A closer look into the ‘black box’ of coaching – linguistic research into the local effectiveness of coaching with the help of conversation analysis

ABSTRACT This paper, located in the paradigm of Linguistic Coaching Process Research, theoretically introduces Conversation Analysis (CA) as a means to explore the ‘black box’ of coaching, i.e., the concrete conversation, and to address pressing research gaps in this field. CA, developed to describe, analyse, and understand practices and patterns of talk-in-interaction, presents a rigorous, systematic, and transparent qualitative method to document and research the sequentially organised conversational structure of coaching as well as the discursive practices of coach and client. The analytical power of CA lies in identifying change-inducing practices and in detailing how these contribute to the local effectiveness of coaching. A case in point are questioning practices, whose transformative power as local agents of change emerges in their sequential set-up as ‘coaches’ question – clients’ response – coaches’ reaction in third position’. In an exemplary analysis using CA, we illustrate the (un-)successful processing of questions in coaching. The two data extracts stem from the ongoing project Questioning Sequences in Coaching. While the paper aims to shed predominantly theoretical light on how to linguistically unpack the ‘black box’ of coaching, it also aims to sensitise practitioners to consider interventions, e.g., questions, as embedded in the locally transpiring, turn-by-turn interaction with their clients.


Introduction: a linguistic perspective on coaching conversations
Alongside more established professional encounters such as psychotherapy, coaching has emerged as an effective helping format to promote clients' intellectual, psychological and/or emotional learning in predominantly work-and career-related contexts (Graf & Spranz-Fogasy, 2018;Grant & Stober, 2006).To support clients' change, coaches rely on a helping conversation as their primary means and method (Graf et al., 2019;Miller & Considine, 2009;Pick & Scarvaglieri, 2022).Helping in coaching is operationalised by means of verbal (and non-verbal) activities (Graf, 2022) and client change, i.e., 'the transformation of one's beliefs, personal convictions and life experiences' (Graf & Pawelczyk, 2014, p. 62), becomes visible and observable on the (micro-)level of the concrete conversation (Peräkylä, 2019;Vehviläinen, 2019;Voutilainen et al., 2018).The ensuing dialogue between coach and client is based on interventions employed by coaches and their clients' reactions to them, which are realised in particular communicative practices.This dialogue functions as the primary vehicle for change and influences the quality and success of coaching and thus its effectiveness (De Haan et al., 2010).Given this centrality of the coach-client interaction, scholars from different disciplines have been calling for more in-depth research into the coaching conversation itself, the 'black-box' of coaching (Ianiro & Kauffeld, 2015, p. 42;Elliott, 2010, p. 124).
In contrast to coaching outcome research and its focus on the global effectiveness, coaching process research is concerned with investigating the 'underlying mechanisms between coach and client' (Graf & Dionne, 2021, p. 6) that allow for this effectiveness.In analogy with the 'change process research paradigm' established by Elliott (2010) for psychotherapy, coaching process research to date spells out in four different designs: the 'process-outcome design', which constitutes the oldest most established approach, the 'helpful factors design', the 'significant event research design', and, finally, and most recently, the 'microanalytic sequential process design' within which linguistic (and particularly Conversation Analytic) research operates (Wegener, 2019).While the coaching process can be investigated from various methodological and disciplinary angles, linguistic research is particularly apt to describe and analyse the concrete 'doings' of coaches and clients on the interactional (micro-)level.Though micro-or interaction analysis (De Jong et al., 2013), such as psychological research carried out by coding particular (deductively pre-established) verbal and non-verbal behaviours of coaches and clients (e.g., Gessnitzer & Kauffeld, 2015;Ianiro et al., 2014) also provides insights into the coaching process, qualitative Linguistic Coaching Process Research (LCPR), focuses inductively on entire conversations in their uniqueness as they locally emerge between coach and client on a momentby-moment basis.LCPR thereby generates detailed insights into basic activities (e.g., coconstructing change), communicative tasks (e.g., defining the concern, managing emotions), discursive practices (e.g., solution-generating questions) and their specific linguistic realisations (e.g., hypothetical wh-questions) (Fleischhacker & Graf, 2023).LCPR thereby responds to some shortcomings identified in process outcome research such as the challenges to empirically deal with the complex and intricate interrelation between participants' communicative actions within coaching (e.g., Schermuly, 2019).
From a linguistic perspective, coaching emerges in-situ, as both coach and client collaboratively co-construct the professional practice 'coaching' in and through their consecutive communicative contributions or turns-at-talk (Graf & Dionne, 2021;Pick & Scarvaglieri, 2022, see, e.g., Sidnell & Stivers, 2013, for a basic Conversation Analytic understanding underlying such a perspective).In so doing, the participants co-construct clients' concerns and possible solutions, manage their interdependent roles as well as their working relationship (e.g., Behn-Taran, 2014;Graf & Jautz, 2022;Scarvaglieri et al., 2022).It is in this consecutive organisation or sequentiality of (coaches' and clients') turns that the local effectiveness, i.e., the gradual transformation of clients' knowledge, emotions and experiences, becomes (analytically) observable (Graf et al., 2019;see Peräkylä, 2019 andVehviläinen, 2019 for psychotherapy) (see Chapter 3).So far, LCPR has made fundamental contributions to the knowledge base of coaching establishing the basic activities of coaching (Graf, 2019) and documenting the communicative tasks coach and client tackle along the process (e.g., constructing roles and identities or managing the interaction) (Rettinger, 2011;Winkler, 2017) as well as zooming in on typical discursive practices such as questions (Spranz-Fogasy et al., 2019;Winkler, 2022) (see Chapter 3.2).Since all of these communicative particularities 'evince the respective interaction type' (Graf et al., 2014, p. 2), linguistic research has also been concerned with differentiating coaching from other helping professional formats (Graf & Pawelczyk, 2014;Kabatnik & Graf, 2021) (for a more comprehensive overview, see Fleischhacker & Graf, 2023).
