POWER AND DIALOGUE: A REVIEW OF DISCURSIVE RESEARCH

This paper reviews the current research that uses discursive analyses of interactions in collaborative-dialogical approaches to family therapy to investigate the exercise of power. The concept of power has been a recurring influence in the development of family therapy. Systemic family therapy was heavily influenced by the cybernetic perspective of Bateson (1987), which viewed families in terms of mindful communicative systems and circular causality. However, Bateson (1987) proposed that power over another invoked a linear relationship, which was epistemologically flawed because he viewed people as situated within and reciprocally constrained by cybernetic relationships (Flaskas & Humphreys, 1993). Such a view was considered to suggest that people were equally responsible for maintaining any specified pattern of behavior, even abuse, which became an indefensible position from social justice and feminist perspectives (Goldner, 1985; Hare-Mustin, 1987; MacKinnon & Miller, 1987).

The 1980s saw an upheaval in family therapy prompted by feminism and social constructionism regarding how relationships and power were conceptualized (Flaskas, 2002). From a social constructionist perspective, meaning and understanding including conceptualizations of power, are perceived as contingent on social processes (Goldner, 1985; Hare-Mustin, 1987; MacKinnon & Miller, 1987). The work of Foucault (1979, 2002), emphasized that power should not simply be perceived as a restrictive force exerted by someone, but also as a constitutive force that is distributed and embedded in hegemonic discourses and social practices. Foucault argued that “power/knowledge” should be perceived as inseparable and suggested that people self-regulate themselves and their interactions with others in relation to all-pervasive power/knowledge. Power was not perceived as the “negative” and coercive exercise of force, but as a constitutive, “positive” force manifested through social practices and interactions. In this view, power is not held by anyone and not something that can be removed from interactions; however, it can be challenged and to some extent be resisted from within power relationships. Despite this postmodern conceptualization of power, family therapists appeared to continue to view power as negative and something to be avoided and family therapy approaches were modified and new approaches developed seeking to offset the power of the therapist within therapeutic interactions (Flaskas & Humphreys, 1993).

In the United States, Anderson and Goolishian (1988) developed the collaborative language systems approach, while in Finland, Seikkula and colleagues developed Open Dialogue (Haarakangas et al., 2007; Seikkula & Arnkil, 2006). Both approaches draw on a social constructionist view of language and meaning (Guilfoyle, 2003) with inspiration from Bakhtin (1981), Shotter (1996, 2006), and Andersen (1987). We will refer to these as collaborative-dialogic approaches (Anderson & Gehart, 2022) and while there are points of difference, we contend that they promote a fundamentally similar orientation to approaching therapy practice, which we will outline below.

In collaborative-dialogic approaches, human relations are defined and organized through active and creative communicative interaction rather than through existing social roles, structures, power relations, and hierarchies (Anderson & Goolishian, 1988). Language does not simply represent reality; rather, language is a resource utilized to construct our views of reality. From a Bakhtinian perspective (1981), a person's identity is developed through interactions with others. To each interaction, people bring their own unique ways of explaining and understanding the world, so by listening to another, our own ideas adapt and change, and new meanings can emerge. Furthermore, it is through interaction that “problems” in a system are defined and consequently it is through a dialogical interchange that new meanings and understandings can emerge and problems can be redefined or dissolved (Anderson & Goolishian, 1988).

In collaborative-dialogic approaches, the therapist occupies two positions representing a tension between collaboration and power (Anderson & Goolishian, 1988). First, the therapist is a participant-observer, mutually co-creating a joint understanding of the problem with clients from “an egalitarian and nonhierarchical position” (Anderson & Goolishian, 1988, p. 9). This includes a position of openness (Galbusera & Kyselo, 2018) or not-knowing (Anderson & Goolishian, 1992) involving a stance of curiosity where therapists position themselves as always in a state of “being informed” by clients (Anderson & Goolishian, 1992, p. 29) who are experts in their own lived experience (Anderson, 2005). Therapists must also contribute to the dialogue through being “public” (Anderson, 2005, p. 500) or “authentic” (Galbusera & Kyselo, 2018) by responding and sharing aspects of their own inner dialogue, which arise in response to the conversation with clients (Seikkula, 2008). This helps to construct a new shared language between all those present (Seikkula, 2008). The therapist and client thus mutually participate in a shared inquiry and the therapist offers their opinions and speculations with tentativeness, respect, and openness to being corrected (Anderson, 1995). Second, the therapist is a participant manager of conversation or a “master conversational artist-an architect of dialogue” (Anderson & Goolishian, 1988, p. 1–2). This means that therapists are flexibly responsible for facilitating a dialogical conversation whereby new meanings and understandings (the “not-yet-said”) can emerge (Anderson & Goolishian, 1988, p. 1) and multiple perspectives are voiced “each from a unique but equal participatory place” (Haarakangas et al., 2007, p. 228) and without directing the content or outcome of the conversation toward change (Seikkula, 2008).

