Ethical practice in participant-centred linguistic research

4.1.1 Informed consent and ethics protocols are dialogic processes

Rather than considering the ethical review process as a bureaucratic exercise before the ‘real’ research begins, we recommend approaching it as an opportunity for dialogue with peers, who can offer valuable insights on the ethical dimensions of a project. This is particularly important for those new to research or without formal supervision but, of course, gaining feedback from colleagues is useful for all researchers. Furthermore, applying for ethical approval is often not a linear process; it can involve back-and-forth discussions and negotiations with review boards, participants, and external organisations. For example, in Sarah’s GP project, the researchers needed to agree a consent process with their university ethics committee whilst also establishing trust and access with an external organisation, the RCGP. The core ethical challenge here concerned collecting and analysing video footage of doctors performing in a stressful, high-stakes examination without causing them additional stress, affecting performance, or creating an undue sense of obligation to take part. Agreeing a process to gather these doctors’ consent involved a long period of discussion with the RCGP (who had been involved in the project since its inception and long before the application for ethical approval) and the university REC, going between each party to identify key requirements whilst also developing a more in-depth understanding of the context.

The university required a Participant Information Sheet (PIS) outlining the research, as well as a consent form for candidates to sign – a standard practice that was not unfamiliar to these participants with a medical background. However, some members of the RCGP had reasonable concerns that asking candidates for formal consent prior to the exam could cause stress, not to mention administrative complications for an in-person exam that was already very complex, involving hundreds of candidates, actors, and examiners. To develop a workable process, the RCGP leadership and exams team had input in the design of the PIS and consent form, with the final version becoming a co-authored document. For example, clarifications were requested around the process for randomly allocating candidates to video recorded rooms, making it clear that no participant would be specifically selected. There was also discussion over the appropriate length of the documents (the PIS and consent form together ran to over a thousand words) with some feeling candidates would be unlikely to read the documents at exam registration. This is perhaps a common tension with university research ethics requirements in external settings: while ethics committees usually expect a comprehensive level of information to be provided, this can represent an administrative burden to participants and organisations. The process eventually agreed, in consultation with the RCGP and REC, involved two stages: participants first received the PIS in advance via email, including a bullet-point summary with key information ‘at a glance’. This provided an opportunity to read and digest the information, ask questions, or even opt out entirely. Then, on the exam day, when candidates registered at the check-in desk, they were asked verbally whether they would be happy to take part, again with opportunity for discussion. At this point of discussion and consent, it was important the study was understood as separate from the partner organisation, with Sarah able to speak as an independent researcher when outlining the risks and benefits of taking part.

This first case study illustrates the dialogic and iterative development of a consent process with a REC and partner organisation. In this project, it would not have been possible to simply decide an a priori consent process without having these extended discussions. Co-producing these documents helped build a working relationship during the early stages of the project. Further, the organisation continued to use versions of these documents in future consent processes, demonstrating an exchange of practices. Whilst establishing a consent process was lengthy, this extended exchange ultimately enriched the research process and partnership. In situations like this, the researcher’s role can become akin to a ‘broker’, negotiating and reconciling differences in the requirements of a university REC and partner organisation. In cases where ethics committees or external partners express reservations, as can often be the case in PCLR, researchers must seek to bridge the gap between established protocols and the needs of their participants, drawing on advice from colleagues with relevant experience where helpful. The researcher must have a sense of boundaries around acceptable research practice, informed by a clear understanding of both research ethics frameworks and the context-specific needs of project partners and participants, to develop a practical, bespoke process.

