Scholarly article on topic 'Methodological challenges of cross-language qualitative research with South Asian communities living in the UK'

Methodological challenges of cross-language qualitative research with South Asian communities living in the UK Academic research paper on "Psychology"

0
0
Share paper
Academic journal
family med commun hlth
OECD Field of science
Keywords
{""}

Academic research paper on topic "Methodological challenges of cross-language qualitative research with South Asian communities living in the UK"

Family Medicine and Community Health ORIGINAL RESEARCH

Methodological challenges of cross-language qualitative research with South Asian communities living in the UK

Manbinder S. Sidhu1, Farina Kokab1, Kate Jolly1, Tom Marshall1, Nicola K. Gale2, Paramjit Gill1

Abstract

Objective: We investigate (1) the influence of ethnic, gender, and age concordance with interviewers and (2) how expression of qualitative data varies between interviews delivered in English and community languages (Punjabi/Urdu) with monolingual and bilingual participants across three generations of the Indian Sikh and Pakistani Muslim communities living in the UK.

Methods: We analyzed and interpreted semi-structured interview transcripts that were designed to collect data about lifestyles, disease management, community practices/beliefs, and social networks. First, qualitative content analysis was applied to transcripts. Second, a framework was applied as a guide to identify cross-language illustrations where responses varied in length, expression and depth.

Results: Participant responses differed by language and topic. First-generation migrants when discussing religion, culture, or family practice were far likelier to use group or community narratives and give a longer response, indicating familiarity with or importance of such issues. Ethnic and gender concordance generated greater rapport between researchers and participants centered on community values and practices. Further, open-ended questions that were less direct were better suited for first-generation migrants.

Conclusion: Community-based researchers need more time to complete interviews in second languages, need to acknowledge that narratives can be contextualized in both personal and community views, and reframe questions that may lead to greater expression. Furthermore, we detail a number of recommendations with regard to validating the translation of interviews from community languages to English as well as measures for testing language proficiency.

1. Institute of Applied Health Research, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK

2. Health Services Management Centre, School of Social Policy, University of Birmingham, Edg-baston, Birmingham, B15 2TT, UK

CORRESPONDING AUTHOR: Manbinder S. Sidhu, BA Hons, PhD

Institute of Applied Health Research, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK

E-mail: m.s.sidhu@bham.ac.uk

Received 1 June 2015; Accepted 2 November 2015

Keywords: Ethnicity; community; concordance; language; South Asian; qualitative; interviews

Introduction

Despite a greater propensity for living with chronic disease, people from minority-ethnic groups are underrepresented in applied health research [1-3]. Increasing participation remains a methodological concern as many people from migrant communities living in developed countries are excluded because of language needs [4,

5]. Given increasing levels of global migration and the development of super-diverse cities [6, 7] across the world, there is a need for qualitative research that is linguistically and culturally representative of study populations [8]. With regard to cross-language research, greater attention has been placed on understanding the challenges of using interpreters

Family Medicine and Community Health 2016;4(2):16-28 www.fmch-journal.org DOI 10.15212/FMCH.2015.0146 © 2016 Family Medicine and Community Health

and translators [9-11]. However, there is a trend toward the deployment of community-based researchers. These are individuals embedded within the community of study with the necessary language skills, who have greater contextual knowledge and understanding of community beliefs/practices as well as social, personal, and economic barriers to healthy living. Community-based researchers are involved in the design, data collection, and analysis stages of research projects, sharing their expertise on language, beliefs, values, and practices which could benefit the community involved [12].

Cortazzi et al. [13] outline a framework of practical considerations for monolingual and bilingual researchers to examine when including participants from diverse linguistic backgrounds into research. These include recruiting interpreters/bilingual researchers and interviewing in either the researchers' or the participant's first or second language or participants using their first and second languages interchangeably. However, there have been some notable criticisms of the use of community-based researchers. First, researchers cannot assume that minority-ethnic participants want to be interviewed by members of their own community, with some having a preference to discuss culturally sensitive issues with those outside their community [14]. Second, languages are subject to dialectical and regional differences that can lead to certain words having multiple meanings depending on the cultural context. Finally, priority is placed on matching researchers and participants on the basis of visual and/or explicit characteristics (e.g., race, religion, ethnicity, and sex), in the belief that those who look the same speak the same [15]. Nevertheless, there are unpredictable ways in which race, religion, ethnicity, and sex interact to produce narratives [16] and how participants may choose to conceptualize them [17].

For minority-ethnic researchers working with their 'own' communities, researcher identities are continually shifting between insider and outsider. Kusow [18] argues that the relationship between researcher and participant cannot be reduced to binary categories of status. Researcher identity is not simply shaped by the nature of the interaction, but is shaped by the sociopolitical space within which the interaction occurs [18]. Ergun and Eredmir [15] strengthen this argument further, stating minority-ethnic researchers are not advocates of the communities they investigate. For example, communities

reflect complex ideas of representativeness and may not wish to disclose information that could taint their identity [19].