A central aspect of collaborative-dialogic approaches is the concept of polyphony or multiple perspectives both within a person and between people (Seikkula, 2008). Therapists seek to elicit the multiple voices of clients within a session and reflect on their own inner conversations (Seikkula, 2008). Rober (1999) describes how the therapist may attend to the internal perspectives of their professional role, which reflects theoretical understandings and hypotheses, and their personal self, involving their experiences, feelings, and associations that arise in the meeting. When responding to clients, therapists are cautioned to reflect on their inner conversation to avoid too much of a focus on the therapist's personal self (Rober, 1999) or to exercise power through promoting normalizing social discourses (Rober, 2008). By balancing their professional obligations and the multiple perspectives of the client and family, therapists are managing multiple positions and their associated power relationships. In this sense, the therapist could be seen as avoiding the use of power to direct the conversation toward a particular outcome and to more evenly distribute power and influence between family members. The role of a collaborative-dialogic therapist thus represents a tension between equality and power; therapist “expertise” involves participating and voicing their opinion from an egalitarian nonhierarchical position while exercising power to facilitate and manage a dialogical conversation but without controlling the content or outcome of that conversation.

Collaborative-dialogic approaches have been critiqued for presenting an oversimplified account of power by conflating power with control and domination and urging therapists to remove power from therapeutic interactions (Guilfoyle, 2003). As demonstrated by Foucault (1979), power is situated within social structures, roles, and practices and therefore cannot be fully removed from therapeutic practice. So, regardless of therapist intentions, power asymmetries persist due to social discourses, which, for instance, position therapists as “expert knowers” with greater relative power and influence than clients who are seeking their help (Guilfoyle, 2006, p. 95). Psychotherapy can also exert power through a normalizing project that shapes client behavior toward public expectations (Cruikshank, 2019; Seikkula & Arnkil, 2006). By adopting a particular therapeutic approach, therapists exercise power over how a therapeutic encounter will progress. Seikkula and Arnkil (2006) suggest that it is the very exercise of power through adopting an Open Dialogue approach that has led to good research outcomes.

Much of the writings on collaborative-dialogic approaches have focused on a conceptual presentation focusing on the therapist's “mindset” (Ong & Buus, 2021) with less attention to the interactive communicative practices and social elements of power that occur in actual conversation. Turner (2005) presents a three-part conceptualization of power that emphasizes its social nature and that a thorough understanding of power requires attention to how power is produced and negotiated within interactions. Persuasion involves convincing another person that a certain position or action is correct or valid, who then acts in accordance with that position by their own volition. In coercion, people are impelled to act in ways against their will. In authority, people voluntarily accept another's legitimate right to determine one's beliefs and actions. Authority is not experienced as coercive domination but is legitimate, desirable, and accepted voluntarily. When applied to therapy, clients may accept the therapist's authority to direct the session as it will hopefully direct them toward some desirable outcome. Authority is jointly created as clients may recognize the authority of the therapist or withhold it by resisting, and therapists may pursue authority through seeking alignment and engagement (Watson, 2019).

Discursive approaches to the analysis of interaction, such as conversation analysis (Heritage, 2013) and discursive psychology (Edwards & Potter, 1992), have a variety of aims but all focus on analyzing real-life interactions. For example, conversation analysis research has used the concept of deontic authority, the right to determine the future actions of others in a particular domain (Stevanovic & Peräkylä, 2012), to represent how power is negotiated within interactions. Deontic authority can be attached to a person's deontic status or the authority connected to their social or institutional position or role. However, this is separate to how a person presents their deontic stance in interaction. For example, a therapist has high deontic status through their institutional role but may use certain tentative markers to present a downgraded deontic stance. Similarly, epistemic authority refers to rights and responsibilities to certain areas of knowledge (Heritage & Raymond, 2005).

Past reviews of discursive research have successfully investigated how theoretical concepts have been applied in practice. Tseliou et al. (2021) reviewed discursive research to examine how change processes occurred in systemic and constructionist approaches to family therapy. Change was displayed through forward moving dialogues and conversational shifts toward a relational perspective or toward nonpathologizing language. Tseliou et al. (2021) also highlighted various ways that power is managed in family therapy sessions. For example, therapists displayed a dilemmic balance between exercising power to direct a session and inviting others to be more active participants. Therapists managed issues of expertise by giving advice and emphasizing the expertise of the family members. Such discursive investigations can inform practice and sensitize clinicians to the dilemmas of authority in their work and reflect on their own conversational practices (Nanouri et al., 2022).

Following Foucault and Guilfoyle, we believe that power is an inescapable element in family therapy interactions. We will take an interactional approach to the concept of power to elucidate how power is produced and negotiated within actual conversations, and demonstrate how it can be conceptualized and theorized as interactional phenomena. The aim of this paper was therefore to conduct an exhaustive literature review of empirical studies examining the discursive interactions of collaborative-dialogic practitioners and to critically examine and synthesize how the concept of power is represented in their findings. We hope that these findings can contribute to a more nuanced understanding of how power operates in collaborative-dialogic psychotherapeutic modalities and inform the further development, theorizing, and practice in these approaches. In reading this article, we therefore invite clinicians to reflect on how their discursive practices position clients and what responses they in turn invite and receive when it comes to power and authority.

