Volume 7, No. 4, Art. 10 – September 2006
Grasping at Context: Cross Language Qualitative Research as Secondary Qualitative Data Analysis
Bogusia Temple, Rosalind Edwards & Claire Alexander
Abstract: Research with people who are not fluent in the dominant language of the research endeavour often involves working with interpreters/ translators or researchers who can speak the relevant minority languages. They conduct the interviews and provide the written data used for analysis in a language other than the original. However, this kind of cross language research is often presented as if it is the analysis of primary data rather than the re-construction of it. We argue that analysis of cross language data shows some strong similarities with secondary data analysis. Questions about the relevance of the context in which data are produced are central to both cross language research and secondary qualitative data analysis. We illustrate our arguments using a research project that examined user views of interpreters and discuss how we dealt with the issue of context in analysing data that were collected by others and produced in languages we did not speak.
Key words: secondary qualitative data analysis, cross language, community researchers, interpretation, translation, methodology
Table of Contents
2. Issues of Context Within Secondary Qualitative Data Analysis
3. Language as Context
4. The Research
4.1 Training and debriefing
4.2 Final interview
5. Links with Communities and Cultural Brokerage
6. Presenting Findings, Representing Voices and Writing Difference
Research with people who are not fluent in the dominant language of the research endeavour often involves working with researchers or interpreters/translators who can speak the relevant minority languages. These people conduct the interviews and provide the written data used for analysis in a language other than the original. This kind of cross-language research, however, is often presented as if it is the analysis of primary data rather than the re-construction of it. In this article we argue that analysis of cross-language qualitative data shows some strong similarities with secondary data analysis. Debates around secondary qualitative data analysis thus can be usefully applied to cross-language research. Most pertinently, questions about the relevance of the context in which data are produced are central to both cross-language research and secondary qualitative data analysis. 
In this article, we initially outline debates about secondary qualitative data analysis and the issue of language as part of the data construction context. We then illustrate our arguments about context drawing on a research project1) (ALEXANDER, EDWARDS & TEMPLE, 2004) that examined user views of interpreters, and discuss how we faced issues of context and social power in analysing data that were collected by others and produced as primary data in languages that we did not speak. Throughout the article we use the term "field researchers" to refer to the researchers who carried out the interviews, and "academic researchers" to refer to ourselves as the lead researchers who are based at universities and, with one exception, did not carry out the interviews. The exception was Bogusia TEMPLE who was both an academic and a field researcher on the project. 
Qualitative research typically aims to collect people's lived experiences and the meanings that they give to them. Secondary qualitative data analysis refers to the re-use of such (often archived) data by researchers other than those who collected the original material. There is increasing interest in the possibilities of the use of existing, "second-hand" qualitative data (see, for example, Forum: Qualitative Social Research 6, edited by CORTI, WITZEL & BISHOP, 2005), and this has led to a debate about the significance of the context of production of the original material. There are methodological and epistemological concerns about the extent to which detailed, situated studies can be subjected to "second hand" analysis, focusing on the relationship between analyst and data. 
In most qualitative research approaches, the research and social context of production is viewed as a crucial and integral element in analysis. Data is regarded as a product of the interaction between researchers and participants, and primary researchers are understood as developing an intimate bond with the material that they have collected, especially where they have also designed the framework, immersed themselves in the field, and drawn on personal grounded insights to make analytic interpretations. This understanding has formed the basis for a quizzical stance towards the possibility of secondary analysis of qualitative data that was originally collected by other researchers (THORNE, 1994). The critique is grounded in the significance of the contextual knowledge that is only derived from involvement in the research at the time of its collection. Martyn HAMMERSLEY (1997), for example, while broadly sympathetic to the prospects for secondary use, refers to the "cultural habitus" that is acquired through direct involvement in fieldwork. He suggests that the key role of this tacit or intuitive knowledge and experience about what is relevant, grounded in cultural assumptions and theoretical presuppositions, limits the usability of other people's data. More critically, Natasha MAUTHNER, Odette PARRY and Kathryn BACKETT-MILBURN (1998) argue that secondary qualitative analysis raises quite fundamental epistemological questions, where the endeavour runs the risk of a "naïve realism" that views the primary data as a discrete, "raw" and somehow authentic entity, with the reconstruction of the details of its contextual production viewed as "background" rather than an integral part of its constitution. Overall then, in this view, "being there" is all important, and the lack of being able to engage in reflexive interpretation is a barrier to secondary analysis. 
