The interpreter as co-interviewer: the role of the interpreter during interviews in cross-language health research

Abstract

Objective

Interviews with ethnic minority patients provide a rich source of data to understand their perspectives of disease and its management. Language barriers are, however, often a problem so interpreters need to be used. We explored the impact of the interpreter on cross-language interviews between researchers and respondents.

Methods

Secondary analysis of four interviews between researchers and patients involving professional interpreters.

Results

Interpreters were actively involved and influenced the interview in several ways: they assumed the interviewer’s communicative role, edited information; initiated information-seeking, took over control of the interview, and took over the respondent’s role. While the interpreter supported the interviewer, they posed risks to the quality of the interview.

Conclusion

Researchers need to be aware of the influence of interpreters. Researchers should instruct interpreters carefully about their roles though they may benefit from interpreters’ strategies to support them.

Keywords

cross-language research interpreters qualitative study

Introduction

Qualitative approaches are used in health services research to explore patients perspectives on disease and disease management. Most of these approaches rely on interviewing but because of language barriers, interview respondents from ethnic minorities have often been excluded so their opinions and perceptions is often lacking. As Squires put it, ‘with words as data, language barriers between researchers and respondents become a potentially formidable methodological challenge’.¹ To yield credible and transferable data, researchers should address language barriers.^1,2 Using an interpreter is one method to assess the experiences of patients who do not speak the dominant language of a country.³ When respondents speak in their native language they perceive themselves as more confident, happy and intelligent.⁴

Existing studies on the use of interpreters concentrate on practical issues such as the choice of an interpreter, matching informant and interpreter, determining seating arrangements, and selecting a mode of interpretation.^1,5,6 There is, however, not much about how an interpreter influences the outcome of an interview and the affect on the quality of the data gathered.⁷ Many cross-language studies tend to render the interpreter as an invisible ‘neutral’ part of the research process, overlooking the interpreter’s role.¹ But as Hsieh showed, interpreters consistently influence the process and content of the practitioner–patient interaction, despite the expectations of interpreters as neutral translators of what was said.^8,9 During the consultation interpreters had different strategies to do so, such as assuming the practitioner’s communicative roles, editing information given by the patient, initiating information-seeking, participating in diagnostic tasks and volunteering medical information to patients. It is plausible that similar mechanisms play a role in interviews between researchers and respondents and that interpreters act as co-constructors of the dialogue.

To address the role of interpreters in cross-language research, we explored how interpreters interfere with general roles of interviewers, such as the way interviewers collect data by asking questions, are in control of the interview and try to build rapport with respondents.

Methods

Design

We used a case study as defined by Hammersley as research investigating a small number of natural occurring cases, as opposed to an experiment or survey.¹⁰ We used theory to underpin the conduct of our case study.¹¹ We felt that the case study was the appropriate approach because we did a secondary analysis of existing interview data and sampling was only possible within the confines of the primary data. This meant that we could only study those cases that were ‘naturally occuring’ rather than strive towards theoretical saturation.

Data collection

We used a secondary analysis¹² of interview data gathered in three different studies about ethnic minority patients’ experiences with health care in in the Netherlands between 2009 and 2014.^13–15 All studies explored patient’s explanatory models about disease and treatment, respectively about child burns and its treatment, about patient safety in hospitals and about colorectal cancer screening. These studies interviewed 21, 22 and 30, respectively participants with an ethnic minority background, living in the Netherlands.

We used secondary analysis two reasons. First, after the primary analysis we were left with questions about the reliability of data from the interviews that were interpreted: questions that our research questions could not answer. Second, we believed that a secondary analysis was the only way to answer our research questions. While we were interested in how interpreters interfere with the interviewers, we needed a real situation in which this was naturally occuring.

In these studies most interviews across language barriers were done without an interpreter but four interviews needed the help of an interpreter. One interview was with parents of Turkish descent whose 2-year-old daughter was treated for her burns (interview 1). The burn centre nurse had just finished her consultation with the parents, using an interpreter by telephone. When the consultation was over, the interpreter was asked to stay on the line and interpret for the interviewer as well. The interpreter was rapidly explained the purpose of the interview. In the second and third interviews two middle-aged Chinese women were interviewed separately about negative health care events in in-hospital care (interview 2 and 3). The interviews were interpreted by a professional interpreter who was present. This interpreter was rapidly explained the purpose of the interview. In the fourth interview, an elderly Moroccan woman was interviewed. At the beginning of the interview the daughter of the woman also joined – with consent from her mother – to support her. When during the interview it became clear that the mother spoke hardly any Dutch, the researcher allowed the daughter to interprete when necessary. The interpreter was rapidly briefed about the purpose of the interview but not about the method of qualitative interviewing.

