“Would you Decide to Keep the Power?”: Reflexivity on the Interviewer–Interpreter–Interviewee Triad in Interviews with Female Punjabi Rheumatoid Arthritis Patients

Abstract

This article presents methodological reflections on the different streams of knowledge that are drawn upon during interpreted interviews and the shifts of power between (1) the interviewer, (2) the interpreter/co-researcher, and (3) the interviewee. Interpreters are increasingly seen as active agents in the interview process, and they act as cultural brokers. Interpretation by a nurse researcher introduces further challenges and benefits to the interview dynamic, which was explored through reflexive discussions with an independent researcher. These challenges include conducting interviews in a clinical setting, where the health professional–patient relationship remains active. A modified discourse analysis was used to examine the subject positioning in the interview situation and the power negotiations that ensued. The main conclusion that can be drawn from these reflexive accounts is that the use of different streams of knowledge (experiential, clinical, cultural, and academic) enhanced the interview interaction, and power relations were successfully negotiated to facilitate rapport and data collection. Reflexivity provides an important tool for identifying, and learning from, the challenges and benefits of working with an interpreter, who is also a co-researcher with multiple professional roles.

The difficulties in recruiting minority groups to participate in research are well documented, including South Asian patients (Hussein-Gambles et al., 2004). It is important to be able to research non-English speaking participants because their English-speaking peers cannot represent them as they have often been acculturated (Esposito, 2001). The least acculturated, with little or no command of English, face the greatest barriers to health care resources, and their specific needs should be researched and understood. There is an increasing recognition of the active role played by interpreters in cross-cultural research. Traditional methods of forward and backward translation to ensure word equivalence have been criticised for weak conceptual equivalence (Larkin, Dierckx de Casterlé, & Schotsmans, 2007). Also, the assumption that translation is an objective and neutral process, in which translators are mere “technicians” in producing texts in different languages, has been rejected (Wong & Poon, 2010). Instead, interpreters are increasingly seen as cultural brokers who balance language proficiency with cultural knowledge to provide a meaning-based translation, with the best conceptual equivalence. Nevertheless, there is little reflection on the influence of interpreters on qualitative research (Temple & Edwards, 2002).

Where interpreters are visible and are engaged in the research process, they provide a resource about the cultural norms of the target community and enable the researchers to respect these norms in the course of their research (Higginbottom & Serrant-Green, 2005). Their knowledge of common idioms and discourses can provide important insights into the social, political, and historical contexts of the participants' lives (Wong & Poon, 2010). It has been recommended that the influence of the social identity and positioning of the interpreter be taken into consideration in qualitative research (Gupta & Ferguson, 1997), just as it is in researchers' reflexivity on how their own social location, or positionality, influences the research process. Positionality allows the lens through which the social world is interpreted to be clearly stated. Maher and Tetreault (2001) explained that “knowledge [is] valid when it takes into account the knower's specific position in any context, a position always defined by gender, race, class and other socially significant dimensions” (p. 22). For example, as much as ethnicity may be a shared identity between an interpreter and a research participant, class positioning and professional status may differ and thus position the interpreter as a relative outsider (Temple & Edwards, 2002). These relational positions, including age, have consequences for participants' sense of control and of power within a research encounter and how the data are constructed by the interviewer–interpreter/co-researcher–interviewee triad.

Less has been written on working with bilingual researchers. Those articles that do address these issues provide few actual examples. As with interpreters, bilingual researchers may differ from the target population, and may be biased by age, education, and gender (Hunt & Bhopal, 2004). In addition, bilingual researchers' intercultural dual representation and role dualism may create ethical complexities (Shklarov, 2007). For example, will participants have greater expectations of empathy and support from them than from a traditionally neutral translator? The bilingual researcher crosses several boundaries, bridging both the language and culture of the participant and the professional knowledge of the researcher.

Qualitative researchers have emphasized that the researcher and participant collaborate in the construction of knowledge (e.g., Gordon, 1998). Nevertheless, we also draw on different streams of knowledge to take part in the interaction (e.g., professional and experiential knowledge). Bourdieu (2001) described how our habitus (how we think, speak, act, etc.) is formed by the access we have to economic, social, and cultural capital. When one's habitus encounters an unfamiliar world, it can change and transform (Reay, 2004), which may initially be uncomfortable. Foucault (1972) showed that dominant discourses lead us to move, speak, and think about ourselves in specific ways. For example, when participating in a research interview, we speak and act in a way that is identified as a research discourse (by us, but perhaps not by the interviewee). Discourses can be understood as power-related structures of how we understand the social world. Power relations are created within an interview situation (Aléx & Hammarström, 2008), and these are more complicated within an interpreted interview.