Those more amenable to, or charged with promoting the practice of, secondary qualitative data analysis, take further the argument that all data and findings are social constructions. Firstly, they argue that the generation of empirical interview material is not merely derived from researchers' aims and assumptions because it is a co-creation between researcher and participant, and those who focus on the research context of production alone are missing the "real" input into the data of participants' own understandings and experiences (VAN DEN BERG, 2005). Secondly, they argue that since interactions between researcher and participant, and between analyst and data, are always mediated through specific cultural and theoretical frameworks: "all analysts whether or not they were "there" at data collection, produce (primary and secondary) analyses which are socially contingent" (HEATON, 2000, p.2). In other words, analytic meaning is always made rather than found. From this perspective, then, contextual reflexivity (albeit it may not equate to "being there") can be "good enough". The socially embedded nature of secondary data means that "good enough" involves not only analysis of the primary data, but also analysis of the context of their production (accessed through grant proposals, correspondence, interview schedules, field notes, reports and so on). Consultation with the original researchers is also suggested as helping to reconstruct the data as socially produced and then integrating this into the secondary analysis. In this way, the particular perspectives from which the data was created can be illuminated as far as this is possible (GILLIES & EDWARDS, 2005; HINDS, VOGEL & CLARKE-STEFFEN, 1997; HEATON, 1998), producing new insights on the original findings (THORNE, 1994), and secondary qualitative data analysis can be a rigorous endeavour. 
As Louise CORTI and Paul THOMPSON have pointed out (2004), while the debates have been conducted in relation to re-use of primary archived data sets that have been generated as part of particular studies in themselves, aspects of practice similar to secondary analysis of data is in fact common in primary studies. These include where research assistants are employed solely to conduct interviews, and others (usually the principal researchers who have designed the framework) analyse the data and write up findings. This practice is not often recognised as secondary analysis, however, and the context of data production is not usually problematised. Nonetheless, it is this similarity that has motivated this article, where we are indeed considering "secondary" analysis of data collected by field researchers. While, as the academic researchers, we had constructed the research frameworks for the study (theoretical and methodological approaches, interview schedules, and so on), we were (with one exception) not "there" when the data was collected. Later on we will detail how, as with secondary analysis of archived data, we attempted to "grasp" the social context of data production through a series of "debriefing" interviews with our field researchers. But importantly, as we will also detail, our field researchers were gathering interview data in one language and translating them into another for us. In this sense, it might be argued that our "primary" data was not merely "second hand" but almost "third hand"—removed from us not only as situated interaction, but also in tongue. Thus, before we go on to discuss our study and attempts to grasp context, we address the issue of language as part of context. 
We have shown above how current debates on secondary analysis of qualitative data raise important questions about the integration of the context under which data are produced, and the possibilities and limits to its re-use. With the increasing interest in cross-language research, lead researchers need to employ people who have particular language skills and, therefore, may not themselves be present at interviews and unable to able to read transcripts in the language originally used by respondents. Debates within translation/interpretation studies mirror those presented above in relation to secondary qualitative, with commentators increasingly arguing against the view that it is irrelevant who interprets or translates a text as long as the translation is carried out "objectively". They view the translator as an author of a new text rather than a technician who re-produces the original in another language (OVERING, 1987; SIMON, 1996; VENUTI, 1995, 1998, 2000; WADENSJO, 1998). For example, Lawrence VENUTI (1998) argues that current, dominant practices of translation result in a re-writing rather than a transfer of meaning. All translators choose between words and concepts to try to reconstruct meaning and there is no single correct choice to be made. The translator and interpreter is part of the context of data production. This approach is not so far removed from the notion hinted at in our discussion on secondary qualitative data analysis—that all researchers are translators and interpreters in their analyses and presentations of their interviewees' experiences and perspectives, even where they share a language. Literal translation, from one language to another, in research makes this process acutely visible, however. 
In the same way as those seeing the potential for secondary qualitative analysis have argued for a "good enough" secondary analytic practice that accesses the circumstances under which the original material was produced, cross-language researchers have begun to develop ways of analysing context where data is collected by others and in a language they cannot understand. For example, two of us have used, respectively, the concepts of "key informants" and "intellectual auto/biography". Rosalind EDWARDS developed the notion of key informants (conventionally used in social research to refer to professionals or lay informants who provide a source of introduction to and information about the field of investigation) to encompass a reflexive exploration of interpreters' social locations, values and beliefs, and understanding of their relationship to the researcher/s and interviewees (EDWARDS, 1995; 1998). Bogusia TEMPLE has drawn on Liz STANLEY'S (STANLEY, 1990) concept of intellectual autobiography as an analytic concern with the grounded specificities of how researchers come to understand the social world—the intellectual construction of knowledge, in examining the influence of interpreters and translators on research (TEMPLE, 2005). Both these analytic concepts highlight the location and influence of all who took part in the research on the final product: participants, researchers, interpreters and translators. 