People of South Asian descent constitute one of the largest minority groups living in the UK [20], yet they differ by ethnic identity, religion, cultural practices, and patterns of migration. We use the definition of South Asian given by Liu et al. [21] - that is, a person with ancestry in countries of the Indian subcontinent, including India, Pakistan, Bangladesh, and Sri Lanka. For the purposes of this discussion we have avoided the use of generic terms such as 'black' and 'minority ethnic,' which conflate issues of race and ethno-religious origin. Such terms can be problematic as they relate to a range of cultural and personal identities which may not be acceptable by all in a single community [22].

In the UK, Muslims from Pakistan and Sikhs from northern India (Punjab) form two of the largest diasporas outside the subcontinent. Yet, both the Indian Sikh and Pakistani Muslim communities are underresearched in applied health research [23]. People of South Asian origin, and particularly men, have a higher risk of developing conditions such as diabetes mellitus and coronary heart disease [24]. Pakistani Muslims and Indian Sikhs are often described as migrant populations, and their composition is changing with a growing number of second-and third-generation descendants. These groups, according to Ballard [25, 26], are constantly affiliating and adapting to multiple identities within various environments composed of changing groups (e.g., home, work, and educational institutions).

Against this background of increased risk and underrepre-sentation we document the methodological challenges faced by two community-based researchers interviewing Pakistani Muslim and Indian Sikh communities across three generations. We argue that concordance by gender, ethnicity, and age along with language influences the nature of the response given by monolingual migrants and bilingual second- and third-generation descendants. In doing so, we contribute to the literature that highlights the importance of understanding cross-language research not simply as a product of concordance but as an embodied and pluralistic practice.

Methods

Study design

Our study design was secondary analysis of qualitative interview transcripts from two community-based research studies

that explored perceptions of risk and/or living with cardiovascular diseases. A summary of both studies is presented in Table 1. Both studies were designed to investigate health beliefs and practices in Indian Sikh and Pakistani Muslim diasporas who are at greater risk of developing cardiovascular disease-related illnesses. Within this remit was the opportunity to address methodological challenges in the context of community-based studies with minority groups. In health services research, significant importance has been placed on researchers having a shared culture and experience with participants to develop a depth of understanding [27]. We acknowledged that research is not conducted in a vacuum and familiarity with cultural norms and values could increase participation [28]. With this in mind, we ensured that researchers shared a common background with participants; subsequently this meant we were unable to accommodate other South Asian communities (e.g., the Bangladeshi community). Secondary analysis was content driven, focusing on depth and detail of narratives opposed to critiquing the structure of discourses. Nevertheless, both datasets recognized the importance of original meaning and contextualization of words in community languages which have no direct translation in English.

Access and recruitment: primary dataset 1

Posters about the Indian Sikh study were displayed in general practitioner practices and attempts were made to arrange informal information sessions in places of worship. However, because of difficulties in working with places of worship, we used a strategy of approaching gatekeepers with an established network of contacts. Written informed consent was obtained immediately before the interview commenced.

Access and recruitment: primary dataset 2

Research advocates were contacted within minority-ethnic business districts to promote the Pakistani Muslim study via posters, word of mouth, and disseminating information sheets. Third-sector organizations and social media were also used to disseminate study details. Those who agreed to participate in an interview were provided with additional information about the study. If they then agreed to participate, interviews were arranged and written informed consent was obtained immediately before the interview commenced.

Sampling: primary dataset 1

A purposive sampling method via gatekeepers [29] was used with the aim of selecting individuals or family units comprising first-generation migrants and second- and third-generation descendants. At least one member of the family had a cardiovascular-related condition. We included both men and women aged 18 years or above. We selected a snowballing sampling method, where initial contacts would facilitate the identification of other, potential respondents.

Sampling: primary dataset 2

As for dataset 1, purposive sampling was used in tandem with snowballing methods to seek potential respondents. We included both men and women aged 18 years or above with or without an illness across three generations. Details of our sample are presented in Table 1.

Data collection: primary dataset 1

Semi-structured interviews (audio-taped in both English and Punjabi) were conducted and translation and transcription were performed by M.S. Interviews with first-generation participants were conducted in Punjabi. Translation was based on achieving conceptual equivalence (i.e., importance was given to what was meant by the participant given the context of the conversation). Transcription (and hence translation) was conducted in English. Interview topics centered on lifestyle practices and the design and delivery of health promotion services for the management of long-term conditions. Interviews were conducted in participants' homes. Each audio recording and transcript was given a numerical identifier to ensure anonymity. Participants were informed that the data would remain confidential within the research team. Ethical approval was obtained from South Birmingham Research Ethics Committee, UK.

