Participants' use of enacted scenes in research interviews: a method for reflexive analysis in health and social care. Social Science & Medicine, 151 . pp. 38-45.

In our study of a workforce intervention within a health and social care context we found that partici- pants who took part in longitudinal research interviews were commonly enacting scenes from their work during one-to-one interviews. Scenes were de ﬁ ned as portions of the interviews in which participants directly quoted the speech of at least two actors. Our analysis in this paper focuses on these enacted scenes, and compares the content of them before and after the intervention. We found that, whilst the tensions between consistency and change, and change management, were common topics for scene enactment in both pre and post-intervention data, following the intervention participants were much more likely to present themselves as active agents in that change. Post-intervention enacted scenes also showed participants' reports of taking a service user perspective, and a focus on their interactions with service users that had been absent from pre-intervention data. In addition, descriptions of positive feeling and emotions were present in the post-intervention enacted scenes. We suggest that this analysis con ﬁ rms the importance of enacted scenes as an analytic resource, and that this importance goes beyond their utility in identifying the impact of this speci ﬁ c intervention. Given the congruence between the themes prominent in enacted scenes, and those which emerged from a more extensive qualitative analysis of these data, we argue that enacted scenes may also be of wider methodological importance. The possibility of using scene enactment as an approach to the validation of inductive analysis in health and social care settings could provide a useful methodological resource in settings where longitudinal ethnographic observation of frontline care staff is impossible or impractical.


Introduction
If practice is the site of knowing and cognition is distributed amongst people in interaction with themselves and the material world (Nicolini, 2011) then our understanding of organizations and work practice in health and social care settings will be strengthened when we shift the spotlight as researchers to interactional data. One way to do this is to conduct ethnographic research in these settings, sampling and recording interactional episodes. However, participants may provide another window on their work practice when they replay 'enacted scenes' (involving the directly quoted speech of two or more speakers) during research interviews. Their choice of scene and their narration on the scene may provide opportunities for researchers to gain a deeper understanding of participant's knowledge and belief about their work practice. Being able to sample the interactional episodes from work practice is especially relevant to the field of health and social care where there is a need to improve the quality of care at the direct point of care during interactional episodes between the healthcare provider and the patient or service user (e.g. DH, 2001;DH, 2010;DH, 2012); in such settings it may not always be possible to record interaction directly. This paper reports on an analysis of 'enacted scenes' as reproduced in research interviews before and after a workforce intervention, and considers the import of these scenes both for understanding the impact of the intervention, and their wider methodological utility in researching the delivery of health and social care.

Background
Individuals vary in their styles of speaking, vocabulary used, length of utterance etc, but all speakers will adjust aspects of their stance in response to the interactional requirements in social contexts. During the context of a research interview the participant will be engaged in interactional work to help the interviewer understand their context and their perspective on it. Baynham (1996) suggests that dramatisation might be used to increase social closeness between participants in order to maintain involvement. One way to identify moments where participants are using dramatisation is to look for moments of direct speech. These can be easily recognized within written transcripts. In using direct speech as a device to re-enact a dramatic scene, the participant is taking themselves and the interviewer on a journey to a moment or issue of real significance to them.
The scenes that participants are enacting in their talk can be defined by the use of direct speech, where two or more speakers are involved. Tannen refers to direct speech as constructed dialogue and states that it is "a means by which experience surpasses story to become drama" (1986: 312). Furthermore, "the creation of drama from personal experience and hearsay is made possible by and simultaneously creates interpersonal involvement among speaker or writer and audience" (Tannen, 1986: 312). Direct speech is one feature of discourse which creates interpersonal involvement (Tannen, 1985: 134), where the focus of the narrative being told is dramatized in a re-telling, recreating the speaker's own initial reaction or prompting a similar reaction in the listener(s) by mimicking the event. The audience is given the opportunity to see the events for themselves, albeit in the manner intended by the reporter, allowing them to both observe and empathise with the reporter. Understanding constructed dialogue as rhetoric, it is shown that the scene will be reported in a way that encourages the listener to interpret the scene as the speaker themselves did (Holt, 2000).
