Working with Knowledge in Clinical Supervision: A Qualitative Study

Clinical supervision is crucial for the professional learning and development of practitioners in many fields. While many have articulated what should happen in clinical supervision, and evaluative research has been undertaken, little is known about the details of what actually happens in clinical supervision. This study addresses this gap, focusing on how professionals (psychologists) work with knowledge while collaboratively working through complex problems of practice. Six psychologists (three pairs) participated, audio-recording five supervision sessions, and giving two interviews each. Transcripts were analysed to uncover what was discussed and how problems were approached. Four ways of working with knowledge (epistemic practices) were found: recontextualising practice knowledge, recontextualising theoretical knowledge, story-telling, and asking expansive questions. These highlight important features of professional supervision practice that have not previously been approached theoretically as epistemic practices.


Introduction
Clinical supervision (hereinafter referred to as supervision) is key for professional development across many professions.It is a practice designed to assist professionals in making decisions, reflecting on practice and supporting their developing expertise in serving the public (Bernard & Goodyear, 2014).Clinical supervision refers specifically to discussion of clinical practice i.e. casework, case conceptualisation, problems of practice etc.This is distinguished from managerial supervision, which focuses on compliance with the requirements of a work role.Professionals learn as they discuss their work, consider problems, and raise concerns with colleagues.In some professions, clinical supervision is mandated for continuing professional development (CPD) purposes, as is the case for psychologists in Australia.
A substantial body of supervision research exists, much of which focuses on the potential value of supervision and how it can be improved, with the intention that this will benefit those who consult healthcare practitioners.The current literature emphasises required competencies, evidence-based practice and accountability (Watkins, 2012), all key trends in contemporary professional practice.This paper explores a less understood area, namely how supervisors and supervisees actively work with knowledge in supervision.
What it means to be a healthcare professional is changing due to shifting institutional and societal conditions and demands, and rapid technological developments (Evetts, 2014;Jensen et al., 2012).Increasingly, "forms of knowledge and ways of knowing now needed by professionals have become more complex" (Markauskaite & Goodyear, 2014, p.82), suggesting that ongoing learning is crucial as professionals' careers progress.In a context of increasingly complex and contested knowledge environments, professionals need to learn to work with and integrate knowledge in new ways (Nerland et al., 2010).Furthermore, those educating future professionals need to help them develop ways to work with, process and contextualise knowledge as they develop in their professional practice.
The study reported in this paper focused on how psychologists work with knowledge as they learn in supervision, foregrounding the epistemic dimensions of supervision.'Epistemic' refers to knowledge, and what it means to know something (Damşa et al., 2010).This builds on recognition that professionals increasingly deal with complex forms of knowledge in their work.In this research, 'knowledge' refers to formal specialist knowledge in a conventional sense, but also encompasses knowing that, knowing how, knowing why, knowing for, knowing what and knowing when (Markauskaite & Goodyear, 2016).'Knowledge work' refers to the actions that professionals undertake as they work with and on the knowledge that pertains to their practice (Hermansen, 2014).Such work requires that professionals adapt knowledge to work on the particular problems they encounter in the course of everyday practice, in order to help their clients/consumers (Nerland & Jensen, 2014).
Although there is a body of knowledge that remains relatively stable in practising a profession, there are also features of knowledge that are uncertain, complex and emergent -not known prior to moments in practice, or prior to supervision (Hopwood, 2016).On the whole, professionals tend to work in knowledge conditions that are characterised by uncertainty, partiality and fragility (Jensen, 2012;Hopwood, 2017), making professional learning especially important.Hence this article's focus on knowledge work and the unpacking of epistemic features of supervision practice.The term 'practice' is used here in both everyday and technical senses.In the everyday sense, it refers to psychology as a form of professional practice, and supervision as something that professionals do in this context.More technically, epistemic practices refer to ways in which epistemic aspects of supervision are enacted -the key point being that this locates epistemic work in concrete says and doings, rather than as purely cerebral or abstract.