Researchers undertaking research involving languages they do not speak will have to make choices about who will carry out the interviews, the processes they will use to try to engage with how data is constructed in interviews, and the translation of the interviews, in a way that does not mask the influence of fieldwork relations and language as context. 
We illustrate these points drawing on research we carried out on users' experiences of interpreters, looking at the understandings of people who speak little English and need interpreters in order to use health, legal, social welfare, education and other services (ALEXANDER, EDWARDS & TEMPLE, 2004). As noted, with one exception, we could not carry out the interviews in the relevant languages and relied on others to provide us with access to the context of interpretation, in the widest sense of the term. Here, we discuss the process and issues involved in grasping at context by focusing on the role of our field researchers as links with the communities we were researching, the way we tried to de-mystify the process of data production and meaning in different languages, and the difficult decisions we made about the presentation of translated texts. 
Our research was based on semi-structured interviews with people needing interpreters from within five different minority ethnic groups: Polish, Chinese, Gujarati Indian, Bangladeshi, and Kurdish. We selected these groups because they represent a variety of experiences of migration to and settlement in Britain, around length of presence, levels of integration or marginalisation, legal status (as citizens, settlers or asylum seekers). They also vary between groups that have been highly visible in terms of academic research and policy formulation (Bangladeshi and Gujarati) and those who often are either overlooked (Chinese), subsumed into generic majority (White) categories (Polish), or present new challenges (Kurdish). Ten semi-structured interviews were conducted with people from each group, in the appropriate first language/dialect, by field researchers. Six field researchers conducted the interviews (two Chinese researchers and one for each of the other groups). The languages used were Polish for the Polish interviews; Cantonese, Mandarin and dialects for the interviews with Chinese people; Gujarati for the Indian group; Sylheti, Bengali and dialects for the Bangladeshi group; and Kurdish, Iraqi, Turkish and Arabic for the Kurdish participants. 
The field researchers were asked to access a range of people to interview, who demonstrated the variety of experiences and needs within each community, for example, around gender, age, family background and responsibilities, and migration history. In selecting people to interview, our aim was not to produce a demographically representative sample, but to provide illustrative depth of the range and kinds of experiences of people who need and use interpreters. This was shaped by significant differences in the profile and interpreting needs of each group. For example, with the Gujarati and Polish samples, the length of this population's settlement in the UK meant that the need for English language support was largely restricted to older people. In contrast, the Kurdish and Polish Roma groups reflected the age and gender profile of the predominantly younger, male and newly arrived asylum seekers. 
Each field researcher accessed communities through a mix of channels—their own personal networks and a range of statutory, voluntary and community sector services. Several of the field researchers were members of local communities, or had work links with them; indeed, three (Chinese, Gujarati and Kurdish) had worked as interpreters for local organisations and in part utilised these networks to access potential interviewees. We avoided completely recruiting our sample though such organisations and agencies, however. This was for two reasons: firstly, we did not want the research to be identified with a particular organisation and their agenda. Secondly, and more importantly, we were keen to use informal networks to allow access to people who did not use such organisations and who, therefore, fall through the gaps of formal provision. 
The interviews themselves encompassed a brief life history, addressing people's lives in their country of origin, as well as their experiences in the UK. This was to enable us to place their perspectives on interpreting provision in the broader contexts of who they were as people and what they felt about themselves and their lives both prior to migration and afterwards. Such issues were to prove very significant in terms of discussing issues around access to "community" support and "trust" in the interpreting process. The interviews then concentrated on encouraging a "story telling" approach to instances when people required interpreters in order to gain access to various services, drawing out specific issues concerning the implications of a lack of English language competence, locating and accessing interpreters from both formal and informal sources, who and what type of interpreter and interpreting they felt best served their needs, and their self-positioning and perception of others within this relationship. This part of the interview allowed us to focus on priorities and issues for interpreting provision from service users' perspectives. 
We (the academic researchers) did not previously know our other field researchers, who were found via adverts and informal contacts. They were employed on full or part time contracts for the duration of the fieldwork. The exception, as we have said, was TEMPLE, who conducted the interviews with Polish people. Her knowledge of the language placed her in a different relationship with the data from these interviews compared with the other two of us, in that, for her, the data was primary, based on interviews she carried out, translated and transcribed. Her interviews, however, became secondary data when they were analysed by the other two of us. 
A key concern in the project was to include all the researchers (academic and field) as part of the context in which data was produced, using the concepts of key informants and intellectual autobiographies (see above). This encompassed two processes: (i) training before interviews and debriefing after each interview; and (ii) a final in-depth interview with each of the field researchers about their life experiences and thoughts on the research itself. 