All studies were separately approved by the medical ethical committee of the Academic Medical Centre/University of Amsterdam. We took every precaution to protect the privacy of research subjects and the confidentiality of their personal information as laid down in the Declaration of Helsinki.¹⁶ We felt that a repeated consent requirement might invoke perceptions of mistrust.¹⁷

Data analysis

The interviews were transcribed, and for the primary analysis the Turkish interview was translated from Turkish in Dutch by a Turkish-Dutch research assistant. The interviews in Mandarin and the interview in Moroccan-Arabic could not be translated due to financial restrictions. We used Hsieh’s theory that interpreters act as co-constructors of the dialogue between doctors and patients.⁸ On the basis of the theory of Hsieh, we predicted that the interpreters would have several strategies to act as co-interviewers, interfering with the interview by editing information for example.

We started the analysis with interview 1 because the interviewer (JS) felt that the interview was ‘out of her control’ but could not pinpoint why, a sign that the interpreter may have influenced the interview. Interview 1 (the interview with the Turkish parents) was read carefully and the role of the interpreter was analysed in terms of the possible strategies (i.e. did the interpreter assume the interviewer’s roles? Did the interpreter edit information? Did the interpreter initiate information-seeking behaviours? Did the interpreter participate in other tasks than merely interpretation? Did the interpreter volunteer information?) All fragments of the first interview were initially analysed in terms of these possible strategies. Table shells (outlines of a table with rows and columns of a data array) were used for the analysis,¹⁸ with in the columns the five strategies of Hsieh, and in the rows, fragments illustrative of a particular strategy. Once the fragments of interview 1 were categorized, we analysed the fragments into more detail looking to what was specifically said by the interviewer or respondent and what was the role of the interpreter in changing or adapting this information (a within analysis). Once this framework of five strategies was ready we used the framework to analyse the other three interviews. On basis of this framework we compared the four interviews and using constant comparison we made systematic comparisons across the different strategies in order to refine them (a cross-site analysis).

Results

We found five strategies that interpreters used during the interview: assuming the interviewer’s communicative role, editorializing information, initiating information-seeking, taking control of the interview, and taking over the role of the respondent.

Assuming the interviewer’s communicative role

In this strategy, the interpreter assumes the interviewer’s communicative role is restricted to posing questions. Thus, information from the interviewer which does not fit this role is not interpreted. All interpreters used this strategy. For example, in the following excerpt the interviewer tried to make a friendly remark to the respondent, in order to build rapport, but this was ignored by the interpreter:

Interpreter [Dutch, translating to interviewer]: Madam has seen four different specialists and she still is not completely cured. So madam think it is quite annoying. She just has to wait for a new operation and can’t do anything.

Interviewer [Dutch]: Yes.

Interpreter [Dutch]: Physically and emotionally madam thinks it is really difficult.

Interviewer [Dutch]: Yeah the insecurity

Interpreter [Dutch]: Yes

The effort of the interviewer to show her concern (‘yeah the insecurity’) was not translated by the interpreter but was seen by the interpreter as a remark to herself which she then answers with ‘yes’: the rapport-building role was ignored.

Editing information

A second strategy was when the interpreter edited information either given by the respondent or by the interviewer. This strategy was used by all interpreters. In the following excerpt the interpreter changed the words of the interviewer as well as the respondent into a more coherent story:

Interviewer [Dutch]: Did you look on the internet for more information?

Interpreter [translating to Turkish]: She asks whether you looked in different ways to find information about burns, on the internet or other ways?

[…]

Respondent [Turkish]: Yes we also looked for information ourselves. And in the hospital parents with children with burns told us that no scars will remain.

Interpreter [Dutch]: Yes we of course informed in different places and by talks with others.

The interpreter changed the closed question of the interviewer about information seeking on the internet to a more general question about information seeking. The interpreter also added ‘about burns’. The interpreter actually improved the initial question of the interviewer by making it more specific. The interpreter also edited information given by the respondent. Cues given by the respondent that may have offered hints to the interviewer, such as ‘in the hospital parents with children told us’, ‘no scars will remain’ were not translated but edited in a general answer. It thus became hard for the interviewer to ask more detailed questions, for example to explore the possibly wrong idea that the child will not have remaining scars because other parents had said so.