This article provides examples from an interview study with United Kingdom (UK) Punjabi women with rheumatoid arthritis and addresses the complex power dynamics within interpreted interviews, where the interpreter is a nurse researcher with multiple roles. Reflexivity identified both a theoretical perspective on the process and the practical challenges and benefits of working cross-culturally with multiple researchers in interview methods.

Biographies

Tessa – At the time of the study, I was a postdoctoral researcher in medical sociology, in my thirties, and interested in the patient perspective of illness and the role of culture on the conceptualisation and experience of (ill)health. Although I had experience conducting qualitative and mixed methods studies on the patient perspective of illness, this was the first time I required interpretation during interviews. I am White British. I was born in Kenya and travelled extensively as a child and an adult, including living with Indian families in India for six months. This led directly to me studying human geography and social anthropology as a first degree. I am married to an Italian and our children are being raised bilingually. At the time of the interviewing, I was visibly pregnant.

Kanta – I qualified from the School of Nursing, Queen Elizabeth Hospital, Birmingham, United Kingdom in 1996. I specialised in rheumatology as a nurse specialist in 2001. In 2003, I took up a post as a nurse researcher and developed a profile of research around the issues related to ethnicity in musculoskeletal practice. I have a track record of publications in this area. I was awarded a Clinical Doctoral Research Fellow award in 2010 for three years from the NIHR. My PhD work focuses on issues related to medication adherence and the influence of ethnicity in patients with rheumatoid arthritis. I am of a British Indian ethnicity and speak a number of languages, including Hindi, Punjabi, and Urdu, all of which I have used in being an interpreter for focus groups, interviews, and questionnaire studies (as a co-researcher or principal researcher).

This study was the first time that Tessa and Kanta had worked together. Tessa kept a field journal to aid analysis and to reflect on the experience of interpreted interviews and researching a different cultural group. These thoughts laid the foundation for this reflexive paper.

The Project

This was an exploratory study to examine the treatment outcomes important to an ethnic minority group and compare them to those of a mainstream White British patient population (Sanderson et al., 2012) elicited from a previous study (Sanderson, Morris, Calnan, Richards, & Hewlett, 2010). Interviews were conducted with 16 Punjabi adult women diagnosed with rheumatoid arthritis, who were recruited in three UK cities with diverse Punjabi communities (Sanderson et al., 2012). Given the complexity of ethnicity and culture, this study focused on one cultural group, defined by language rather than country of origin. Purposive sampling was used to obtain a diversity of characteristics: disease duration, age, generation of participant, work status, English fluency, and religion. Potential participants were recruited through rheumatology departments in the three cities. Patient information was available in written English and an audio Punjabi version (because few Punjabi people are able to read the script). Ethics approval was granted by the Southmead Research Ethics Committee (ref. 10/H0102/10).

Tessa was chief investigator and the interviewer (1 of the triad) across the three sites. Kanta is a nurse researcher and provided interpretation (2 of the triad) where participants did not have sufficient proficiency in English to participate in a research interview. Interpretation was usually provided after participants had spoken three to four sentences to enable some fluidity of speech. Where the meaning of participants' speech was not clear to Kanta, she used clarifying questions without recourse to Tessa first. Five of the 16 participants fell into this category, although it was not always evident in advance whether interpretation would be required because conversational English was not sufficient to conduct a research interview. Interviews began with an open invitation to “Tell me about your health,” and prompts from a topic guide (Sanderson et al., 2012) were used to ensure important areas identified from the literature were covered. Interviews were digitally recorded and later transcribed verbatim. Where participants were interviewed in Punjabi, Kanta produced the English translation of the transcript. A sample of these interpreter interviews was verified independently to reach a consensus on nuances in meaning. Analysis for the treatment outcome data was conducted using Framework (Ritchie & Lewis, 2003), a tool that allows comparison across participants in addition to enabling the individuals' stories and contexts to be considered.

During the process of conducting the interpreted interviews, the importance of reflecting on the interview dynamics became apparent. Although we (Tessa and Kanta) had informally debriefed after each interpreted interview, we decided to formally reflect on our experience. When the data collection had finished, a researcher (Laura), who was independent from the study, conducted separate reflexive discussions with us to elicit our experience of conducting interpreted interviews where the chief investigator (Tessa), the interpreter/co-researcher (Kanta), and the patients had diverse backgrounds and experiences. Examples from the interview topic guide include:

1. How have these interviews been different from previous ones?

2. Describe the dynamics in these interpreted interviews.

3. What have you learned about interviews that require interpreters?

4. How were Kanta's multiple roles managed?

5. Do interpreted data have implications for analysis?

6. For Tessa: How have these participants reacted to you as someone who is White British?

7. For Kanta: Does the interpreter have a protective role towards the participants and their community? How?

The reflexive discussions were recorded, transcribed verbatim, and analysed using a modified discourse analysis. During an interview, discourses may develop around specific power-related dimensions such as gender, class, or ethnicity (Davies & Harré, 1990). Within these discourses, subject positioning allows a way of analysing the production of power and how dominance may shift situationally. Nunkoosing (2005) argued that as an interviewer, one's identity is not limited to that of an interviewer, but other identities are also drawn upon to legitimize actions. Subject positioning can therefore offer an explanation of an individual's ability to occupy and move between various identities (subject positions) within an interaction, such as an interview, depending upon the power dynamics and context (Sundin-Huard, 2001). In this article, the term actor is used to indicate the possibility of multiple identities performed by any of the individuals involved in the interview interaction, rather than individuals having one static identity. Tessa and Kanta analysed both reflexive transcripts and identified similarities and differences of experience.

The following material draws mainly on the analysis from the reflexive discussions, but also uses excerpts from the original interviews with the female Punjabi rheumatoid arthritis patients (Sanderson et al., 2012).

Analysis and Reflections

The dynamics of the interpreted interview can best be understood through unravelling the different streams of knowledge (see Figure 1), and the power associated with them, that the actors drew upon during the interview.

Figure 1.

The different streams of knowledge affecting the interview process.

Four themes were identified from the reflexive interview data that illustrate the subject positioning based on the different streams of knowledge: challenges to control, outsider positioning, juggling roles and streams of knowledge, and interplay of discourses.

Challenges to Control

Tessa's reflections.

Control over a situation is an indicator of power. Control during the interpreted interviews was challenged for different reasons for Kanta, the interviewees, and me. My power originates from being the grant holder for the project and directing the research. Figure 1, however, indicates how my knowledge is restricted to that related to research, which consists of research skills, the literature related to the project, and my previous research that led to this particular project. Direct cultural knowledge is limited to travelling through the Indian Punjab region in the year 2000. In comparison with previous interview studies, I was aware that I would have to relinquish some control over the interview process because of the need for interpretation and interviewing through an intermediary. Part of relinquishing control was sanctioned formally at the outset through inviting Kanta to ask questions that addressed issues that I may have missed during the interviews (particularly through lack of cultural knowledge, but also as evident from her clinical experience).

This invitation arose from recognition of her multiple identities and different streams of knowledge:

Kanta is a co-researcher, rather than simply an interpreter, professional interpreter, she's, you know, got her own thoughts about what's coming out of the data, her own questions she wants to follow up. (Tessa, lines 434-6)

Nevertheless, I had not anticipated that some of my control would be lost unwillingly, and this was disconcerting. Control was lost when I felt that questions were being clarified without being able to form those questions myself and when opportunities were missed to probe on talk that was pertinent to the research because of a delay in interpretation. This loss of control is evident in this reflection:

You can hear the sort of backwards and forwards between the participant and her [Kanta], and she's asking questions, clarifying, and you think, “Hang on, hang on, I don't know what's going on.” You know, “Let me back in.” (Tessa, lines 447-9)

Over the course of the interviews, debriefings between Kanta and me clarified how we worked together and as a result I felt more comfortable “giving away control” (Tessa, line 564), rather than having a sense of losing it. I saw evidence in the transcripts that Kanta focused on accurately interpreting the meaning of the talk, even when specific words were not interpreted during the interview because of necessary précising (but were there in the translations she made later). I also realised that the iterative process of using analysis of earlier interviews to inform later ones enabled me to explore interesting data and emerging theory with later interviewees.

Kanta's reflections.

A challenge to my control of the situation was instigated by coming out of my professional role to establish a friendly relationship in an unfamiliar situation for the interviewee (i.e., the research interview) and acting differently towards that person than in my clinical role. This was a potentially vulnerable position, which I wanted to limit in order to then be able to re-establish a professional identity:

You're coming out of your professional role slightly, because you're calling them Auntie, so therefore you're proposing that this is now a friendly relationship. But at the same time you can change that very quickly then to be more formal once you start the interview. (Kanta, lines 130-4)

It was important that the interviewees knew I was a health professional and a researcher in addition to acting as an interpreter for the duration of the interview. None of the participants were previously or currently under my care, but communicating a professional identity was important in case of future clinical contact. It also explained my role, for example, in escorting participants through the hospital to the interview setting. From using interpreters in clinical encounters, I was aware that patients experienced most interpreters as disinterested, and therefore I wanted to disassociate myself from them and that perception of them:

The patients, usually, are well aware of the hospital interpreters. Some expressed a view that they're very cold, very rushed with time, down to the point. (Kanta, lines 147-8)

It appeared that the participants also experienced a challenge to their sense of control of the situation. This was initiated through the act of participating in research, which was an unfamiliar interaction. Although the interviews took place in the rheumatology department where the participants attended appointments, an unfamiliar room was used:

This body language is there from patients: “OK now you're taking me to a separate room. That's not, you know, my normal environment.” (Kanta, lines 60-1)

In addition, signing a consent form and having the interview recorded were unfamiliar occurrences that signalled a shift of control:

Some of them get very alarmed about um, “God, there's a recorder, you're going to record me, what is all this for?” And although you tell them over the phone, or [in] the pre-interview sort of discussion you have with them, I think it's very real for them, especially when they haven't done an interview before. (Kanta, lines 30-3)

Reiteration of the reasons for obtaining consent and recording their voices was a vital part of the interaction. Reassurance was given that respondents' negative views would not impact their care. The voluntary nature of participation was emphasised to signal that they were able to cease the interview at any point without consequence, and therefore they maintained control over their participation. The key factor throughout the interview was to ensure that patients felt comfortable enough to talk about issues that they had perhaps not spoken about before. The reassurance and presence of the co-researcher allowed the patients an opportunity to discuss openly about their well-being and disease-specific issues. Patients became very aware that the setting gave them permission to hold conversation through the researcher. At the end of the interviews, we asked whether there was anything that was important to the participant that we had not raised. This enabled the interviewee to take some control at the end of the interview to communicate what was most salient to her.

Fourteen of the interviews took place in rheumatology departments and two at participants' workplaces. A decision was made not to conduct interviews in the environment of their homes, where it may have been difficult to focus on the participants' accounts in households where other family members would be present. The research setting, therefore, may have affected the participants' sense of control, by removing them from a potentially non-threatening home space to one (in the case of the rheumatology departments) where their control may be challenged by increasing the power of the “healthy” interviewers over the “ill” interviewees. However, Tessa was visibly pregnant at the time of the interviews and this provided a common source of talk before and after the formal interview, where participants were interested in the researcher's body and feelings, showing an awareness of a changing body, if not an ill one.

Outsider Positioning

Tessa's reflections.

In the reflexive interviews, it is evident that I positioned myself as an outsider from the outset due to not speaking the Punjabi language and not having “lived” or “embodied” cultural knowledge. I therefore relied on Kanta to act as a cultural conduit to enable data collection with those that spoke no or minimal English and to provide contextual support for data analysis across all participants. I noticed a number of practical consequences resulting from the language barrier and Kanta providing a cultural gateway. First, it was a challenge to build verbal rapport with the non-English speaking participants through her rather than directly. This was difficult because where I would normally engage in informal conversation while we settled into the room, the need for an interpreter appeared to formalise the interaction and make small talk stilted. Second, I relied more heavily on my interpretation of their non-verbal communication and reciprocation of it to build rapport and understand the emotional tone of talk. However, this was not always reliable. For example, it was difficult to make a timely response to sensitive information or issues because of the delay of interpretation. The most uncomfortable experience was where I understood after interpretation that the interviewee's husband had died. Although the interviewee was aware that I would understand only through interpretation, it felt awkward to me to have a delay in response to an emotional disclosure of this magnitude:

Sometimes you feel like you haven't reacted in the same way, or it's very delayed, and you think, “how is that person going to react to me sitting there going, ‘Oh yeah,’ smiling,” you know. “Oh,” and suddenly you realise the severity of what they're telling you. (Tessa, lines 74-8)

Third, I was concerned about offending participants if I asked about issues that might be sensitive or stigmatised in their community. For example, the stigmatisation of illness and disability within the Punjabi community arose in the interviews (Sanderson, Calnan, & Kumar, in review), and I was unsure whether participants would be comfortable about or open to talking to me as an outsider about these issues. As the project progressed, I realised that it was the sensitivity and appropriateness of questions that was important, rather than my outsider status with these participants. A useful strategy was to transfer the source of the question to previous interviewees. For example, saying “Others in your community have talked about the stigma of illness as a problem. What is your experience?” Fourth, advantageously, I explicitly used my status as an outsider to elicit explanations about cultural values and norms where understanding is assumed. For example, a participant was explaining how daughters are relied upon for support in Asian families and said to Kanta, “Well, you know how it is with us,” and I was able to ask her to explain in detail for me as an outsider. In the reflexive interview, I described how by positioning myself as “Other,” I was able to elicit explanations about values and experiences that may be left unspoken between those who assume they share the same culture:

Sometimes I felt, because I wasn't from that culture, it was easier for me to ask certain questions actually, than it would have been for her [Kanta] because, “Well, why are you asking me? You should know this.” (Tessa, lines 541-5)

The transcripts of the interpreted interviews show that I made attempts to show that I had some relevant cultural knowledge, such as knowing the names of Punjabi dishes. On reflection, this was to establish my legitimacy for my interest in conducting research with this specific group. Therefore, my lack of cultural knowledge was a potential threat to my control of the research but also could be used advantageously.