Due to its recent emergence, Linguistic Coaching Process Research (LCPR) still has major research gaps, especially in comparison to linguistic research on other helping professional formats such as psychotherapy. 1 Based on findings of the first scoping review on LCPR by Fleischhacker and Graf (2023), one such gap concerns the contribution of the moment-bymoment co-construction of the coaching conversation to client change.This represents a shared object of interest with existing psychological outcome and process research and is the primary focus of Linguistic Change-Process Research, a newly developing strand of LCPR (see Figure 1).Though LCPR draws on a variety of qualitative (linguistic) methods such as Discourse Analysis, Critical Discursive Psychology, or integrative discourse analytical approaches, Conversation Analysis (CA) as the 'best developed qualitative sequential process method' (Elliott, 2010, p. 129) is particularly suited to address this prominent research gap.It constitutes a rigorous, systematic, and transparent qualitative method to describe the sequentially organised conversational structure and the discursive practices employed by coach and client, as well as the (local) transformational processes that contribute to an overall successful outcome of coaching.Above all, CA provides in-depth insights into the turn-by-turn development of potentially change-inducing interventions such as questions (Graf, Künzli, et al., 2023;Peräkylä, 2019).It can illustrate how and why local effectiveness is achieved in the interaction, offering concrete evidence of causal relations between coaches' and clients' communicative actions.As such, it can serve to develop tools and recommendations for professionals regarding 'how' and 'when' interventions such as questions are most effectively embedded in the coaching conversation (Antaki, 2011;Lester & O'Reilly, 2018).
To illustrate how CA can be used to further explore the coaching process and the effectiveness of coaching interventions, this paper first introduces its general assumptions, aims, and methodological procedure before we elaborate its concrete potential to researching the coaching process as part of Linguistic Coaching Process Research (Chapter 2).In Chapter 3, we theoretically describe CA's contribution to documenting and analysing the locally transformative, i.e., change-inducing, potential of sequencesthe goal of Linguistic Change-Process Researchthereby focussing on questions.Finally, we illustrate this via an exemplary analysis of (un-)successful) questioning sequences from authentic coaching data (Chapter 4).
The paper ends with summarising the main arguments in favour of CA as a fruitful approach for researchers and practitioners of coaching alike.Overall, this methodological-conceptual paper aims to shed predominantly theoretical light on how to linguistically unravel the 'black box' of coaching.Its goalis to raise awareness for the sequential, turn-by-turn co-construction of the coaching conversation and to sensitise practitioners to consider their interventions, e.g., questions, in this local embeddedness.While our recent scoping review (Fleischhacker & Graf, 2023) has given an overview of the current state of the art of linguistic coaching (process) research, this paper uses a CA-based approach to illustrate one methodological approach as well as its benefits to research and practice.
2. Conversation analysis: a rigorous qualitative method to analyse helping professions such as coaching … with notable exceptions … , the coaching activity itself, the interactions of the dyad, including the elements of listening, questioning, clarifying, reflecting, challenging and thinking, have not been researched (Fillery-Travis & Cox, 2018, p. 527) Conversation Analysis offers itself as the most promising approach to address previously identified gaps in quantitative and qualitative research regarding the concrete interaction between coach and client and to face empirical challenges that arise when analysing the in-situ, turn-by-turn coaching interaction (De Haan, 2019;Fillery-Travis & Cox, 2018;Fleischhacker & Graf, 2023;Graf & Dionne, 2021;Schermuly, 2019).Conversation Analysis with its long-standing research tradition on both mundane and professional interactions such as psychotherapy (e.g., Davis, 1986;Edwards, 1994;Peräkylä et al., 2008), doctor-patient interaction (e.g., Deppermann & Spranz-Fogasy, 2011;Heritage & Maynard, 2006;ten Have, 1991;West, 1984) or counselling (e.g., Hutchby, 2002;Peräkylä, 1995;Vehviläinen, 1999) (for a general overview see Heritage & Clayman, 2010), represents a rigid and transparent procedure to generate systematic and stable findings (Sidnell, 2010).In particular, CA allows to document and analyse discursive practices such as questions or (re-)formulations in their local emergence and sequential contribution to client change.
