‘I’m sure at some point we’ll be switching’: planning and enacting an interview language policy with multilingual participants

ABSTRACT

When conducting interviews with multilinguals, researchers make (often invisible) decisions about the interview language(s). Whilst the research design may require a particular approach in some cases, linguists generally recommend giving participants a choice or interviewing them in their first language. There are ethical and methodological reasons for considering this, such as the implications for self-expression – including emotion communication – and therefore data generation and analysis. This paper offers methodological reflections about planning and conducting a research interview in which the researcher and participant knowingly share two languages, shining a light on the process of building linguistic flexibility into a study. The case study is an interview conducted in French and English, which explored a bilingual client's language use in psychotherapy. The paper gives practical insights into offering a choice of language(s) and planning for the possibility of a multilingual interview (i.e. code-switching). It considers how to mitigate language insecurities before illustrating how the interview language(s) may be negotiated in interaction. I argue for researchers to set clear interview language policies which foreground inclusivity, and show in the process that interviews can become multilingual exchanges, in which both interlocutors experience linguistic freedom.

RÉSUMÉ

Les entretiens avec des participants multilingues donnent lieu à des décisions (souvent invisibles) sur le choix de langue(s). Quoique certaines études dictent une approche particulière, en général les linguistes recommandent de laisser le choix aux participant.e.s ou de les interviewer dans leur première langue. Ceci pour des raisons éthiques et méthodologiques, telles que l’influence sur l’auto-expression – y compris la communication des émotions – et par conséquent sur la génération et l’analyse de données. Cet article propose une réflexion méthodologique sur la planification et le déroulement d’un entretien de recherche lorsque chercheur.se et participant.e partagent sciemment deux langues, illustrant comment mettre en pratique le principe de la flexibilité linguistique. L’étude de cas est un entretien qui se déroula en français et anglais, traitant des choix langagiers d’une cliente bilingue en psychothérapie. L’article offre un aperçu des démarches à faire pour proposer un choix de langue et se préparer à l’éventualité d’un entretien bilingue (avec alternance codique). Il examine comment limiter l’insécurité linguistique et illustre comment les langues de l’entretien sont constamment négociées. Je souligne l’importance d’établir une politique langagière qui met en avant l’inclusivité, tout en démontrant qu’un entretien peut devenir un échange multilingue, où les deux interlocuteur.rice.s jouissent d’une liberté linguistique.

KEYWORDS:

• Interview
• multilingualism
• language policy
• code-switching
• linguistic repertoire
• translanguaging

MOTS-CLÉS:

• Entretien
• plurilinguisme
• politique linguistique
• alternance codique
• répertoire linguistique
• translanguaging

Introduction

Many research guides have been devoted to interviewing, which is a versatile qualitative method of enquiry. One dimension which is less frequently discussed than others is that of the interview language(s), which requires careful consideration when researching with multilingual participants. Scholars in applied linguistics have advocated for offering a choice of language and enabling code-switching (CS) within interviews (Pavlenko 2007; Mann 2011). Moreover, Holmes and colleagues (2013, 2016, 2022) provide an overarching theoretical framework for researching multilingually. Nevertheless, the concept of an interview language policy appears to be absent, and the body of examples which researchers can draw on to inform decisions is still relatively limited. Indeed, the language(s) of interviews is not typically stated or analysed within research reports (Cormier 2018; Cortazzi, Pilcher, and Jin 2011) and practical advice tends to focus on post-interview aspects, such as translation (e.g. Halai 2007). It is not well understood, for example, whether interviews between multilinguals are typically conducted monolingually or multilingually. Reflexive accounts regarding the impact of language practices, including code-switching, are still an exception rather than the norm.

This article seeks to contribute to the field by providing an account of how I made decisions when preparing for and conducting an interview with a fellow multilingual. The exchange took place within the context of researching multilingual clients’ language practices in psychotherapy sessions and their implications for self-expression (Rolland, Dewaele, and Costa 2017; 2021), using mixed methods. Research has shown that multilinguals’ expression of emotions, autobiographical memory recall and identity perception can vary according to the language(s) used (Dewaele 2010; Pavlenko 2005), suggesting that the language(s) of a research interview may influence what is communicated, and how (Rolland, Dewaele, and Costa 2020). Thus, I aimed to allow interviewees as much linguistic freedom as possible and to attend to language choices, for both ethical and methodological reasons. This was articulated in my university ethics application, forming the kernel of what I will refer to as my ‘interview language policy’. I therefore considered how my own language repertoire (Busch 2017) – which includes fluent English and French – intersected with potential interviewees’ languages, ultimately deciding to include a French participant. Juliette (an alias) had reported mainly speaking French in her ongoing UK-based psychotherapy, with occasional switches to English, and therefore interviewing her in French if she wished might have several benefits.

In practice, as I prepared a French version of the guide used in the preceding (English) interviews, I was uncertain how to proceed in the event that Juliette code-switched: would I switch between two versions? Moreover, was it (in)appropriate for me to initiate a code-switch? These considerations presented a tension between giving the participant full agency – rather than assuming the power to select a language as interviewer – and the more natural approach among multilinguals of switching as each saw fit (Gardner-Chloros 2009). I also realised that I felt anxious about conducting an interview in French, which was no longer my academic language and somewhat attrited after nearly two decades in the UK (Schmid 2018).

This account, despite being limited to a single interview shaped by our specific linguistic repertoires (Busch 2017), offers a candid window into the methodological process of attending to questions of language when interviewing multilinguals and an emerging framework for other researchers. It is hoped that such descriptions will become more common and be given more space within reflection and guidance on methods, as well as within research reports themselves.