METHOD

We conducted systematic searches of databases for studies involving discursive analyses of Open Dialogue and dialogical practices. The first stage of the search utilized the databases PsycINFO, PubMed, and CINAHL. We conducted two block searches. The first used the search terms “open dialogue” OR “dialogical” and the second used “power” OR “deontic*” OR “authority.” We combined these two block searches using “AND” resulting in a total of 1207 references after duplicates were removed. We used Covidence to collate, manage, and screen references. References were screened by two independent reviewers against the inclusion criteria: (a) empirical discursive research with data from real-life psychotherapy sessions reporting that therapists utilized a dialogical approach; (b) individual, couple, or family therapy; and (c) some form of discursive analysis, which focused on conversational interaction, including transcripts of conversation with a line-by-line analysis. Some studies used additional analytic processes such as interviews after participants re-watched a therapy session video (e.g., Lidbom et al., 2014, 2015). These were excluded due to a focus on the interview responses rather than on the discursive interactions. Other articles (Guregård & Seikkula, 2014; Rober et al., 2006) using a similar approach were included as the discursive analysis was emphasized and distinct enough to be considered independently. Following title, abstract, and full-text screening, four articles met the inclusion criteria. We then conducted systematic chain searches of each article's reference list for additional articles. This search was completed for each new article included. The article searches resulted in a total of 14 additional references comprising a total of 18 articles for review (see Figure 1 PRISMA statement).

FIGURE 1. PRISMA statement. Adapted from Moher et al. (2009). Figure not reproduced.

The first author reviewed all the studies and collated general information on geographic location, therapy mode, and analytic approach. He also extracted sections from each article's analysis and discussion that related to power. Following Turner (2005), power can be reflected through influencing others through knowledge or directing the actions of others whether legitimate or otherwise. In examining and selecting data for this synthesis, the first author looked for any part of the analysis that reported differences in the relative influence of the therapist or family members particularly in relation to knowledge or actions. These extracts from the articles formed our dataset for analysis and were recorded within Covidence's data extraction tool and exported to NVivo for further analysis. We did not use the transcribed extracts of conversations themselves as sources of data for our own re-interpretation as we did not have sufficient understanding of the extracts and their conversational context or original recordings for a vigorous analysis. We did not evaluate research quality.

Data analysis, conducted by the first author, began with inductive open coding of the extracted data using NVivo. The codes were then compared, contrasted, and aggregated into larger groupings of themes providing a broad overview of the data. The construction of themes focused on the semantic or surface meanings of the coding providing a more descriptive approach to themes that are later interpreted and analyzed for their assumptions, meanings, and implications for dialogical practice. Our construction of themes therefore represents a “scientifically descriptive” form of thematic analysis (Finlay, 2021, p. 105) rather than a reflexive interpretative approach (Braun & Clarke, 2022). However, we acknowledge that our construction of themes represents our interpretive perspective influenced by our aims to provide a more nuanced understanding of power than has been provided in current theorizing in the dialogical literature.

Our group of authors included some with research and practice experience in the Open Dialogue approach (BO and NB). They therefore had a close affinity with dialogical approaches and their promise of a more egalitarian approach to health care. However, they also took the position that there are several areas where there has been a lack of detail and consideration of how dialogical ideas are integrated into conventional health systems, such as the concept of power. The other authors (ET and TS) had extensive experience in family therapy and discursive research. We anticipated that the combination of authors could provide a balance between a research and practice perspective to expand the interpretive frame of the analysis. All authors discussed the key findings, their interpretations, and their significance and implications for the study aims.

FINDINGS

A total of 18 articles were included in this review published across a 20-year period from 2002 to 2022 (see Table S1). Most of this research occurred within the last 10 years with 16 articles published since 2012. Ten articles studied couple therapy, seven studied family therapy, and one was not stated. There was not a diverse range of data in the studies: Six studies utilized the same single dataset of a couple therapy, five of these were published in one book (Borcsa & Rober, 2016); three articles used a different couple therapy dataset; and three studies came from the same family therapy dataset. Thus, 12 of the 18 studies utilized only three different data sets. The research was also heavily focused on Europe with 15 studies from European settings, with international co-authors, and three studies from Australia. There was a wider range of analytic approaches, with Dialogical Investigations of Happenings of Change (DIHC) and Conversation Analysis (CA) used in seven studies each, Discourse Analysis in three studies, and Discursive Psychology, and Dialogic Sequence Analysis were reported in two studies each. Narrative process coding system, narrative analysis, interpersonal process recall, and thematic analysis were each reported once. Five studies reported using multiple types of analyses.