Literature Review
A great deal of literature on clinical supervision has been published, particularly in the last 30 years.Consideration of this literature provides a sense of how supervision has been studied, and where consensus and dissent are manifest.Studies have focused on a variety of areas such as defining supervision, understanding its functions and benefits, investigating what makes supervision more or less effective, how to optimise outcomes for clients, the significance of the supervision relationship, and outlining different models of supervision practice.
Several authors have called for more studies that make the content of supervision explicit, describing and analysing what actually goes on in supervision (Milne, 2007;Pearce et al., 2013;Pollock, 2017;West & Clark, 2004;Wilkens et al., 2017).Without such research, information about supervision content tends to be taken for granted, based on experience or anecdote.Such content is assumed to be broadly understood as covering key areas such as case conceptualisation, intervention strategies and ethical issues (Barletta, 2009).However, there is a need for more empirical detail about specifically how supervisors and supervisees work with the problems of practice in supervision (Gottschalk, 2020).
Some studies do examine the content of supervision, although they tend not be located in the Psychology field.Pearce et al. (2013) reviewed the literature in nursing and allied health with a view towards understanding what supervision content might be most appropriate and associated with better outcomes.They found four themes, namely reflective practice, task-oriented content, diversity of content and stress management.They concluded that research into content "is limited and of low quality" and that more research is needed (2013, p.139).In a mixed-methods study, McKenna et al. (2010) used phone and mail surveys to canvass nurses in New Zealand about supervision provision and content.They found that nurses in supervision most frequently discussed reflection on clinical work (78% of the sample), professional development (51%), interpersonal issues (38%) and organisational/management issues (33%).However, this study relied on self-report rather than observational methods, as does most research in this area (Wilkens et al., 2017).Further, in both of these studies, categories of content are broad and do not give details of what specifically was discussed and how this was approached.
Wilkens et al.'s social work study highlighted the lack of evidence "about what happens when managers and child and family social workers meet to discuss casework" (2017, p.942).These researchers audio-recorded and analysed thirty supervision sessions between social workers and their line managers.They found a similar pattern and structure across sessions, namely a "verbal deluge" by the supervisee, followed by a discussion to identify the problem, and concluding with a solution given by the manager (Wilkens et al., 2017, p.947).A further finding was that supervision focused on managerial oversight, with little time for reflection or discussion of the nuances of clients/cases.The educational and supportive functions of supervision (Proctor, 2011) were thus not being adequately fulfilled, despite managers stating that their priorities were for supervision to be reflective, supportive, analytical and helpful.

Aim of this Study
This study aimed to understand what takes place in supervision through a distinctive, fine-grained approach.Studies such as those discussed above tend to focus on the what of supervision content, rather than how supervisors and supervises go about discussing issues.This is addressed in the present study, which focused on uncovering both what is discussed in supervision and how practitioners work with knowledge in grappling with complex problems of practice.This paper emphasises the latter, outlining four key ways of working with knowledge that emerged.The research posed the question: 'How do psychologists work with knowledge in clinical supervision?'

Conceptual Framework
In order to understand the 'what' of supervision in more granular detail, and grapple more explicitly with the 'how' of supervision, it was necessary to operationalise a focus on knowledge.To do this, concepts of epistemic practices and recontextualisation were used.

Epistemic Practices
Epistemic practices are the means by which knowledge is produced, validated, legitimised and shared in a field of expertise (Cunningham & Kelly, 2017;Nerland, 2016).They involve ways in which problems are unpacked and analysed, means of verifying knowledge and strategies for finding solutions (Nerland, 2018), and focus on the "practical 'know-how" of knowledge", which is specialised for a contextual purpose" (Jensen & Strømsø, 2019, p.345).An epistemic practice approach shifts the focus from what knowledge 'is' to how knowledge is 'done' (Nerland, 2016, p.131).As Hopwood & Nerland explain, this perspective attends to how "knowledge is activated and becomes actionable, how reasoning is made explicit, how positions as knowers are taken in interaction, and how connections in action are formed on epistemic terms" (2019, p.3).
The concept of epistemic practices magnifies knowledge in use (Hopwood & Nerland, 2019), examining knowledge-related interactions as professional practice unfolds.Despite the relevance of this to professional learning in supervision, research into epistemic practices has not investigated supervision, instead focusing on the 'main' practices in professions including engineering (Cunningham & Kelly, 2017); child and family nursing (Hopwood & Nerland, 2019) and law (Enqvist-Jensen et al., 2017).
While the analysis in this study involved identification of actions in supervision, this was as part of an approach that took epistemic practices as a unit of analysis.
Practices comprise multiple different actions, including variation in the specifics of their enactment; this is consistent with the work of Jensen, Nerland, Hopwood and others (Hopwood & Nerland, 2019;Jensen, 2012;Nerland, 2018).