But direct speech offers more than just a description, it provides "a type of demonstration" (Clark and Gerrig, 1990: 764) which pertains to authentic information, a form of evidence that carries an "air of objectivity" (Holt, 1996: 242) and to which the reporter, as a first-hand observer, has a "greater fidelity" (Li, 1986: 41). The idea of evidentiality is explored as an aspect of stance (Chafe and Nichols, 1986;Aikhenvald and Dixon, 2003), indicating one's basis and reliability for making their assertion. The enacted scene, however, is not characteristic of a stand-alone evidential in the way that researchers have tried to index stance and, similarly, modality through semantic and grammatical markers (Downing, 2001;K€ arkk€ ainen, 2003;Aijmer and Simon-Vandenbergen, 2004;Du Bois, 2004). Rather as an interactional evidential its meaning, function and role as evidence is realised in the conversation, is "mobilized interactionally across turns" (Clift, 2006: 583), and the audience still has the opportunity to interpret the scene, and to question its quality or impact as evidence.
As an 'auteur' of the scene re-enactment, the reporter manages stance at the three levels identified by Biber et al. (1999), that is, epistemic (concerning certainty, doubt, knowledge, imprecision), affect (concerning states, evaluations, emotions) and manner (relating to style of speech). Researchers have found displays of strong affect (Kochman, 1981) and a performative power ( Alvarez-C accamo, 1996) in giving voice to one's own or another's words. It is argued that a speaker's selective depiction (Clark and Gerrig, 1990) of scenes comes from a store of moments in their memory and scenes are selected and depicted as appropriate to the assertion they are trying to make. Whatever their stance is in the moment, the resources for scene enactment remain largely unchanged and so the very selection of a particular scene is indicative of their stance. Furthermore, scenes can be re-imagined in light of new evidence or knowledge, and a moment can have new meaning for the reporter. In this way, given the level of stance that is potentially invested in this re-enactment, the episode is highly personal and can offer a deep insight into the reporter's thoughts, views and feelings about its content.
It can be argued, then, that in using direct speech to re-enact scenes the participants are, in effect, providing us with a virtual ethnography. If the researchers were the ethnographers the selection of scenes for analysis would be at least partly made by the researcher. Placing an analytic focus on the scenes enacted by the participants puts them in the role of ethnographer because they are choosing what to highlight for the researcher. The scenes can potentially tell us something about their individual stance, but also about the organisational culture. In previous work (Pilnick and James, 2013) we have used Goodwin's (1994) ideas about professional vision to show how transforming culture and practice within communities is a socially constructed endeavour. According to Goodwin, the enactment of coding, highlighting and representing information can become ways of shaping perception, of showing others 'how to see' a particular object or event.
Additionally, we propose that scene re-enactment through direct speech is both a novel and relevant phenomenon to explore in the specific context of workplace intervention research. In the intervention underpinning this study, the context of the interviews, the interviewer and the job roles of the interviewees remained constant before and after the intervention, thereby minimising other potential influences on interaction. If the ways in which the participants enacted scenes from their work setting changed, we argue that this may provide evidence for the impact of the intervention itself. Other researchers interested in the interactional manifestations of authority, responsibility and entitlement have proposed that these can be linked to specific features of talk (e.g. Fox, 2001;Heritage, 2012). For example, Fox (2001) proposes that people making a bid for authority produce few or no evidential markers for their assertions. It follows that pre-and postintervention scene enactment could be explored from this perspective, to identify any changes in the ways that workers position themselves within sites of practice through their talk. We have chosen to use the term 'scene enactment' as opposed to 'vignette' because it indexes what we perceive as the speaker's degree of commitment to the talk, privileging the dramatic portrayal that characterises these episodes.
The purpose of this paper, then, is to show how we set about identifying scenes where interactional evidence was played out during a research interview, in order to demonstrate a) the purposes for which participants use the reporting of direct speech in this setting and b) to assess how this use of direct speech changes pre-and post-the study intervention. We end by reflecting on the relationship between our findings here, and the findings of the wider qualitative study from which these data are taken, and by considering the wider methodological utility of enacted scenes involving direct speech for health and social care research.