Initiating information-seeking

A third strategy which all interpreters used was when they took the initiative to gather information but did so without informing the researcher. For example:

Interviewer [Dutch, to respondent]: Do you find it difficult that your daughter will have permanent burn scars on her arm?

Interpreter [translating to Turkish]: Is it difficult to accept that your daughter will have permanent burn scars on her arm?

Respondent [Turkish]: She first had a lot of scars on her arms and face. So if she will have only a few scars I don’t mind. She really had a lot of scars.

Interpreter [Turkish]: So if a few scars remain on her arm you don’t mind so much?

Respondent [Turkish]: No. She only has a few scars now. So we just wait and see.

Interpreter [Dutch to interviewer]: He says that first the girl was really bad. Particularly in her face, but that will heal well. And about the arm they have said that small scars will last, but he cannot really tell now. Time will tell he says.

The interpreter took upon himself the initiative to gather new information when he was not satisfied with the intitial answer of the respondent. However, when he translated the respondent’s answer at the end, he ignored contradicting information given by the respondent: are the scars on the child’s arm a problem or not? The translation suggests that they have hope of total recovery, something the respondent had hinted at already (in the previous extract), but ignores the respondent’s original answer, in which he seems to have a different opinion (‘if she will have only a few scars I don’t mind’). This makes it also harder for the interviewer to explore parent’s perspectives further.

Participating in interviewer’s control of the interview

A fourth strategy, used by all interpreters, was when interpreters took over temporarily the control of the interview. For example, in the following the interpreter hinted that the interview was over:

Interpreter [translating the respondent’s turn in Mandarin]: The children did not like to interpret, it was an extra difficulty for the whole family.

Interviewer [Dutch]: Yes of course, for the whole family.

Interpreter [Dutch to interviewer]: Are we finished?

Interviewer [Dutch]: Yes, if there is nothing for you to add. If we have discussed it all. Yes?

The closing of the interview by the interpreter made it difficult for the respondent to add extra information or for the interviewer to ask an additional question. This is not a problem when respondents agree that everything has been said, but some respondents offer important information at the closing of the interview or interviewers keep the most sensitive questions for the last.

Taking over the role of the respondent

A final strategy was when the interpreter took over the role of the respondent. We found this strategy only in the interview which was interpreted by the informal interpreter (the daughter of the respondent):

Interviewer [Dutch]: What would you do when you would find a letter from the colorectal cancer screening organization on your door mat?

Respondent [Dutch]: I understand that it is about cancer, I understand that. […] But I think it is an advertisement

Interpreter [Dutch]: But I think she will be scared.

Respondent [Dutch]: No I know it is an advertisement

Interpreter [Dutch]: Yes she is scared.

In this excerpt, we see that the initial question was answered by the respondent but then corrected by the interpreter signaling she was not satisfied with the initital answer. Other excerpts from this interview showed a similar pattern in which the interpreter repeatedly offered additional information to the extent that at times the interviewer was interviewing the interpreter rather than the respondent herself.

Discussion

Main findings

We found five strategies of professional interpreters’ interference with the interview between interviewer and respondent: assuming the interviewer’s communicative role; editing information; initiating information-seeking; participating in interviewer’s control of the interview, and taking over the role of the respondent.

Most strategies were similar to those identified in the study by Hsieh, except for one.⁸ Hsieh found that interpreters of the dialogue between practitioner and patient had a fifth strategy where they volunteer medically related information to the patients, for example such as giving medical advice or taking away worries of the patient. We found a different strategy of an informal interpreter, who, because she knew the respondent so well, tended to take over her role.

We found that all interpreters acted as co-interviewers and were actively involved in the interview, rather than merely translating what was said.^8,9 Interpreters interfered in all of the interviewers’ roles. Often there were disadvantages, such as when information about the patient’s perspective was transformed into a coherent story by the interpreter and cues for the interviewer were not translated. Prompting therefore was difficult.⁴ There was also loss of small talk: talk that is designed to foster relationships rather than to establish facts.¹⁹ The interpreters seemed to have a ‘limited’ idea of the interviewer’s role (merely as someone who asks questions) and neglected other important aspects such as rapport building and being in charge of the course of the interview.²⁰ Although many strategies may be attributed to the interpreter’s effort to conserve the interviewer’s time,⁸ they also posed risks to the quality of the interview.