2.1.Conversation analysis: underlying assumptions, aims and methodological procedure CA was established in the 1960s by Harvey Sacks, Emmanuel Schegloff and Gail Jefferson.Inspired and influenced by Erving Goffmann and Harold Garfinkel's sociological work, the researchers developed 'an approach within the social sciences that aims to describe, analyse, and understand talk as a basic and constitutive feature of human social life' (Sidnell, 2010, p. 1).CA assumes that society is rooted in a web of practices and orders that organise social interaction.Conversation, i.e., talk-in-interaction, thereby represents a basic form of social interaction and as such is based on practices, patterns, and orders, too.By means of close observation and in-depth analysis of authentic talk-in-interaction in everyday and professional/institutional contexts, these patterns can be documented and understood.Most generally, CA aims to determine: What (organisation) principles and linguistic resources are used by participants themselves in a conversation and how do participants co-create their reality?What practices do they employ to solve communicative tasks and problems to reach conversational goals?How do participants orient to, react to and understand each other's conversational contributions?How do they co-create a mutual understanding of their interaction?(Heritage & Clayman, 2010;Sidnell, 2010).The fundamental question underlying CA-based research is: Why that now? (Sacks, 1992): Why do participants such as coach and/or client produce a particular utterance or perform a (non-)verbal action in a particular conversational context in a particular form?To address these issues, CA has developed a core set of methodological steps and analytical procedures for collecting, organising, describing, and analysing patterns in talk that has produced 'a large body of strongly interlocking and mutually supportive findings' (Sidnell, 2010, p. 20) across both mundane and professional interactions.
First of all, in its close observation of talk-in-interaction, CA is radically empirical and data-driven.It relies on video-or audio-recordings of authentic, i.e., naturally occurring, talk involvingin the case of coachingreal coaches and clients with authentic concerns.Only such recordings allow to observe the wide range of linguistic phenomena (e.g., pauses, overlaps, intonation, hesitations, etc.).Recordings of authentic interactions reveal that 'apparently counterintuitive and unexpected things actually do happen' (Sidnell, 2010, p. 20) and often attest to a contrast between talk-in-theory (how people assume they talk) and talk-in-practice (how they actually talk) (Stokoe, 2012).The recorded talk is transcribed as transcripts build the basis of analysis.For this, Jefferson (and others) developed linguistic transcription systems that consider, in addition to the verbal content, a wide range of vocal and interactional phenomena such as intonation, pauses, or overlaps to capture the nuances and details of talk-in-interaction.All of these could be consequential for the participants in their interaction and, thus, for the analysis (Janusz & Peräkylä, 2021;Sidnell, 2010).Since CA constitutes a descriptive-phenomenological approach, the analysis itself is not theory-but data-driven; data is not approached with pre-established theories or hypotheses, 'but with an open heuristic to develop categories inductively from the material itself within an exploratory design' (Graf & Dionne, 2021, p. 10).Theory building is based on a careful (re-)examination of the data; analytical categories and interesting interactive phenomena emerge from the material itself.This way, relevant 'seen but unnoticed' (Garfinkel, 1967) phenomena can be discovered and described, too.However, such analysis is neither random nor without focus, but guided by a clear research interest and research questions that may, however, be adapted along the research process.While classical CA-questions aimed at uncovering universal, context-free mechanisms like turn-taking, most research now focusses on communicative practices in specific contexts such as questions in coaching, their sequential development and organisation principles (Sidnell & Stivers, 2013).The ultimate goal of a CA-based analysis is to carve out participants' own understandings of the practices under scrutiny and how their communicative behaviour illustrates their understanding of what is happening in the interaction (Janusz & Peräkylä, 2021).Once a relevant interactional phenomenon or a target action (Peräkylä, 2019) is identified, the analysis proceeds from case-by-base examination to making collections of instances (Janusz & Peräkylä, 2021).These collections serve to uncover patterns and recurrent underlying structural features of the phenomenon or target action across a particular data set.Finally, the collected phenomena are categorised and are assigned a categorical name that allows their theoretical discussion.Over time and across a multitude of interactional contexts, a number of core principles and (analytical) concepts have emerged in CA research as underlying and organising talk-in-interaction.Some of these will be illustrated below.

Conversation analysis: basic concepts and their possible contribution to coaching research
As established by Conversation Analysis, talk-in-interaction is the product of 'multiple, simultaneously operative and relevant organisations of practice' (Sidnell, 2010, p. 5) or 'intersecting machineries' that participants continuously orient to and which can be used for analysis.These sets of practices structure entire encounters (e.g., a conversation is started and ended with greeting practices), organise turn-taking (i.e., conversations are based on participants taking turns at talking), and coordinate actions and action-preferences (e.g., a question by one participant is ideally followed by the other participant's response).Moreover, in case of conversational trouble or misunderstanding, these will be immediately indicated by participants via hesitations or silences and addressed by means of repairs, where speakers correct their own or their partner's talk (Sidnell & Stivers, 2013).A final organising principle concerns word selection, which is mainly influenced by 'recipient design'.Speakers select particular words to construct appropriate and relevant contributions for the person being addressed.All of these 'intersecting machineries' are context-dependent, i.e., they occur at a particular moment and place, with the involvement of a particular group of people, and after a particular point in the conversation (Schegloff, 2007;Sidnell, 2010).Based on these organising principles of talk-in-interaction, core (analytical) concepts were developed in CA.These include turns and sequentiality (Figure 2) (e.g., Sidnell, 2010;Sidnell & Stivers, 2013), adjacency pairs and sequences of action (Figure 3) (e.g., Schegloff, 2007) as well as intersubjectivity (Figure 4) (e.g., Schegloff, 2007;Sidnell, 2010).
Together with its methodological procedure, these organising principles and basic concepts guarantee both the generation of systematic, observable and stable findings and, more generally, the empirical rigour of CA-based research (Janusz & Peräkylä, 2021).