Method
For the purposes of this paper, we conducted a new analysis on participant interviews that were obtained for a main study (James et al., in press) evaluating the impact of a coaching-style workforce development intervention. The participants in the research were employees at a social care/education organisation where Video Interaction Guidance (VIG) was being implemented, and came from the health, education and social care sectors. The guiding principles of the VIG intervention and more details on the way in which it was carried out in this setting are provided in box 1. All the participants volunteered to receive the intervention and knew that this intervention aimed to support them in their work roles. The intervention is based upon person-centred goals, and all the participants had articulated their own personal goals for change. An example goal from the study is "I want to know that what I'm doing at work is making a difference to the children I work with". Working with this goal, the VIG guider would aim to understand what success might look like to the participant, and then agree a context for video recording everyday practice that would allow this to be addressed. The guider and participant watch the footage back together in a shared review; VIG employs a strengths based model so that positive evidence is focused upon and used as a basis for discussion about future practice. Such evidence may be found in very small moments of interaction, and may be entirely non-verbal. Successful use of VIG aims to enable the participant to develop new ways of understanding interaction and the interactional work that he or she does routinely, and to build on existing strengths. Given the fact that participants in this setting were employed in a range of job roles, there is no 'typical' interaction with a service user and so each individual examines only their own practice as part of the intervention, though they may choose to share their findings with others. Participants were interviewed before and after the VIG intervention took place.
The business case for the research funding required us to evaluate the impact of the intervention on practice. The main desired outcome was to support the creation of a more personcentred culture within the organisation. We therefore opted to conduct a pre-and post-intervention evaluation. As a component of this, an independent researcher conducted in-depth interviews, with 10 participants, focussing on three main areas: the intrapersonal level, investigating how people were talking about themselves; the inter-personal level, investigating how people were talking about their everyday interactions with peers and service users; and the societal level, investigating how people were talking about the general organisational culture, interactions with senior management, and their own role within the company. Whilst these three areas were used as a guide during the interviews, they were not explicit questions. The interviews were loosely structured around these three areas, and conducted in a person-centred manner with the interviewer following the participant's lead. Each interview lasted between 20 and 40 min and was videotaped and transcribed verbatim. Any individuals referred to by name were rendered anonymous. Participants were randomly allocated into two groups. Participants in group 1 were interviewed before and after receiving the intervention. The participants in group 2 acted as an initial control; they were interviewed at the outset, and were then interviewed again 4e6 months later, during which time they did not receive the intervention. This was to test for whether there was a substantial change in their talk that might be due to a sense of familiarity with the interviewer; in other words to test whether the interview was an intervention in itself. Participants in group 2 were then interviewed after receiving the intervention as per group 1. However, for the purposes of this paper we have not included data from the second pre-intervention interviews because these data were so similar to the first interviews, and added nothing new to the thematic analysis that was undertaken, the results of which are reported elsewhere (James et al., 2013, in press). Whilst participants had undertaken a variety of training prior to involvement in the intervention, they were not actively receiving any other training or intervention during their involvement with VIG.
In identifying enacted scenes, we read all the transcripts from the pre intervention and post intervention transcripts. We identified all instances of direct speech where the participant reported the speech of at least two speakers. It is worth clarifying here that there were times when participants reported the speech of themselves in isolation, or the speech of another person in isolation. These segments were not included in the analysis presented here because they did not provide the strong sense of replaying a scene from their work practice which was evident in the segments that contained the direct speech of both (or more) individuals in the interaction. The following example illustrates an instance of twoperson direct speech as reported by an interviewee: you have like evidence of everything that you do [sure] and that's placed in a placement plan [uh-huh] which is sent in for every single child, erm, and, you know, anybody who'd want to see, I dunno, their social worker can say, "well why's he doing that". "Well we thought we'd try it and look, it's really positive".
In this example, the participant is reporting the speech of the social worker and a member of the staff team in explaining the processes of formal knowledge exchange. The participant may be enacting a scene that actually took place, or it may be hypothetical. In this example the participant is using the direct speech within a description of individual placement plans. To take the direct speech out of the context and examine it in isolation would leave, "well why's he doing that". "Well we thought we'd try it and look, it's really positive". The direct speech in isolation does not enable us to

Boxed text the knowledge transfer intervention
The intervention was delivered as part of a knowledge transfer partnership (KTP). The aim of the KTP was to create a new person centred culture in the organisation by providing the workforce with a new way of seeing their work role. A video-based intervention was used that focuses on relational attunement known as Video Interaction Guidance (VIG) was used (Kennedy et al., 2011).