The term “power” was not often directly used as part of the analyses. Instead, authors used different terms depending on the type of analytic approach. The articles utilizing DIHC and dialogic sequence analysis favored the term dominance (e.g., Avdi et al., 2015; Seikkula, 2002; Vall et al., 2016), with distinctions between semantic dominance (who determines topical content), interactional dominance (who determines speakership), and quantitative dominance (length of time speaking). The articles using CA tended to refer to power through the concept of authority. This included claims of knowledge (epistemic authority) and the ability to determine the future actions of others in the session (deontic authority). Studies using other analytic approaches used various terms such as therapists shaping the conversation (Avdi, 2016) or how active the therapist was in their questioning (Laitila, 2016).

In this review, we developed two main thematic areas, the power of the client and the power of the therapist. While studies did generally analyze the therapist and client interactions together, their division here illustrates how the studies were more heavily weighted toward investigating the actions of therapists. The power of the therapist is further divided into subthemes of the therapist as the catalyst of dialogue, which includes the minimal and active ways that therapists use power, and the therapist constrained by institutional obligations referring to the institutional and social demands affecting therapists' use of power. Figure 2 represents our themes and the relevant articles contributing to each theme.

FIGURE 2. Summary of Themes and Supporting References. Figure not reproduced.

The power of the client

A client's power was notably present in the attribution of blame for problems in a relationship (Avdi, 2016; Wahlström, 2016). We considered blaming an exercise of power as it represented a superior claim to describe another's behavior as problematic. Blame was associated with and supported by reference to a moral order or dominant discourses about relationships. For example, in a study from couple therapy, power was associated with a partner's ability to construct a dominant story congruent with discourses that relationships require emotional closeness and talk between partners. By adhering to this dominant discourse, a partner could ascribe certain appropriate rights and duties to her partner and consequently place her partner in a contrary and accountable counter-position if he was not meeting these expectations (Päivinen & Holma, 2016). Another study described dominant stories as invoking a moral responsibility for the partner (Wahlström, 2016). For example, a partner described the problem of not being able to talk to her partner as a “need” that was not being fulfilled: “It is her ‘need’ to talk, not her ‘wish’ or her ‘demand’. For the partner, to refuse a need is a morally questionable act, which is quite different from refusing a wish or even more so a demand” (Wahlström, 2016, pp. 156–157). One partner could also attribute blame toward a partner to justify their own nonaccountability and low agency to change the situation (Vall et al., 2014). Thus, while blame invoked a powerful claim about another's responsibility for problems and need to change, this blaming also involved the blamer's own lack of power and ability to change the situation.

Clients also expressed power through the structural elements of conversation. For example, through introducing topics for discussion and thus guiding the topical direction of the session (Rober et al., 2006) or by taking more speaking time and limiting the talk of others (Vall et al., 2016). This power was somewhat limited by the power of the therapist who could ultimately either accept, reject (Rober et al., 2006), or avoid topical initiations by clients (Avdi, 2016; Guregård & Seikkula, 2014). Similarly, clients could accept or reject the formulations of therapists. For example, Avdi (2016) describes how a therapist provided a formulation that emphasized a couple's resourcefulness and agency, which was rejected by the couple.

These examples demonstrate how clients were active contributors to a session and displayed a degree of power in allocating responsibility and blame, shaping the topical direction of the session, and resisting therapist formulations. However, client power was most often presented as something that was conferred upon, promoted, or allowed by therapists (Avdi et al., 2015) rather than something that was directly claimed by clients.

The power of the therapist

Particular importance was placed on the work of the therapist, who was presented as central to the promotion of dialogue. The importance of the therapist was most explicitly stated by Vall et al. (2014, p. 277) where in dialogical approaches “the therapist is included as a prime focus.” We have summarized the findings of these studies as describing two themes around the therapist's power: the therapist as the catalyst of dialogue including minimal and active shaping of the conversation, and the therapist constrained by institutional demands.

The therapist as the catalyst of dialogue

The occurrence of dialogue was presented as a direct outcome of therapists' conversational practices. That is, the therapist was presented as essential in constituting a dialogical conversation and promoting other desirable outcomes such as reflexivity and agency in clients (Seikkula & Olson, 2016). The studies detailed how the therapist can produce dialogue in two ways: through minimal or active shaping of the conversation.

Minimal shaping

Minimal shaping was represented by a therapist's nonchallenging acceptance of clients' talk (Laitila, 2016) or a “spare responsiveness” (Seikkula & Olson, 2016, p. 67) and included the use of continuers, repeats of the clients' prior talk, and simple evaluations (Avdi, 2016; Laitila, 2016). These responses displayed a therapist's listening and involvement in the conversation and an agenda of encouraging clients' elaboration of richer narratives, new meanings, and the discussion of difficult experiences and emotions (Avdi, 2016). Laitila (2016) describes this as “responsive, nonchallenging joining, and attunement” (p. 37). It is argued that these minimal responses allowed clients' control over the direction of the conversation and facilitated a transition from monologue to dialogue and consequently from positions of victims to agents (Avdi, 2016; Avdi et al., 2015; Olson et al., 2012). Even simple therapist questions such as asking the clients what they wanted to talk about were considered as “constructing agency.” For example:

…the therapist then asked the partners to define how they wanted to use the remainder of the session… the therapist's question positioned the couple as competent agents in charting the course of their treatment. It further reinforced the dialogicality of the dialogue by not only giving voice but also constructing agency. (Olson et al., 2012, p. 430)

Therapists' minimal responses were thus presented as a powerful influence on the conversation through promoting dialogical responses and agency.