Recontextualisation
Linked to the notion of epistemic practices is the idea of recontextualisation.Guile (2014) regards recontextualisation as the enactment of professional knowledge in practice.Knowledge generated and practised in one context is changed when put to work in a different context (Evans & Guile, 2012).In the process, theory and practice "commingle" as practitioners actively work with knowledge and make professional judgements (Guile, 2014).Knowledge cannot simply be transplanted, inserted or transferred to another context.In order for knowledge to do work in a different context to that within which it was generated, it must be recontextualised (Evans & Guile, 2012).This suggests that a process takes place within supervision whereby knowledge is worked on and with to make it fit for purpose in practice.
Part of a supervisor's role is to link theory with practice (Holloway, 1995) by collaboratively offering and translating relevant theory into insights or actions that the supervisee might find useful.In order to do this, recontextualisation is key, yet this relevant concept has not been taken up in the supervision literature.Recontextualisation involves more than a simple transfer of knowledge.Guile suggests that professionals "are making conceptually-structured professional (i.e.practical) judgements in context-specific circumstances, rather than applying their theoretical knowledge practically or taking practical decisions that lack any conceptual content" (2014, p.89).Recontextualisation thus involves a complex interrelationship between theoretical understanding and professional experience, both of which constitute forms of knowledge on which professional judgements rest (Guile, 2014).

Method
The complexity of supervision poses a challenge to meaningful research (Holloway & Wolleat, 1994).This study adopted a distinctive approach in order to address: a lack of observation of "what actually happens in supervision" (West & Clark, 2004, p. 20); a shortage of longitudinal studies (Watkins, 2014a); a tendency to investigate supervision after it occurs (Watkins, 2014b); a reliance on self-report (Wheeler et al., 2011); a lack of attention to context (Rapisarda et al., 2011); and an over-representation of students or trainees as participants (Forshaw et al., 2019).In seeking a robust way to record what actually happens in supervision and to overcome these methodological dilemmas, this study adopted a qualitative approach.It involved close, detailed recording of the situated practice of qualified practitioners in real time, across time.
Three pairs of psychologists in regular supervision (in Australia and New Zealand) were participants in this study.To be eligible, participants had to be registered psychologists engaged in regular supervision, and provide informed consent to par-ticipate.All six who joined the study were women, with between 16 and 25 years of experience of practice.Details of the participants are shown in Table 1 below.
Recruitment of participants proved challenging, with few practitioners willing to expose their supervision practice to a researcher through being audio-recorded.This seemed to be due to the privacy of the supervision space and the perceived risk of evaluation.This left a small sample size, which was not a problem given the qualitative, detailed focus of the research.Thus, although participants may not represent supervision in general, they provide three in-depth and authentic instances of how supervision is practised amongst experienced psychologists in Australasia.Participants were recruited using an arms-length strategy (i.e. the researcher did not ask potential participants directly), drawing on the researcher's extended professional networks.Ethics approval was sought and granted by the UTS Human Research Ethics Committee.Participants provided written consent and understood that they could withdraw from participation at any time.
Each pair audio-recorded five consecutive supervision sessions, which were emailed to the researcher.Each participant was individually interviewed twice: after the second supervision session, and at the end of the five sessions.Interviews were developed by the researcher, based on supervision literature and the focus of the research.Interviews focused on reflections on the supervision sessions and what was discussed, thoughts regarding the supervision relationship and the impact of supervision on practice.The data included 15 recorded sessions of supervision and 12 interviews, spread over an 11-month data collection period, and equating to 224 pages of transcripts.During transcription of the data, identifying details were changed and data was anonymised.
Data was analysed using a combination of MAXQDA software and spreadsheetbased approaches.Data was coded using MAXQDA, which helped to identify key themes in the data.Following Braun and Clarke (2021), this treated themes more as 'stories' based on shared meaning, rather than 'buckets' based on shared topic or con- tent of particular bits of data.Srivastava & Hopwood's (2009) framework was used to manage a priori theoretical purposes (guided by research questions focusing on how supervisors work with knowledge) with a need to respond to the data and allow themes to emerge.This framework keeps the questions of 'What are the data telling us?'And 'What do we want to know?' in dynamic interplay.In this way themes were developed iteratively around what was being discussed and how it was being handled.The process was largely inductive, with broad categories/themes generated as coding progressed.Once the data was coded, 11 themes were identified.This was aided by a simultaneous process, whereby data was summarised in spreadsheet form using broad questions as categories ('What is happening here?','What is happening epistemically here?' and 'Anything else interesting?).The combination of the thematic approach and the more granulated analysis of what knowledge work was being undertaken and how it was done uncovered the four epistemic practices discussed below.Hopwood (second author) reviewed the coding structure and examples of the data allocated to each code, and any queries were discussed, bolstering the trustworthiness of the data (Shenton, 2005).