A CA-approach based on sequential, i.e., turn-by-turn, analyses of particular target actions and phenomena serves coaching process research in various ways: The local and processual co-construction of coaching-specific actions, themes, meanings, and mutual understandings can be documented and (critically) examined as they unfold (Graf & Dionne, 2021).A sequential analysis of the concrete coach-client interaction makes it possible to trace interventions such as questions and how they are understood and processed by coach and client (Spranz-Fogasy et al. 2019;Winkler, 2022).This allows to establish more or less successful interventions in solving communicative tasks (Graf, 2015).Such research can also unravel how established success factors like the working alliance emerge locally and how they influence the coach-client interaction (Graf & Jautz, 2022;Scarvaglieri et al., 2022).By yielding detailed descriptions and analyses of recurrent structures and practices in coaching, its specificities relative to other helping formats, but also the particularities of different coaching approaches, can be assessed on the concrete interactional (micro-)level (Graf et al., 2014(Graf et al., , 2019)).In sum, CA-based qualitative research allows to gain empirical, fine-grained insights into the 'black box' coaching that cannot be obtained within the methodological paradigm of quantitative research (De Haan, 2019;Fillery-Travis & Cox, 2018;Schermuly, 2019).Moreover, it offers coaching practitioners the tool Stein asked for: 'What if coaches had a tool where they could listen to, or look at transcripts of, their conversations with clients and be able to name what they were doing in every sentence or phrase that they said?' (Stein, 2007, p. 163).

Conversation analysis and the local effectiveness of (questions in) coaching
People come to coaching for lots of different reasons, but the bottom line is change (Whitworth et al., 1998, p. xix).
Apart from documenting and analysing discursive practices and how they interactionally unfold, the focus on the sequentiality of interactions (see Figure 2) provides analysts with 'the microscope' to observe the moment-by-moment change processes (Peräkylä et al., 2008, p. 16), i.e., the local effectiveness of interventions.Such a focus on the raison d'être of helping professional interactions, i.e., client change, has recently gained momentum in applied CA (Graf et al., 2019;Vehviläinen, 2019) and has initiated a new strand within LCPR.Our final chapter elaborates how change processes become observable and analysable in (coaching) conversations via Linguistic Change-Process Research (Chapter 3.1) (see Figure 1) and then illustrates its procedure focusing on questionsthe central intervention in coaching (Chapters 3.2 and 3.3.).

Analysing change processes/the local effectiveness in coaching
While change represents the desired outcome of helping interactions such as coaching, to this date, there is little empirical research regarding what exactly brings about such change on the local, communicative (micro-)level.CA-based applied research, however, is geared towards gaining an understanding of changes in clients' experiences, feelings, and attitudes by analysing the observable part of their change processes, i.e., the concrete conversation with the coach.The underlying axiom is that change (of perception, feelings, thinking etc.) becomes visible on the linguistic level as 'topical and emotional shifts' (Vehviläinen, 2019, p. 192) but also in changes of expressions, of relating to individuals, to objects etc. which all index a transformation in progress (Graf & Dionne, 2021;Pawelczyk & Graf, 2019).According to Vehviläinen (2019), such research addresses two relevant time spans: While a focus on the micro-level sequential analysis of concrete interventions renders insights into the local effectiveness of key change-inducing practices, studying macro time spans yields information on the gradual variation, reshaping of experiences, and co-construction of change over time across sessions and processes (e.g., Voutilainen et al., 2018).
Local effectiveness in CA terms is understood as an interactionally accomplished process between helper/coach and helped/client with moment-to-moment changes that impact the ensuing conversational development (Pawelczyk & Graf, 2019).According to Peräkylä (2019), local change takes place through 3-part sequences of action (i.e., transformative sequences) in which a target or initiating action (e.g., a coach's question) creates the context (or slot) for a particular response (e.g., answer) by the client and in which the third position constitutes the reaction to the response (e.g., follow-up question, ratification, etc.) where the coach assesses its appropriateness.In addition, the actions prior to the target action create the necessary affordances and relevancies for e.g., the coach's question.In the course of such a sequential set-up, transformation is empirically observable in a change of emotions, (coach-client) relations and / or referents (or content) (see Figure 5).
Based on the concepts of sequentiality and intersubjectivity (see Figures 2 and 4), each utterance is a documentation of these three 'realms of experience' and each following utterance again orients to and provides an understanding of the previous one, though necessarily something has changed in the process.This utterance-by-utterance interplay of understandings regarding content, emotions, and relationship allows for the local effectiveness of the coaching interaction andif successfuleventually brings the helping or change process forward.In sum, sequentiality becomes the vehicle for the professional agenda and change (Peräkylä et al., 2008, p. 16).