Active shaping

In contrast, therapists could take a more active role in selectively shaping the conversation through questions, reflections, and topic and speaker selection. In an example of questioning, the therapist asked a wife who had begun crying: “What are your thoughts about what you're listening to right now?” (Vall et al., 2018, pp. S53–S54). The wife responds that her husband does not understand and that she should be able to feel and say anything in the home without feeling afraid. It is argued that the therapist's noticing of the wife's crying and the design of their question, prompted a reflexive meta-dialogue where the “the spouses are no longer talking and arguing about something that has happened in their lives, but instead are reflecting on their own speech and taking their own stance in the therapy setting (they are able to position themselves)” (Vall et al., 2018, pp. S54–S55).

In reflections, therapists speak to each other while clients are instructed to listen without interruption. This process allows therapists to introduce their own thoughts into the conversation to promote polyphony and reflexive responses in clients. Reflections also represent therapists' power in their ability to direct the session and occupy privileged and uninterrupted speaking time (Ong et al., 2021a). However, during these reflections, therapists explicitly downgrade the epistemic authority of their assertions by presenting their reflections as defeasible “thoughts” or “wonderings” rather than objective truths about the family (Ong et al., 2022; Schriver et al., 2019). For example, therapists used “I think” in reflections to mark their interpretive comments about clients.

“I think” downgraded the knowledge claim in the statements. This displayed an orientation to the reflectant's limited access to the participants' minds… [this] helped the reflectants show discretion and sensitivity, making sure not to postulate anything that they did not have “the rights to”. (Schriver et al., 2019, p. 29)

Consistent with theoretical principles (Galbusera & Kyselo, 2018), therapists present their thoughts to clients in a tentative fashion to encourage further dialogue. Therapists thus appear mindful of the power of their knowledge claims and sought to downgrade their statements and defer to the knowledge position of clients.

Ong et al. (2021b) and Rossen et al. (2020) describe how therapists select topics and speakers via a practice of eliciting stances. This involved a process of eliciting a stance position from one client and then selecting another client to comment on some aspect of the prior stance position. This is represented in the prototypical question form of “Y, what do you think about what X just said?” (Rossen et al., 2020, p. 719). The therapist exercises power in selecting what topics are relevant for discussion (also see Avdi, 2016) as well as selecting who can speak and when. Rober et al. (2006) identifies a similar phenomenon where clients brought up topics, which were analyzed as “invitations” which were only explored further if “accepted” by the therapist. Despite this exercise of power, the therapist worked to design their stance eliciting questions to minimize their influence on the types of responses they may receive. Therapists do this by designing their questions in very general terms with multiple components and using phrases such as “I wonder” to present their thoughts as defeasible. This is said to promote a responsive flexibility as clients can choose which aspects of the therapist's turn they can respond to and which to ignore:

So, although the therapist is directing the session through speaker, topic and question selection, “I wonder” combined with the non-specificity of the stance-eliciting question removes a certain degree of the directness associated with a more specific question… The therapist's question including “I wonder” and the non-specificity of asking about “what's come up” provides the mother with the freedom to respond to an earlier part of the conversation that was not specifically selected by the therapist. The therapist thus exerts the deontic authority to select the next speaker while mitigating their authority to dictate the content of that speaker's turn. (Ong et al., 2021b, p. 125)

Therapists thus have a delicate role of exercising power to control the conversation while also minimizing their influence over how a client may expectedly respond. Such an exercise of power is not necessarily inconsistent with a dialogical approach as it has been proposed that the therapist's directive approach is what makes a dialogical interaction possible (Ong et al., 2021a, 2021b; Seikkula & Arnkil, 2006; Vall et al., 2016, 2018).

The findings of these studies suggest that there is a pattern to how therapists utilize power within and across sessions. At the beginning of sessions, therapists prioritize minimal responsiveness, that is, repeats, continuers, and repair initiators to encourage client elaboration of detailed narratives and the expression of emotions with minimal therapist input. However, these minimal responses by the therapist are not neutral and guide the clients' narratives while also promoting some flexibility in how clients may respond. Later in sessions, therapists take a more active role to ensure that new perspectives are being voiced or if there is a “stuckness” between the participants (Ong & Buus, 2021). These more active practices include reflecting conversations to introduce the therapists' thoughts, asking questions to focus on particular topics, selecting speakers to balance speaking times or voice alternative viewpoints, to address safety or other institutional concerns, or to reinforce positive changes or resourcefulness. Therapists are thus more actively utilizing interactional dominance or deontic authority at these times through regulating speakership and topical focus. Therapists also make clearer epistemic claims but use numerous downgraded forms of verbal and non-verbal practices suggestive of “tentativeness.”