As with the GP project, ethical procedures in the Marginalised Families Online (MFO) study were navigated in tandem with a university ethics committee and external partners – in this case, Adoption UK (AUK) and the Donor Conception Network (DCN). These charities were involved in discussions around the nature and scope of the project before it began, and both were instrumental in the participant recruitment process. This meant that, like Sarah, Jai was required to balance the demands of her research partners – with whom a relationship was important for the viability and impact of the project – alongside the requirements of her institution. She therefore worked with both charities to agree appropriate processes of informed consent. This incorporated the details to be included in the PIS and consent form and when these should be shared with participants. Researchers at DCN were particularly keen to make it clear from the outset that participation would involve a high level of commitment and information sharing. Jai therefore agreed processes that would ensure prospective participants had multiple opportunities to find out about the project and understand the level of participation required. As with the GP project, the PIS was given in advance, via email. It was then verbally discussed in a short video call, before individuals decided whether to take part. The consent form was not discussed and signed until the start of the first interviews, at which point participants had a good deal of time to consider their willingness to take part and the nature of their participation. Both documents committed to a high level of anonymity and data security, and they promised to put control over information sharing (both in interviews and digital media) in participants’ hands. During this extended process, most participants wanted to discuss and ask questions around sharing their digital media data. Reassurance that they would have control over what they shared was a significant factor in their agreement to participate.

Data collection processes for the MFO project involved partnership between researcher and participant, in keeping with long-established approaches to participatory research that allow for greater participant involvement, autonomy, and power over decision-making. These processes were supported by a research design whereby no private social media data was shared until after the second interview, once a relationship of mutual trust and understanding was established. Ultimately, many participants decided not to share certain media, for example direct message exchanges through WhatsApp or Messenger. There were also occasions where participants offered to share data but Jai declined because it would be unworkable or unethical to do so, as where one participant suggested sharing personal emails from individuals he was no longer in contact with. The process of data selection and sharing for this project therefore involved a careful negotiation of personal and ethical priorities and boundaries for both researcher and participant. This negotiation of data-sharing also meant Jai’s institutional ethics application could not detail the precise nature and origin of digital media data that would be collected. Instead, the initial application foregrounded the nature of the participant-centred approach, gave examples of the kind of data that might be collected, and demonstrated the researcher’s awareness of ethical issues that could arise in relation to a range of digital media platforms and contexts.

In the LGBTQ+ youth project, all ethnographic and interview data collection took place within established youth groups run by qualified youth workers. Like the MFO project, consent forms were not introduced at the first point of contact with prospective participants. Given the age and vulnerability of her participants and the nature of her ethnographic research, Lucy gave the young people even longer to digest her presence before attempting to make any formal agreements. Like Sarah and Jai, she worked with relevant external partners – in this case, the youth group leaders – to negotiate consent in a way that was appropriate for her participants, gaining initial (verbal) permission to visit the groups, then sharing an accessibly worded PIS with young people who attended this first meeting. After that, she further negotiated the parameters of the ethnography with each lead youth worker, once they had been able to check how group members felt about the project. The ethnographic process thereafter involved Lucy attending each youth group every week for an agreed period, and gradually getting to know as many of the young people as possible. Through informal conversation, she explained her role as a researcher and tried to glean which individuals might be comfortable playing a more central part in the study. Where young people seemed happy to talk, Lucy would make a point of chatting with them the following week, asking if they thought they might be interested in being interviewed. She waited until some rapport had been developed before asking individuals directly whether they would like to take part and gaining written consent to begin recording their interactions. Where young people were clearly disinterested, she did not press them.

Lucy’s approach developed organically. The young people tended not to engage with the PIS, and so asking them to formally consent to her presence at the group would have been meaningless. It took time and one-to-one conversations for the young people to understand her reasons for being there, and they seemed more comfortable asking questions about the study as their trust and familiarity grew. Lucy therefore relied on her awareness of ethical research practice to draw a line between research activity which required formal consent and that which did not, with consent forms required only from young people involved in recordings. Since it was not possible to gain consent for her participant observation, Lucy was careful to focus her ethnographic notes on general patterns and observations rather than in-depth discussions of individuals who had not yet provided written consent. This ethical requirement – to draw a line between general observations and formally consented interactions – distinguishes Lucy’s research from studies (including Jai’s and Sarah’s) where no formal research activities take place until consent forms have been signed. It also demonstrates the importance of the researcher understanding the ethical boundaries of their specific project.