After surveying relevant literature and describing the study context, I examine the multilingual features of the interview, analysing excerpts from the transcript and my reflexive research journal (Roulston 2010a). Taking a constructionist perspective (see Roulston 2010a, 2010b), I approach interviewing as ‘social practice’ (Talmy 2011, 25), i.e. with an interest in how the interaction developed, including ‘the interviewer's work’ (Roulston 2010a, 125). I show how offering a choice of interview language(s) unfolded, using tools from Conversation Analysis. The language practices within the interview are then presented, with code-switches analysed through a psycholinguistic lens (Dewaele 2010; Grosjean 2008; Pavlenko 2005). Two aspects which I found linguistically challenging follow. Finally, I discuss the findings, identifying the role played by my interview language policy and lessons which may be useful to other researchers.

Literature review

Language policies in interviews with multilingual participants

The language variable is often neglected within literature on research interview methodology, even in applied linguistics. Guidance regarding multilingual subjects typically deals with populations with low foreign language proficiency, requiring bilingual interviewers (Roberts 2006) or interpreters (Andrews 2013; Temple and Edwards 2002). Yet researchers necessarily make choices about which language(s) to interview fluent multilinguals in. As Pavlenko (2007) argues, ‘a single language […] signals an assumption that stories and interviews are simply descriptions of facts’ (172). These methodological decisions and the rationale behind them are seldom made explicit in research reports (Cortazzi, Pilcher, and Jin 2011; Mann 2016; Pavlenko 2007; Temple and Young 2004), owing to factors ranging from monolingual ideologies or linguistic and budgetary constraints, to the constraint of word counts or a lack of precedent.

Recommending a reflexive approach to researching multilingually, Holmes et al. (2016) proposed a three-stage framework: ‘realization, consideration, and informed and purposeful decision-making’ (90). Establishing an interview language policy, then, can be viewed as part of the consideration stage, which includes ‘deciding which language(s) to use when generating data’ (91). However, specific guidance and practices in applied linguistic interviews, where they are reported, vary. Reviewing doctoral research with Chinese-English bilingual participants, Cortazzi, Pilcher, and Jin (2011, 516–517) found that some interviewers ‘made the choice for the participants, others gave the choice of language to the participants’. Similarly, Catalano (2016) and Halai (2007) chose to conduct interviews in participants’ first language (L1), while Prior (2016, 15) recommends that ‘the choice to use (and even not use) a particular language or languages should be left to the interviewee’. Zhou, meanwhile, regarded participants’ L1 Mandarin as more reliable but upheld an ethical imperative to respect any participant's preference for English (Stelma, Fay, and Zhou 2013). Certain research questions understandably dictate a specific linguistic design, such as Koven’s (1998) study of bilinguals’ presentation in each of their languages. Ultimately, there is no single ‘“right way” to interview (when the interviewer and interviewee do not share the same L1)’ (Mann 2016, 218). However, Attia (cited in Mann 2016) recommends asking oneself: ‘With multilingual researchers and participants, why use one language over the other?’ (65).

In addition, when offering a choice, researchers sometimes make assumptions about participants’ preferences or take-up. L1 interviews are not universally welcomed (Cortazzi, Pilcher, and Jin 2011; Polo-Pérez and Holmes 2023; Schembri and Jahić Jašić 2022). In fact, Kalocsányiová and Shatnawi (2022) reported that a majority of participants chose English to discuss language learning and integration in Luxembourg, reflecting ‘a sense of pride and accomplishment’ (221). Moreover, languages which have been stigmatised may not be taken up, as illustrated by L1 Welsh participants deferring to the interviewer rather than choosing between Welsh and English (Madoc-Jones and Parry 2012). The language initially used by researchers also sets the tone (Nemouchi and Holmes 2022). Thus, far from being straightforward, language use can be a delicate negotiation affected by the social context and power relations.

Researchers’ focus often appears to be on establishing a single interview language, even though some communities may consider switching between two or more languages as the norm (Gardner-Chloros 2009). Mann (2011) proposed that ‘code-switching practices in interviews’ (15) should be attended to within the research methods literature. In his book, he advises that if ‘interviewer and interviewee(s) do not share the same L1 but they are both reasonably comfortable in communicating in each other's L1 […], they might codeswitch during the interview’ (Mann 2016, 218). Yet CS is not prominent within interview reports, making examples hard to find. Ali (2023) reported initiating interviews with first-generation immigrants to Barcelona in Spanish, but noted 61 instances of CS – mostly to English, but also to Catalan and Urdu – and mainly attributed them to difficulties using Spanish. CS also appears in reflexivity statements, such as Gkonou (cited in Mann 2016, 289) and Prior (2016, 15) who described the presence and type of participant CS within L1 and L2/3 interviews, respectively. Pietikäinen (2014), meanwhile, interviewed couples who use English as a lingua franca and analysed their switches. Mann (2011) further proposed invitations to CS as an interviewing technique: ‘is it worth encouraging interviewees to code-switch if they feel that an explanation can be fuller in L1?’ (15). Yet the language practices of multilingual interviewers seem absent from methodological guidance. In her account, Halai (2007) emphasises that both she and the teacher she interviewed switched from Urdu to English and back when using scientific vocabulary and borrowings, but also because mixing in English words was common practice.