The therapist constrained by institutional obligations

Collaborative-dialogic therapists do not operate in a vacuum, they are embedded within health organizations and social discourses, which have expectations and demands in addition to the practice obligations arising from collaborative-dialogical theory. In the reviewed studies, there was evidence of therapists exercising power in sessions to meet these institutional demands, which at times conflicted with a collaborative-dialogical approach. These institutional demands included completing mental health assessments, promoting safety, and orienting to psychotherapy discourses.

A therapist's orientation to institutional demands is demonstrated through question-and-answer sequences of talk where the therapist is seeking information about client symptoms like in a psychiatric assessment (Avdi et al., 2015; Guregård & Seikkula, 2014; Seikkula, 2002). For example,

… [the therapist] mostly assumes the position of a clinical interviewer, in the sense that he shapes the interaction in a question-answer format; although he seems interested in [the client's] experiences, his questions mostly orient toward a framework of symptom description… this monological interaction does not create opportunities for emotional expression or new understanding. Moreover, it could be argued that it is associated with [the client's] position of reduced agency. (Avdi et al., 2015, p. 334 & 336)

The authors argue that this form of therapist questioning conflicts with a dialogical approach as it positions the client in a nonagentive position with little power to decide on the topics of conversation or to engage in dialogue.

In couple therapy for psychological intimate partner violence, therapists oriented to a responsibility to maintain safety and accountability in and outside of sessions. Vall et al. (2014) demonstrated how a therapist explicitly raised the topic of safety and the possible exacerbating effects of therapy on violence. They argued that by raising the issue of safety and discussing it in a dialogical fashion the therapist can “pre-empt any possible bad effect” and “the couple become more aware of the importance of safety” (Vall et al., 2014, p. 287). Therapists also worked to promote accountability and responsibility in clients by addressing a person directly: “The therapist tries to make the husband an accountable agent through the invitation to explain the stress he feels in the domestic situation” (p. 284), and via multiple addressees: “questions of this kind allow the clients to feel addressed without the addressing being done directly. This means that the person who is indirectly addressed may feel less under pressure, and at the same time, may move toward accepting accountability” (p. 293).

Vall et al. (2016) stressed the importance of therapists directing a session to disrupt dysfunctional patterns from being repeated in the therapy session: “Here the use of interactional dominance shows its merits, since it allows the therapists to regulate the speech and minimize the couple dominance and ‘power and control game’” (p. 230). Similarly, Rober et al. (2006) showed how therapists avoided a topic about violence as the children in a session were regarded as being too vulnerable at that point in time. In these examples, the therapist appears guided not only by collaborative-dialogical principles but also by responsibilities to ensure safety and accountability, disrupt behavioral patterns, and to protect children deemed vulnerable.

Another therapist orientation to institutional demands is the promotion of positive changes. Laitila (2016) and Avdi (2016) described how therapists use more active forms of questioning which focused on positive changes or outcomes for the clients and the use of their own resources in achieving these changes. For example,

In the final session, however, active asking took a different form. It emphasised differences, change and outcome, empowerment and resource orientation as compared to the first session and sessions subsequent. The focus on resources and the positive tone were present serving the consolidation and stabilisation of the changes made, as well as crediting the couple for these, and for the continuous therapeutic process of the client couple while terminating the therapy sessions. (Laitila, 2016, p. 41)

Therapists thus displayed an orientation to the importance of consolidating positive changes as well as emphasizing personal responsibility for those changes. Similarly, therapists used reformulations of clients' prior talk. Reformulations took up aspects of prior talk and represented them in institutionally congruent ways. Avdi (2016) identified how therapists reformulated the conflicts between a couple as something that is expected in their situation: “the therapist puts forth the view that the problems the couple are facing are part of the normal course of ‘building a new family’” (p. 83). This functioned to “normalize” the experience of the couple and to offer an alternative and more positive reinterpretation of their situation. Such reformulations although downgraded amounted to therapists' claims of knowledge to describe the clients' experiences in preferred ways.

Therapists also oriented to more abstract discourses about psychotherapy. For example, Päivinen and Holma (2016) showed how a therapist implicitly sides with the wife's position about the importance of talking:

From their institutional position, the therapists actively ask detailed questions to help the couple in reconstructing the dominant story that has been constructed during the therapy. By doing this exploration, they accept the dominant story constructed by Victoria and acted in accordance with the dominant discourse of the psychotherapy world: experiences and feelings have to be shared. (Päivinen & Holma, 2016, p. 97)

So, not only did the therapist side with the wife (Victoria) in this example, but this position also reflected a social discourse that talking is helpful in psychotherapy. The therapist thus appeared to guide a session influenced not only by the principles of a collaborative-dialogical approach, but also by discourses about the role of therapy.