A key issue with gaining consent in the context of LGBTQ+ youth groups, of course, concerns the participants’ age. Ordinarily, researchers might request that consent forms for participants under a prescribed age (often 18) are signed either by both the young person and a parent or guardian, or just by a parent or guardian. However, in this case, many young people attended the LGBTQ+ youth groups without their parents’ or guardians’ knowledge. Requiring permission from the responsible adult in these young people’s lives may have therefore posed a risk to their welfare, given it would necessitate ‘outing’ them. Indeed, many ethical review boards now recognise that, in circumstances like these, a signed consent form may actually introduce more risk for participants than not having a form at all (Guta 2018). In Lucy’s case, she had the advantage that the youth groups were led by responsible adults: qualified youth workers with a duty of care to the children and young people attending, who could give permission in place of a parent following careful discussion with the young person in their care. Knowing this would deviate from long-standing practice, in her REC application for each study she cited relevant scholarship in the social sciences that supported this approach (e.g. Taylor 2008). Where ethics committees expressed additional concerns, Lucy revised aspects of the process in partnership with the lead youth workers at each group, thus meeting the needs of both the institution and the young people she was working with. This example demonstrates how research ‘in the real world’ does not always fit with the ideal hypothetical research context.

4.1.2 Expectations around confidentiality and anonymity can shift

Our second principle builds on the point that regulatory ethics procedures may provide a useful starting point, but the immersive and shifting nature of participatory research means they should not be treated as fixed outcomes or immutable rules. Indeed, in PCLR the researcher must be attuned to participants’ shifting needs throughout data collection and be prepared to modify their plans should the unexpected happen; this may require further negotiation with an REC or external partners, or decisions may need to be made spontaneously in the field. Either way, to prepare for the unexpected, researchers must be clear on the boundaries of their research practice and should manage all parties’ expectations carefully.

For the MFO project, Jai could not know in advance who her participants would be, what kind of social networks and digital practices they would engage with, or what media they would be willing to share. Similarly, before entering each youth group setting, Lucy did not have first-hand knowledge of members’ typical practices, so could not predict how natural her presence as an ethnographer would seem or how straightforward it would be to carve out time for interviews. In both cases, it was necessary to take a flexible approach which (as discussed above) prioritised the developing relationship with participants. Being clear on the aims of the project and the key ethical principles that should be maintained – including participant autonomy, informed consent, and anonymity – helped each researcher to be adaptable whilst also ensuring they remained focused in their data collection processes.

For the MFO study, collecting participants’ personal digital media data across a range of contexts raised several ethical concerns. Giving participants a degree of autonomy and control was essential for addressing ethical concerns around privacy, anonymity, and the potential for undue intrusion. However, this flexibility meant that Jai had to accept high levels of uncertainty and variability in the data collection process and be willing to adjust the parameters of her research as the project progressed. One outcome of this approach was the varied scope and amount of digital media shared by participants, because they each used digital media in different ways, and had different expectations and feelings around what they would share. For example, one participant shared only six pieces of digital media data from two platforms (Messenger chat and Facebook groups), whereas another shared 198 pieces from four platforms (blog, Facebook group, Facebook page, and Twitter).

It also became clear through Jai’s interviews that some participants were uncomfortable sharing digital media data that involved other interlocutors. Some were concerned that even asking for consent from friends and acquaintances could harm important relationships or affect their reciprocal trust and legitimacy within communities that were important to them. As a result, Jai took the default position (contrary to her original plans) to only collect participants’ own posts, with no interactional data (e.g. multi-party chats, replies, or comments) unless participants explicitly expressed a desire to the contrary and sharing was deemed ethically viable. Two participants, however, were keen to share selections of their mobile messages (specifically, extended chats on Messenger and WhatsApp). These participants acted as mediators, seeking informed consent from secondary participants. Whilst the decision to largely exclude interactional digital data limited the scope of this research, giving participants a high level of autonomy and control had significant benefits for the project, fostering strong relationships of trust and enhancing the project’s ethical viability (for a similar example, see Spilioti and Tagg 2021).