Results
Analysis revealed four key ways of working with knowledge, elaborating the specifics of how psychologists grappled with the problems and issues that they faced.These four forms, conceptualised as epistemic practices, were: recontextualising practice knowledge; recontextualising theoretical knowledge; storytelling; and asking expansive questions.

Recontextualising Practice Knowledge
The psychologists used what was known and familiar to resource what was unknown, in so doing recontextualising knowledge to make it fit for purpose in another context or in relation to a new problem.One way in which they did this was to draw knowledge from various practice-related sources, using their years of experience as a repository, in conjunction with their professional judgement.These sources are summarised in Table 2 below, along with an explanation and example of each drawn from the supervision session data.Table 2 reveals the variety of source contexts for practice knowledge recontextualisation.All three pairs regularly recontextualised practice knowledge, but they did not draw on the same knowledge sources.Pair 1 accessed the widest variety of practice experiences in recontextualising knowledge, while Pairs 2 and 3 made use of only some of these sources.Pair 3 were the only pair who recontextualised 'typical' practice knowledge, drawing on a more diffuse, broad set of references compared to Pairs 1 and 2. Pair 1 were the only pair to draw on prior knowledge of a client -neither of the other pairs discussed long-term clients who had been previously discussed in supervision.Pair 1 were also the only pair to access analogous knowledge.

Recontextualising Theoretical Knowledge
Theoretical knowledge was recontextualised in two ways.Firstly, general psychological theory was applied in various forms and secondly, theory was activated to categorise, primarily diagnostically.
The use of a variety of theoretical resources enhanced learning, reflective of Eraut's claim that "the interpretative use of an idea in a new context is itself a minor act of knowledge creation" (1985, p. 130).The psychologists developed new knowledge using psychological literature, commingling theory and practice (Guile, 2014) as The supervisor drew on her own experience of practice to offer ideas and suggestions that the supervisee might pursue The supervisor in pair 3 offered the supervisee a strategy for engaging isolated children that she had observed in a school she had worked with 'Typical' practice Drawing on general knowledge based on experience of practice over time, constituting a form of psychological pattern recognition facilitated by years of working with clients and problems The supervisee in Pair 3 was highly experienced in assessing children.In one case discussed in supervision, the child's parents were hoping for a diagnosis of autism, leading her to comment that, "normally I get parents who don't want a diagnosis".This facilitated her understanding of the current, atypical set of parental expectations