However, the role of the interpreter should not be seen as merely disruptive to the interview. We agree with Raymond that interpreters actively try to improve the quality of the interview.⁹ For example, by actively screening the information the interpreter seemed to help the interviewer with the task. This can be to the benefit of the interview when the interpreter reframes questions and answers more clearly.

Limitations

This study has several limitations. The study was based on a secondary analysis of data in which we were not able to sample until theoretical saturation.¹² We based our results on four interviews in the Netherlands and generalizations should be made with caution. Still, secondary analysis is thought to be of benefit in situations where sensitive topics are being discussed²¹ and our study raised several interesting sensitive issues that would not have come forward if, for example we would have only interviewed interpreters. We believe that by this case study we have opened up the door to new research that can further strengthen the methodology of doing cross-language research. We therefore would like to suggest this study be repeated with a larger sample of interviews. In addition, observational methods during the interpreted interviews and interviews with researchers, interpreters and respondents about their aims during the interview might be useful to explore how interpreters are actively involved in the interview.

Implications for practice

While we would like to warn that generalizations to other contexts should be done carefully, we have learned from our experiences and in our future research we will take into account the following recommendations:^7,20 familiarize interpreters with the different roles they have during the interview and explain their natural tendencies to assume the interviewer’s role, to edit information, to initiate information-seeking and to participate in controlling the interview. Furthermore, instruct interpreters comprehensively about the aims of the research and their role, and inform them about the interview schedule and key terms that may be used during the interviews, so that their assumptions of the interviewer’s role (such as rapport-building) and intentions are concordant with the interviewer’s expectations.²² During the interview, pause the interview when problems arise (for example when the interpreter undesirably takes over the role of the interviewer).²² When the interview is finished the interpreter can debrief the researcher about the interpretation of what respondents said and reflect on their roles in the interview.

We would recommend informal interpreters in research interviews should not be used, even if their presence can bring a safe and pleasant atmosphere for respondents. Informal interpreters may volunteer information that the respondent would rather keep secret or, in case of sensitive and delicate issues, decide to omit or change: a strategy that we did not find in our analysis but that is found as a strategy of informal interpreters in the patient–doctor consultation.²³ Statements about the use of professional interpreters should be included in medical ethics guidelines for conducting qualitative research.

While we did not find differences between interpretation by telephone and face-to-face, we would recommend to use a face-to-face interpreter if possible because this allows the interviewer to use non-verbal cues to instruct the interpreter.

We recommend to have the interviews rapidly translated and transcribed in order to check the role of the interpreter. In instances of inconsistencies between speakers, interpreters may be instructed differently or another interpreter may be hired.

Given the possible drawbacks of using interpreters in interviews, we think that in case of mild language barriers an interview can be conducted without the aid of a professional interpreter.²² This type of interview demands experienced interviewers, additional time, rephrasing of questions if the initial question is not understood and checks for mutual comprehension but may in the end deliver more credible data.²² An additional benefit is that the flow of the interview is not disturbed by the frequent breaks for translation. Otherwise, bilingual researchers may be used. The advantage is clearly that researchers keep control of the interview and are not dependent on an interpreter to translate their questions and respondents’ answers. This adds to the reliability of the results. Furthermore, bilingual researchers can provide cultural information about the communities of interest²⁴ and may be of large benefit in recruitment of participants. However, bilingual researchers may also lead to socially desirable answers from respondents who may fear disapproval after disclosing sensitive information that does not follow a group’s cultural norms. Furthermore, assessment of proficiency in both languages can be difficult and hiring many different bilingual researchers may be logistically impossible or too expensive.

A mixed-method design may be useful, where researchers combine different types of interviews (with and without the aid of a professional interpreter) to validate findings. For this, Lee et al.²⁴ developed a table of the disadvantages and advantages of professional interpreters and bilingual researchers. For example, specific issues to take into account when choosing between different types of interviews are financial issues, sensitivity of the topics discussed, recruitment of respondents and diversity of languages to be included. Also, the choice of type of interpreter may be dependent on the research design.²⁵ For example, for phenomenological studies bilingual researchers may be the best solution, as such studies ideally are conducted in the language of the participant and using interpreters is too disruptive. For narrative analysis, interpreters may be useful because experiences of participants can still adequately be captured.