Alongside this flexible approach to digital data collection in the MFO project, the parameters and expectations around anonymity also shifted in response to participants’ circumstances. The initial PIS and consent form stated that participants’ confidentiality would be respected through the anonymisation of data and removal of identifying factors such as places and names. However, there are a range of factors that make definitive promises of anonymity impossible, especially when participants have public online accounts and profiles. For example, with open and accessible digital media data (for example public profiles on X [formerly Twitter], or public blogs), it may be possible for a third party to search text or images shared by the researcher, and find the original, non-anonymised source (as outlined in BAAL 2021). This is a particular problem within linguistic and discourse analytical research, where analysts generally quote sources verbatim. It is important that participants are made aware of such barriers to anonymity. In this case, the issue was raised with MFO participants when sharing the PIS and completing consent forms, and again at the point of collecting ‘searchable’ media of this kind. However, the care taken when selecting data to be shared, coupled with the relatively small readership of academic texts, was enough to reassure both researcher and participants that sharing and collecting ‘public’ digital media data posed acceptable levels of risk in this case.

The interview process for the MFO project also revealed a second, unanticipated, barrier to anonymity: several individuals were very well known within their communities. For example, one worked for a prominent adoption charity, was well known amongst adopters on social media, and had written a book about education for adopted children. Whilst it was unlikely that members of the general public could identify these anonymised participants, it was likely that members of their own communities could recognise them from what would usually be quite innocuous details about their lives, or from excerpts of their digital media data. Jai was therefore called to raise this issue with affected participants, pointing out the details that could make them identifiable in some contexts and giving them the opportunity to withdraw some of what they had shared. This process of re-negotiating informed consent was undertaken in collaboration with the ethics committee. After approaching the committee with this information, it was agreed that, once affected participants had been adequately briefed, they would sign a new consent form which explicitly stated anonymity could not be guaranteed. Ultimately, all participants decided to leave their data as it stood, since they did not feel the need to be anonymised to the degree Jai had originally promised. Having clear boundaries around appropriate and acceptable data-sharing and confidentiality was crucial here; Jai used her judgement to navigate satisfactory levels of risk in relation to participants’ preferences, her own sense of responsibility to these individuals, and institutional requirements. The involvement of an institutional ethics committee at an advanced stage in the research process, further, demonstrates how regulatory frameworks can support researchers in navigating micro-ethical shifts throughout the life of a project.

In Lucy’s research with LGBTQ+ young people, it was even more imperative for her to carefully consider, both with and on behalf of her participants, the potential consequences of their information sharing. As an adult collecting data from potentially vulnerable children and young people, she had to be sure her participants understood that they had autonomy and that, just because they had agreed to take part in the research, they were not obliged to share private details about their lives. Making it clear that participants can (a) share as much or as little as they want, and (b) withdraw or retract anything they have said after the event, is a standard component of many ethical procedures, designed to minimise harm to participants. However, it is particularly important for the researcher to emphasise this when working with young people who may have had less opportunity to think through the impact of sharing personal or upsetting information. To support young people in understanding her expectations of the interview, Lucy was particularly careful not to ask direct questions about their own experiences. Instead of asking if they had been bullied or about their coming out experience, for example, they were asked how easy they thought it was to be LGBTQ+ in the UK today, giving them control over the direction of the conversation. Inevitably, most of the young people opened up with personal stories despite not being explicitly asked for them, although many chose not to give specific details.

During one of Lucy’s project interviews, however, a young person under 16 shared that they had been sexually assaulted by another pupil at their school. When asked, they said they had not told a teacher or another adult about this. This created an ethical quandary for Lucy, who wanted to respect the young person’s privacy but was also deeply concerned that something so traumatic had happened and they had not been supported by a responsible adult. Pausing the interview, Lucy told the young person she felt it was important they talked to someone and asked if they would be happy for her to inform one of the youth workers; this was delicate, given the young person had been told that the interview would be confidential. Fortunately, despite this shift in what they had expected in terms of privacy, the young person was comfortable with a youth worker being informed. Had they said no, this would have created an even more difficult situation, as Lucy would have been obliged to go against their wishes in order to prioritise their safety over their right to confidentiality (and, inevitably, their trust and participation in the research). The young person was subsequently supported by their youth worker and was ultimately willing to continue participating. Following further discussion with Lucy, the young person felt, on reflection, that they would rather not have disclosed this information in the context of a research interview. It was therefore agreed that this moment be permanently deleted from the master copy of the recording and not used in the research. Had it not been deleted, however, Lucy would still not have included it in any published analysis; the nature of discourse analysis, whereby participants’ conscious and subconscious interactive moves are scrutinised for the meanings they create, inevitably requires data to be reproduced verbatim. Publishing this alongside the ethnographic detail of the youth group would have increased the risk of the young person’s confidentiality being broken. Furthermore, creating a permanent record of such a traumatic moment in the young person’s life would have risked them reliving it at a later stage.