In the project application of VIG the practitioner took a brief (15e20 min) video film of a typical interaction between the staff member and a service user. The VIG practitioner then analysed the film using a set of contact principles to find moments of relational success between the staff member and a service user. From this analysis, the practitioner selected a few brief clips to feedback with the staff member. In a shared review of the video footage the practitioner and staff member discuss the clips. The VIG practitioner used a coaching style of communication to allow the staff member to explain and explore his or her perception of the successful moments in relation to his/her work role and specific goal for change. The social constructivist foundation of VIG means that it is focuses on the relational aspects of communication rather than viewing communication as a competency of the individual.
The approach was embedded within the organisation with a cascade model for training staff from across the organisation. The intervention approach meant that staff members were provided with a positive way of re-experiencing themselves in their work role using a person-centred intervention. The proposition was that this would equip them to craft person-centred approaches with the people they work with and, if the way of seeing is persuasive, change in perspective might arise as a result of receiving the intervention.
identify the work practice that the participant is evidencing using direct speech. For this reason, throughout our analysis we have considered direct speech in the context of the enacted scene, rather than extracting the reported talk alone.
Our decision to focus our analysis in this way came about initially because of the notable prevalence of enacted scenes in the data, and a resulting desire by the authors to better understand the uses to which this type of talk was being put in this context. Preliminary reading of the enacted scenes, alongside findings from the existing literature described above, gave a sense that they represented particularly significant moments or issues for individual participants. As the literature described above shows, reported speech has been an area of interest for researchers working from a number of closely related methodological perspectives that share an interest in the close analysis of talk, and also a focus on its action orientation: applied linguistics, discourse analysis and conversation analysis. Our own desire to explore this further was strengthened when preliminary thematic analysis of the enacted scenes showed a strong similarity to the results of the more extensive qualitative research that had been carried out around the intervention, suggesting that participants were using this kind of talk to present and demonstrate aspects of practice that were of particular significance to them. Table 1 shows the prevalence of enacted scenes in both pre and post intervention interviews, and the ways in which these map on to the thematic coding of the intervention study. Whilst there are a small number of codes from the wider study that were not represented by enacted scenes, the general 'fit' was noteworthy. We also noted that there were no instances of enacted scenes which did not align with the coding framework from the wider study.

Ethical considerations and consent
Ethical approval for the study was granted by Newcastle University Research Ethics Committee. Participants were asked to provide written consent to being interviewed, which stated that they understood that anything they disclosed regarding named individuals would treated anonymously except in the event of any safeguarding issues becoming apparent, in which case a senior manager of the organisation would be made aware. All participants were asked to provide written consent to the use of video in order to receive the intervention (for further information see James et al., in press).

Results
Our first, and significant, finding was that all speakers used direct speech at some point during their interviews. We were confident therefore that the use of direct speech during interviews was not a stylistic feature of one or two participants, or a response to an individualised or unusual situation. Instead, it was common practice for participants to enact scenes and report interaction using direct speech during the in-depth interviews. Our second finding was that there was no observable difference between the quantity of direct speech used in the pre-intervention interviews compared to the post-intervention interviews. Having undergone the intervention did not impact on how frequently participants used these scenes to convey a sense of their work and work practices. Our third area of interest was whether the nature of the enacted scenes was consistent both pre-and post-intervention, or whether participants might instead place emphasis on different kinds of interactions. It is to exploring this third area of interest that we now turn. Our analysis here is underpinned by insights from existing literature in the fields of applied linguistics, discourse analysis and conversation analysis.

Pre-intervention interviews: enacting the organizational achievement of consistency and change
The most prominent topic in the enacted scenes contained in the pre-intervention interviews, which spanned a number of themes, was in relation to consistency; this commonly arose in terms of the way in which a need for consistency around the service users also led to a potential for the narrowing of the scope of the professional role. Depending on how it was described by participants, this might be coded under, for example, 'managing change', 'formal processes of knowledge exchange', or 'management and leadership'. This latter category is exemplified by one participant who reports the speech of multiple speakers on this theme.