Finally, we want to stress that reflexivity is essential in case of cross-language research.^2,19 The researcher should continuously reflect on how the next interview can be improved, which method would be the best possible to address the language barriers in a specific context and do the very best of what is possible, given time or financial barriers.

Footnotes

Acknowledgements

We would like to thank Kirsten Langeveld for her critical comments on an earlier version of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Squires

. Language barriers and qualitative nursing research: methodological considerations. Int Nurs Rev 2008; 55: 265–273.

Ogilvie

Burgess-Pinto

Caufield

. Challenges and approaches to newcomer health research. J Transcult Nurs 2008; 19: 64–73.

Bischoff

Perneger

Bovier

. Improving communication between physicians and patients who speak a foreign language. Br J Gen Pract 2003; 53: 541–546.

Murray

Wynne

. Researching community, work and family with an interpreter. Community Work Fam 2001; 4: 151–171.

Wallin

A-M

Ahlstrom

. Cross-cultural interview studies using interpreters: a systematic literature review. J Adv Nurs 2006; 55: 723–735.

Ingvarsdotter

Johndotter

Östman

. Lost in interpretation. The use of interpreters in research on mental ill health. Int J Soc Psychiatry 2012; 58: 34–40.

Plumridge

Redwood

Greenfield

. Involving interpreters in research studies. J Health Serv Res Policy 2012; 17: 190–192.

Hsieh

. Interpreters as co-diagnosticians. Overlapping roles and services between providers and interpreters. Soc Sci Med 2007; 64: 924–937.

Raymond

. Conveying information in the interpreter-mediated medical visit: the case of epistemic brokering. Patient Educ Couns 2014; 97: 38–46.

10.

Hammersley

. What’s wrong with ethnography?, London: Routledge, 1992.

11.

Yin

. Case study research. Design and methods, 3rd ed. Thousand Oaks/London/New Delhi: Sage, 2003.

12.

Hinds

Vogel

Clarke-Steffen

. The possibilities and pitfalls of doing a secondary analysis of a qualitative dataset. Qual Health Res 1997; 7: 408–424.

13.

Suurmond

Uiters

de Bruijne

. Negative health care experiences of immigrant patients: a qualitative study. BMC Health Serv Res 2011; 11: 10–10.

14.

Suurmond J, Bakker A, Dokter J, et al. Psychological problems in ethnic minority parents of preschool children with burns: a mixed method study. In 2nd review.

15.

Woudstra A, Dekker E, Essink-Bot ML, et al. Knowledge, attitudes and beliefs regarding colorectal cancer screening among ethnic minority groups in the Netherlands – a qualitative study. Health Expect. Epub ahead of print 17 November 2015.

16.

World Medical Association. WMA Declaration of Helsinki – ethical principles for medical research involving human subjects. http://www.wma.net/en/30publications/10policies/b3/ (accessed 1 May 2015).

17.

Yardley

Watts

Pearson

. Ethical issues in the reuse of qualitative data: perspectives from the literature, practice, and participants. Qual Health Res 2014; 24: 102–113.

18.

Miles

Huberman

. Qualitative data analysis: an expanded source book, Thousand Oaks/London/New Delhi: Sage, 1994.

19.

Aranguri

Davidson

Ramirez

. Patterns of communication through interpreters: a detailed sociolinguistic analysis. J Gen Intern Med 2006; 21: 623–629.

20.

Shimpuku

Norr

. Working with interpreters in cross-cultural qualitative research in the context of a developing country: systematic literature review. J Adv Nurs 2012; 68: 1692–1706.

21.

Fielding

. Getting the most from archived qualitative data: epistemological, practical and professional obstacles. Int J Soc Res Methodol 2004; 7: 97–104.

22.

Kosny

MacEachen

Lifshen

. Another person in the room: using interpreters during interviews with immigrant workers. Qual Health Res 2014; 24: 837–845.

23.

Seale

Rivas

Al-Sarraj

. Moral mediation in interpreted health care consultations. Soc Sci Med 2013; 98: 141–148.

24.

Lee

Sulaiman-Hill

Thompson

. Overcoming language barriers in community-based research with refugee and migrant populations: options for using bilingual workers. BMC Int Health Hum Rights 2014; 14: 11–11.

25.

Squires

. Methodological challenges in cross-language qualitative research: a research review. Int J Nurs Stud 2009; 46: 277–287.