Data collection: primary dataset 2

Semi-structured interviews (audio-taped in both English and Urdu) were conducted by and translation and transcription were performed by F.K. Discussions centered on barriers to healthy living, social networks, and generational differences in beliefs, values, and practices with regard to disease prevention. Interviews were conducted in participants' homes or

Table 1. Description of studies and sample

Primary dataset 1- Indian Sikh study

Research aim

Methods and content covered

Researcher background

Description of participants

To understand how members of a faith-based community interpret the risk of developing and/ or living with a chronic disease (diabetes and cardiovascular disease), and what forms of self-care support they were accessing Interviews were used with the aim of generating 'stories' covering a range of areas (healthy living, self-care, cultural practices, and religion) with the use of a topic guide

M.S. is a qualitative researcher with a background in sociology working in applied health research. As a male, second-generation descendant of Sikh parents from the Punjab, India, he identifies with both a British and a Sikh identity Trigenerational families were recruited across the West Midlands with one person living with diabetes and/or cardiovascular disease. The sample comprised 17 participants, of which nine were bilingual (English and Punjabi) and eight were monolingual (Punjabi).

First generation (born in the subcontinent) 8

Age range (years) 47-77

Male/female 2/6

Occupation Retired (4), manual worker (4)

Chronic diseases

Second generation (born or received formal education from the age of 5 years in the UK)

Age range (years)

Male/female

Occupation

Diabetes (type 1 and type 2), hypertension

18-43 4/3

Student (2), manual worker (2), professional (3)

ORIGINAL RESEARCH

Primary dataset 2- Pakistani Muslim study _

To explore how social capital (social networks, trust, and cultural norms) affects the prevention of cardiovascular disease within

the Pakistani community CD_

Interviews were conducted to collect data on social network structure, sources of support, information, and influence on lifestyle factors pertaining to diet and exercise. Atopic guide was used alongside the convoy model diagram to illicit responses F.K. is a qualitative researcher with a background in psychology, health psychology, and social research. As a female, second-generation descendant of Muslim, migrant parents from Pakistan, she identifies with a British and Muslim identity

Trigenerational families were recruited across the West Midlands using word-of-mouth and snowballing techniques as well as lay-led posters and information sheets.

In total, 42 participants (22 female, 20 male) were recruited from diverse occupational and educational backgrounds. Twelve participants were monolingual (Urdu) and 30 participants were bilingual (English and Urdu)

21-70 14/10

Retired (2), manual worker (6), housewife (4), professional (9),

unemployed (2), student (1)

Hypertension

18-45 7/8

Student (6), manual worker (3), professional (6)

at the University of Birmingham. Interview transcripts were translated into English and transcribed by F.K. Participants were informed of their right to withdraw from the study at any time and were assured that their personal details would be kept confidential. Ethical approval was obtained from the University of Birmingham. All participants were given a numerical identifier to ensure anonymity.

0 v> ra ra

0 v> ra ra

a o Z;

M ■ó

Data analysis

We used two stages of analysis. In the first stage, we adopted a qualitative content analysis approach to compare accounts on a similar range of topics separately for each study. Transcripts were descriptively coded by M.S. and F.K., and a sample was verified by members of the research team (N.G., P.G.). Potential biases in data analysis (e.g., drawing conclusions heavily from a single participant from the sample) were resolved through monthly meetings with the research team for each respective study. Transcripts were coded with NVivo 10. Transcripts were re-read for methodological reflection, and that led to the second stage of analysis, identifying illustrations across both studies where language and the nature of the topic discussed influenced participant responses across three generations. Moreover, we investigated how the style of questioning altered the depth of response from participants. M.S. and F.K. presented data regarding emerging themes from both datasets, providing each collaborator with the opportunity to ask questions and address analytical disagreements. We have used the analytical approach of Cortazzi et al. [13] from a study of interviewing monolingual and bilingual Chinese participants as a guide to interpreting and presenting our findings. Their intention was to elicit examples where researchers were insiders/outsiders, where there were nuances of cultural communication, and where meaning may be unsaid but is applied.