Before they undertook the interviews, we conducted training for our field (non-academic) researchers. As well as research aims, interview tools, ethical practice and so on, this included discussion about the importance of pinpointing possible differences across languages in concept and word meaning. Then, as each interview was carried out and translated, we held a debriefing session with the field researcher where we discussed the situated interview interaction and context. Each interview debriefing also offered an opportunity for us to investigate the field researcher's position in their communities and the influence this had on the way the written text was translated. It also allowed a concrete, rather than abstract, discussion of translation issues around each interview. Our aim here was to "grasp" the context in which the interviews were produced, in place of "being there" ourselves. 
At the end of their work on the project, we interviewed our field researchers about the research process. These final interviews addressed their position in local communities, the informal networks and organisations they used to access people to interview, insights they felt they had developed on the research topic itself, the process of conducting the interviews, and issues about the process of translating and transcribing the interviews. Again, these final interviews were intended to allow us to engage analytically with the field researchers as integral to the constitution of the original data, as primary for them and secondary for us. 
In the rest of this article, we focus on how we approached two aspects of "grasping" at the context of our research data production: that of the position of field researchers in relation to minority communities, and the role of language in producing accounts. 
There has been growing interest amongst academic researchers in working with field researchers who have links to a particular community or communities, often because of their language proficiency. These people are commonly called "community researchers", and they are seen as a way for researchers outside of those communities to gain open and unmediated access to them. In this sense, community researchers are positioned as having a direct and "authentic" link into communities that they can then extend to other, non-member researchers, and as able to act as "cultural brokers" between the community and non-member researchers. This brokerage process is taken-for-granted and is invisible, in which community researchers are implicitly placed as part of the identified community's supposed homogeneous bubble of existence, separate from and parallel to wider society, containing a static and finite monolithic culture (ALEXANDER, EDWARDS & TEMPLE, 2007). 
There is some concern, however, about the often unquestioned nature of these links between communities and selected field researchers. We argued above that all researchers are integral to the construction and production of data, and this includes field researchers employed to act as "cultural brokers". Researchers have shown that the social background of such brokers influences their perspectives on what constitutes community, whom they include from communities and their views on the issues that are covered by the research (BOWES & DAR, 2000; JAN-KHAN, 2006; SCHICK, 2002; TEMPLE, 2002; TWINE, 2000). 
Definitions of community can be based on, and are cross-cut by, a range of factors, including race, religion, social class, language, friendship and place. This means that no one person can represent the views of everyone in an identified community. Elsewhere (ALEXANDER, EDWARDS & TEMPLE, 2007), we have challenged the notions of community that lie at the heart of current discourses around nation and citizenship. We discuss the different ways in which the idea of community is imagined and constructed, including around institutions and personal social networks of family and friends. People do not belong to discrete bubbles of community straightforwardly defined by markers such as ethnicity, language and place. This point can be extended to field researchers employed because of their links to community defined solely on language lines. Thus, situating field researchers within their social contexts helps to disentangle whom it is they are supposed to be representing, what pool of people they can access to interview, and what methods can be used to ensure that a range of views are represented in the research. 
Part of the complexity of community and the ability to "find" it and the people who comprise it, can be seen in how our field researchers attempted to access participants for our research. One of the Chinese field researchers felt that a Chinese church was a valuable source for contacts, and himself had an informal contact who introduced him to the key worker at the church. He explained to us that people would trust him if he was introduced through a church. While he continued with his chosen method of access, he did acknowledge that this meant that, in effect, the key worker had selected people to take part in the research within the parameters that he had presented to her. The other Chinese field researcher similarly began to approach people through a pivotal figure running a community centre, but felt more strongly that this person was trying to control whom she spoke to and withdrew. Nonetheless, she continued to access people through other centres. In the final interview with her, however, she reflected on the advantages and drawbacks of working through such centres as compared with using less formal networks:
"Because I am introduced by their centre manager or by their worker so they trust their Chinese centre to know me. Because I'm not employed from the Centre I'm from outside so they a bit frightened to have an interview with a stranger. So if I'm interviewed by someone they know, well that would be better.
Academic researcher: Do you think if you'd gone through people and not through a centre they would have been less worried about that [saying anything negative on tape], or still worried, maybe?
They worry less." 