Erm, the biggest one is normally inconsistencies [right], yeah. If you've got somebody, a key worker doing something and they particularly want their key child to be doing something and it's not getting done, you know, they can say, "well listen, you know, I have asked for this to be done when I'm not on shift and it's just Table 1 Longitudinal data codes with number of scene enactments pre and post intervention in brackets.

Intra
Inter Social *the first number in the bracket is the number of quotes in that code in the pre intervention interviews that contained a scene enactment, the second number in the bracket is the number of quotes in that code in the post intervention interviews with a scene enactment. Codes that only occurred in the post intervention interviews are at the bottom of the table and the numbers in brackets therefore refer to the number of quotes in that code that contained a scene enactment in the post intervention interviews. An absent number indicates that no enacted scene corresponded to a particular code. not happening". And then either myself or my line manager will say, "well why isn't it happening", you know, because that's one of the most vital things, you must be consistent at all times with our boys, it's-it's hugely important. Anderson (2005:63), in his study of organizational discourse, argues that what he calls 'represented voice' can be used to "discursively translate between a single situated utterance and organizational processes"; in other words it can serve as an example of what is 'typical' in a setting. In this example of an enacted scene staff members are specifically identified by role: key worker, line manager. These specifics are used to illustrate a larger challenge: that of inconsistencies in practice with service users. The final reported exchange exemplifies a challenging peer interaction, and the invocation of authority by management in resolving this is justified with reference to the 'vital' need for consistency. In our analysis for the wider qualitative study from which these data are drawn, a common narrative has emerged through which participants assert that stability can be achieved only if consistency is practiced (James et al., in press). Key workers for a child are seen to have a particular role in this achievement of consistency. What this enacted scene sheds additional light on are the kinds of interactions, and situated practices, through which the search for consistency may be practically carried out. In addition, the position of the teller is supported by the offering of evidence from a named third party: the key worker.
Given the emphasis placed by respondents on consistency, the issue of changing practice, and how change can be achieved, might be expected to be a tricky one. The balance between consistency and an openness to trying new ideas or strategies was also one which recurred in participants' enacted scenes. The interviewee below has been asked how s/he feels about change initiatives that originate beyond her immediate workplace, and responds as follows: Erm, but sometimes you do feel it's like well when they say oh well, erm we're going to do A, B and C. And you just think well hang on a minute, can we do this? And it's like no, that's what it's going to be, and you just think mmm … But you need to come over here and have a look.
In this extract, the interviewee invokes external management figures, who say "oh well … we're going to do A, B and C". These plans are described as being made without regard for the specific context of the service, but at the same time any attempt at dialogue about practicality of change from those working within it ("hang on a minute, can we do this?") is rejected. Through the enactment of this scene, and the dramatization of her experiences, the interviewee draws a sharp contrast between local and managerial knowledge in relation to change implementation, and makes a case for the site-specific nature of workplace practices; she makes a distinction between 'what is known generally' and what she knows through personal experience in this specific context (Macintyre and Oldman, 1977). The end result, however, is that change is foisted upon her, although it may ultimately be unsuccessful.

Post intervention interviews: enacting oneself as an agent of change
As we have already noted, in both sets of interviews, participants enacted scenes which specifically invoked the issue of consistency, and where change, and barriers to change were the basis of the enacted dialogue. However, post-intervention, there was a marked difference in how interviewees positioned themselves in relation to this change. Whereas the extract directly above positions the interviewee as a passive recipient of change handed down from on high (albeit one who can foresee problems with this change), post-intervention scene enactment depicted interviewees as agents of change themselves. The interviewee below is describing a disagreement between two members of staff about the way forward with a particular child, and her subsequent handling of the matter. What is immediately evident in the extract above is the degree of agency that the respondent affords herself. Her reported dialogue is prefaced with 'I said' and 'I says', marking clear ownership of the words that were spoken. The actions that were taken to resolve the issue and restore relationships within the staff team are described emphatically and through the use of voiced imperatives ("It's got to stop" and "In the office, the pair of you"). Recent work in conversation analysis has focused on the area of epistemics in interaction: examining how claims to knowledge are produced, understood and contested through talk (Heritage and Raymond, 2005;Heritage, 2012). One element of such analyses is epistemic authority: whether participants have the right (or primary right) to particular elements of knowledge, or to know how the world "is" (Stevanovic and Per€ akyl€ a, 2012). Stevanovic and Per€ akyl€ a (2012:298) expand on this further, by suggesting that there is a useful distinction to be made between epistemic authority and deontic authority: if "epistemic authority is about knowing how the world "is"; deontic authority is about determining how the world "ought to be"". From this perspective, the interviewee's talk in the extract above illustrates an enactment of deontic authority, by using her knowledge to determine what the consequences of the disagreement ought to be, and putting this into action. As a result, her agency in this scenario is foregrounded.