A focus on bilingual interviewing, however, is found in Velásquez' (2010) retrospective analysis of bilingual Spanish-English interviews conducted with Latino high school students, examining ‘code-switching as an expression of [students’ and interviewers’] identities’ (5). Analysing eight interviews about students’ Canadian school experiences, involving eight students and three interviewers (not herself), she highlighted frequent participant CS. This included language alternation between turns, such as when students responded in English to a question asked in Spanish, indexing an ‘English-speaking school identity’ (50). Interviewers also switched between turns: Carolina repeated Iván's response in English for clarification, before reverting to Spanish to request further examples: ‘¿Algo más?’ (52). Again, Velásquez (2010) ascribed this to identity negotiation, however it is possible that interview management played a role since the main questions were asked in Spanish. Velásquez (2010) further describes how interviewers sometimes repeated a question or statement in both languages to emphasise that the interviewee ‘may answer in the language of his choice’ (78), demonstrating flexibility through their own language use and thus playing ‘a fundamental role in language choice’ (79).

Another practical aspect concerns translating the interview guide, or generating a ‘bilingual interview guide’ (Rolland, Dewaele, and Costa 2020, 285). Jagosh and Boudreau (2009) highlight an ecological approach to translation, which goes beyond seeking linguistic equivalency. However, technical questions remain regarding linguistically fluid interviews: how will the interviewer follow the guide if the participant switches language partway through? Should the document present each question bilingually, or simply juxtapose separate language versions? Such models and their advantages or disadvantages are absent from the literature.

Finally, the researcher's confidence in interviewing multilingually (Rolland, Dewaele, and Costa 2020; King 2023) should be considered. Indeed, unless their institution is officially bilingual, researchers typically receive training in a single academic language (Andrews, Fay, and White 2018), which is also the language in which they read academically and plan their work (Andrews et al. 2020; Holmes et al. 2013) – and this may be an L1 or a later learned language (LX). Even a L1 may become attrited while living abroad (Schmid 2011), such that employment interviews become challenging (Schmid 2018).

Implications for data collection

There are several reasons for paying attention to the language(s) which are offered and used in interviews. Multilingual participants, when asked to give autobiographical accounts of specific phenomena or events, as is typical in interviews (Mann 2016), have been shown in both sociolinguistic and psycholinguistic studies to express themselves differently, and even recall past events differently, according to the language used (Javier, Barroso, and Munoz 1993; Koven 1998; Marian and Neisser 2000). Language, then, may influence which events are remembered and retold, the level of detail and emotion, as well as self-presentation. Cortazzi, Pilcher, and Jin (2011) conducted consecutive interviews in different languages and concluded that L1 interviews were richer in data and more reliable.

Moreover, the interview is an interaction between interviewee and interviewer (Talmy 2011), in which any language which is spoken or signed (or omitted) contributes to the power relations within the dynamic (Rolland, Dewaele, and Costa 2020; Cormier 2018; Roberts 2006; Temple and Young 2004). The interview takes place within a social reality which may lend certain connotations to languages. In specific contexts, for example, English may index colonisation and oppression (Phipps 2019). In such contexts, offering the possibility to be interviewed in a local language may thus be particularly appreciated as a mark of ‘respect for the country's national language and culture’ (Tesseur 2022, 38). If the researcher is not fluent or confident in the participant's language, Phipps (2013, 2019) argues that embracing the vulnerability of language incompetence is part of decolonising research with multilingual participants.

Interview background

Multilingualism in psychotherapy

The study investigated multilingual clients’ language practices in psychotherapy and their implications. The research identified that inclusive language practices can help multilingual clients to regulate emotion and navigate identity issues (Rolland, Dewaele, and Costa 2017, 2021). Of 109 multilingual adults who completed a web survey, five participated in a follow-up interview.

The interview guide, which had been approved by an ethics committee at Birkbeck (University of London), evolved slightly with each semi-structured interview and according to participants’ questionnaire responses. The main topics were: language patterns in psychotherapy sessions, their impact on the therapy and the therapeutic relationship (see Rolland, Dewaele, and Costa 2020, for an abbreviated version). The final interview, described here, was carried out remotely using video conferencing software. I transcribed the audio recording and kept a reflexive journal (Roulston 2010a) during the process.

Interviewer linguistic background

I am a ‘simultaneous bilingual’ (Li Wei 2007, 6), inheriting English (L1a) from my mother and French (L1b) from my father. I grew up in France, speaking both languages at home and following a mixture of French and bilingual schooling. I moved to the United Kingdom to pursue university education and, having settled there, used French less and less in my daily life and inner speech. Fifteen years on, this attrition felt like an erosion of my French identity. Although my Belgian supervisor often initiated French conversation, I lacked confidence in projecting my adult and professional selves in a medium which I had not used academically (Grosjean 2008; King 2023) since my school years.

Participant linguistic background

Juliette was a ‘consecutive bilingual’ (Li Wei 2007, 6): born to a French family in France, she had studied English from the age of 11 and moved to the UK with her French husband twenty years earlier. She used both languages daily in her research (on bilingualism) and private life, including with her children and in her diary. Juliette had chosen a French-speaking therapist and described her ongoing therapy as mainly French, albeit with frequent switches to English.

Language policy and preparation

My decision to take a bilingual approach to collecting data (Dewaele et al. 2023; Holmes et al. 2016) influenced the language choices I made when communicating with Juliette, who volunteered to participate in my study at a doctoral training workshop. I followed up our discussion – which I recall as being in English due to our surroundings – with my standard email invitation (in English), signing off in French to index our common background (Negrón Goldbarg 2009). Juliette then completed the English-medium questionnaire, noting her interest in the topic and recommending several French-language works. Within the 14 emails we exchanged regarding the questionnaire and interview, Juliette wrote exclusively in English (aside from an accented letter in her name), while I consistently signed off with ‘Cordialement’ or ‘Merci!’.