This situation demonstrates that, sometimes, the researcher must decide on the most appropriate course of action in the moment. These on-the-spot decisions are far more manageable if the researcher has clarity on potential risks to participants, and the limits of acceptability regarding those risks, before in-depth research begins. In Lucy’s case, this was also an important moment of learning; in all interviews that followed, she was sure to explain the parameters of the confidentiality agreement before beginning to record and to confirm that the youth workers would step in and provide immediate support if required.

4.2.1 The researcher-participant relationship is complex and variable

Managing the complexity of researcher-participant roles and relationships during the twists and turns of fieldwork requires careful consideration of the relational and professional boundaries that shape our decisions, interactions, and behaviours. Identifying and maintaining boundaries, however, is not always straightforward – roles and identities can inevitably become blurred when engaging with participants and co-producing research. In participatory research models, for example, Macfarlane and Roche (2018: 57) suggest that the centrality of relationships with participants can mean “boundaries between community and academic spaces, between personal and professional spaces and between formal and informal interactions may become […] less defined”. This is often true in the case of institutional settings, where professional and relational boundaries can be complex. Researchers often gain access to organisations through ‘insiders’ but must be sensitive to the hierarchies in which these individuals are situated, and take care not to damage participants’ position within their communities. Such loyalties to individuals might typically be considered a problematic ‘conflict of interest’ in many research ethics frameworks. However, navigating the challenges of these multiple roles and interests is often central to participatory models (Guta 2018: 158).

In Sarah’s GP project, feedback sessions were used with participants, with the aim of collaboratively analysing video data. These sessions were conducted with GP examiners in the early stages of the project (those with the responsibility for grading candidates in the exam) and GP trainees in the latter stages (those preparing to take the assessment). The researchers ensured a range of high-performing candidates, including those from ethnic minority and overseas backgrounds, were represented in the video clips used for feedback sessions and workshops, ensuring these discussions were not contributing to a narrative of deficit around international medical graduates. The sessions drew on a ‘video reflexive feedback’ approach (Iedema and Xyrichis 2021), where practitioners look at videos of workplace practices as a way of focusing on and questioning what might be habituated activities in their day-to-day work. This involved playing short video clips of consented candidates and were intended to be comfortable, open discussions around the exam, in which examiners and trainees were asked to focus on small features of communication. In linguistics, engaging participants in such close analysis presents opportunities to gain insight on members’ understanding of language and communication. However, this can present ethical challenges. While gaining members’ views was the aim, focusing so closely on language could sometimes give rise to comments about people in the recordings that felt unduly critical. For example, one candidate’s video clip (although they were successful in the exam) was heavily criticised by one participant, who cited features such as filled pauses (“umm”) and non-standard grammar in the talk. There was an ethical discomfort in an individual being criticised over language in a way that the researchers might not have viewed as problematic and which the individual might never have anticipated when consenting to the research. As a means of mitigating this in later feedback sessions, Sarah developed a policy of setting some ground rules at the outset, asking participants to maintain a respectful level of critique in their analyses. As outlined in the background to this paper, perceptions of language are socially and culturally laden and research participants, as language users themselves, will bring particular backgrounds and linguistic ideologies that may not align with the researcher’s (BAAL 2021: 3). Engaging with these differences and acknowledging our own positionality is part of good PCLR but there is also an ethical obligation to protect participants in these real-time research interactions.

The video sessions with GP trainees could also become a forum to critique the exam itself and Sarah felt a sense of her professional boundaries during such discussions – for example, at the point a group of GP trainees critiqued the authenticity of a simulated interaction, jokingly saying: “I realise we’re going on a bit of a rant now (laughter). It’s probably quite cathartic though? I hope I don’t get shot for saying it (laughter)”. Though said in humour and received as such, this was clearly framed as a critical position. Sarah judged it inappropriate to participate in this discussion, both in terms of being an ‘outsider’ to their professional group but also as a researcher working in partnership with the organisation. While she was keen to give participants this space, and not shut down their conversations, Sarah was also aware of her own institutional roles and obligations in a research partnership. Behaving professionally in this context meant she drew some relational boundaries in how far she could align with participants.