This focus on reporting a specific incident where the interviewee takes clear control contrasts with the more generalised depictions of change occurring that were seen pre-intervention.
However, whilst the above extract goes some way to demonstrating the differences in scene enactment post-intervention, perhaps this phenomenon is most clearly illustrated by comparing pre-and post-intervention data from the same participant around the theme of change. The following two extracts compare talk from the same interviewee, before and after the intervention.
Erm, it's like, "oh no, it's got to be this way, this is tried and tested, this works", it's like, "yeah, but we're trying this", so erm some people do find it hard to change. I suppose in some respects if it was a certain -if it was something I didn't like, I wouldn't voice and say, "no, I don't like that, that's wrong". But some do.
In the first line of the quote here she describes pushing back at her colleagues ("yeah but we're trying this"). In the later part of the quote she uses direct speech again, imagining how it would be if she were to fully assert herself. Whilst she acknowledges that other team members may do this, for her the words enacted in this scenario have not yet been played out.
When it comes to the post intervention interview the same staff member recounted a lived experience at work, recalling a conversation with her peers about a service user who had been displaying problematic behaviour and who had been excluded from group activities as a result.
I've done everything differently to them, I've gone about everything differently, we've stripped everything back and we've sat down and I said, "right why do you do these things with [L]", this is with the teacher, and they went, "well because she does this, this and this", and I went "right ok then so when does she hit out", "well if she goes near anybody", "so have you shown her how to do it properly"," no", "why", "because we don't let her near anybody", "well then why use a member of staff"," well because staff would get injured", "yes but if you're gonna use a member of staff", I says "fine [J] can bring her over and we'll initiate something with myself", and that's what we did.
In this quote she is an agent of change; there is a similar use of 'I said' and 'I done' that we have seen previously, denoting ownership of utterances and actions. She is asserting herself in the role of advocate for the child. Her language is emphatic 'right, why do you do these things with ....', and she presents herself as challenging the other speaker's assertions. She ends the scene with an imperative followed by a declarative ('J can bring her over/and we'll initiate something with myself'). As Heritage and Raymond (2005) have shown, declarative formats mark the taking of a knowledgeable epistemic stance, and building on this Cornillie (2009) has described the degree of confidence grammaticised into an utterance as its 'epistemic modality'. The staff member's unmitigated talk here marks herself as both knowledgeable and confident (Fox, 2001) and indeed the upshot of her story is that her peers did what she asked them to do. In contrast to the pre-intervention scene, then, at post intervention the participant is replaying a scene where she is successfully able to reject her colleagues' assertions.
At face value, the change in the assertion of this person with her peers over the well-being of the service users could be attributed to the impact of an intervention that was designed to develop confidence and self-efficacy in the participants. We might therefore have expected more confident interactional exchanges as a result of the intervention and so be inclined to see the second quote as an outcome of new behaviour as a result of the intervention. In fact, what is interesting is that the scene that she chose to play out in the post-intervention interview actually took place over a year before either of the interviews was conducted. Based on this evidence, what changed was not her behaviour itself, but the scenes that she chose to enact in the interview. This leads us to hypothesise that the impact of the intervention should not be considered solely in terms of observable or reported behavioural change at work, but also in terms of the way participants subsequently view and reflect on their workplace practices and their place within an organisation.

Interaction with service users
A second key difference in the post-intervention data is the presence of enacted scenes directly involving service users. Whilst all the pre-intervention interviews did contain enacted scenes, as we have previously reported, none of the scenes involved the voicing of service users themselves. The scene below is being enacted by a support worker, following a question from the interviewer about what the participant thinks is her greatest achievement over the course of the intervention; she describes her improved relationship with a particular service user, L.