My policy – albeit not communicated to Juliette until the interview – was to offer the option of interviewing in her L1 French, to facilitate emotional expression and mirror the discussions in her therapy. My intention was to translate the English-language guide used in the four previous interviews, in full. In practice, due to time constraints and my impression that Juliette was favouring English, I only translated the high-level questions. Thus, while the French version was a single page covering the seventeen overarching questions and selected sub-questions, the English version ran to eight pages with scripts and prompts for various situations and space for notes. Despite considering the possibility of code-switching in interview, it did not occur to me to merge them into a single, step-by-step bilingual guide.

The act of translating key questions made me aware of keywords and differences between the two languages (Jagosh and Boudreau 2009), such as the French distinction between ‘langue’ and ‘langage’, which is not made in English, or the term for code-switching (‘alternance codique’). I also had to choose a term of address, opting for the formal ‘vous’ rather than the informal ‘tu’.

My linguistic insecurity at the prospect of interviewing in French, which was exacerbated by Juliette coming across as a French intellectual, may also have influenced this lack of prioritisation. On discussing my feelings of imposter syndrome, my second supervisor suggested explaining that I was unfamiliar with interviewing in French when making the language invitation. Whilst I did not anticipate any problems with comprehension, I expected that I might sometimes struggle to find the right expression, particularly in a formal academic register.

Conducting the interview

Turning to the interview itself, I present our opening discussion concerning the interview language(s). Next, I describe code-switching patterns in the interview, by both participant and researcher. Finally, I reflect on two language-related challenges, namely aspects of turn-taking and navigating between two versions of the interview schedule.

Initial language selection

My exchange of greetings with Juliette went as follows, according to my research journal:

she greeted me with ‘Bonjour! Hello!’ and I responded ‘Hello!’ (hadn't prepped my interview intro in French) and we continued in English, checking camera set-up & sound

Thus, although Juliette opened with a bilingual greeting, my response points to English as my default academic language. Despite my overarching language policy, I had only considered using French within the formal interview once/if agreed with the participant, rather than in its informal margins.

Once I pressed record with Juliette's consent, the first topic in the guide was to discuss the interview language(s). I planned to open with the principle that ‘the language of the interview can be flexible’, before asking if Juliette had a preference. In order to emphasise that code-switching was acceptable and even expected, I had scripted: ‘We can start in one language and see if either of us feels the need to switch … ’.

Excerpt 1 below shows how the dialogue¹ unfolded on the day.

Excerpt 1. Opening the language discussion

I: [Let's] start by talking about language for the interview since we both speak French and English. Erm, so: (.) whatever you’re (.) comfortable in, do you have any preference? P: Not really. Er, I’m, I’m fine with both languages. I’m sure at some point we’ll be, you know, switching from French to English because you (understand it) as well so [we’ll see …  I:                  [Yeah, ok, yeah. P: … It will be interesting to see if it shows up in the conversation or not! I: Exactly, erm, so: let's just see how it goes, then.

My initial focus on language selection could be construed as inviting a monolingual interview in either language, however, Juliette quickly voiced her expectation that we would code-switch. The fact that she expressed no preference at all unsettled me, since I did not wish to impose my own choice. With hindsight, the starting point may not have mattered to her. Yet, I hesitated, using pauses and fillers (‘Yeah, ok, yeah’., ‘erm, so:’). From a methodological perspective, I preferred to use the participant's L1, but for my own comfort as interviewer it was tempting to seize this opportunity to continue in English. I delayed making an explicit decision further, as if code-switching would solve the problem (‘let's see how it goes’), and raised my lack of French-language interviewing experience instead (see Excerpt 2 below).

Excerpt 2. French-language competence

I: Erm, one, er, sort of health warning, I’ve not done an academic interview $in French before, so it’ll be interesting (.) to see (.) how that goes$. Erm, but if, so if it is feeling a bit, you know, ‘off’ for you, erm, then [just let me know. P: [If you want to conduct it in French, we can do it in French. It's fine!

Here my discomfort is apparent in my attempt at a humorous tone when raising that I might be a little ‘off-key’ interviewing her in French, and through the euphemistic use of the word ‘interesting’. Again, Juliette gave an unexpected response, seemingly thinking that I preferred to interview her in French and agreeing to this in order to be helpful.

In my final turn (see Excerpt 3 below), I again validated the idea of code-switching and tried to repair her misunderstanding regarding my preference. I was still unwilling to take the lead and invited Juliette once again to make a choice (‘it's up to you’). I was also anxious since Juliette did not appear to register my nervousness regarding L1b attrition. However, after detecting no further hint of an inclination at her end, I tentatively came to a decision, albeit hedging with ‘maybe’.

Excerpt 3. Selecting an initial language

I: E:rm, I think it would be interesting to, yeah, go in and out. I don't know whether I have a preference for starting. Maybe, erm. Yeah, maybe because the first question is really about, erm, your thoughts […], so, erm, I guess it's up to you what language you would say those thoughts in, I can ask you in English or French, erm, really, well, maybe I’ll ask it in French, e:rm …

At this point, I finally took the plunge and articulated the first formal interview question in French.

