation in this particular way, it is helpful to observe her use of available discourse, mainly her use of instructional texts. What draws analytical attention to such texts in the first place is that Anna seems to use them by establishing a rather firm rhetorical structure of her introduction. She does this in the mode of speech. First, she provides information on who she is, then who the audience is, and finally, what they are going to do. The rhetorical structure makes sense in relation to the overarching CAT of a rhetoric class.
As introduced in section 4.1, students are facing (at least) two different texts: the template text and the study guide. Figure 4 shows part of the template text that covers the introduction of an instructional presentation. At the top of the figure, the Swedish original text is shown, with an English translation below.
In the template text (Figure 4), examples of how an instructional speech may be structured is presented. The first sentence in the introduction box states: 'tell what you are going to do'. In light of this, it is possible to understand Anna's framing of her presentation as a response to the request from the template text. It is noteworthy that the explicit role distribution observed in lines 01-02 is not part of the template text, even though Anna does this verbally in the action of setting the scene. A possible explanation for this is that Anna Figure 4: Excerpt from the template text given to student teachers by their tutor prior to session two. combines the use of the template text with another textthe study guide (see Appendix A)which provides an overview of the assessed tasks that the student teachers need to accomplish in order to pass the rhetoric class.
In the study guide, the task specified for the instructional presentations gives the student teachers information on role combinations regarding presenter and audience. Furthermore, this text provides information on the outline of a situation analysis, which the student teachers are supposed to provide to their tutor prior to each seminar. Here, under the heading 'Me', the rather interesting formulation 'Who am I and in which role am I speaking?' can be found. Anna, then, could be said to be answering part of a question laid out in the study guide in her presentation, thus making the text a participant in the interaction through a question-answer sequence. At this point, she uses the study guide to interact with a larger institution.
The study guide is recontextualised through Anna's action of distributing roles and is thus transformed to a useful tool in her role shift. While framing her presentation as being in a professional setting outside the seminar room, Anna simultaneously gives the presentation an institutional framing through recontextualisation of institutional texts. I interpret this as an example of integrated hybridity between institutional and professional activities (cf. Rönn 2009), triggered by Anna's use of discourse.
Moving on, the analysis concentrates on a focal point in the interaction, which happens at the end of line 02 in Excerpt 1a, immediately after Anna distributes roles and sets the scene. This instance is transcribed in further detail in Excerpt 1b below. At this point, Anna initiates the main action of changing perspective. Excerpt 1b shows how Anna pauses for 1.8 s before her next utterance. She shifts her gaze from looking at the audience to her manuscript, and back to the audience again. Her manuscript is located on the table in front of her, behind the basket that can be seen in the images in Excerpt 1b. The basket belongs to the tutor and contains what could be called "teaching utensils", such as whiteboard pens and books. After Anna's presentation, the tutor notices the impractical placement of the basket in front of Anna and removes it. In Excerpt 1b, the modes of gaze and posture appear as important features. Anna's gaze turns to the manuscript 0.7 s into the pause, and while still focused on the manuscript, she produces the prolonged sound 'em: ' (equivalent to 'um' in English). Following this is a micropause, during which Anna's gaze quickly returns to the audience, and she moves her head from left to right in a short motion. She then adjusts her posture by taking a small step backwards with her left foot, uttering 'well e: as you all'. One of the pragmatic aspects of the linguistic sign um is its function as a discourse marker, where it refers to, e.g., timing and intention to speak (see Shiffrin 1987). In this case, Anna uses it and immediately follows with 'yeah', which appears to have the function of contextualisation cue.
Anna's use of the manuscript is interesting. The manuscript is obviously a written text, but in comparison to the recontextualisation of the template text and the study guide observed in the analysis of Excerpt 1a, the manuscript also seems to function as a physical object used in the interaction as a point of orientation for Anna's gaze. After returning her gaze from the manuscript to the audience, Anna takes a small step backwards, starting her next utterance with 'as you'. I interpret this as the start of her third main action of performing the new role, where Anna, through the modes of movement and gaze, initiates the next main action. The data does not show whether Anna actually read the manuscript, but her gaze returning to the audience from the manuscript may be understood as a closing of the action of looking at the manuscript, thus her utterance beginning with 'as you' may be interpreted as a breakthrough into performance (Hymes 2004(Hymes [1981) of the new role.
In summary, Anna produces a chain of actions coming together as a role shift. Each main action has a different function for the role shift, where the action of setting the scene has the anticipatory function of framing the situation, and the action of changing perspective represents a multimodal practice linking the previous main action with the next main action of performing the new role.