Analogous knowledge
Referencing what happens in other, similar, non-psychological contexts/situations The supervisor in Pair 1 used mothering as an analogy to explore how the supervisee was working therapeutically with her client.sessions unfolded.Approaches to theoretical recontextualisation differed among the three pairs.Pair 1 were guided by the use of schema therapy1 , which they recontextualised in almost all case discussions, offering a shared language for communicating about clients.Pairs 2 and 3 tended to adopt a broader approach to theoretical recontextualisation.Pair 2 drew on a smorgasbord of theory and background knowledge when discussing clients, tapping into a diversity of knowledge sources.Their theoretical references varied depending on the client and his/her problem.Pair 3 shared a specialisation in Educational Psychology, hence their theoretical recontextualisation was characterised by a broad underpinning body of knowledge, rather than a specificity in knowledge referencing.
The construct of countertransference2 provides a specific example of how theoretical knowledge was not simply used, but was more deliberately put to work.Pair 1 recontextualised countertransference as an explanatory construct, which was not apparent with the other two pairs.Possible countertransference reactions were evident in Pair 2 and 3's sessions, but they did not construe them as such.Pair 1's supervisor highlighted the supervisee's feelings, identifying them as countertransference and working with them as a source of information about her and her client/s.One example occurred in discussing the supervisee's work with a long-term client, which was leaving the supervisee feeling exhausted.The supervisor picked up on this as follows: Supervisor: That exhaustion that you find … if you succumb to the countertransference and work reflecting on it, thinking about the way you are now, what other people in her life might do is just, she's too hard work and so they would just pull back and that would confirm for her.Supervisee: Oh absolutely.Supervisor: And that withdrawal would confirm.Supervisee: Most people in her life would not have tolerated what I have tolerated, they would get out of there.
Here, the supervisor's characterisation of the supervisee's exhaustion as a countertransference response makes her feelings an epistemic resource that offers potential information about the client rather than about the supervisee.In her second interview, the supervisee referred to this instance in explaining how combining an understanding of her countertransference reaction with the theory she accessed for understanding her client (schema therapy) provided insights that would not have occurred to her without the supervision session.She described this as "you try and marry your emotional experience of the client with some of the evidence and the facts, and you come up with something that's completely different".
In contrast, in one of Pair 2's sessions, the supervisee revealed her feelings of frustration and anger about how her client allowed her partner to treat her.The supervisor did not conceptualise this as countertransference or interrogate it as such.The fact that therapist feelings did not appear to be a knowledge source for Pairs 2 and 3 in the same way that they were for Pair 1 highlights the different ways in which supervision participants may identify and activate possible resources for knowledge work and theoretical recontextualisation.
All three pairs regularly engaged in theoretical categorisation, or the recontextualisation of diagnostic knowledge.Knowledge of diagnosis, which is a key knowledge resource for psychologists, was applied to the clients discussed, comparing individual clients with diagnostic categories.This involved epistemic movement between the particular and the general, zooming in on the client and zooming out to the category, an "oscillation between categorisation and particularisation" (Makitalo, 2003, p. 499).Further, categorisation provided the psychologists with relevant knowledge in relation to understanding and interacting with category members (Mattila, 2001).
The key resource for diagnosis is the Diagnostic and Statistical Manual of Mental Disorders (DSM).Diagnosis as categorisation provided an example of how knowledge might be used as an attempt to stabilise (Engestrom, 2007) within complex problem-solving situations i.e. knowing where a client 'fits' diagnostically can help the supervisee find ways to work with him or her more effectively.This assumes that naming something provides a sense of "stable thinghood" (Knorr Cetina, 2001, p. 193).However, at times this had the opposite effect.In one of Pair 1's sessions, they grappled with the case of a young man who presented with symptoms that approximated psychosis3 , but he did not fit the typical picture, with the supervisor commenting that "this is sounding very different, it's usually the opposite with patients".The more they compared this client to the category (psychosis), the less he fit the norm, which stimulated creativity about how best to treat him and brought to the fore the dynamic between confirmatory and exploratory dimensions of practice.
Lastly, in terms of recontextulisation, it is worth noting that at times the knowledge from past practice is already commingled theoretical and practical knowledge -experience from practice is not held in an epistemic vacuum from theory.Thus, when reaching judgements, sometimes psychologists were further recontextualizing previously comingled knowledges in new situations, while in others theory and practice were being brought into new connections.