Results

o ■c

■o h

« ■Ö

Language proficiency as a demonstration of authenticity

Although both researchers shared a number of similarities with first-generation participants, there were key differences in linguistic communication, primarily an underdeveloped understanding of particular words and meanings in Urdu or

Punjabi. There were words and meanings that were entrenched in concepts such as home, migration, and an authentic identification of oneself as Punjabi Indian/Pakistani or Sikh/Muslim. Furthermore, migrant participants' command of their first language ranged from unrefined colloquial dialect to more educated. Regardless, first-generation participants attempted to give direct conceptual and linguistic translations in English to help researchers understand their accounts:

What do you call it? Metabolism, in Punjabi we call it Aagani, so in some people their Aagani is much faster than what it is in others so it makes no difference to them. (Male, first generation, 52 years old, Punjabi)

I have been listening to some translations of the Quran, and it says in there that you have 2 or 3 Nafs, one is the good and bad one, do right or wrong... the other talks about eating. (Female, first generation, 48 years old, Urdu)

Here the participant dictates the position the researcher has on the insider-outsider-researcher-participant spectrum. Although interviews were conducted in the participants' first languages, participants doubted the researcher's proficiency and subsequently provided a greater explanation for their response. First-generation interviewees wanted to demonstrate their knowledge of both Western and cultural/religious concepts while simultaneously adopting a position of authority which warrants respect from the researcher to listen and acknowledge.

Concordance: ethnicity, gender, and age

Interviews with participants across all three generations were influenced by ethnicity, age, and gender, particularly in the context of marriage. A number of participants were of an age similar to that of the researchers (20-30 years), encountering social issues such as completing education, finding employment, and suitable partners for marriage. There was an explicit assumption of a shared cultural knowledge of the practices of modern day 'arranged introductions' for marriage and expected cultural conformity to participate in this tradition:

I always had a really positive response which I think has surprised my family, and you know when they've said to

people she's coming to age, she studied and all the rest of it, we'd like to start looking, nothing too serious lets just get the ball rolling, she's got to be married by the time she's thirty that kind of a thing [laughs]. (Female, third generation, 24 years old, English)

The final phrase that concludes the excerpt "that kind of thing" illustrates a shared experience between the researcher and the participant, where issues such as ethnicity, gender, and age are acknowledged within a cultural space, increasing rapport where the relationship becomes closer and personalized. Conversely, establishing a close relationship with participants is conflated with trustworthiness. In contrast to this shared parity of dialogue, first-generation migrants took an informative stance where they felt comfortable commenting on the physical appearance of the female researcher:

Yes, my daughter is as skinny as you are and I am always doing this to her [grabs researcher's arm] and she is skinnier than you are and she doesn't eat much and I was like that when I wasn't married, when [she] wears my clothes from before I was married its perfect how they fit her and I was skinny as well. (Female, first generation, 46 years old, Urdu)

She'll [wife] have a long woolly jumper up to here [knee length] and her trousers, and you can't see anything of her. Just like you are dressed. I don't see anything wrong with it. (Male, first generation, 58 years old, English)

Sharing a common caste identity became an essential prerequisite in understanding the context behind the behaviors Sikh participants practiced. Although Sikhism rejects the existence of caste, the male researcher shared the same caste identity (Jatt [agricultural background]) with 15 participants. This shared identity allowed participants to convey the importance of culturally embedded lifestyle choices, such as drinking alcohol, although this is religiously prohibited:

Us Jatts we don't let a person leave until they have fallen over and can't take anymore. This has always been the case, even people before us [previous generations] they

were also like this, "pick it up, pick it up" [alcohol], these were the practices of Jatt people. (Male, first generation, 51 years, Punjabi)

Concordance by ethnicity, particularly having a South Asian background, defined the boundaries of the topic and the positions interviewees adopted. The nature of relationships quickly moved away interviewer-participant to a familial nature such as 'mother-daughter,' generating narratives entrenched in cultural practices passed from one generation to another. However, caution was taken not to assume knowledge of such traditions, and participants presented their narratives on the basis of their own conceptualizations and experiences.

Cross-language illustrations

To interpret the nature of variation or similarity across transcripts, we used the method of Cortazzi et al. [13] of identifying examples where information was the same in both English and Punjabi/Urdu; there was greater divulgence or expression in English, and greater divulgence or expression in Punjabi/ Urdu. We have included both questions asked by interviewers and responses given by interviewees. In Tables 2-5. responses given in Punjabi (and correspondingly in English) are from the Indian, Sikh study and those in Urdu are from the Pakistani, Muslim study.

Responses that showed little or no difference in English or Punjabi/Urdu. Table 2 contains examples where participants conveyed similar information in English and in Punjabi or Urdu. Although there are differences in the length of the response, there is little variation in content.

In both examples, the information is almost identical: discussion centered on diet and exercise. There was a distinct variation in the manner in which the questions were framed in English or Punjabi/Urdu, with questions asked in the latter taking much longer. This was a result of (1) the researcher's limited proficiency in community languages, (2) translating questions accurately using terminology and concepts which may not exist in English (or vice versa), and (3) framing questions in respect of cultural, religious, and gender values to avoid causing offence.