Joel
In light of Anna's role shift, it is interesting to compare it to the corresponding sequence of another student teacher's presentation. Thus, Joel's presentation is primarily used as a comparative case to Anna's presentation. Joel is studying to become a high school teacher with religion as his main subject and has chosen to give his presentation as a teacher talking to his high school pupils about the importance of ethics. His introduction is shown in Excerpt 2a. The structure of Joel's introduction shows a recognisable pattern in relation to Anna's presentation. He first sets the scene by distributing the anticipated roles in a rather general way (lines 01-02)he is a teacher and the audience are high school pupils. However, Joel signals a potential role shift without framing his presentation with a specific activity suitable for the anticipated roles, hence deviating from the instruction in the template text that states 'tell what you are going to do'. It is not obligatory to structure the presentation in accordance with the template text, but Joel not using it is a rather interesting feature in relation to the potential implication of resisting potential institutional constraints when performing assessed tasks in an educational setting. At this point, Joel recontextualises the study guide in the action of role distribution, thus framing his presentation by distributing the roles to himself and the audience. Consequently, the main action of setting the scene anticipates a rather strong focus on role performance. The similarities between Anna and Joel's introductions appear to follow Scollon's (1998) idea of maxims of stance, meaning that, when opening an interaction, a social actor attends to channels of communication and relationships between social actors before introducing a topic. It is visible in the data as both Anna and Joel point to the roles of themselves and their audience almost immediately at the beginning of their presentations. Making role distribution explicit through the mode of speech is likely due to the instructional texts and institutional framing (the fact that the oral presentation is an assessed activity).
After setting the scene, Joel initiates the main action of changing perspective through his first micropause on line 02. This is transcribed in further detail in Excerpt 2b. As in Anna's case, Joel uses different modes to produce the main actions, which in combination with the use of contextualisation cues changes the perspective. As previously noted, he is using his body in an interesting way through the mode of movement, from left to right, that can be seen in images 1-3 (Excerpt 2b), where he seems to be moving bodily into the new perspective. This is similar to Anna in Excerpt 1b, but more visible in Joel's case. When he stops, he looks up and starts speaking in a new voice. Shifting between informal and informal talk can function as a marker for a new voice (Sveen and Magnusson 2013: 91), and a change in voice is one way to express a new role. All of this points to this being Joel's first utterance in the role of teacher, an action that is multimodally produced through the modes of speech (shifting voice), gaze (looking towards the audience), and movement (stopping). Considering this, Joel's change of perspective appears rather distinct.
In summary, Joel's role shift shows many similarities to Anna's role shift. However, Joel does not recontextualise the template text in his first main action of setting the scene. In his second main action, he uses his body in a relatively large sideways movement, distinguishing his change of perspective from Anna's. This movement consequently transitions to his use of formal speech, linking change of perspective to the third main action of performing new role.

Discussion and conclusions
One important aspect of the article is to relate the analysis of the two student teachers' role shifts in oral presentations to the question of whether role shifts can be conceptualised as CPs, and the potential fruitfulness of doing so. In order to initiate this discussion, I will draw on the three criteria (see section 3) based on Linell's (1998Linell's ( , 2009Linell's ( , 2010 research and apply them in light of the empirical results. The criteria will be discussed in turn. First, a CP is planned or anticipated with the aim of completing a communicative task. In the case of Anna and Joel's oral presentations, the role shifts could be considered to be anticipated. Firstly, the role shift is institutionally established in the task of performing the presentation; secondly, the student teachers themselves make it anticipated in the first main action of setting the scene, where they distribute roles to themselves and the audience. Despite Linell (2010: 39) pointing to CPs as sometimes being vaguely planned, the role shifts studied in this article appear to be rather firm and well anchored in institutional discourse, mainly through the student teachers' use of instructional texts. The choice Anna and Joel make to perform their presentations as teachers consequently makes shifting to the role of teacher a potential institutional constraint, and thus a communicative task that needs to be resolved. Anna's and Joel's role shifts complete institutional goals that are expected to be accomplished in the presentation.
With respect to the second criterion of process and accomplishment, one of the key characteristics of role shift is its constitution of a set of actions, chained together in order to achieve a function (Stec et al. 2016: 3). Based on the analysis of the data, highlighted in the examples of Anna and Joel performing their oral presentations, it seems that role shifts occur through what could be identified as three overlapping main actions: (1) setting the scene, (2) changing perspective and (3) performing the new role (see Figure 3). The three main actions of role shift are not, as Figure 3 might suggest, separated entities, but instead the student teachers' role shift comes through as a process where the subsequent main actions are linked together (cf. the scales of actions, Norris 2017). The action of setting the scene is used to make the anticipated role shift explicit in interaction and to frame the presentation, and is, in the cases of Anna and Joel, carried out partly by recontextualisation of different texts into different modes. According to my interpretation, drawing on the ethnographic fieldwork, the most relevant type of discourse for this action would be the different texts that student teachers have at hand. The second main action is described as changing perspective and shows the peak phase of the role shift. In this action, the student teachers combine spoken discourse markers with other modes such as gaze, posture and movement (cf. Helisten 2017 for a similar result).