Linguistic change-process research based on CA centres on documenting these sequential relations of content, emotion, and relationship between the contributions of coach and client to make the (discursive) change process empirically observable.It thereby carves out potentially change-inducing practices by, for instance, paying close attention to topical continuities and discontinuities or thematic threads across sessions (Vehviläinen, 2019).It also traces forms of successful interventions in their sequential set-up as well as their occurrence within the process or session.Questions of 'what' (i.e., which intervention), 'where' (i.e., where in the local unfolding of the conversation) and 'how' (i.e., the intervention's verbalisation) regarding the use of interventions can be addressed.Fundamental change-inducing practices identified for helping professional encouters include initiatory interventions (e.g., questions) and responsive interventions (e.g., formulations, interpretations, extensions) (Peräkylä et al., 2008;Spranz-Fogasy, 2020;Vehviläinen et al., 2008).The primary research interest is to empirically reconstruct how these contribute to the desired outcome of the helping professional format, by inquiring about the relationship between linguistic form and function in connection with their change-inducing potential.Change-process research thereby responds to critiques on qualitative process research as merely descriptive and unrelated to the overall purpose of the interaction, i.e., clients' change (Graf et al., 2019).In its focus on the local, interactional accomplishment of effectiveness it helps to close one of the most pressing research gaps in coaching and allows for an intersection with psychological coaching research on global change processes (Fleischhacker & Graf, 2023;Graf, Dionne, et al., 2023).The approach is particularly valuable for practitioners in terms of when, which and how linguistic interventions are successful and induce change (Graf et al., 2019).In the remaining part of this article, we will illustrate these concrete potentials for coaching (research and practice) by zooming in on questioning practices as local agents of change.

Questioning sequences as local agents of change in coaching
Questions and their change-inducing power have been widely researched across mundane and professional interactions (see e.g., Tracy & Robles, 2009).According to Köller (2004), questions have the potential to initiate processes of (self-)reflection and imagination and to trigger a change in perspective; questions may topicalise or localise knowledge deficits to manage mutual understanding or allow for linking cause and effect, present and future etc. Questions thereby help to (re-)build (buried) information and to generate new insights and thus support establishing new and re-structuring existing experiences, ways of thinking, or behaviours etc.These general characteristics are particularly pronounced in helping professional interactions (Graf et al., 2020; Graf & Spranz- Fogasy, 2018).Yet, despite the fact that coaching practice has labelled questions the guarantor for successful coaching and the central and most powerful intervention or steering tool, empirical insights into the (successful) use of questions in coaching are still very scarce; a stark contrast exists between the vigorous discussion of questions in practice literature and training manuals and the dire research situation (Graf et al., 2020;Graf & Spranz-Fogasy, 2018;Graf, Dionne, et al., 2023).
The international and interdisciplinary research project 'Questioning Sequences in Coaching' (QueSCo) (I499-G) (https://questions-in-coaching.aau.at/)sets out to change this situation.It develops an empirically-based, systematic, coaching-specific typology of question types and questioning sequences (questionsresponsesreactions in third position) (see below) on the basis of authentic data (Graf, Dionne, et al., 2023).Using video-recorded (systemic,solution-oriented, work-related) coaching processes from Germany, Switzerland, and Austria, the focus is to gain empirical insights into authentic questioning practices.The corpus consists of 14 dyadic coaching processes, 52 sessions and approximately 3,000 questioning sequences. 2The interdisciplinary research team (linguists and psychologists) investigates both the local effectiveness of questioning sequences, i.e., how questions in their sequential, moment-by-moment set-up contribute to change (linguistics), as well as the relation of successfully completed questioning sequences to global effectiveness, i.e., to the overall satisfactory coaching outcome, and to established phases of change (psychology).
The linguistic part of the project is premised on the following CA-based assumptions: Questions have an immanent transformative potential due to their sequential set-up as questioning sequences: As prototypical adjacency pairs, questions as an initiating first pair part make a response (second pair part) conditionally relevant (see Figure 3).In professional interactions, it is usually the professional who asks the questions and thus, any question coaches ask makes a reaction by the client necessary.Due to a normative (social) pressure to respond, unanswered questions create a disruption in the conversational flow, i.e., a relevant absence.Furthermore, questions do not only make an answer conditionally relevant, they also systematically pre-structure the kind of (re-)action they require in terms of preferred action, content or topic, agenda, and structure depending on the form of the question, the action logic (e.g., requesting confirmation vs. requesting information), the topical agenda, and the presuppositions they contain (Graf et al., 2020;Graf & Spranz-Fogasy, 2018;Hayano, 2013;Läpple et al., 2021).For instance, a polar interrogative such as 'Have you been practicing mindfulness?' makes a yes/no-answer conditionally relevant thereby formally creating a specific answer preference, i.e., yes.It also presupposes that 'mindfulness' is a familiar concept to both interactants, and that practising mindfulness has been a mutually established goal. 3 Wh-questions or content questions either ask for a particular piece of information (e.g., what, when, where), or invite much longer and detailed answers allowing for a broader range of responses (e.g., 'What has changed since our last session?').Finally, declarative questions (often in combination with response-mobilising features such as questions tags, see Stivers & Rossano, 2010) again restrict responses since they only require ratification or rejection of a proposition (e.g., 'You have been working a lot lately, haven't you?').