Responses that showed greater divulgence and/or expression in English: Table 3 gives two examples of where participants speaking English give greater personal biographical detail in their accounts. In the first example, the second-generation participant addresses concerns regarding 'risk' because of a family history of chronic disease, generating a narrative centered on diagnosis and dealing with expectations. Narratives in English used more emotive language or words such as 'risk,' 'hate,' and 'positive.' In addition, community-based researchers confidently used their insider status whereby

Table 2. Responses that showed little or no difference in English or Punjabi/Urdu

Question and response in Punjabi

Interviewer: In your own opinion what would a healthy lifestyle be and what would a healthy person eat or behave like?

Interviewee: Firstly that person should have the right diet, and secondly if that person exercises that is also very good, because one should always control their diet, what they have to eat and drink. In my opinion if a person does these two things then that person can be healthy

Question and response in Urdu

Interviewer: So in your opinion what is a healthy routine?

Interviewee: A healthy routine is eating properly and being observant of your diet and walking and going to the gym, that is a healthy routine but we don't do this.

Question and response in English

Interviewer: Right, in your opinion what consists of a healthy lifestyle?

Interviewee: It's erm, positive mental health, its eating the right foods, exercise in moderation ... and it's ... yeah, that's it those three things really.

Question and response in English

Interviewer: Yeah, so what in your opinion is like a healthy lifestyle?

Interviewee: Um eating good, staying good, staying fit, staying healthy it's a nice lifestyle I don't think it's anything bad it's not slacking.

Table 3. Responses that showed greater divulgence and/or expression in English

Question and response in Punjabi

Question and response in English

Interviewer: I want you to go back to the start and tell me the time when you first were diagnosed with diabetes, tell me what happened?

Interviewee: At first I had no illness then the doctor gave me some tablets and that caused my diabetes, that particular medication.

Interviewer: How did you feel when you were first diagnosed, what sort of thoughts were going through your head when they told you that you were diabetic?

Interviewee: Well I mean there was an element of ... the back of your mind because you know you know that you have a family history and because of my dad I did some digging up on diabetes and those sort of things as well and then you realize and the you sort of have a medical check-up, I mean I've been having medical check-ups for a good few years and you do those through work anyway right and because you've been having those diabetes has always been recognized as a risk so at the back of your mind you've been looking at that and that makes you half prepared it was still a bit of a shock and this and that and finally it's been diagnosed but it's probably a bit of a soft landing because in one way you've been expecting.

Question and response in Urdu

Question and response in English

Interviewer: So do you think that your family or friends or people you know share this opinion [of health]?

Interviewee: Opinion, my child I will think that in my opinion, majority of our Pakistani people, they have the same opinion but I don't know, maybe some are tough and they don't care about anyone else but in my opinion you can't be inhumane and not look after your wife, kids and relatives, and if someone is ill then he will feel for it and when he feels it and if he is an old one so he will get to it.

Interviewer: So why do you think people, or your friends, have that kind of attitude towards health?

Interviewee: with my friends, at first I had a running buddy with me that was really big and then we kind of got her stamina really high and we did a couple of races and we were really good but then she put on the weight again and she said no I can't run any more, it's too much for me it's bad for my back so she stopped doing it and then you got, I've got white friends who stuck with me with running because they grew a passion for it then I've got the Asian girls who are getting married in a couple of months' time so they want to start running to lose it to get a flat stomach but it doesn't happen, you get tired, you might think I'm crazy but I get up at 8 to go running on a Sunday right, that's because you look forward to it, like I hate waking up, I know it sounds really bad but if I'm going swimming in the morning I look more forward to work, because yeah I'm going swimming in the morning and then I'm going to work and when I'm at work I'm finishing work and yeah I'm going for a good work out, I look forward to it because I have a positive relationship with working out because I know I would release those endorphins and get runners high.

Table 4. Responses that showed greater divulgence and/or expression in Punjabi/Urdu

Questions and response in Punjabi

Interviewer: In your thoughts whose responsibility is it to manage one's health?

Question and responses in Urdu

Interviewer: What would you say about your health right now?

Interviewee: Initially like I said when I left the army, the routine was disturbed and I put on some weight but since being ex-army and having an army background I felt that it's not good being overweight and being unhealthy so I tried my best to go for physical activity but keep checking on my daily intake as well, whatever I have, that is even more important than your daily jogging.

Questions and response in English

Interviewer: What aspects of your health do you feel most responsible for?

Interviewee: Of my health ... a bit of everything really you've got to be conscious of what you eat, my lifestyle, a bit of everything really you've got to look at what you want to eat, look at what you do, your sleeping patterns and things like that, a bit conscious of everything really not just one main area.

Question and responses in English

Interviewer: Mm. So what kind of ... what would you say about your health right now?