Sarah’s case study also shows how research boundaries and relationships can become blurred when working closely with partners and stakeholders. In the GP project, some members of the organisation were closely involved in the design and conduct of the project but also became a type of ‘participant’ in the research themselves – in ethnographic terms, ‘key informants’, people who provide privileged access and a detailed understanding of the setting (Atkins 2023: 8–10). A particularly complex ‘key informant’ role the researchers had to consider was the position of a co-researcher on the team, who was also simultaneously a senior member of the organisation. This co-researcher provided invaluable insights and interpretation of data from a practitioner perspective. However, at times this also placed the co-researcher in a difficult intermediary position between the organisation, which was invested in defending aspects of the assessment against criticism, and the openly enquiring and sometimes critical stance of the research (see Roberts 2021: 273–279). This conflict came to the fore during publication of the research report, when considering how best to incorporate feedback from the organisation with the expression of findings. While all findings were presented accurately, avoiding professional harm to the co-researcher involved ongoing judgements, checking in through ‘backstage’ discussions on how aspects of the research were likely to be received, as well as the potential negative impact this might have on them within the organisation. Such a position, where the concerns of an individual are taken into account, perhaps cannot be claimed as truly independent, but in institutional settings like these, researchers inevitably need to make relational judgements about the impact of their involvement on those who work within the organisation.

For the MFO project, Jai was required to navigate relational and professional boundaries in very different settings. Most interviews for this study took place in the intimate space of participants’ homes, and many individuals talked about intensely personal experiences. However, the intrusion of bureaucratic and documentary tools into that space, such as the consent form and audio recorder, was a continual reminder of the professional boundary between participant and researcher. Furthermore, although Jai did talk about her family and personal experiences if prompted, she prioritised attentive listening over reciprocal sharing. These markers of a more distanced, pseudo-professional relationship, however, did not seem to impede participants’ openness or trust. Indeed, as in Miles’s (2019) research, several participants mentioned that they felt able to talk about events and experiences that they would never mention in other situations. For example, Jenny, a single adopter of two children, spoke in her first interview about her experiences of child-to-parent violence. After detailing a time her son physically and verbally abused her, Jenny explained that she did not usually talk about this kind of event, even with close friends and family members, because she felt they wouldn’t understand. Reflecting on why she was happy to talk with a researcher she had never met before, Jenny described the interview process as “almost … artificial” and “almost a therapy”. Jenny maintained this perception of interviews as therapeutic instruments throughout the research process, saying halfway through our third interview “I’m getting well off the point it’s like therapy”. Jenny’s words suggest that rapport between participants and researchers is not necessarily due to the researcher’s efforts to generate a sense of trust, reciprocity, or to ‘flatten’ power differentials. Indeed, in line with Miles’s (2019) findings, it may be the institutionally authorised role of the researcher, the responsibility that comes with this power, and the artificiality of the research interview, that leads participants to trust and feel comfortable with the research processes of which they are a part. Furthermore, established ethical strictures around confidentiality and anonymity can free participants to talk about all aspects of their lives without fear of reproach.

4.2.2 Close attention must be paid to power dynamics within the research setting

Power (im)balances present several complicating factors for researcher-participant relationships, particularly in institutional settings. In ethnographic research, which seeks to understand experience and practice from the perspective of participants, researchers will often work to reduce perceived barriers. Yet researchers still have an ethical responsibility to ensure they do not ignore or dismiss their own power, alongside other aspects of their positionality. Lucy’s work with LGBTQ+ youth groups provides an example of how these dynamics can be negotiated in practice. For this research, Lucy endeavoured to break down barriers between herself and her participants by sharing with them that she was also a member of the LGBTQ+ community. This helped her to explain her motivation for the study and allowed her to position herself as a relative ‘insider’, despite being much older than her participants (being the same generation as many of their parents) and not living in the same city as them.