These studies showed that therapists acted in accordance with institutional responsibilities as well as promoting positive change and disrupting violence in accordance with broader norms and values. This again involved a tension between collaboration and power through a unilateral exercise of power by deciding upon the topic of conversation while also conducting the conversation in a way that invites dialogue and discussion.

DISCUSSION

The studies in this review suggest that power is not antithetical to collaborative-dialogic practice. Instead, power is pervasive and present in multiple forms utilized both by clients and by therapists. The studies regularly reported that therapist utterances lead to immediate changes in clients suggesting that the exercise of power is not only consistent with collaborative-dialogic practice, but it may also in fact be necessary for the promotion of dialogical conversations.

The studies represented an overwhelming focus on the power of the therapist, less focus on the power of clients, and almost no attention to the power between family members. This focus on the therapist suggests that this research is primarily directed toward a clinical audience and providing evidence of therapists' practices that may be useful for clinicians to apply in their work. Indeed, much of this research has been conducted by practitioner–researchers. In addition, past theoretical descriptions of collaborative-dialogic practice focused on the therapist's mindset with less attention to how this mindset plays out in actual interactions (Ong & Buus, 2021), and this review reflects this continued interest in the therapist over the client. This focus risks an oversimplified view of how therapy progresses and overemphasizes the impact that a therapist has by minimizing the contribution of clients. Future research could consider how the client and family members experience the different expressions of therapist power in sessions. For example, how are the minimal responses by therapists received by clients, or how more directive statements and exercises of power are experienced.

There was a notable absence of research on the power relations between family members. A few studies examined differences in clients' perspectives in couple therapy for psychological intimate partner violence (e.g., Vall et al., 2014, 2016, 2018); however, they tended to focus on the power of the therapist and how they mediated the relative semantic and quantitative dominance of a wife and husband. This is somewhat surprising considering the early critiques of power in family therapy focused on issues of abuse within families and gives the impression that the clients' perspectives are relatively equal especially when mediated through the power of the therapist. This relative “blindness” to the power differentials between family members could be due to the relatively small number of research articles in this area and the peripheral role of collaborative-dialogical approaches in current practice. Collaborative-dialogic practice is still marginal in comparison with other forms of psychotherapy, and it may be that researchers are trying to establish the approach by focusing on the conversational outcomes that therapists are able to achieve. Whatever the reasons for this imbalance, researchers have downplayed the role and the voice of clients in collaborative-dialogic practice, which is itself an exercise of power. Future research could redress this imbalance by focusing more on how therapists and clients mutually develop a dialogical conversation.

Another aspect of power not addressed in this research is the choice of using a collaborative-dialogic approach. The choice of therapeutic approach adopted by a therapist is not something that clients generally have influence over as it is predetermined by the therapist before even meeting a client. Some clients, however, prefer therapists to exercise expertise via their clinical judgment and the prescription of directive interventions (Buus et al., 2021). A predetermined dialogical approach may be potentially as challenging to clients as a medicalized one albeit with different implications. A collaborative approach suggests a joint decision-making process about treatment, which is not applied to decisions about the treatment approach itself.

This review highlighted how the practice of collaborative-dialogic therapists is influenced by social obligations and discourses. This is reflected through therapists orienting to institutional obligations, such as completing assessments of client symptoms and emphasizing resourcefulness and positive outcomes as well as adopting a position against violence. As argued by Guilfoyle (2003) and Foucault (1979, 2002), as distributors of knowledge and expertise therapists are a means and end point through which power is enacted in society. Family therapy is not value neutral and therapists may unwittingly function as members of a system that directs clients toward ways of self-regulating in accordance with social values. These values can be relatively uncontroversial and productive such as a stance against family violence, but these norms can also represent a form of oppression by transforming an aspect of self, such as self-esteem, resourcefulness, positive thinking, or powerlessness into an object of social reform (Cruikshank, 2019). Therapists are in a sense social reformers and thus need to recognize this aspect of their role and consider if their actions are for the benefit of clients or derived from a potentially paternalistic position that perpetuates forms of social control or oppression.

Toward theorizing power in collaborative-dialogic approaches

Guilfoyle (2003) stated that dialogical therapists should not deny and therefore conceal the existence and workings of power but should examine how power operates and theorize how it is to be incorporated into collaborative-dialogic approaches. This review suggests that there are (at least) two main areas where collaborative-dialogic approaches intersect with power. First, there is power over who has privileged access to knowledge and truth. This has been variously described as epistemic authority or semantic dominance and connects with the theoretical positions of a “not-knowing” and “openness” stance. This research suggests that therapists generally defer knowledge to clients through focusing on eliciting client elaboration. When therapists do introduce their interpretations about clients, they are tentative and use downgrading features and invite clients to respond and disagree. However, when issues of safety or violence arise therapists have an obligation to take a clearer stance. This position can also be elaborated through the concept of conversational or discursive ethics (Smoliak et al., 2018; Strong & Sutherland, 2007). This perspective recognizes that clients are not passive participants but are active in shaping the conversation and the therapist's responses (Strong & Sutherland, 2007). The role of the therapist concerns how to respond to differences or disagreements and what moral values they espouse in their interactions. Discursive ethics promotes opportunities for clients to express different opinions (Smoliak et al., 2018) even where meanings are contestable and negotiable (Strong & Sutherland, 2007). Therapists therefore invite contestations and are appropriately welcoming and responsive to these contestations when they do arise.