Such (structural and thematic) projections build up strong expectations for the next speaker, i.e., the client, to react in a specific way.The coachee, then, is confronted with binary options (yes/no, answer/no answer, accepting the presupposition or challenging it, etc.) whereby one option is (socially and interactionally) preferred, and the other is dispreferred.Such preference organisation is easily observable in interactions, since preferred responses are produced unmarked without delay, mitigation, hedging, accounts, repairs etc. (Hayano, 2013;Läpple et al., 2021).It is because of this invasive and demanding quality of questions and their interactional consequences that Steensig and Drew (2008, p. 7) claim '[a]sking a question is not an innocent thing to do'.Hayano (2013, p. 396;p. 409) calls them both a 'versatile resource' and a 'powerful tool to control interaction' due to their compelling force to solicit responses and various constraints they put on possible answers.Since knowledge, obligations, and responsibilities in professional interactions such as coaching are asymmetrical, questions are an important steering element for the professionals.They allow coaches to pursue their professional agenda assuming a leading responsibility to support clients in solving problems and reaching goals (Graf & Spranz-Fogasy, 2018) via their control over the interaction and the thematic development.Questions also offer the possibility to address delicate matters or express hypotheses and assumptions in a less direct way since they entail a recipient-tilted epistemic asymmetry (Graf et al., 2020;Läpple et al., 2021), i.e., only coachees can answer them.This way, questioner and respondent enter a particular communicative and cognitive relationship, which suggests openness and interest and allows for the above-mentioned potentials of questions to unfold (Läpple et al., 2021).
Since questions make a response necessary, it is essential to also focus on how respondents (in the second pair part) react to the multiple constraints that questions impose on them and how they address the informational and epistemic gaps suggested by the questioner.Though interactants usually conform to social norms of solidarity and cooperation agreeing with the questions' agendas and terms, there are multiple ways they may resist the constraints.Apart from providing no response at all or challenging the question per se, various reactions are possible, not all of them representing a proper answer to the initial question (Läpple et al., 2021;Lee, 2012).Stivers and Hayashi (2010), for instance, discuss 'transformative answers', which indicate problems with the question and slightly adjust them without disrupting the questioning sequence, e.g., by replacing an unwanted term used by the questioner.This way, respondents, i.e., clients, claim epistemic authority, control, and independence and may pursue their own agenda.
In accordance with Peräkylä's (2019) transformative sequence model (Figure 5), the coach's reaction to the response in the third position is essential, too.It is where coaches must take a stance regarding the client's answer or reaction.The third position is the place to accept a client's understanding or repair it, to insist on an unanswered question, to align with or ratify an answer, or to flag it as inappropriate or insufficient (see Figure 4 and 5) (Sidnell, 2010).The coach has various options (e.g., seeking specification, changing the topic or focus, initiating repair, ratifying and moving forward in the coaching process) depending on their professional agenda, i.e., the co-constructed coaching goal, their working alliance with the client, and whether the answer was preferred or dispreferred, etc. (Läpple et al., 2021).Both the coach's contribution in third position as well as client's response modify, elaborate, and transform previous talk in ways that are relevant to the conversational goals by how they treat, select, or take up previous conversational material (Sidnell, 2010).Coach and client thus equally contribute to the conversation and drive the change process forward.
4. An exemplary analysis of (un-)successful questioning sequences In the following, we will use two extracts from the QueSCo corpus to illustrate how a sequential CA analysis of questioning practices in coaching allows to observe and describe their (un-)successful processing.In this way, both the moment-by-moment changes of contents, emotions, and relations, i.e., client transformation, but also interactional troubles and the participants' orientation towards them emerge on the linguistic micro-level of the coaching conversation.The excerpts are taken from the first session of an authentic, online coaching process between a male coach and a female executive.The data has been transcribed according to (a simplified version of) CA transcription conventions (as outlined, e.g., by Jefferson, 2004). 4For the purpose of this article, the original German data has been translated into English, questions are highlighted in bold.
The experienced coach applies a solution-oriented short-term coaching approach.The female client has just taken on new leadership responsibilities in her company.The goal that both have defined for this coaching session is to find ways to feel more at ease and to experience less stress and self-doubt in her new management position.To generate solution strategies for this concern, they are in the process of addressing a member of her 'inner team', prior to the first extract.They have already entered into a dialogue with the client's inner team member 'easiness'. 5  Extract 1: Successful questioning sequences.In lines 6 and 7, the coach formulates two wh-questions ('what would that be' and 'what would you perhaps still like to ask') whereby the second question immediately specifies the (possibly unclear) deictic reference 'that' and clarifies the projected action, i.e., to formulate a final question directed at the client's 'easiness', so as to secure intersubjectivity and mutual understanding.The latter is also supported by the question preface in which the coach yet againhumorouslyframes the ongoing activity and metapragmatically introduces his request for information as the last question directed at the client's inner team member.Despite the long silence of more than 47 seconds (line 8), which by far exceeds the 'standard maximum' allowance for silence within a conversation (Jefferson, 1989), the coach does not expand his turn or initiate repair thus interpreting the client's pausing as a 'creative' silence, i.e., as suggesting reflection or thinking, not interactional trouble (Lynden & Avery, 2016).Indeed, after breathing in audibly and another turn-initial delay, the client starts giving a response (in line 11) 'I think I would ask her how she can be more consistent', which she explainsencouraged by the coach's continuer 'hm' (Schegloff, 2007; line 15)as her easiness becoming a 'faithful and constant companion'.Though the client first assumes a low epistemic stance (Heritage, 2012) via 'I think' and conditional 'I would', she eventually fulfils the projected action by formulating a specific question she would like to ask (lines 19-20) and in which she directly addresses her 'easiness' with the personal pronoun 'you'.She then indicates the completion of her turn with 'yes' (line 22) (Sidnell, 2010).The client's direct question formulation allows the coach to ask an immediate follow-up question in his third position ('and what answer does she give you', in line 27), which he connects to the client's turn with 'and' as well as 'to that' (i.e., your question).The coach can thus make use of the client's contribution to initiate an intra-personal change in perspective.His next wh-questionwhich constitutes a third position and yet again initiates a new questioning sequencemakes relevant a response in which the client assumes the perspective of her identity part.Another long turn initial delay and audible breathing ensues, indicating the cognitive and imaginative complexity of the task.In line 31, the client, however, produces a response 'that I need to change my perspective' and then adds 'my perspective and my focus'.Her answer is unproblematic in that it provides a response without repetition of the initial question and with grammatical and morphosyntactic congruency (Thompson et al., 2015).Another pause suggests further reflection or introspection, but the client decides not to add anything else, which she again indicates by finishing her turn with 'yes (line 37).This constitutes an important development in the process, since the (more concrete) behavioural and cognitive changes required to achieve the client's goal are first topicalised here.After various delays and hesitation markers (lines 38-27), the coach decides to pose yet another wh-question in third position, which requests an emotional stance evaluation (Rossen et al., 2020) from the client regarding this important realisation.