The tensions between trusting someone because they were linked to known organisations such as community centres and fearing that these same links were a cause for concern in themselves (which indeed were echoed in the accounts of the people who were interviewed for our research, ALEXANDER, EDWARDS & TEMPLE, 2004) lead to a querying of any homogeneous "community" to which a member of an ethic and language group has privileged and direct access. Underlining this issue further, the one of us who was both academic and field researcher had no long term links with Polish community organisations in the area she was working in, and some of her time was spent developing these because her personal experience led her to feel such organisations were very important for some Polish speakers. Her informal networks helped her to access these initially, and her experience of growing up in another Polish community helped to make her aware of different views within formal organisations and the need to approach various "community leaders" even within one organisation (TEMPLE, 1999). Her current networks also made her sensitive to some people's dislike of being part of organised groups. Furthermore, she felt that her lack of identification with Polish organisations in the area helped in securing interviews with people seeking asylum and sometimes with young people who recently arrived in England and were not welcome at, or who had decided for a variety of reasons not to belong to, these organisations. 
In contrast, the Kurdish field researcher started from his own informal networks and then moved on to formal Kurdish organisations. His awareness of the political and social divisions between and within the organisations he was approaching, and his concomitant need to situate himself in relation to these divisions, especially politically, however, made him feel that informal snowball sampling was an easier process. Moreover, his discussion of his links with other Kurdish people problematised any view of him as representative or a member of "the" Kurdish community in any simplistic or straightforward way:
"I don't have a particular position in Kurdish refugees. I am usually global man! (laughs) … I am just a humanity as a human. Because I lived in a Kurdish area in the countries still you feel some things more than people who are not in them. So I tried to be with the peoples as a member of them. Before the interviews I tried to have a conversation with them and try to share things with them. And so this give them confidence of he is one of us." 
This field researcher's view of himself as a "global" human being who nonetheless may be differentiated from some other Kurds in his national origin and life experiences, has to be placed in the context of, in particular, the politics and complexity of definitions of being "Kurdish". One concrete demonstration of this was the range of languages he had to use to converse with research participants (Kurdish, Iraqi, Turkish, and Arabic). His discussion also illustrates the differential positioning of particular people who act as field researchers within "their" communities, and the need for him to "give" something of himself to participants before they gave something of themselves to him for the research. 
Another example of differential positioning and strategies to deal with this is our Bangladeshi field researcher. He had long-standing involvement in a local Bangladeshi community organisation, and felt this made it easier to access participants for our research. He told us that people would be embarrassed to admit to a stranger that they had difficulty speaking English. He saw himself as very much part of, and active within, the community, which he felt was represented by the range of organisations set up in the area. Moreover, he said that although he found it uncomfortable interviewing women and young people within this community, he knew enough about their culture to be able to develop a strategy to include them:
"Some were very nervous and I was quite nervous. Because you meet someone you have never met and it is like culturally very sensitive as well … I tried to treat them like sisters. If they were mother figures then it is all right, you know auntie. So when you call someone sister that is the most appropriate relationship. That really helps. As in all these things. How to approach young people and how to talk to them. The same with old people … As a worker (within a community organisation) I know what language to use for the right people. Professional or like academic people won't know this." 
The final interviews with the field researchers were not only a valuable resource for "grasping" at the context of accessing research participants, but also revealed other integral aspects of context. These related not only to the field researchers' means of access and revelations of differential positioning within their respective communities, but also their understanding of their social position in society generally. Our Kurdish field researcher, for example, felt that generally he was seen as a refugee with few relevant skills, in the face of his own self-identity as an "educated man":
"But from my view and my life experiences, I found it very very difficult to be a refugee. I have leave to remain status here, I don't have any problems. I am an educated person, but still I have many problems, for instance to regain my qualifications, to have another driving licence, to learn the language … Another thing, it's difficult to mix with the people, educated respected peoples. You don't have a chance. You are all the time in the urban area and you have a chance to mix with street peoples. I find it's more easy than, as an educated man, to mix with respected people who can respect you, can respect you and understand you and you understand them. But we don't have this access actually. I find this only here in the University after I start to work." 
This case raises the complex issues we have tried to draw attention to, as against seeing field researchers as a straightforward means of linking other researchers into homogeneous communities and acting as a cultural broker for them. It also illustrates our field researchers' sometimes uneasy positioning between the two. This positioning was not only social but also linguistic. 
When academic researchers employ "community researchers", we have argued that they should be seen as key informants and co-producers of research findings. We have explored our field researchers' links to and positioning within communities as part of reflexively grasping the context of their access and data production in the face of not "being there" ourselves. As well as the contextual question of fieldwork relations, however, in cross-language research, exploration of field researchers' co-production of the data needs to extend to the process of translation. This aspect of cultural brokerage needs to be opened up so that researchers who are using such secondary data can examine the context under which the data was re-produced in English. 