These three extracts illustrate how language selection for a research interview can be far from straightforward. The participant's stance may be variously interpreted as offering flexibility between multilinguals with a shared repertoire, deference towards me as the interviewer or helpfulness towards a student she perceives as needing interview experience in French. Juliette's lack of take-up on both occasions, however, suggests an unwillingness to align with one language or the other, which may be attributed to identifying as multilingual. My own hesitation reveals both unease at taking a unilateral decision (wishing to neutralise power relations with the interviewee) and linguistic insecurity. Eventually, I put these concerns aside and prioritised data quality, since I preferred to interview in the participant's L1, which was also the main language of her therapy.

Code-switching in interview

The interview transcript shows extensive code-switching between French² and English, illustrating that we were in a bilingual language mode (Grosjean 2008). I counted 52 occurrences within a body of approximately 13,500 words. The majority of switches (and utterances) were by the participant. These were mainly brief switches for a word or phrase, but there were also several switches to the base language (Grosjean 2008). Moreover, although fewer, there were some notable switches by the interviewer. The purpose and significance of the switches are discussed below with supporting examples.

Brief participant code-switching

Juliette's numerous short switches (n = 44) were exclusively from French to English. She typically switched to use English vocabulary, especially academic or technical (e.g. medical) terms. A recurring example was her reference to ‘mon PhD’, which would have been everyday language for her as she studied in the UK. She also used the word ‘code-switching’, unaware of the French term ‘alternance codique’ until I used it. Other words did not translate easily to French, including ‘screensaver’, ‘one-to-one’, and ‘closure’.

In addition, Juliette made five quotative switches, such as her account of being asked ‘Where is home?’. Another switch was for a longer phrase, as she described her empathy for an international student struggling to express herself in English: ‘ je pense que je peux le sentir parce que je sais qu’elle … (I think that I can feel it because I know that she …) she struggles to find the right word to, to express what she's trying to express! Et euh’. This switch may have been prompted by the fact that the relationship with the student in question was in English (see another example discussed below).

Finally, two of Juliette's switches can be classed as borrowings (Grosjean 2008, 119), taking ‘a lexical item from the guest language and [integrating] it phonologically and morphologically into the base language’, as she added French verb endings to English verbs: ‘prompté’ and ‘switcher’.

Brief interviewer code-switching

My own brief switches were also from French into English. Two were motivated by difficulties in translation when using psychotherapy terminology which I did not know in French (e.g. ‘to contain’). The third, and last, was to translate the linguistic term ‘alternance codique’, since Juliette did not understand it.

Switching the base language

As mentioned above, I asked the first interview question in French and the conversation followed in French, but for Juliette's frequent switches to use an English word or two. In the second half of the interview, a different pattern emerged: on several occasions, one of us switched and remained in the other language. This changed the base language and carved the interview into six sections alternating between French and English. First, Juliette switched to English mid-sentence while describing her emotional response to a student's work: ‘j’avais du mal à … [makes sound/gesture], I had problem to cope with the emotion she was carrying in what she was saying’. She immediately remarked: ‘I’m switching between English because that's the way I speak to her’. Thus, expressing herself in English matched the internal language of retrieval when thinking about this person (Altman, Schrauf, and Walters 2013). Accordingly, I asked my next question in English, querying an earlier statement by Juliette in French, about a breakthrough in communicating with her mother.

In her response, Juliette switched to French (see italics):

And I always felt, you know, this separation, abandonment and this feeling of er, ce, ce sentiment d’abandon, c’est, c’était quelque chose de, qui a toujours été très très très vif chez moi, tout le temps, toujours! Je peux partir, mais je ne supporte pas que les gens me quittent.

(this, this feeling of abandonment, it's, it was something which, which I've always felt very very very keenly, all the time, always! I can leave, but I can't bear for other people to leave me)

This is not attributable to translation difficulties, since initially Juliette repeats words she has already uttered in English (e.g. ‘abandonment’). Instead, the switch to her L1 appears to be driven by the intensity of her emotion (Dewaele and Costa 2013; Rolland, Dewaele, and Costa 2017): Juliette's difficult relationship with her (French) mother is at the heart of her psychotherapy. While transcribing the recording three days later, I noted: ‘CS feels like it is necessary to speak in French to express the feelings in a torrent of words. The emotion is evident in the speeding up of her speech and emphasis on certain words’. It also seems significant that in this case Juliette did not stop to notice or comment on the fact that she had switched to French. The interview continued seamlessly in French, since I followed Juliette's lead when asking my next, unscripted question.

A short while later, talking about supervisors who are critical of international students’ academic writing in LX English, Juliette suddenly switched to English (see italics):

des tuteurs monolingues qui ne sont pas capables de voir la richesse (.) que des gens qui écrivent avec un mode de pensée différent, lié à la façon dont ils s’expriment dans leur première langue, en fait, ça, ça, it, it kind of displace the way English works.

(monolingual tutors who are not capable of seeing the richness (.) that people who write with a different way of thinking, linked to the way in which they express themselves in their first language, in fact, it, it …)

Here the switch may be due to the topic (writing in English) and the language associated with the tutors whose approach Juliette is criticising, or to the need to use the word ‘displace’ rather than a French equivalent, because as she has stated elsewhere, English is her language for critical reflection.

The language of the last two sections of the interview was dictated by my own switches. First, I explicitly announced – in French – that I was reverting to French to ask a question quoting her earlier comment:

je reviens en français parce que je, vous avez dit quelque chose qui, qui me reste en tête. Vous avez dit plusieurs fois ‘entre’, ‘passer entre les langues’.

(I’m coming back to French because I, you said something which, which sticks in my mind. You said several times ‘between', ‘passing between languages’.)