Storytelling
The narration and elaboration of the story of what had transpired for the client and the supervisee emerged as a key aspect of knowledge work for two of the pairs (2 and 3).Narrative played a major role in supervision, wherein the telling of the story set up particular kinds of knowledge-sharing and actively shaped the knowledge work that unfolded in the session.Storytelling selected relevant knowledge and laid the informational groundwork for collaborative knowledge development.In these narra-tions, the supervisees made choices about what to report, selecting information that might productively resource the supervisor's responses.The need to convey a substantial amount of information in a short space of time took skill and relied on adept epistemic decision-making.This points to a different quality in supervisee talk from the "verbal deluge" that has been reported elsewhere (Wilkens et al., 2017, p. 947).
Objective aspects of the narrative were interwoven with psychological interpretation, supporting Berkenkotter & Ravotas's (1997) claim (in studying therapists' notes) that clients' descriptions of their situations become translated into psychological language, with knowledge being transformed as it moves further from the first person account.The same might be true of verbal accounts in supervision, given that the client is not present.Hence the story must be skilfully told to mitigate against the potential loss of richness and context that can result from a second-hand account (Berkenkotter & Ravotas, 1997).The language of 'translation' and 'transformation' in this quotation suggests that concepts akin to recontextualisation are apparent in other disciplines, yet might be named differently.
Further, although supervisees narrated the story, supervisors re-authored or reshaped the narrative through their responses, emphasising the two-way, interactional nature of supervisory communication (Luke & Gordon, 2012).Although the supervisee's storytelling influenced the supervisor's perspective, the fact that this knowledge was entirely new to the supervisor freed her up to see the situation differently, without the preconceptions that the supervisee might hold.Not knowing the client provided insights that may be less available to someone acquainted with them.Thus, the supervisor both shaped and was shaped by the knowledge that was worked with over the course of the session.
Sometimes, the process of storytelling brought insights simply through verbalisation, without direct input from the supervisor.An example occurred with Pair 2. As the supervisee was telling the client's story, she commented: She's just started taking some anti-depressants and that actually could be why she's feeling better now that I think about it … I don't think they'd have that much quick effect, but quite possibly there's something going on … ah I didn't think about that actually.
The supervisor in Pair 3 explained in her first interview that she saw her role primarily as facilitative, expressing that: I think I see my role as a facilitator of her inner wisdom, or knowledge, her expertise … to facilitate her being able to access and reinforce the ideas and the knowledge that she already has and to elicit things that she hasn't thought of.
Hence, giving her the space to tell the story in her own way was important.This supervisor's role in the storytelling thus focused on asking questions to flesh out or elaborate the narrative (see below).The supervisee conveyed substantial amounts of information within a relatively short amount of time, providing a firm informational base from which to undertake epistemic work.One characteristic of this pair was that the supervisee's goal for each session was not articulated.Hence, her rationale for choosing to bring a case to supervision was expressed in how she related the story and what she chose to highlight.This, in turn, contributed to the topics and themes that emerged in the session and the knowledge resources that were brought to bear.