Writers such as Sherry SIMON have argued that translation is not a case of choosing words from dictionaries but is about making decisions on equivalence:
"The solutions to many of the translator's dilemmas are not to be found in dictionaries, but rather in an understanding of the way language is tied to social realities, to literary forms and to changing identities. Translators must constantly make decisions about the cultural meanings which language carries, and evaluate the degree to which the two different worlds they inhibit are 'the same'. These are not technical difficulties, they are not the domain of specialists in obscure or quaint vocabularies. They demand the exercise of a range of intelligences. In fact the process of meaning transfer has less to do with finding the cultural inscription of a term than in reconstructing its value." (SIMON, 1996, pp.137-138) 
Those choices help form the data that is passed to other researchers for secondary analysis. Researchers not privy to the translation need to set up processes for opening up discussion about the basis upon which those choices were made if they are to have a full grasp of the context of the production of meaning. Without this, researchers are perpetuating a method of access to unmediated, raw primary data, where actually it is secondary data and meaning is made rather than directly found in the language. 
As part of debriefing sessions after each interview field researchers were asked to point out parts of their translation where choices around equivalence of meaning were evident to them. Examples of choices that they identified included how to translate words that had no direct equivalent or where they felt that the equivalent word in English did not convey quite the right "meaning" in a different language, and decisions about the extent to which to stick to the sentence construction that interviewees used in their own language in rendering these into English. 
Our Bangladeshi field researcher's discussion of the choices he had to make to convey meaning will be recognised by many interpreters and translators:
"And there is a common phrase that people use. Use that. But if you try to back-translate raw Bangla it has not got any sense. … [That phrase] means courts and offices. Although people have never been to courts and offices, it is a phrase people use when they say 'all the public offices are closed today'. They mean all the public offices when they say it…so… when they say 'I have been to … the courts and offices', but they have never been to court …Even like when we drink ... when we say in village language [we] say we 'eat' even if you drink. We say we eat." 
There were many examples throughout the interviews of these kinds of choices being made by the field researchers. Such meaning-making work is often invisible in cross-language research. But more than this, the process also involves an effortful production of selves through language, both the self of the research participant and of the field researcher/translator. As our Kurdish field researcher explained:
"During the interpreting time I had to focus everything, writing, listening, watching the word as I write. So it was very difficult. Two or three hours of interpreting with listening to it used to make me dizzy and very tired. … Most of the time I can't find myself exactly with my own words. You have to think in your language and speak in the English language. And all this make you bring the lack of the confidence. It's not reflecting your exact personality. All the time you feel, even to yourself, underestimated." 
This crossing from one language to another is not only an integral part of the context of the production of data for research, but is also a process that was used in context by our field researchers. Our Gujarati field researcher described how she used Gujarati and English on different occasions:
"I think in both ways. If I am doing something in the community centre and then with those members I would think in cultural issues in Gujarati. That would be better. I would think more in Gujarati.
Academic researcher: What about when you go shopping? What language do you think in then?
Oh in English. I think you have to speak in English." 
As a field as well as academic researcher, TEMPLE also recognised the different ways in which she used her languages and the way in which she felt more comfortable using and thinking in Polish in particular contexts. There has been very little written on the different ties bilingual/multilingual researchers have to their languages and the effects of differences in the way they are bilingual. These differences shape the way people interpret and translate primary into secondary data, and form part of the context of the research that needs explication (TEMPLE, 2006). 
Although few researchers would deny that the choices of words and concepts in translation influence how respondents are "heard", fewer acknowledge the effect of the ironing out process that is translation. Our Bangladeshi field researcher's graphic description of the implications of different sentence construction across languages will be recognised by many who have worked as translators but are rarely asked how they translated:
"I found from the research that people speak back to front in the words. Some words are back to forward …." 
Ultimately translators—and indeed researchers—are expected to produce easy-to-read English texts in which the process of production is not apparent. Writers within translation studies argue that the way people's words are translated and then presented in English is a form of colonialisation of meaning or stripping of any other context other than that provided by the English language. For example, Gayatri SPIVAK (1993, pp.399-400) describes standard translation practices as a form of "translatese" where everyone sounds as if they speak perfect English, whatever the language of origin. The re-writing of interviews into this neat English is in effect a "domestication" (VENTUTI, 1998) of text in which the fluency of the final version masks the benchmark of English meaning. VENUTI (1998) points to the role of the academy and publishers in the drive to conceal the language context in which texts are produced and the attempt to re-contextualise into English. He sees fluency as assimilationist. Rather, he argues the need to re-introduce language and cultural context and to "send the reader abroad" (1995, p.34). He calls for translation where the text is non-fluid or estranging in style and is designed to make the translator visible (see also SPIVAK, 1992; 1993; WADENSJO, 1998). This position is recognisable in the arguments above on reflexivity and grasping context in secondary data analysis. 