Excerpt one showcases two successful questioning sequences (lines 3-27 and lines 27-47) in which the client readily fulfils the projected actions, i.e., provides the requested information and performs an intra-personal change in perspective, allowing the coach to build on this in his follow-up turns.This co-construction leads to an important change in referents, i.e., an insight regarding necessary changes in behaviour to reach the client's goal, to which the coach can then respond with a focus on the client's emotions.
Extract 2: Unsuccessful questioning sequences.Prior to extract 2, coach and client have been working on a possible solution strategy (i.e., following in her male colleagues' footsteps by exhibiting more self-confidence and less hesitation and self-doubt).The implementation of said strategy, however, was postponed by the client until the beginning of the next year, a vague and distant future.This leads the coach, in line 11, to start asking a polar interrogative question (Hayano, 2013) regarding an earlier solution implementation by giving space to her 'easiness'.However, the question formulation is interrupted in line 16 ('is there in the near future very soon an opportunity where you can maybe also'), which is indicated by a micropause, and it is only continued via repeating part of the question in line 23 ('where you would have the opportunity').Additionally, the question is followed by an explanation as to how such 'giving space' might look like (lines 24-34).In his insertion (lines 17-23), the coach uses contingency labelling 'almost an experiment' (Muntigl et al., 2020) and other forms of downgrading ('very small', 'not in full') to construct the task as less intimidating or significant and to work towards changing the client's first vague solution implementation in the distant future.Overall, the coach's turn is characterised by hesitation markers, intra-turn delay, pauses and self-repair (Sidnell, 2010), which altogether suggests that he might be addressing a delicate matter (Peräkylä, 1995) possibly infringing on the client's deontic authority (Stevanovic, 2021) to decide when to implement a solution.The polar interrogative question makes relevant a 'yes/no' response, though a 'yes' is preferred here (Hayano, 2013).Rather than answering, the client declines her turn, remaining silent for 1.1 seconds (line 35) indicating some problem or difficulty to respond (Yao & Ma, 2017).The failure to produce a second pair part creates a relevant absence (Schegloff, 2007), blocking the progressivity of the sequence.In this context of interactional trouble, it is the coach's responsibility as questioner to determine its source (Pomerantz, 1984), which he seems to attribute to a lack of clarity.He initiates repair by repeating his question, this time in a much more concise and clear manner in line 36.However, yet again, the client refrains from answering the question and instead initiates an other-repair (Kitzinger, 2013) asking the coach for a clarification of the 'opportunities' mentioned, i.e., whether they can relate to professional or private contexts and whether they can be 'small situations', too.She thereby continues to indicate problems with the original question which suggested a purely professional context and initiates another delay in her response, again suspending the ongoing action by breaking 'the contiguity between the adjacency pair parts' (question-answer) (Kitzinger, 2013, p. 253).Since repairs may signal upcoming dispreferred responses or disalignment but allow speakers to re-examine their turn (Kitzinger, 2013), the coach now enthusiastically welcomes also (small) examples from the private sphere with 'oh yes' (in line 46).This then leads the client to, finally, provide an answer by naming possibilities (starting in line 55) to assume an alternative, and more relaxed conduct in particular private situations.
In contrast to Extract 1, Extract 2 features two unsuccessful questioning sequences (lines 11-35 and 36-46) in which the client refrains from any conduct whatsoever, i.e., remains silent, in response to the original question and then initiates an other-repair once the coach has (more concisely) reissued the question.The client thus blocks the progressivity of the ongoing interaction and does not allow the coach to build on her input.Only the coach's concession in adjusting the original scope of the question thereby allowing a solution implementation in the private context, leads to an answer from the client.
This exemplary analysis illustrates that discursive practices such as questions require focusing on their sequential set-up, in this case 'questionresponsereaction in third position' (e.g., Stivers, 2013).The inherent sequentiality and transformational power of questions, made observable via a CA-based linguistic analysis, ensures local effectiveness and (eventually) becomes a driving motor for the process of change in a more macroscopic time frame, i.e., across sessions and entire processes (Graf et al., 2019;Graf et al., 2020;Graf & Dionne, 2021;Graf & Spranz-Fogasy, 2018;Voutilainen et al., 2018).