This suggests that I felt the need to justify or explain the switch, which was planned and occurred at the start of my turn. Juliette acknowledged the quote with a quick ‘Oui’, allowing me to segue into the question (in French).

Finally, as she was speaking in French, Juliette's connection cut out. When she called back on Skype to resume our conversation, she opened with a French-sounding ‘Euh, Louise?’, to which I responded ‘Yes’. Juliette then accommodated to me by offering in English to send me her telephone number. However, as my post-interview note explains, I used the hiatus and my switch to English – which I recall as deliberate – to follow the ‘end prompts’ in my interview guide, concluding the late-running interview:

Her wi-fi disconnection served to end interview! Calling back provided an opportunity to start again or rather approach conclusion in English where I was more familiar with the phrases I wanted to use.

Here, the English language served both to give me the confidence to be decisive (see the turn-taking discussion below) and to run through my ‘end interview’ protocol, which I had neglected to translate into French. This contained various cues such as asking how the interview went for the participant and checking whether they would like anything removed from the record.

Summary

Overall, code-switching occurred frequently within the interview, at the initiative of both participant and researcher. The function of the brief switches varied from accessing vocabulary to referencing relationships or quoting speech acts which were embodied or pronounced in the other language. Switches to the base language, meanwhile, were linked to increased emotionality in Juliette's case and interview management in my own. Thus, the bilingual environment created in our initial discussion of interview languages, allowed each of us to use our linguistic resources freely (Pavlenko 2007). The switches were unscripted – although some were self-acknowledged and explained – and required flexibility from both parties. Each time one of us switched for a whole turn, the interlocutor cooperated so that it seamlessly became the new base language. This did present challenges in keeping track of interview topics, however (see below).

Regarding the relevance to my study, Juliette's language practices in the interview provided further insights into her account of how she communicated with her French-speaking therapist, triangulating the results of my thematic analysis. She enacted some of her stated reasons for CS in our exchange, a multilingual context with a different purpose and dynamic but with notable parallels (Mann 2016). For instance, although Juliette reported frequent switches to English in her therapy, she could not recall specific examples. The unmarked code-switching in the interview suggests that it is her speech style and therefore most switches are not memorable (Pietikäinen 2014). Meanwhile, her emotional switch to French when discussing her mother implies that French allows her to express deep feelings about the relationship more closely, so that using French might be important for communicating and processing these emotions in psychotherapy also.

Linguistic challenges

Feedback and turn-taking

In the interview, I frequently signalled my listening stance, agreement or encouragement by giving lexicalised feedback, such as ‘mm-hm’ (in both the French and English sections), ‘oui’/‘yeah’ or ‘d’accord’ (meaning ‘ok’, ‘all right’). Although these response tokens appear to be consistent with conversational French (Prévot, Bigi, and Bertrand 2013), in post-interview notes I described feeling off-key in French:

I noticed early on that I wasn't going ‘Mm’, ‘Mm-hm’ in response to her points as I do in English. I tried it a couple of times and it felt unnatural. […] How do I do this in conversations in French?

In particular, these French utterances seemed ineffective when I wished to take the floor for a follow-up question or to steer the conversation onto a new topic. The fact that the same words can either ‘signal topic shifts’ (Bailly et al. 2016, 2905) or on the contrary indicate ‘success and encourage continuation’ (Bailly et al. 2016) may have led to ambiguity for Juliette. Whilst I used additional cues such as tone and body language, it was challenging to communicate subtle shifts across the computer screen.

However, a closer look at the transcript to compare patterns in the French and English sections, suggests that I was only slightly more successful at turn-taking in English than in French. In the opening French section, I shifted from ‘Mm-hm’ to a more urgent-sounding ‘Mm’ when I wished to intervene, before interrupting after a more decisive ‘D’accord’. In the first English section, I used a similar pattern with a slightly quicker progression to interrupting. Even so, Juliette interrupted me back, leading me to reflect in my journal that I felt ‘caught in responding mode, as her audience’ even when we spoke in English.

In the moment, however, I felt a greater strain when trying to manage the interview in French:

How to take a turn while she was in full flow was even more difficult in French, I couldn't find the word or sound that would let me express/signal my intention to participate in the conversation and ask the questions.

Overall, it seems that Juliette wished to hold the floor and I had been moved to interrupt to progress the interview in both base language contexts. My initial impression could be attributed to linguistic insecurity, which had ideological roots in addition to my lack of practice, since I felt Juliette to be a more legitimate native speaker of French than me. It follows that my closing switch to English (see previous section), whilst motivated by the need to conclude within the agreed time-frame, could also be interpreted as a power move (Cormier 2018). In addition to reaching for my comfort zone as an English language interviewer with a set script, I may have been unconsciously indexing the language of my academic institution in order to invoke my role as the interviewer in charge of proceedings. Finally, and this is an uncomfortable position to consider, I may also have been attempting to ‘disarm’ Juliette by addressing the ‘battle’ for the floor (as I perceived it in that moment) in what was my dominant and first language.³ This potentially left Juliette at a disadvantage since she had commented that she could not always enter into the fray as quickly in English.

Navigating between two different language versions

Language alternation was challenging for me as an interviewer since, rather than a truly bilingual guide, I had two monolingual protocols before me, containing different levels of detail. This impacted on the visibility of my scripted questions and transitions (such as starting and ending the interview) as well as note-taking. Since we started discussing the study topics in French, I reached for this version, writing notes in the margins of the single page. Two days later, while transcribing, I commented on the brevity of these notes:

they started out with key words and arrows in the margins of the French schedule summary, certain words getting circled for importance & follow-up […]. I ran out of space […] and turned it over to make notes […] I filled the page and then moved on to the first page of the English-language schedule […].