Asking expansive/open Questions
The questions asked in supervision worked to elicit stories about problems and possible solutions, hence questioning was intertwined with storytelling.An expansive or open question is one where the answer requires more than a single word or short phrase in response.The term expansive questions refers to questions that were open, but more than just not closed: they invited new ideas and ways of thinking and thus expanded the terrain upon which an issue might be considered.This has some resonance with Engeström's (2016) notion of expansive learning, but the term is not used in the technical activity theory sense here.In the data, open questions were commonly used to encourage the supervisee to expand, explain and/or reflect.Along with covering relevant terrain, such questions served to position the supervisee as the expert on her practice.
An example drawn from Pair 2 occurred when the supervisee related the story of a 'mistake' she made when she gave an opinion on a client's situation.In response to the initial story the supervisor asked: "what was her response?".This raised the possibility that things may not be as dire as the supervisee implied, since she conceded that the client "was very gracious".The second question followed the supervisee's characterisation of the incident as a "rupture" in the therapeutic relationship.The supervisor directly contested this -"doesn't sound like it's a rupture at all" -asking: "what's going on for you around it?".This opened up the discrepancy between the supervisee's self-blame and an alternative appraisal.In referring to the supervisee's desire to address the issue with her client, the supervisor then asked: "what do you think is going to be the outcome of this interaction?"This got the supervisee talking about "what I'm hoping to achieve…".The last key question involved the supervisor asking for evidence that the relationship had been damaged.This prompted the supervisee to draw on her knowledge of her client, shifting her into thinking about the client's contribution to what had transpired and what that implied for the issues that her client traditionally grappled with.Through this process, constructive knowledge was built that (as was reported in the interview) allowed the supervisee to work creatively with her client in helping her achieve her therapeutic goals.
Open questions prompted supervisees to think out loud, sometimes talking to themselves as much as to the supervisor.They positioned the supervisee as expert, compelling her to draw on her own wisdom and knowledge.Thinking aloud in the presence of the supervisor offered something qualitatively different to thinking alone, enabling supervisees to tolerate the uncertainty of not knowing exactly how to proceed.Yet through this process, a sense of clarity started to emerge.This illustrated the role that careful interjection of questions can have in facilitating the unfolding of the story, making it a two-way process, rather than something relying only on the supervisee.Even though the supervisee held the knowledge about the client, she needed to convey it in a way that allowed the supervisor to ask pertinent questions that elicited possibilities, demonstrating the interconnectedness between the epistemic practices of story-telling and asking expansive questions, as well as the collaborative nature of supervision practice.Interviews with participants revealed that in a number of instances the supervisee came to see something new or different in relation to their practice through a process of joint epistemic exploration prompted by expansive questioning.

Discussion
The study revealed a number of key insights, which are highlighted below.Firstly, knowledge work was found to be prevalent in all the supervision sessions, regardless of their purpose, context or content.The fact that supervision involves knowledge work is taken for granted in the literature to the extent that it is rarely articulated and/ or is subsumed by reference to supervision's formative/educational purpose (Proctor, 2011).This study has made knowledge work overt, teased out its components, and opened up possibilities by going beyond the literature's current understandings.The study highlights that supervision can aptly be described as an 'epistemic space' i.e. a local site "of knowledge use, reuse, production and learning" (Markauskaite & Goodyear, 2016, p. 120).Supervision, as the literature extensively outlines, encompasses requisite knowledge and skills, relies on a productive working relationship and depends on how ingredients for effective supervision are operationalised.(Bernard & Goodyear, 2014;Borders, 2014;Culbreth & Brown, 2010).Yet it can be seen as more than this -as the ability to undertake knowledge work that augments these well-recognised dimensions.
Results revealed detailed information about the content of supervision, highlighting how the psychologists worked with knowledge in four key ways, referred to as epistemic practices.The identification and elaboration of these practices constitutes a novel way of conceptualising supervision and provides a framework for understanding knowledge work by psychologists (and other healthcare professionals) in supervision.This also contributes to the body of knowledge on the detailed content of supervision, in so doing illuminating supervision as a site of knowledge sharing and knowledge creation.Such knowledge has local relevance e.g. with regards to a supervisee's practice with a client, but goes beyond this in its potential recontextualisation to other clients and practice in general, illuminating how knowledge is diffused into practice.
Recontextualisation of knowledge was crucial to knowledge work for all three pairs, although there were differences in its enactment among the pairs.It involved the commingling of theory and practice, thereby making knowledge from one context relevant to another, a challenging process that required effort.The participants' recontextualisations skilfully moved knowledge continuously between the general and the particular, acting to "'make universals from particulars'" (Nerland, 2018, p. 251) and vice versa, while staying focused on the needs of the client, case and supervisee.The constant recontextualisation was evidence of ongoing learning through supervision, as learning emerged in relation to new and changing circumstances in practice (Kemmis et al., 2017), such as a new client or a challenge in the workplace.
A stable knowledge base is important, but these findings demonstrated how supervisors and supervisees continually adapted and extended this knowledge to meet the requirements of clients and supervisees in situated contexts (Fenwick & Nerland, 2014).The literature recognises that supervision requires different types of knowledge -declarative knowledge (competencies), procedural knowledge (best practices, or the implementation of competencies) and reflective knowledge (based on both experience and theory) (Borders, 2014).However, more attention is needed into how knowledge work takes place in action in supervision sessions.
This study has revealed that knowledge work in supervision is an intricate craft demanding nuance, skill, agility, flexibility and effort, while simultaneously calling forth a tolerance for not-knowing and ambiguity, something which may prove challenging to teach to aspiring professionals.Such craft was evident in the skilful use of recontextualisation when working with cases or issues.The findings exemplified how applying knowledge to individual cases is not a straightforward task (Bradley, 2009).Rather, it takes work and effort.Although the participants may not be conscious that they are doing this knowledge work, it does not happen 'naturally'.Findings demonstrated in the supervision context (as researchers have shown in other contexts) how "professional knowledge is created in use as professionals, faced with ill-defined, unique, and constantly changing problems, decide courses of action" (Sergiovanni, 1985, p. 15).The nuances of such craft are difficult to operationalise and measure, posing a challenge to a conceptualisation of supervision solely in terms of competencies or best practices.
The epistemic practices were found to work together in an entangled fashion, building knowledge and contributing to creative practice in the context of collaboration between the members of the supervisory dyad.Although the epistemic practices could be teased out for analytic purposes, in the action of practice they worked together, building on one another, amplifying each other, and interacting with the supervisory relationship.Rather than being separable or sequential, they knitted together in a tapestry-like fashion.The knowledge work accomplished relationally through their use relied on their integration and entanglement.This created a powerful collaborative mix, contributing to simultaneous knowledge development and relationship-building.As this happened, so the interplay between confirmatory, stabilised practice and exploratory, expansive practice unfolded.This intriguing interplay could be taken up as a focus of detailed analysis, but is beyond the scope of the current paper.