Conclusion
In 2018, Fillery-Travis and Cox (p.527) have argued that ' … with notable exceptions … , the coaching activity itself, the interactions of the dyad, including the elements of listening, questioning, clarifying, reflecting, challenging and thinking, have not been researched'.While we agree with the authors regarding this major research gap, our recent scoping review (Fleischhacker & Graf, 2023) has shed light on the growing body of linguistic process research addressing this gap.Nevertheless, further empirical research is needed investigating, for instance in the context of questions, particular question types in coaching (e.g., Fleischhacker et al., under review; Fleischhacker et al., in prep.), client's reactions to questions (Dionne et al., 2024;Spranz-Fogasy & Moos, in press), or coaches' reactions in third position (Dionne & Graf, in prep.).In continuation of our scoping review, which collected and synthesised existing research, the aim of this conceptual-methodological article was to describe and illustrate a promising linguistic methodology to further amend the unsatisfying research situation.The idea was to raise awareness and understanding of the benefits of such an approach and of the potential insights of its empirical application.Conversation Analysis allows us (i.e., researchers and practitioners) to take an indepth look into the 'black box' of coaching, i.e., the coaching process and its local transpiring on the interactional micro-level.Within Linguistic Change-Process Research, CA provides an analytically sound means to meticulously trace how change in coaching is conversationally co-constructed by coach and client via particular discursive practices (e.g., questions).Through such research, conversational vehicles for change become observable in transformative sequences, since the coach's actions are not considered in isolation but as part of a conversational architecture in which prior and following contributions are linked by thematic, interactional, and social relevancies.In line with Peräkylä's (2019, p. 257) succinct assessment, we want to end by stressing the double benefit of CA research for coaching: On one hand, it indicates how the machinery of interaction is adapted to serve the institutional tasks of coaches and clients, and, on the other, CA reveals how the process of clients' behavioural, cognitive, or emotional learning takes place in and through smaller and larger chunks of the coach-client interaction.
Notes 1. CA drawing on linguistic terms and insights has been widely applied to describe and analyse key interactional and change-inducing practices of therapeutic talk, focusing primarily on the therapist's responsive (e.g., formulation, interpretations, extensions) and initiatory (e.g., questions) actions and their sequential organisation and achievements in a particular therapeutic context (see Peräkylä et al., 2008).2. Written informed consent was obtained from all participants (coaches and clients) for the publication of anonymised data as part of the QueSCo project (https://questions-incoaching.aau.at/en/).Persons, organisations, places etc. referred to within the coaching, including names of coaches and clients, have been replaced in the transcription.3. Despite the critical attitude towards closed questions in coaching practice literature, research has shown that answers to such 'closed questions' are not necessarily shorter or less detailed (Deplazes, 2016) as clients may account for or explain their answers or use the question as a topic proffer to start a narrative (Schegloff, 2007).4. A note on transcription conventions: .h/ .hh/ .hhh= in-breath with increasing length; h./ hh.
/ hhh.= out-breath with increasing length; (.) = micro-pause; (0.5) = refers to the length of pauses, 0.5 meaning half a second; [...] = indicates overlapping talk; ((...)) = gives information on para-/non-verbal actions; _ = indicates latching talk; CO = coach; CL = client; the numbers, e.g., CO7/CL1, refer to the specific coaching dyad. 5.When working with the "inner team", a client's personality traits (positive or obstructive with regard to the coaching goal) are identified and named, thus allowing concerns to be address on the intra-personal level.Clients should then enter into a dialogue with these parts, either strengthening positive or appeasing negative ones (see Graf 2019).

Figure 3 .
Figure 3. Adjacency pairs and sequences of actions: illustration and description (own figure).

Melanie Fleischhacker ,
BAKK.BA MA has been working as a writing coach and writing instructor for several years.She hast been a guest editor for the topical collection 'Interventions in Writing Coaching' of the journal Coaching | Theorie & Praxis.As an applied linguist, she has been interested in gender and sociolinguistics, as well as professional interaction.As part of her MA thesis in English and American Studies, she investigated (multimodal) representations and discourses of gender and sexuality in EFL textbooks.In ongoing projects, she focuses, among other things, on the identity construction of adolescent female football players and develops alternative (gender-sensitive) teaching materials.As part of the linguistic project team of 'Questioning Sequences in Coaching', she is involved in the development of a coaching-specific typology of questioning sequences and will write her dissertation in the context of the project.She is focusing on solution-generating questions, their sequential organisation and prior actions.Dr.Eva-Maria Graf is a professor of applied linguistics working on, among other things, helping professional interactions with a special focus on coaching.She is a founder of linguistic coaching process research.Amongst numerous other publications in this field, she has authored the first linguistic monograph on coaching ('The Pragmatics of Executive Coaching').In addition to the overall project management of the 'DACH' research project 'Question Sequences in Coaching', she leads the Austrian team based in Klagenfurt.She has already been grantedalso by the Austrian Science Fund (FWF)a three-year 'Elise-Richter Habilitation' Fellowship.Eva-Maria Graf has been working as a coach for many years focusing on coaching in academia.She is also a senior coach and trainer.Together with Hansjörg Künzli (as well as Elke Berninger-Schäfer and Robert Wegener), she is editorin-chief of the journal Coaching | Theory & Practice.