"… Nine times out of ten, that's how she greets me in a morning [right] so as soon as she comes in it's like she'll come up and I say "now then Mrs" or I'll say something to her and it's straight away the arms are round me, the cheeks in and I'm like "Thphthph", blow a raspberry and then she's "eeeeee"! And I'll say aye morning and then she trots off and puts her things away and gets herself sorted but. Or if I'm not there she'll get herself sorted but as soon as she sees me she comes over and if I have a day off, she'll hover for about an hour, she'll follow me around and I'm not allowed to move too far away from her or she will get hold of me by the hand and sit me down at the side of her and say "right you left me yesterday, you're not leaving me today".
Although this service user (L) is largely non-verbal (i.e. she does not use speech and language as her primary means of communication), the support worker imagines her dialogue in order to create an enacted scene of greeting, which emphasises the interactional nature of their current relationship. In considering this kind of imagined constructed dialogue produced on behalf of an other with limited verbal skills, Webb et al. (2015) have noted the lack of prefaces such as "I think she's saying …", that might be expected where the thoughts of another could not definitively be known. They suggest that such dialogue does not require these kinds of prefaces when it is a) offered in a context which makes clear it is a possible rather than a definitive suggestion, or b) when the professional producing the dialogue can be seen as having special warrant to do so, through the nature of their relationship with the service user. In the extract reproduced above, then, it appears that the improved relationship the interviewee reports with L provides the warrant for an unmitigated assertion of L's thoughts.
Whilst the extract above provides insight into a changed interactional relationship with an individual, post-intervention scene enactments involving service users also shed light on broader work practices, and the rationales that underpinned them. In the following quote a teaching assistant paints a vivid picture of a busy classroom: I had three students the other day and I was saying, "I need more hands". N saying, "can you help me with this?" And C's saying, "can you do that?" And J was coming up and trying to sit on my knee. And I was like, "aah" because I try not to ignore any bids for communication. I try to respond to everybody who speaks to me. Which spreads you a bit thin … .. erm because they're nonverbal and because they do, so, a pull for your hand. And if you're just ignoring, "no don't do that", you've missed something that could potentially be quite fabulous and wonderful. And sometimes it is, and sometimes it's just "go and get me a drink". But sometimes it's really really special.
This participant is replaying a scene from her work practice in which we get a strong sense of the complexity of her work role. The complexity of this role is portrayed through interactional exchanges between herself and the children she is with. As in the previous extract, she reports the children's intentions in direct speech even though the children she is with may themselves be non-verbal, and again we would argue that this denotes a particular epistemic stance adopted on the basis of her relationship with them. However, the way that she narrates around the direct speech gives us a broader insight into her work practice, "because I try not to ignore any bids for communication. I try to respond to everybody who speaks to me". From this description we get a sense of how she sees her role in the classroom beyond her designation as a teaching assistant for specific childrene where she will try to respond to anyone who approaches and not just children with whom she is in a key-worker relationship.
This extract also touches on another theme that is evident from a more detailed analysis of the post-intervention interviews: a greater focus on positive feeling states and emotional responses. The respondent describes the way in which responding to a bid for communication can result in a moment that is 'really really special', whereas non-response could lead to missing something that is 'fabulous and wonderful'. Insofar as expressed feeling states have been considered as a phenomenon within talk, they have tended to be analysed in terms of epistemic status (Heritage and Raymond, 2005), and the fact that something characterised as a feeling is marked as experiential, and hence having a lower confidence stance than something that is known or thought. However, we argue here that invocation of feeling states can also be used as a means to highlight the impact of specific moments and their significance within a workplace context. Through this example, a member of staff describes how she puts a philosophy or belief about her work role into practice, through interaction with others, and the positive consequences that ensue.