Conclusion
This novel study of clinical supervision offers a fresh approach to understanding its knowledge dynamics.The research aimed to illuminate what actually happens in supervision, particularly in relation to how psychologists work with knowledge.It built on the growing amount of research into what happens in supervision, focusing attention less on what is discussed and more on how this is done.In doing so, it offers an original framework through the idea of epistemic practices and the related concept of recontextualisation, foregrounding the epistemic and emergent dimen-sions of supervision.This framework advances a line of work initiated by Jensen, Nerland and others that took up a premise that epistemic aspects of professional practices warrant more attention, not as cerebral and abstract workings of professionals' minds, but as enacted concretely in what professionals do.The four key epistemic practices shed new light on how practitioners work with knowledge in supervision, and that what they are doing constitutes 'work', as knowledge is recontextualised and actively made fit for purpose.Activities such as working with theory, drawing on practice, outlining narratives and asking questions might previously have been taken for granted but are viewed here in a new, epistemic light.
There are limitations to the study.Three pairs is an adequate but small sample, not designed to be exhaustive but rather to explore supervision in depth.Although the participants may not represent supervision in general, they provide three detailed and authentic instances of how supervision is practised amongst experienced, generally registered psychologists in Australasia.Different professions and supervision contexts might elicit different kinds of knowledge work, but this study offers a start in how to go about investigating these.
These findings suggest the value of studying supervision up-close and in-depth, using observational methods.They highlight the importance of research that unpacks the local character of supervision and approaches it from innovative empirical angles.Further insight into what happens in supervision could lead to increased transparency amongst practitioners and productive information for those who supervise.The epistemic practices identified in this study could be further explored to investigate whether they are apparent in other supervision contexts, and with larger sample sizes, as well as made relevant for supervisor training programs.Also, this research focused on experienced psychologists.Because supervision plays a critical role in the training and early development of practitioners in many professions, future research could focus on how knowledge work happens for students, interns or early-career professionals, and how this might develop once practitioners become more experienced.

Table 1
Research participant details Pair no.
* These practitioners engaged in peer supervision, acting both as supervisor and supervisee.Each presented a case/problem in turn.

Table 2
Sources of practice knowledge recontextualisation Knowledge Source