I later reflected that despite the space limitations, I appeared to have followed up on those keywords. I also valued listing the key questions on one page (which I only did in French) so that I could check where I was topic-wise at a glance. Thus, I found advantages to using a shorter version of the schedule. When I found myself missing ‘my “repères” [points of reference] in terms of the flow of the interview’, I quickly copied across the approximate time limit for each sub-section (marked in the English version). A full French translation or bilingual version – providing each question in English and French in turn – may not have made a significant difference. Indeed, I did not complain in my reflexive notes about these gaps. However, as the code-switching analysis above shows, a translation of the closing protocol would certainly have been useful and other situations which I had anticipated in the English version could just as feasibly have arisen in French.

Summary

In the analysis above, I have shown that offering a choice of interview languages did not lead to a straightforward decision. There may be unspoken expectations at play, or the starting point may seem unimportant to the multilingual participant, if code-switching is anticipated. Indeed, the language of this interview shifted to a great extent as the conversation flowed and we both reached for the language which suited our narrating or questioning. As for challenges for the multilingual interviewer, preparing a high-level translation of the interview guide helped to prevent certain ‘technical’ difficulties (such as linguistic vocabulary) but did not entirely alleviate the insecurity caused by L1 attrition or provide a truly bilingual protocol to match the language alternation which occurred in practice.

Discussion

I have described my interview language policy, the rationale for setting it, and how it provided a framework for a multilingual interview with both expected and unexpected linguistic features. In developing this policy to offer more than one language to participants whose languages I shared, I was answering a call to be more reflective and transparent about linguistic decisions and practices when conducting applied linguistic interviews (Cortazzi, Pilcher, and Jin 2011; Cormier 2018; Mann 2011; Pavlenko 2007; Temple and Young 2004) and regarding the methodologies of researching multilingually (Holmes et al. 2013, 2016) more generally. Offering a choice, where possible, felt like an ethical commitment (Dewaele et al. 2023; Rolland, Dewaele, and Costa 2020; Schembri and Jahić Jašić 2022; Stelma, Fay, and Zhou 2013; Zhou 2010) given the impact of language on multilinguals’ self-expression (Dewaele 2010; Pavlenko 2005) and the importance of examining linguistic positionality within the research relationship (Cormier 2018). It also mirrored my investigation of language practices in psychotherapy (Rolland, Dewaele, and Costa 2017, 2021).

In the course of these post-interview analyses, I realised that rather than preparing for a ‘bilingual interview’, I had considered two different monolingual interview modes: English and French, with possible short incursions into the other language. This reflects the type of accounts which are available to researchers in the literature, even when the focus is on researching multilingually (e.g. Halai 2007; Mann 2016). It affected how I decided to translate and organise the interview guide, as discussed further below. Prior to the interview, I also made assumptions about my participant's likely preference: first her L1 (Cortazzi, Pilcher, and Jin 2011) and then her L2 once we had established email contact. I did not envisage a lack of preference, nor that Juliette might defer to me (Madoc-Jones and Parry 2012), as transpired when I offered to interview in the language of her choice (Prior 2016). Moreover, I found that conveying linguistic insecurity about ‘performing’ as an interviewer in French was not straightforward. Researchers might better prepare for language discussions by communicating what is available within their language policy at the recruitment stage (Schembri and Jahić Jašić 2022), within a multilingual study information sheet for example, allowing participants to consider their linguistic preferences in advance. This would then be discussed in person as part of the interview preliminaries, and could include an exploration of linguistic positionality (Cormier 2018), such as acknowledging any linguistic vulnerability or asymmetry. As Cormier states, ‘whether they are linguistic insiders or outsiders, all qualitative researchers should address linguistic power issues by firstly being aware of their linguistic positionality and then by being empathetic towards their participants’ (329).

As for the code-switching observed within the interview, Juliette appeared comfortable to use whichever language felt appropriate in the moment, drawing on her multilingual resources (Pavlenko 2007), whilst I happily accommodated to her base language switches. At the same time, I felt able to CS as well, which is consistent with several reports (Halai 2007; Mann 2016; Velásquez 2010). In particular, I switched base language in response to interview tasks such as following up a statement made in the other language (Velásquez 2010) or reverting to my English-only script to close the interview. This mutual linguistic freedom or flexibility was explicitly raised by Juliette and acknowledged by myself in our initial language conversation, after my invitation to choose an interview language. Having set these multilingual norms for the exchange, the nature of the data is arguably unsurprising. It is consistent with the multilingual data analysed by Velásquez (2010), which also included language alternation between turns, as well as the more targeted CS mentioned in other interview studies (Gkonou, cited in Mann 2016; Halai 2007). Many of our switches appeared to relate to the material under discussion belonging to a language domain which mapped to the other language (Grosjean 2008), which surely occurs in numerous interview situations. In particular, interviewing multilingually offered an opportunity for increased emotional expression around difficult experiences, likely resulting in richer interview data than if we had been constrained to one language.