Kanta's reflections.

The feeling of being an outsider was applicable to the research study itself. I initially positioned myself as a translator and therefore relatively distant from the research aims, but later I exhibited an interest about the content of the patient conversation and revealed my researcher identity to Laura:

So my role in this particular project has been as a translator. (Kanta, line 7)

When Tessa gives you the opportunity to say, “Do you want to ask anything?” and I say, “Oh yes please. Can I ask the patient? I really want to ask her this, this and this.” You know, so I'm sort of trying to get things out and tease things out. And I will say to Tessa, “Oh this was said actually in our previous studies,” or whatever, so you're making the links. (Kanta, lines 614-9)

My request for permission to ask a question signals that I perceived Tessa as holding the dominant questioning power. However, during the period of data collection a shift occurred in my position, from interpreter only to co-researcher, as we learned to work together. This is evident in our decision to let the participant talk more freely, without “breaking the conversation,” because interviewees were becoming frustrated:

I could see the patient taking a big sigh to say “I'm trying to remember where did I leave off?” (Kanta, lines 216-7)

Instead, we agreed that I would interpret larger amounts of talk, focusing on conveying what appeared to be most important to the interviewee and in relation to the research aims during the interview, and then I would translate the transcript fully later. This required a continuing dialogue between us as researchers about the unfolding themes and theories in the data. This also enabled me to re-position as an insider in relation to the research study. I had worried that “Tessa had a second-hand story” (Kanta, line 607) because of my interpretation of the interviewees' answers. Nevertheless, through working closely and developing trust between us, it was evident that rich interview data were being obtained.

Interviewees appeared to position themselves as outsiders to the academic discourse and the research setting. This was most explicit where unfamiliarity with the research process led to initial feelings of uneasiness, as discussed above. Another way in which interviewees positioned themselves as outside of the academic discourse was through a perceived lack of education. This manifested in the interviews as statements of not knowing the answer to questions because of a lack of education or being unsure of the “correctness” of their talk:

Interviewee B10: [In Punjabi] Well I don't know much. I have said what I know…. I don't know any more [pause] I am not that educated. (Interview B10, lines 283-5)

Interviewee B10: [In Punjabi] Well I hope I have told you everything that I know I hope you found my story helpful. (Interview B10, line 594)

Participants' unfamiliarity with reflecting on their lives appeared to result in difficulty in answering more abstract questions, particularly ones not in their first language:

Kanta: She says she does speak English.

Interviewee B10: [In Punjabi] I can but I don't know how to construct sentences, my sentences properly. I can speak a bit, but this was very deep conversation. (Interview B10, lines 469-72)

Practically speaking, this meant that open questions, such as “Tell me about your health,” that had been used successfully in previous research (Sanderson et al., 2010) made the interviewees uncomfortable. Perhaps they were more familiar with closed questions and found silence, meant to enable them to think about the question, awkward. As a result, more reassurance about the value and importance of their experience was required:

In this particular group, we needed, both Kanta and I needed to give more reassurance about the fact that what they were telling us was valuable. (Tessa, lines 644-5)

Although we gave prominence to the patients' experiential knowledge, through the act of conducting research on this topic and stating explicitly to them that we valued their experience of living with rheumatoid arthritis, it appears that these particular interviewees did not themselves value this experiential knowledge. This is in contrast to previous research where White British patients claimed expertise through their experiential knowledge and even valued it more highly than health professionals' training and biomedical knowledge (Sanderson & Angouri, in review). Instead, by asking whether they were giving the correct answers they presumed we had greater knowledge and thus diminished their potential power in the interaction.

Juggling Roles and Streams of Knowledge

Tessa's reflections.

A decision was made in the design of the study to use a co-researcher with the appropriate language skills, rather than a professional interpreter. The three roles that Kanta brought into the interview interaction (interpreter, researcher, and nurse) were reflected in the streams of knowledge drawn upon in the interviews. First, there were times during the interview when she would draw on her clinical experience and knowledge, for example using a medical term when the participant had used a less technical phrase and/or adding a clinical commentary:

Interviewee B1: [In Punjabi] My GP has said that my raised blood pressure [phrase in English] isn't a disease as such, it's through your RA. I am on a lot of medications.