Some writers (including RIESSMAN, 2000; SPIVAK, 1992, 1993; TEMPLE, 2005; VENUTI, 1995, 1998, 2000) discuss strategies such as the inclusion of words in the source language, the use of notes to discuss possible meanings, and resistance to an English language grammatical structure for non-English texts. In sum, these authors suggest that the language of the original text is part of the context of data production. 
These methods come with their own set of issues (TEMPLE, 2005), however, related to the arguments presented above. For example, the translator's choices of meaning are only one amongst other possibilities since no one person can represent entire communities and language has no simple deterministic tie to meaning. In our research these contextual issues of presentation of possible tensions within translations were brought out in the debriefing after each interview. These debriefings, and the final interviews with field researchers we discussed earlier, were part of our attempt to "grasp at context" in operating at one remove from the primary data for our project, that is, we were not present when the data was collected. But they were also, as we explained earlier in discussing the concepts of key informants and intellectual autobiography, an attempt to make visible and acknowledge our fieldworkers' part in the production of this data, both as primary and secondary. 
Two factors meant, however, in the writing up at least, that we rendered our field researchers invisible and hid their part in the linguistic production of the data. In writing up, we tidied up the quoted data we presented to conform to English readers' expectations. There were two reasons behind what could be regarded as this betrayal of context. Firstly, the demands of academic publishing and the expectations of funders meant that we experienced strong resistance to "untidy quotes". Secondly, issues of social power and hierarchy as these are intertwined with forms and styles of language come into play—another contextual issue shaping the production of research that has to be grasped. There is evidence that strategies that reveal the translation can be self-defeating because they result can in a stereotypical "pigeon English" that makes non-English speakers seem incoherent (HALE, 2002). Jan BLOOMMAERT argues that the:
"… process of (re)structuring talk into institutionally sanctioned text involves a dynamic of entextualization that is based on power asymmetries. This process also involves the problematic of the availability and accessibility of linguistic-communicative resources—an often overlooked 'context' of talk." (2001, p.415) 
Ignoring the language an interview was carried out in, obscuring the process by which it was translated and by whom, is de-contextualising and elides social power. But there are also knotty choices here in that it is that very social power that shapes researchers' ability to and decisions about revealing the process of linguistic production in cross-language research. We acknowledge that some may feel that our practice was not "good enough" in this respect, and that we surrendered to these power relations too easily for the final product. 
Tying together debates about secondary qualitative data analysis and cross-language research provides many fruitful avenues for comparison and enlightenment relating to how secondary and primary researchers grasp at context in order to construct meaning. It opens up discussion about this context, both in terms of the relationship between researcher and data: as situated fieldwork and the implications of the language used. For secondary researchers of whatever form, it is never possible to have the same relationship to the data as "being there" but—as for some writers on secondary analysis—we have argued that it is possible to examine context in cross-language research as part of "good enough" practice, and to debate how this shapes research. 
Cross-language field researchers produce a new (secondary) text of which they are an integral part. Researchers can debate influences on the texts they get, including the effect of moving from source language to final dominant language text. We have argued that principal researchers are often dealing with secondary data, and that cross-language researchers may be looking at data that can be described as third hand in that not only were the researchers not present when the data was produced but they usually analyse the interviews after they have been transformed into another language. This data production process is not neutral; it is shot through with social power relations and positioning in terms of field researchers' place in the communities to which they are linked and in wider society. 
We have suggested that the field researchers' position as cultural brokers and key informants is complex rather than unmediated and straightforward. We have shown how using debriefing sessions after each interview and carrying out a final interview with field researchers can begin to open up and make accessible (albeit partially) the context of production for cross-language research, which is second (if not even third) hand linguistically and analytically. This process produced data that enabled us, as academic researchers who, to some extent, were secondary analysts, to grasp at the context of data production in a way that we feel is "good enough". As we have also discussed in relation to writing up, however, actually revealing that context of production in a final product (accounts of the research findings) is a difficult judgement that is itself shaped by that context that is grasped at. 
1) The "Access to Services With Interpreters: User Views’" project was funded by the Joseph Rowntree Foundation and we are grateful to them for their support. <back>
Alexander, Claire; Edwards, Rosalind & Temple, Bogusia (2007, in press). Contesting cultural communities: Language, ethnicity and citizenship. Journal of Ethnic and Migration Studies.
Alexander, Claire; Edwards, Rosalind & Temple, Bogusia with Usha Kanani, Liu Zhuang, Mohib Miah & Anita Sam (2004). Access to services with interpreters: User views. York: Joseph Rowntree Foundation.