Both the linguistic challenges which I noted, namely a lack of confidence in turn-taking in French, and working with two monolingual versions of the interview guide, could have been mitigated by conducting a pilot interview with a French speaker rather than solely in English. Indeed, piloting in each of the languages offered for interview (Jagosh and Boudreau 2009; King 2023), with attention to other aspects envisaged in the interview language policy such as code-switching, could help researchers to identify any linguistic, cultural and practical issues associated with cross-linguistic or multilingual interviewing within their specific context. In particular, practising how to handle difficult topics or vulnerability (of both participant and researcher) in a safe and sensitive way (Rolland, Dewaele, and Costa 2020), as well as how to maintain boundaries in the relevant languages could be important aspects of becoming at ease as a multilingual interviewer. Indeed, such skills require training and practice even in one's dominant language. Nevertheless, when researchers are in difficulty, they may still automatically reach for familiar scripts in their dominant language and are at risk of co-opting any linguistic asymmetry to their advantage. Any impact on the exchange (e.g. the participant feeling less comfortable) would then need to be explored.

To support this, I strongly recommend that researchers translate their interview guides in full – including any ice breakers and other situational scripts or prompts – into each of the languages offered, rather than omitting sections or outsourcing translation (Jagosh and Boudreau 2009). Novice interviewers especially might rely on formulations suggested in research manuals (Braun and Clarke 2013; Mann 2016), which are not necessarily easy to translate on the spot. Reading interview literature or textbooks in each language would also help to highlight any language-specific or culture-specific conventions (Holmes et al. 2013). Indeed, although interview types and styles vary, the interviewer(s) should use materials of a similar standard in each language and be aware of cross-linguistic differences. It is also important to consider how to organise versions in relation to each other, perhaps merging them into a multilingual guide (to be tested through piloting) in case the base language of an interview changes. The lack of guidance within the research methods literature suggests a need for further reporting and research on which formats researchers find helpful for interviewing and fluid note-taking in multilingual contexts.

Finally, I acknowledge that because of the constraints of my linguistic repertoire, I only offered a linguistic alternative to two of the five interview participants in this study (all of whom were fluent speakers of L2 English). I encouraged the others to CS if they were moved to, explaining that I would need to rely on them to translate or explain themselves post hoc. Accordingly, one participant made brief switches to Spanish, but other languages were absent from two further interviews. As a self-funded doctoral student, I had no formal access to multilingual collaborators (see Lorette 2023 and Selleck and Barakos 2023, for alternative ways of working). Where possible, and particularly when targeting multilingual participants such as migrants or refugees, I urge researchers to consider teamwork (Andrews 2013; Reilly et al. 2023; Roberts 2006; Selleck and Barakos 2023 ). Factoring multilingualism into research budgets and planning would help to extend linguistic hospitality to interviewees irrespective of their linguistic repertoires (Busch 2017). Expanding to a team with multiple languages and perspectives brings additional challenges, of course: these need to be weighed up against the possible effects of limiting the participant's language of expression (aside from what self-translation can offer).

Conclusion

In this article, I have argued for the importance of deciding on and articulating an interview language policy which informs the planning and conduct of interviews (and subsequent stages not discussed here), since researchers in most settings may come across multilingual participants. Through concrete examples, I have attempted to offer ‘a reflexive recognition of the situated accomplishment of the interview, the co-construction of data, and thus, a problematisation of the ideologies of interviewing’ multilingually (Talmy 2011, 33–34). I suggest that researchers need to consider how they will handle the potential for interviews to include more than one language, mindful that choices to include or exclude participant languages have a complex impact on data generation, far beyond questions of fluency. It may be particularly important to recognise this when planning sensitive research or working with vulnerable participants, such as refugees. Whilst this example focused on two shared languages, showing linguistic hospitality can also extend to other languages, since it is possible to empower participants to communicate something relevant in the language in which it comes to them, relying on post hoc translation (Rolland, Dewaele, and Costa 2020).

The policy should include which languages researchers (as an individual or team) are able to interview in, and the flexibility which both researcher and participant will have to move between languages. This will inform the creation and piloting of multilingual interview materials – from the invitation, information sheet and consent form to the interview guide itself – and prepare the ground for a frank discussion with the participant. I argue that due to the co-constructed nature of interviews, rather than offering a binary and fixed choice of interview languages at the start, a more authentic way of engaging with multilingual participants is to identify an initial interview language (which both interlocutors are comfortable with) and to create an environment in which switches to other languages – as triggered by the communicative context – are welcomed or encouraged. This is important for expressive purposes in fields beyond applied linguistics, and where multilingualism is the topic of study the approach also provides additional insights into language use. When linguistic repertoires differ, researchers prioritising the participant's language choices may find themselves feeling incompetent (Phipps 2013), especially in the academic domain, and vulnerable. However, my account shows that these feelings can be mitigated to a certain extent through preparation and by accepting and being honest with the participant about one's own linguistic insecurity or limitations, and embracing the joint production of meaning.

Acknowledgements

I would like to thank my dear friends and colleagues Hannah King and Pernelle Lorette, as well as both reviewers, for their detailed and constructive comments on earlier drafts. I am indebted to my supervisors, Jean-Marc Dewaele and Beverley Costa, for their unfailing guidance and support throughout my PhD. I am also grateful to the School of Arts at Birkbeck for continued access as an Honorary Research Fellow. Finally, without ‘Juliette’, this article simply would not be. Merci pour cet échange!

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 Transcription conventions are adapted from Conversation Analysis (Jefferson 2004): $word$ indicates speech produced in a laughing voice; (.) a brief pause; vowel: an elongated vowel sound; (word) my best guess at unclear speech; underlining indicates emphasis.

2 Each excerpt is presented as it was said, with code-switching highlighted in italics, and my translation of any French sections provided in brackets.

3 Although I have two L1s (as described in the background section), English was my dominant language and was not the participant's L1.