Kanta: [Translating to Tessa] She does suffer from hypertension, but then she doesn't feel like that's a condition as such, she feels that she's on numerous medications and that's probably a side effect of one of them. (Interview B1, lines 259-64)

Kanta: [Translating] She [Interviewee B2] was on medication, she's not quite saying the name, but she was on injections. [To Tessa directly] I guess it was probably methotrexate or gold in [those] days. (Interview B2, lines 67-9)

In addition, Kanta occasionally switched into a clinical role when there was a particular care need identified and she was conscious of this identity and role shift. We agreed that this was a necessary and appropriate response. For example:

Kanta: You need to come to a different session, you need to have a chat to somebody. (Interview B3, line 561)

Second, Kanta could be seen as enacting a research identity by making links between the current data and her research knowledge, particularly her own previous studies:

Kanta: [To Tessa directly] That's interesting. That's the same concept as in my beliefs study. [Translating] So she says no, no they haven't changed hugely. (Interview B10, lines 534-5)

Kanta: [Translating] Arthritis is less than cancer, cancer is a serious disease. [To Tessa directly] That's in the delay study. (Interview B10, lines 689-90)

Kanta's reflection on her own previous research was helpful for me during the interviews to provide context for the participant's illness experience, but it had to be carefully managed during analysis to distinguish who had contributed what in the co-construction of the interview data. I was aware that Kanta was involved in a “juggling act” (Tessa, line 581), and in retrospect it would have been beneficial to discuss managing potentially conflicting roles prior to commencing and during the data collection period.

Kanta's reflections.

My three different identities (interpreter, co-researcher, and nurse) were expressed during the interviews and had implications for positioning myself in relation to Tessa and the interviewee. For example, in the reflective discussion, I was deliberating balancing the need to make the interviewee feel comfortable with the need for effective data collection, which included knowing when to shift control of the questions to Tessa:

We're trying to ease the patient…. if it needs to be that, um, the power just needs to stay here because I'm getting rich data, then would you decide to keep the power? This is what I've been thinking about. Perhaps, you know, um, perhaps it's not power, but it's sort of the patient's comfort zone. And so would you keep the comfort zone because they're connecting with you? Or would you throw the comfort zone into somebody else's, um, ball court for them to connect with that person? (Kanta, lines 671-9)

It could be argued that the interpreter acts as a conduit of power, transferring the opportunity to speak from the interviewer to the participant, and back again. As primarily an intermediary in the interview interaction, I was balancing Tessa's needs (and research aims) with those of the interviewee. In practice, the flow of the conversation was maintained with the interviewee until they completed their answer. Tessa then understood the interviewee's answer via my translation. Upon understanding what the patient had expressed, she then would ask further questions on the same topic or generate a new question, and thus flow was maintained throughout the interview. In some cases, I was required to continue probing as I felt that some patients gave short answers and were hesitant to expand on them. Once the interviewees were made to feel comfortable, they were able to connect with me and, by extension, Tessa, and in-depth conversation took place.

Interplay of Discourses

It was apparent from the reflective interviews that specific discourses were utilised to facilitate communication as “bridges” to make connections between actors. These bridges enabled an actor positioned on the outside of one discourse (e.g., academic) to be invited into the interview interaction more effectively through the utilisation of other discourses (e.g., age). This process made explicit the different social positionings, and it recognized the power differentials of diverse streams of knowledge and looked to address them through an interplay of (temporarily) making one more dominant than another. Two examples are provided.

Kanta's reflections.

On meeting the interviewees, I focused on showing respect for the elder patients (those that spoke Punjabi only were all first generation). I used cultural markers of respect, such as calling them “Auntie” to indicate the social capital that an elder woman in Punjabi culture holds:

Some you refer to “Auntie” and so, to make them feel comfortable. (Kanta, lines 97-8)

They expect that from you automatically, because that's the respect that one has to have for the eldest. So you have to sort of have that concept. “Oh Auntie,” you know, “I hope that you've reached here OK. And was the taxi ride OK?” (Kanta, lines 100-3)

Through this change from being a research participant situated in an unfamiliar process to a respected elder, a favourable shift of power was achieved and a sense of ease was established in the relationship between us. This was achieved through our shared cultural knowledge:

The process of easing begins from the minute that you meet them. (Kanta, lines 45-6)

For those non-English speaking ones, you have to try and reassure them that I will be there and I will be translating and so I'm not going to leave you. (Kanta, lines 108-9)

Interviewee B3: [In Punjabi] I don't think they want to understand [pause].

Kanta: They?