Interviewee: It's okay. I dunno what I'd class it as. I wouldn't class myself as a healthy health type of person or a really unhealthy, just. just normal, I'd, I'd say, but I'm not sure.

Interviewee: Our own, other people shouldn't be responsible for you. If I tell your uncle that he is unwell he won't do anything about it, one has to do things for themselves. If one is able then they should look after themselves, once one falls then they cannot help themselves someone has to take care of them. When I was ill your uncle had to do everything for me because I was unable. I want to do things for myself; I don't believe in others, I don't like food that has been made by someone else. I want to be able to walk and do things for myself, being able to eat and feed myself.

Table 5. The presence of bilingualism

Response in Punjabi or Urdu

Response in English

Interviewee: I don't know, but I think there is an organ within the body that stops making sugar. Some people say it is with eating sugar, it can't be with eating sugar can it?

Interviewee: Well praise be to Allah, my mum and mother in law I saw that they had blood pressure and sugar and heart problems till now I don't have that.

Interviewee: It's conformity isn't it, it's what other people would perceive to good seva [selfless service] for them and you don't want to disrespect anyone that comes around, well that's what you're brought up with and it's something that you internalize and you don't challenge the status quo because the last thing, over any food or health is you don't want people to walk away with a bad taste in their mouth about the way you've treated them when they've come to your house you know, so it's not a battle worth having

Interviewee: yeah, communities so you've got your dads community like you say beradri [kinship], like we've got my dads who are from the village in Pakistan and then we've got my mums who are quite lenient cause they're from the city but that's like a contradiction within itself, then you've got the whole British thing and you go to work and there's only three brown people who work with you so, you're pulled in all directions and you've got a passion for something and all the Muslim girls that want to do it are like oh you run, so it must be fine obviously its doing good for you but do you think its Islamic, its like I don't know I'm not a molvi [religious clerk] I don't know so like just get on with it you're going to die soon anyway (both laugh) its true though! Although, isn't it?

topics such as race were discussed without censorship. In contrast, topics such as diagnosis or family/friends may be considered too personal for first-generation migrants to discuss with members of their own community, where researchers remain outsiders to their personal beliefs.

Responses that showed greater divulgence and/or expression in Punjabi/Urdu: The examples in Table 4 show first-generation interviewees speaking in Punjabi or Urdu spoke of their personal accounts but also how they would advise members of their own community with regard to healthy living. Here, researcher and participant roles are clearly distinct; however, responses in English were much more generic rather than incorporating the participants' own experiences to support their opinions.

The presence of bilingualism: The extent to which participants changed from speaking one language to another was limited to second- and third-generation bilingual participants, with only occasional references made by first-generation participants to particular foods or concepts - for example. "sugar" (diabetes). Bilingual responses (Table 5) referred to accounts regarding religion, culture, and caste. There was a distinct lack of bilingualism on topics such as exercise, dietary behavior, and illness. Allowing for bilingual accounts, however fraught, increases the credibility of participants' narratives, using the most accurate terms to describe their experiences, rather than relying on terminology which they feel the researcher, or a wider audience, would be able to understand.

Given the presence of bilingualism was reduced to topics of religious or cultural significance, the advantage was the ability of researchers to draw on such shared knowledge, probing the participant further. Using researchers from the community led to greater nuanced data that leads to greater exploration. Hence, if participants have greater awareness of the availability of community-based researchers who can better understand their narratives, this may lead to greater participation from minority groups in applied health research.

Discussion

The nature of accounts with Indian Sikh and Pakistani Muslim communities differs according to the language and topic.

First-generation migrants when discussing religion, culture, or family practice were far likelier to use group or community narratives and give a longer response, indicating greater familiarity with or importance of such issues.

Our findings show the importance of a household approach to researching migrant communities as well as the role of inter-languages (i.e., the language system a learner of a language constructs out of the linguistic input that he or she has been exposed to) [30]. Hence, participants shifted between second and first languages to construct their narratives, with bilingual participants placing themselves at the center of their socially contextualized accounts, emitting a strong individualistic identity. These differences may be a result of the socioeconomic progress and acculturation of second-and third-generation descendants compared with first-generation migrants. For instance, accounts were entrenched in 'personal development' with regard to healthy living, not adhering to shared community values and norms on diet, cooking practices, and exercise.

Notably, our findings highlight qualitative data collection with minority-ethnic communities needs revisiting. Temple and Young [10] discuss the importance of agreement at the source (i.e., the participant) with regard to the accuracy of translation to remove any potential bias. Moreover, 'racial matching' of interviewers and participants can be problematic where the researcher/translator role is dictated by the social positioning in the community of interest [31].