Lucy’s differences from her participants, in combination with her regular attendance and relaxed but interested approach in the young people’s interactions, meant that she was typically perceived to have the same role as other adults in the youth service context and therefore to be in a position of responsibility or authority. Despite being repeatedly informed that she was a researcher, the young people tended to treat Lucy like any other youth worker; they would ask her permission to do things or ask her practical questions to which she did not have an answer, and on occasion would alert her to concerns about their peers. On the one hand, this misunderstanding provided a useful framework for her participants to understand how to interact with her, and gave her a sense of legitimacy that engendered trust from the young people. On the other hand, to ensure participants were able to give their informed consent, it was important not to mislead them by reinforcing their misconception. Lucy therefore drew a clear ethical boundary on this issue: she accepted the fact that she was perceived by those on the periphery of her research – i.e. the young people not actively involved in her study – as a youth worker. However, she felt that her key participants must understand the significance of her presence as a researcher. Lucy therefore prioritised and underlined her role as a researcher when it came to arranging recordings and obtaining consent, ensuring her participants understood that their membership to the youth group and access to support was not contingent on their involvement in the project. More broadly, she also took care not to take on more responsibility than she was qualified for; in the case of young people asking for her help or permission during the youth group sessions, then, she would always respond by finding a trained youth worker to support them.

For the MFO study, Jai also felt it was important to tell participants about the aspects of her personal experience and identity that might align with theirs, to emphasise her position as a relative ‘insider’. In the PIS, she introduced herself as “the child of a single, lesbian parent”, and “a bisexual mother with two children”, noting that her own experiences had driven her interest in the challenges faced by underrepresented family groups. During the interview process, Jai found that different aspects of her identity and experience came to the fore with different participants. For example, her experiences of family and sexual identity came to the fore at times during her interview with one single, bisexual parent, who was interested in Jai’s experience as both a parent and a child. Far more participants, however, foregrounded Jai’s role as an academic and a researcher. Two participants were researchers themselves, two had written informative, evidence-based books, and several had taken part in research on previous occasions. These participants tended to show particular interest in, and understanding of, the research process. Explaining her role as a researcher, then, tended not to be a concern in the same way that it was for Lucy. Furthermore, all Jai’s participants were adults above the age of 29, all of them worked, or had worked, in professional roles, all were white and cisgender, and could be broadly positioned as middle class. Jai therefore shared far more with her participants than she had initially anticipated. Whilst this relative homogeneity had some undesirable consequences, in terms of a lack of diversity in some domains, their shared sociocultural experiences meant rapport and trust could be built quickly, and they shared a similar sense of personal and professional boundaries.

Despite their familiarity with the research process, some of Jai’s participants compared the research interview with therapy, as we saw with the example of Jenny in Section 4.2.1. In this case, Jenny was orienting to the asymmetrical therapist-client structure that has some parallels with the researcher-participant relationship. This comparison is not entirely surprising: like a therapist, Jai was completely focused on her participant’s words and needs, and she engaged in a process of active listening that is also used in counselling and therapeutic contexts. Nevertheless, therapists and researchers are bound by very different rules and regulations, so this positioning could point to a disjuncture between participants’ expectations of (and feelings about) interviews, and what actually happens to those interviews once they become ‘data’. This point of conflict can be a particular concern for linguists and discourse analysts, especially those who conduct close qualitative analyses of data. Even when participants know (as they did in this case) that the researcher is interested in their language use, it may not be at the front of their mind in the moment of sharing.

When Jenny framed the research context as therapy, Jai was required to make a snap judgement about whether this positioning was a cause for ethical concern, in terms of a potential misunderstanding of the researcher’s role, or overstepping of relational boundaries. In this case, Jai felt it was not problematic because, firstly, she could see that the comparison highlighted the beneficial aspects of the interview process, as well as the trust and rapport they had built. Secondly, Jenny was one of several participants who had taken part in research before, and was a professional with experience of evidence, information gathering, and analysis herself. Jai was therefore confident that Jenny fully understood both the researcher’s role, and the kind of analysis to which her words would be subjected. Questioning or correcting Jenny’s framing, on the other hand, may have affected their trust and rapport. Further, labouring the point about the level of scrutiny involved in linguistic analysis would make any participant unacceptably self-conscious about what they were saying. Nevertheless, had Jenny requested advice, or asked for therapeutic insights, this would have likely crossed an ethical and professional boundary. Furthermore, had discussions of mental health difficulties arisen, Jai would have referred Jenny to appropriate sources of support, just as Lucy directed young people to youth workers when they needed help or advice. In such unpredictable moments, the researcher must make well-informed decisions on the spot, according to their personal and professional boundaries, understanding of their participants’ situations, and the nuances of the research context.