Interviewee B3: [In Punjabi] Our Indian people. My relatives even, when I go to the temple I can't sit for long so people start asking you “Ohhh, what's wrong with you?” (Interview B3, lines 518-23, emphasis added)

A shared South Asian cultural knowledge, an “ethnicity” discourse, was also commonly utilised to establish a rapport. However, one interviewee asserted a religious boundary between us, apparently deciding my Hindi heritage from my name distinguished me from her Sikh heritage:

Interviewee C4: [In Punjabi] What's your name again?

Kanta: Kanta Kumar

Interviewee C4: [In Punjabi] Well, in Sikhism we believe … (Interview C4, lines 569-71)

This reinforces that although we shared some social positioning (e.g., ethnic minority and female), there were differences (e.g., age, religion, and education), and some boundaries were asserted.

Tessa's reflections.

A couple of the interviewees, where an interpreter was necessary, still tried to communicate directly with me using a few words of English. This could be interpreted in different ways, but in these interactions this direct communication by the interviewees appeared to be concerned with asserting their illness experience and their patient “voice,” by emphasising the impact of the rheumatoid arthritis to the primary researcher rather than to the intermediary, and thereby asserting a social hierarchy:

Kanta: [Translating] She's giving an example of herself, where she was very disabled with her hands.

Interviewee B2: Can't open hands [sentence in English].

Kanta: [Translating] Very much curled up like this [curling fingers] and making the chapatti, you're very much required to open your hands and actually tap it and she couldn't do it. (Interview B2, lines 294-308)

Kanta: In what way do you feel uncomfortable…?

Interviewee B3: [In Punjabi] If I go out I can't eat properly. Too much problem [sentence in English]. If I go out to a wedding I can't eat well because I can't hold things properly. (Interview B3, lines 512-5)

In this way, it appears that the interviewees were using experiential knowledge, or an “illness” discourse, to communicate their priorities for the academic discourse of which they were outsiders, and they could understand this academic discourse held greater influence on the outcome of the research. Through the interview interaction, it appears that the attitude of these elder Punjabi women had changed towards research from one of unfamiliarity to appreciation for the bridges that were being constructed:

Tessa: I really appreciate the time you've spent with us.

Interviewee B3: I appreciate your help as well. (Interview B3, lines 638-9)

To end the reflections on a positive note—my experience was that by closely listening to and valuing the interviewees' experiential knowledge, the power differentials created by educational, professional, and cultural discourses did not obstruct the interview interaction, but were constantly negotiated to enhance collaboration through mutual respect:

Often I got the impression that it didn't matter who I was, it was the fact that I was interested in them and their experience that counted. (Tessa, lines 296-8)

Incorporating Reflexivity Throughout Cross-Cultural Research: Practical Suggestions

Following the analysis and reflections that have been presented, we offer some practical suggestions about how to incorporate reflexivity throughout your study. First, the importance of having a research briefing and a cultural briefing prior to and post data collection, and shorter debriefings after each interview, is emphasised. Some points for reflection at the briefings are provided in Figure 2, which are relevant for working with interpreters and co-researchers in interviews and for projects with multiple researchers conducting separate interviews. It is by no means a complete list, but makes explicit the issues that were most relevant from our experience.

Figure 2.

Incorporating reflexivity into regular research and cultural briefings.

Second, how non-verbal cultural information (e.g., body language, tone of voice, etc.) will be recorded should be discussed and consideration should be given to how the method of recording may be experienced by the cultural group being interviewed. In this study, non-verbal cues were not formally recorded during the interviews but were discussed post-interview and noted where possible in the transcripts. Because of interviewees' initial discomfort about the research setting it was decided that no notes (on the verbal cues) would be written during the interview, but this may have lessened our recording of them. Third, reflexive accounts could be written or formal reflexive discussions conducted by those engaged in data collection prior to analysis. This would allow an examination of how the different streams of knowledge and power dynamics may have influenced data construction. This enhanced reflexivity is particularly important in conducting culturally safe interviews with those from minority groups or other hard to reach communities.

Discussion

The interviewer–interpreter/co-researcher–interviewee triad provides challenges and benefits in the process of data collection. This article reflected on how power dynamics affect interpreted interviews. The social positioning of the actors was influenced by the different streams of knowledge on which they could draw to legitimate their role and voice in the interaction. Each actor's habitus (how we think, speak, act, etc.) was transformed through its encounter with an unfamiliar world (Reay, 2004)—for example academic, clinical, or cultural, and through finding ways to overcome discomfort and enhance communication. Furthermore, by engaging with different discourses (e.g., age), the actors were able to utilise different sources of cultural and social capital to facilitate the interaction. In this way, the participation of these Punjabi women in interpreted interviews was successful, and the giving of socially acceptable answers, as documented in research with Black respondents (Fielding, 1994), was avoided.