Interviews are not culturally neutral despite South Asian participants having a greater preference for talk-based methods. The use of bilingual researchers, by promoting a shared language and cultural background, allows the possibility of the use of a sharing stories approach to elicit data that are contex-tually relevant to understanding beliefs and practices. Hence, community-based bilingual researchers become producers of research data and shape analysis through their identity and experiences as well as those of the participants [32]. As a result, we propose the following recommendations when one is collecting 'talk-based data' from migrant groups using bilingual community researchers:

1. First-, second-, and third-generation individuals from minority-ethnic communities should be given the opportunity to be interviewed in their first and second languages.

2. The language proficiency of bilingual researchers should be tested/measured for linguistic ability.

3. Throughout interviews, researchers need to address concerns of conceptual equivalence in order to comprehend the meaning of interviewee narratives.

4. If feasible, interviews conducted in minority languages and transcribed into English need to be checked for conceptual equivalence by an external member of the research team.

5. Talk-based data collection methods used with minority populations need to allow for greater time, for example, for translation and framing of questions during interview.

Community-based researchers bring a range of skills and knowledge within qualitative health research that are difficult to replicate with interpreters or link workers used in previous studies [32]. Community-based researchers bring knowledge of how to recruit participants, interview participants, and interpret data from migrant communities, and the closeness of the relationship that can be developed with interviewees can increase the likelihood of participation in applied health research. In addition, research methods, such as interviewing, are not culturally neutral, with the question and answer style alien to many minority-ethnic communities[34]. Ultimately, researchers need to be flexible in their interview style to incorporate the interwoven social, cultural, and religious nuances in participant narratives and generate findings as a single perspective.

The strengths of our study are the number and varied background of the participants recruited from two 'at risk' migrant communities. We have included a range of examples to highlight how concordance and language can influence the presentation of narratives. Our findings may be transferable to other 'at risk' communities living in developed countries. Methodologically, we have contributed to the existing literature by identifying that certain cultural, social, and religious concepts/discussions could happen with more nuance and depth in community languages and others could happen with more nuance and depth in English. Such a method of analytical interpretation could be further added to and developed in future studies with minority groups.

However, including members from migrant communities from other parts of the world and not just the subcontinent may have highlighted different topics that could alter the depth and expression of response in English and other languages. We have focused only on using interviews to collect data, where community-based researchers using other methods (e.g., observation) may encounter different outcomes. Our sample size is small and our findings may be transferable to but not general-izable to other minority communities. The nature of the topics addressed throughout interviews may have influenced the depth, detail, and diversity of the narratives developed between the researcher and participants. Our findings may have been different if the transcripts were not translated into English and then subsequently coded, as importance was placed on conceptual and not linguistic or structural equivalence.

Conclusion

The recruitment of community-based researchers plays an important role when one is attempting to recruit participants from marginalized groups. Further methodological research needs to explore how the researcher/translator epistemological position influences insider/outsider status, whether or not the order and frequency of interviews with monolingual/bilingual participants influences the nature of emergent themes, and the consequence of end-of-interview clarification with participants relating to the meanings of words and/or concepts. This article strengthens the argument for adapting existing talk-based data collection methods, reconsidering the current format of how to ask questions and interpret exchanges throughout qualitative interviews. Acknowledging the presence of different narratives can lead to nuanced analysis of participant accounts in second languages, which may lead to a better understanding of the health needs of minority groups living in developed countries.

Conflict of interest

The authors declare no conflict of interest.

Funding

This work was funded by the University of Birmingham and the National Institute for Health Research (NIHR) through the Collaborations for Leadership in Applied Health Research and Care West Midlands program. Primary Care Clinical Sciences

is a member of the NIHR School for Primary Care Research.

The views expressed in this publication are not necessarily

those of the NIHR, the Department of Health, NHS Partner

Trusts, the University of Birmingham, or the CLAHRC-WM

Steering Group.

References

1. Hussain-Gambles M. South Asian participation in clinical trials: the views of lay people and health professionals in Health Policy. Health Policy 2006;77(2):149-65.

2. Gill P, Plumridge G, Khunti K, Greenfield S. Under-representa-tion of minority ethnic groups in cardiovascular research: a semi-structured interview study. Fam Pract 2013;30(2):233-41.

3. Redwood S, Gill P. The under-representation of ethnic minority groups in research studies. Br J Gen Pract 2013;63(612): 342-3.

4. Jolly K, Lip GY, Taylor RS, Mant JW, Lane DA, Lee KW, et al. Recruitment of ethnic minority patients to a cardiac rehabilitation trial: The Birmingham Rehabilitation Uptake Maximisation (BRUM) study. BMC Med Res Methodol 2005;5:18.

5. Rugkasa J, Canvin K. Researching mental health in minority ethnic communities: reflections on recruitment. Qual Health Res 2011;21(1):132-43.