Overall, the three main actions are highly multimodal practices, and the student teachers use different modes to efficiently produce the three main actions, as well as link them together. As Norris (2011: xiii) observes: "When taking all of the communicative modes into consideration that people use in their everyday lives […] the connections between actions and belongings, between individual and society, and between the hidden and the overt begin to make sense". In the analysis, we see that the action of performing the new role appears as a closing phase of the role shift and can be marked, as in Joel's introduction, by a change in voice combined with movement. Thus, paying attention to different modes of communication is necessary in order to understand the students' role shifts. The combination of contextualization cues and actions in other modes, in order to construct larger multimodal practices, is also something that has been noticed by other researchers (see, e.g., Adelswärd 2004;Helisten 2017).
Finally, considering the third criterion of CPs as being dialogical, the audience can be perceived as respondents, involved in understanding and producing minimal responses towards what is being said (Bakhtin 2006(Bakhtin [1986). But since the audience does not actively 'play their part' in the role-play activity (due to constraints in their institutional role of students in a rhetoric class), Anna and Joel use institutional discourse to potentially interact with virtual participants (Adelswärd 2004), making them habitants of a discourse world (Linell 2009, 98) where they can be respondents in interaction. Moreover, the use of institutional discourse, particularly Anna's use of the template text, where she answers questions from the text in her presentation, points to students being involved in interaction with larger institutions such as the department within which the rhetoric class is held. Thus, we see how different institutional and professional discourses meet in the oral presentations, taking place in the hybrid activity type of a rhetoric class. Considering all of this, the studied role shifts show several dialogical aspects, despite being performed in what appears to be a monological manner by single speakers. From an MDA perspective, the CPs studied in this article may be understood as higher-level social actions (Norris 2004) of shifting from the role of student teacher to teacher, in which other higher-level social actions (the three main actions) are embedded. Furthermore, the action of perspective change, for example, incorporates several lower-level actions in different modes (moving from the right to the left, gazing downwards and upwards, etc.). Thus, when using MDA as a main theoretical outline, the notion of CP may complement the analysis of social action, when interested in its goal orientation and communicative aspects.
The analysis shows that the studied role shifts from student teacher to teacher consist of several subsequent actions, some of which could potentially be considered CPs themselves, e.g., the main action of setting the scene. Thus, role shift, as represented in this article, can be interpreted as an overarching CP in which other CPs may be embedded. In turn, each CP consists of several mediated actions produced in different modes, such as moving from one place to another and verbally introducing a new topic. Conceptualising role shift as CP brings some clarity to another relevant issue: by highlighting how student teachers use institutional discourse in the form of texts to engage in interaction, we can see how they simultaneously maintain and navigate the constraints characterising the CAT of the rhetoric class through social action.
With respect to practical implications of the results, and thus changing the nexus of practice, the analysis indicates that the role play aspect does not appear to be the core activity (Linell and Thunqvist 2003) of the student teachers' oral presentations. The role play aspect is performed in a one-sided fashion, i.e., the audience does not actively perform their designated roles, possibly since they are institutionally constrained by other tasks. Based on the notion that contextualisation is identity making (Blommaert et al. 2018), it would be fruitful for educations to facilitate discourse related to a future profession as a teacher. In other words, if there is an interest in making role-played oral presentations more relevant to the future profession of student teachers, they perhaps ought to be more dialogical, in the sense that the audience would more actively perform the roles distributed by the presenter to a larger extent. Using role play in teacher education would point to the anticipated future as teachers, and since such use of role play is used in many vocational education programmes, this would be a motivated experimental change for the studied practice. The potential effect of such a change on student teacher development of teacher identity would be an interesting topic for further research.
In summary, the results points in several directions: firstly, the phenomenon of role shift appears to be a highly multimodal practice, where student teachers (cf. Stec et al. 2016) combine, e.g., gaze and movement with speech. Furthermore, it shows that role shifts have the properties of a CP as it is presented in this article. Studying them as such, with a focus on how discourse is used, provides insight into how texts regulate institutional constraints and are simultaneously used as tools for framing the presentations. Based on the analysis, I argue that viewing social actors' role shift as communicative projects may help us understand how they use discourse, such as texts, to manage institutional constraints.
Appendix A: Excerpt from the study guide Situationsanalys och manus Till varje uppgift ska du göra en situationsanalys som visar hur du tänkt kring uppgiften.