Looking to the GP project, the feedback sessions and workshops described in Section 4.2.1 offered a helpful means of bringing ‘insider’ perspectives into the data analysis. The researchers did not hold a fixed position of ‘expertise’ here. All participants, whether examiners or trainees, were healthcare professionals who shared equal or perhaps more senior status to the researchers; it was made clear in these sessions that it was their practitioner expertise being sought, to help interpret video data. However, there were also moments where participants oriented to the researcher’s ‘expert’ footing in linguistics or, as outlined in Section 4.2.1, where the researcher stepped in to moderate discussion around individuals’ language use. This kind of relationship in institutional ethnography has been observed by Roberts and Sarangi (2003: 342), who comment on the learning needed on both sides to understand each other’s “epistemological worlds”, as well as the two-way process of language socialisation (Roberts 2021: 279). The wider controversy around differential attainment, language, and ethnicity in the assessment also had a bearing on these relationships, with participants frequently articulating their own backgrounds. As a white British woman herself, Sarah was aware of the privileged cultural lens through which she herself looked at language and communication in the assessment. Participants’ ability to articulate criticisms of the assessment structures and even their own evaluations of ‘good’ performance suggests feedback sessions did enable some to voice controversial opinions that did not always align with the official position (see Section 4.2.1). Nevertheless, the broader power dynamics of the organisation were still present and many participants were concerned to express criticism cautiously. As one examiner noted in a feedback session, “I’ve got to say this rather politely. I was very conscious that the majority […] they were all old and male doctors”. Making space for participants to discuss the exam therefore involved navigating, but not eliminating, complex power dynamics and asymmetries. Holding the sessions outside the organisation and ensuring anonymity may have enabled more critical discussion, but participants were still talking to groups of colleagues and were therefore alert to the potentially damaging consequences of criticism. Power relations in these types of settings are inevitably complex and, for all the design that goes into the planning of feedback groups, they must often be managed interactionally during the research itself.

These studies reinforce the point that power dynamics are nuanced, situation-dependent, and may shift across time and context. The goal of flattening hierarchies to empower those being researched, on the other hand, is not always practical or even desirable – indeed, we have shown that participants may find it helpful or reassuring to embrace asymmetrical structures they are familiar with, as we saw in Jai’s and Lucy’s studies. In more institutional settings, such as the youth group (Lucy’s study) or RCGP (Sarah’s study), there may be perceptions of authority for the researcher to manage. These examples also show how, in attending to power imbalances, PCLR practitioners must exercise critical awareness around their own bias, and how their participants may be affected by these biases. A lack of critical awareness in this respect could result in participants being misled, and potentially delegitimise the process of gaining consent.

More broadly, however, it is important for researchers in PCLR to engage reflexively with their subjectivity throughout the research process, considering how our own experiences, identities, and feelings, and thus our own assumptions and perspectives, impact our interactions with participants and subsequent analytical interpretations (see Pillow 2010). It is imperative that PCLR researchers acknowledge their privilege and status when seeking to understand and represent marginalised participants. This includes acknowledging the colonial perspective of most research in PCLR whereby white researchers’ understanding of what is sociolinguistically interesting or marked has often been prioritised over that of their participants (D’Arcy and Bender 2023). Awareness of these dynamics is certainly helpful but it may not be a straightforward matter of ensuring equality and/or reciprocity between researcher and participant. Instead, the researcher must develop clear boundaries that are appropriate for, and sensitive to, the research context, and everything that falls within it (including themselves). Where possible, we advocate for these reflections to be included in published accounts of PCLR work; such transparency can support future researchers, who learn from the experiences of others, and ultimately help to create new ethical standards.