Bourdieu (1992) speaks of a “room of possibilities,” which implies that it is possible to “achieve good contact between people despite their different access to economic, social and cultural factors” (Aléx & Hammarström, 2008, p. 173). In our study, the unfamiliarity of the research “culture” appeared to be more of a challenge to interviewees' participation than the interviewer's ethnicity. This was due to the brokerage of the interpreter/co-researcher through enacting a shared cultural discourse. In actuality, the interviewer's outsider status to Punjabi culture was advantageous because embedded cultural meanings could be questioned (Tsai et al., 2004). Use of “our” and “as you know” identified the co-researcher as an insider in the South Asian community, although one participant demonstrated that their shared culture had religious boundaries. The co-researcher differed from these interviewees because she was UK-born and second generation, thus differing in terms of age and acculturation. This has parallels to comments such as “you know” that were common in Temple's interviews with Polish people, which indicated her inclusion as an insider in the Polish community because of her background, despite differences in what constitutes a “Polish” way of life and time of settlement in the UK (Temple & Edwards, 2002). This illustrates that there is an off-setting of discourses, where decisions are made about insider/outsider status with respect to the other actors in a given situation.

There were several methodological implications. In studies with bilingual researchers (e.g., Hunt & Bhopal, 2004; Shklarov, 2007), extra time in the research process is required to allow for detailed research and cultural briefings. This is an artificial separation in the briefing because the two aspects are interrelated, but the distinction is useful here to signify the multiple roles of the bilingual (nurse) researcher. Negotiations occurred between the interviewer and the interpreter/co-researcher about their active roles in the construction of the interview interaction, which resulted in an effective team approach. Since a bilingual researcher is actively contributing methodologically to the process, in addition to interpreting meaning, discussions about cultural and intellectual perspectives are required in order to achieve “culturally safe” interviewing (Higginbottom & Serrant-Green, 2005) and maintain a shared focus on the research aims. The (de)briefings are important for remembering that “no one person can represent whole communities” and that “people employed because of their cultural or language expertise have perspectives of their own, which are woven into their social interactions” (Temple, 2002, p. 853). This study benefited from collaboration with a patient research partner of Punjabi origin, who provided a second cultural voice in the analysis. In health research, interdisciplinary project teams are increasingly common and the issues raised in this article will be relevant in discussions about the design of data collection and analysis, if not the interview itself (e.g., if a bilingual researcher alone conducts the interviews). Our example illustrates a specific interview dynamic; however, the reflections are applicable to any interviews where multiple professionals are involved in cross-cultural research.

A modified discourse analysis on the reflective interviews provided an insight into the social positioning of the actors involved. Tsai and colleagues (2004) questioned the use of analytic approaches that rely heavily on semantics, exact word usage, or the structures of narratives (e.g., discourse analysis and narrative analysis) in cross-language research because translation may have changed the “structure” of the original language. These approaches would not have been effective on the interpreted interview data collected because the process of interpretation fragmented and changed the structure of the narratives. In fact, the transcripts of the interpreted and non-interpreted interviews look different, with the latter having longer stretches of uninterrupted speech from the interviewee. This article focused on the dynamics during data collection and not on how this impacted on analysis. More research is required on analytical approaches to interpreted data and combining data from different languages (e.g., interviews conducted in English and Punjabi).

This article used reflective discussions with the interviewer and interpreter/co-researcher, and this formalised the reflexive process. The interviewees' social positioning is extrapolated from comments in the original data collection. It would be beneficial to examine the interviewees' experiences of taking part in interpreted interviews, perhaps using focus groups to also elicit shared fears about participating in research among this underrepresented group (Hussein-Gambles et al., 2004) and elicit recommendations for recruitment. Non-verbal cues were not formally recorded during the interviews because it was decided that this may be off-putting to the participants; however, these data may have provided further insights about the social positioning that was occurring. The main caution about conducting a study with interpreters and multiple researchers is that the amount of extra time required for (de)briefings may be grossly underestimated, especially immediately following data collection in the field. The main recommendations are 1) to discuss shared and disputed meanings emerging in the data and to test these in subsequent interviews and 2) to explicitly reflect on how positionality may be affecting data collection and consider employing strategies to alter this during the course of the research (e.g., changing the research setting if appropriate).

In conclusion, the challenges and benefits of the interviewer–interpreter/co-researcher–interviewee triad have been illustrated. Different streams of knowledge (experiential, academic, cultural, and clinical) were drawn upon to facilitate effective communication in the interviews. Power dynamics resulted from the shifting dominance of different knowledges, but in this study power relations were negotiated to benefit rapport and data collection.