6. Vertovec S. Conceiving and researching diversity. MMG working paper 09-01, 2009.

7. Phillimore J. Approaches to health provision in the age of super-diversity: accessing the NHS in Britain's most diverse city. Crit Soc Policy 2011;31(1):5-29.

8. Esposito N. From meaning to meaning: the influence of translation techniques on non-english focus group research. Qual Health Res 2001;11(4):568-79.

9. Temple B. Crossed wires: interpreters, translators, and bilingual workers in cross-language research. Qual Health Res 2002;12 (6):844-54.

10. Temple B, Young A. Qualitative research and translation dilemmas. Qual Res 2004;4(2):161-78.

11. Plumridge G, Redwood S, Greenfield S, Akhter N, Chodhury R, Khalade A, et al. Involving interpreters in research studies. J Health Serv Res Policy 2012;17(3):190-2.

12. Israel BA, Schulz AJ, Parker EA, Becker AB. Review of community-based research: assessing partnership approaches to improve public health. Annu Rev Public Health 1998;19:173-202.

13. Cortazzi M, Pilcher N, Jin L. Language choices and 'blind shadows': investigating interviews with Chinese participants. Qual Res 2011;11(5):505-35.

14. Squires A. Methodological challenges in cross-language qualitative research. Int J Nurs Stud 2009;46(2):277-87.

15. Ergun A, Erdemir A. Negotiating insider and outsider identities in the field: "insider" in a foreign land; "outsider" in one's own land. Field Methods 2010;22(1):16-38.

16. Archer L. Constructing minority ethnic middle-class identity: an exploratory study with parents, pupils and young professionals. Sociology 2002;45(1):134-51.

17. Johnson-Bailey J. The ties that bind and the shackles that separate: race, gender, class, and color in a research process. Int J Qual Stud Educ 1999;12(6):659-70.

18. Kusow AM. Beyond indigenous authenticity: reflections on the insider/outsider debate in immigration research. Sym Interact 2003;26(4):591-9.

19. Herzfeld M. Anthropology and the politics of significance. Soc Anal 1997;41(3):107-38.

20. Office for National Statistics, 2011. Census: standard area statistics (England and Wales). ESRC/JISC Census Programme, 1 Census Dissemination Unit, Mimas. Available from: www. statistics.gov. uk/hub/index.html.

21. Liu JJ, Davidson E, Bhopal RS, White M, Johnson MRD, Netto G, et al. Adapting health promotion interventions to meet the needs of ethnic minority groups: mixed-methods evidence synthesis. Health Technol Assess 2012;16(44):1-469.

22. Johnson MRD. Ethnic minorities, health and communication. A research review for the NHS Executive and West Midlands Regional Health Authority, Research Paper in Ethnic Relations 1996, No. 24, Centre for Research in Ethnic Relations, University of Warwick.

23. Yancey AK, Ortega AN, Kumanyika SK. Effective recruitment and retention of minority research participants. Annu Rev Public Health 2006;27:1-28.

24. Gill PS, Kai J, Bhopal RS, Wild S. Health care needs assessment: black and minority-ethnic groups. In: Raftery J, Stevens A, Mant J, editors. Health care needs assessment. the epidemiologically based needs assessment reviews. 3rd ed. Abingdon: Radcliffe Publishing Ltd.; 2007. pp.

25. Ballard R. Middle class neighborhoods or 'African Kraals'? The impact of informal settlements and vagrants on post-apartheid white identity. Urban Forum 2004;15(1):48-73.

26. Ballard R. 'Slaughter in the suburbs': livestock slaughter and race in post-apartheid cities. Ethnic Racial Stud 2010;33(6):1069-87.

27. Archer L. Change, culture and tradition: British Muslim pupils talk about Muslim girls' post-16 "choices." Race Ethnic Educ 2002;5(4):359-76.

28. Serrant-Green L. Black on Black: Methodological issues for black researchers working in minority ethnic communities. Nurse Res 2002;9(4):30-44.

29. Denzin NK, Lincoln YS. Handbook of qualitative research. London: Sage Publications; 1994.

30. Lie Mabel LS. Methodological issues in qualitative research with minority ethnic research participants. Res Pol Plan 2006;24(2):91-103.

31. Twine F. Racial ideologies and racial methodologies. In: Twine F, Warren J, editors. Racing Research researching race:

methodological dilemmas in critical race studies. New York: New York University Press; 2000. pp.

32. Adamson J, Donovan JL. Research in black and white. Qual Health Res 2002;12(6):816-25.

33. Flores G. The impact of medical interpreters services on the quality of health care: a systematic review. Med Care Res Rev 2005;62(3):255-99.

34. Redwood S, Gale NK, Greenfield S. 'You give us rangoli, we give you talk': using an art-based activity to elicit data from a seldom heard group. BMC Med Res Methodol 2012;12:7.