The second aspect of power concerns who has power to direct the conversation. This is described in the discursive literature as deontic authority or interactional dominance and aligns with the theoretical ideas of the therapist as an “architect of dialogue” (Anderson & Goolishian, 1988, p. 1). Our findings suggest that the therapist maintains control over the directing of the conversation although it is mitigated. That is, therapists invite other speakers (Ong et al., 2021b; Rossen et al., 2020), ask questions (Laitila, 2016), and provide reflections and formulations (Avdi, 2016; Vall et al., 2016). Such use of power seems to influence the conversation in a way that invites participation, focuses topics, and promotes the voicing of multiple perspectives that is accepted by clients as the legitimate directing of group actions (Turner, 2005). The power connected to directing a conversation is omnipresent because conversation is structured around a turn-taking mechanism of adjacency pairs whereby initiating turns make relevant appropriately fitted responsive turns (Kendrick et al., 2020). In essence, adjacency pairs means that there will always be a speaker exercising some power through producing an initiating turn at talk. The relevant question to collaborative-dialogic therapy is how do therapists design these initiating turns and how often are these initiating turns shared between therapists and clients? For example, how do therapists respond to client initiatives to shape the conversation such as through silences and introducing new topics? Another question concerns the extent that the therapist actively shapes the conversation. For example, a form of active engagement is therapist persistence or maintaining a particular conversational topic or project, maintaining a particular stance, or pursuing some outcome (Sutherland, Dienhart, & Turner, 2013; Sutherland, Turner, & Dienhart, 2013). Therapist persistence incorporates a responsiveness to the conversation and can include the therapist's posture in maintaining a dialogical stance and/or pursuing a topic of enquiry to develop depth of understanding. Such persistence seems consistent with collaborative-dialogical principles in that it develops deeper understandings and the expression of polyphony; however, it may also border on the therapist directing a conversation away from topics of interest to clients. It thus remains an open question for future research on how therapist persistence (as well as other conversational practices) is consistent with and promotes dialogue and further understanding.

Collaborative-dialogic therapists are thus balancing how to shape a process to promote inclusiveness while not pre-determining outcomes. Using Turner's (2005) description, therapists hold authority to facilitate the conversational actions of the family therapy group but avoid attempts at persuasion. Dialogical therapists thus appear to implicitly acknowledge that guiding the structure of a session through topic and speaker selection is appropriate, while guiding what people should think and do, within certain ethical limits, is not.

Limitations

The first stage of our literature search only retrieved four articles and was largely supplemented by our chain searches of reference lists and citing articles. This reflects the newness of the field as we relied on free-text searches, which are less reliable than controlled subject headings. We therefore opted for a broader search strategy that resulted in a higher number of search results and a more laborious process of screening. We included articles that stated that clinicians had utilized a dialogical approach. Currently, there is not a widely agreed upon fidelity criteria for dialogical approaches. Therefore, adherence to a dialogical approach is largely based upon the original seven principles or key elements of Open Dialogue (Olson, Seikkula, & Ziedonis, 2014; Waters et al., 2021). Continuing research on interaction will likely contribute to developing clearer fidelity guidelines but at present, researchers must rely on clinicians who self-identify as using the approach. This means that there may be a broad range of practices that are considered “dialogical,” thus creating potential difficulties in drawing conclusions about the approach. Our analysis of the articles and subsequent synthesizing of themes naturally reflects our interpretive process. We understand that all research has an interpretive element and its acknowledgement is a strength and characteristic of qualitative research (Braun & Clarke, 2022). We strove for a rigorous process with input from multiple authors to add depth and reflection to the analysis and acknowledge that other researchers will likely have different interpretations of the data. We have been transparent in our assumptions going into the analysis as described in the introduction and method sections. This review focused on observable interactions that show how power is visibly enacted in therapeutic interactions. We acknowledge that power is not always visible and that there are larger institutional and societal influences at play that warrant further investigation.

Conclusion

To date, there has been a relatively underdeveloped conceptualization of power in dialogical therapies. We hope that this review contributes to a more detailed conceptualization of how power is present and utilized by therapists utilizing collaborative-dialogical approaches. We propose that in collaborative-dialogical approaches, therapists acknowledge the power of their position and utilize it in ways that guide a conversation to promote the voicing of multiple perspectives and provide opportunities for all participants to speak. In addition, we propose that therapists voice their own perspectives but in a way that does not claim greater importance than other participants and explicitly invite comments and even disagreements. Our review highlights how therapists and researchers are also social agents representing social and institutional morals and values. Accordingly, a clearer statement of their ethical position and values may be necessary for transparency in clinical practice.