Discussion
The analysis presented here underlines the importance of enacted scenes as an analytic resource. Through examination of these enacted scenes, we have identified how, though themes of change and consistency are common, the respondent's interactional location of themselves in relation to this change becomes significantly different post-intervention. We have also shown that, post-intervention, there was a focus on reporting scenes of interaction with service users; this focus was absent from the preintervention data. Post-intervention, respondents also described positive feeling states in relation to the work practices they enacted through reported scenes. As we have noted, one possible explanation for all these differences is a simple relationship between participating in the intervention, and becoming a more confident and engaged member of the work force. However, since we are dealing here with reported behaviour, any straightforward relationship of this kind is difficult to map. Nevertheless, we would argue that this post-intervention invocation of agency in relation to change, interactions with service users, and the inclusion of descriptions of feeling states gives an insight into how staff can attempt to incorporate their underpinning beliefs and philosophies into practice in the workplace. As Anderson (2005) has argued, considering how organizational change is formed and constructed requires a close examination of the language used at particular occasions in the change process.
One other analytic possibility that we needed to discount was that participants' choice of post-intervention scenes was influenced by the video footage of their practice that they had been shown during the workforce intervention. We looked closely at the extracts to see if the scenes that they were replaying in the postintervention interviews were merely those which they had remembered from watching. There were two cases where participants had replayed scenes from their experiences during the intervention, but in all other cases the scenes were replayed from their general work practice.
Alongside the analysis presented here, we also considered the relationship between the themes arising from enacted scenes, and the themes that arose from the more comprehensive qualitative analysis of the entire corpus of interview data, which was conducted using Grounded Theory (Glaser and Strauss, 1967). We found that the enacted scenes the participants chose to foreground for the interviewer showed the majority of the same overall themes that were present based on the inductive analysis of all the data. One way to interpret this would be that the analysis of enacted scenes 'validated' the main themes of the prior analysis, but did so in a way that was rooted in the participants' perspectives of their work practice. The participants' selections of the enacted scenes they shared with the interviewer gave them the opportunity to choose which areas of their work practice to highlight, and to position themselves in relation to that practice. In this sense, the positioning of the service user at the centre of post-intervention scene enactment is particularly significant. The need to get the person at the centre of care is an aim for policy in aging research (DH, 2009), intellectual disability (DH, 2001, family health services (Kennedy, 2010) as well as generic policy initiatives around patient choice (DH, 2012); on the evidence presented here, this intervention appears to have had an impact on the prominence of this aim in this setting.
We conclude with a consideration of the wider methodological implications arising from this research. We have suggested above that, when analysed in isolation from the wider data set, a focus on scene enactment and direct speech produced a validation of the overall study findings. That this should be the case is an idea perhaps already implicit in existing literature, where researchers writing from a range of closely linked perspectives have concluded that respondents use reported speech as a means of emphasising the crucial parts of a narrative, and evidencing key claims (e.g. Buttny 1998;Myers 1999;Wigginton and Lafrance 2014). However, the findings presented here suggest that reported speech may have even greater analytic significance than has previously been recognised. Whilst our findings are currently based on a single study, it is evident that an approach focussing on scene enactment and direct speech used in research interviews could have particular advantages for research in health and social care settings. In the UK, since September 2011, revised guidelines mean that Research Ethics Approval from an NHS Research Ethics Committee is no longer required if a project consists only of interviews with health and social care staff. 1 By contrast, because service users would also be observable in the context of an ethnographic study, full approval would be required even for a project that sought to focus on staff actions and interactions. It is not our intention to argue that such approval is to be avoided, but instead to note that, as a necessarily and properly rigorous process, it can be time consuming, complex (for example where some service users may lack capacity to consent) and in some circumstances, impractical. What we have presented here is a potential alternative approach, which may enable researchers to capture significant workforce-relevant themes and issues from more readily accessible interview data. Such an approach could be widely used, and would have potential methodological advantages over other reflexive methods such as member checking that are both practical and theoretical. Further work is needed to establish the utility, and the validity, of such an approach.

Declaration of conflicting interests
The authors declared no potential conflicts of interests with respect to the research, authorship, and/or publication of this article.

Funding
Funding for this research was provided by the Knowledge Transfer Partnership with funds from ESRC and Technology Strategy Board. Support was also provided by National Institute of Health Research via core funding to the Biomedical Research Unit in Hearing at the University of Nottingham where Deborah James was employed (2009e2011) and where Luke Collins was funded for his doctoral research.