Bloommaert, Jan (2001). Investigating narrative inequality: African asylum seekers' stories in Belgium. Discourse and Society, 12(4), 413-449.
Bowes, Alison & Dar, Naira (2000). Researching social care for minority ethnic older people: implications of some Scottish research. British Journal of Social Work, 30, 305-321.
Corti, Louise & Thompson, Paul (2004). Secondary analysis of archived data. In Clive Seale, Giampietro Gobo, Jaber F. Gubrium & David Silverman (Eds.), Qualitative research practice (pp.327-343). London: Sage.
Corti, Louise; Witzel, Andreas & Bishop, Libby (Eds.) (2005). Secondary analysis of qualitative data. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(1), http://www.qualitative-research.net/fqs/fqs-e/inhalt1-05-e.htm.
Edwards, Rosalind (1995). Working with interpreters: access to services and to user views. In Gail Wilson (Ed.), Community care: Asking the users (pp.54-68). London: Chapman & Hall.
Edwards, Rosalind (1998). A critical examination of the use of interpreters in the qualitative research process. Journal of Ethnic and Migration Studies, 24(2), 197-208.
Gillies, Val & Edwards, Rosalind (2005). Secondary analysis in exploring family and social change: addressing the issue of context. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(1), Art. 44, http://www.qualitative-research.net/fqs-texte/1-05/05-1-44-e.htm.
Hale, Sandra (2002). How faithfully do court interpreters render the style of non-English speaking witnesses' testimonies? A data-based study of Spanish-English bilingual proceedings. Discourse Studies, 4, 25-47.
Hammersley, Martyn (1997). Qualitative data archiving: some reflections on its prospects and problems. Sociology, 31(1), 131-142.
Heaton, Janet (1998). Secondary analysis of qualitative data. Social Research Update, 22, http://www.soc.surrey.ac.uk/sru/SRU22.html.
Heaton, Janet (2000). Secondary analysis of qualitative data: A review of the literature, ESRC Research Report R000222918.
Hinds, Pamela; Vogel, Ralph & Clarke-Steffen, Laura (1997). The possibilities and pitfalls of doing a secondary analysis of a qualitative data set. Qualitative Health Research, 7(3), 408-424.
Jan-Khan, Manawar (2006). The community leader, the politician and the policeman—a personal perspective. In Bogusia Temple & Rhetta Moran (Eds.), Doing research with refugees: Issues and guidelines (pp.97-109). Bristol: Policy Press.
Mauthner, Natasha; Parry, Odette & Backett-Milburn, Kathryn (1998). The data are out there, or are they? Implications for archiving and revisiting qualitative data. Sociology, 32(4), 733-745.
Overing, Joanna (1987). Translation as a creative process: the power of the name. In Ladislav Holly (Ed.), Comparative Anthropology (pp.70-87). Oxford: Basil Blackwell.
Riessman, Catherine Kohler (2000). Even if we don't have children [we] can live: Stigma and infertility in South Indian narrative and the cultural construction of illness and healing. In Cheryl Mattingly & Linda Garro (Eds.), Narrative and the cultural construction of illness and healing (pp.128-152). Berkeley, CA & London: University of California Press.
Schick, Ruth (2002). When the subject is difference: Conditions of voice in policy-oriented qualitative research. Qualitative Inquiry, 8(5), 632-651.
Simon, Sherry (1996). Gender in translation: Cultural identity and the politics of transmission. London: Routledge.
Spivak, Gayatri (1992). The politics of translation. In Michele Barrett & Anne Philips (Eds.), Destabilising theory: Contemporary feminist debates (pp.177-200). Cambridge: Polity Press.
Spivak, Gayatri (1993). Outside in the teaching machine. London: Routledge.
Stanley, Liz (1990). Moments of writing: Is there a feminist auto/biography? Gender & History, 2, 58-67.
Temple, Bogusia (1999) Diaspora, diaspora space and Polish women. Women's Studies International Forum, 22(1), 17-24.
Temple, Bogusia (2002). Crossed wires: Interpreters, translators, and bilingual workers in cross-language research. Qualitative Health Research, 12, 844-854.
Temple, Bogusia (2005). Nice and tidy: Translation and representation. Sociological Research Online, 10(2), http://www.socresonline.org.uk/10/2/temple.html.
Temple, Bogusia (2006). Being bilingual: Issues for cross language research. Journal of Research Practice, 2(1), Art. M2, http://jrp.icaap.org/content/v2.1/temple.html.
Thorne, Sally (1994). Secondary analysis in qualitative research: Issues and implications. In Janice M. Morse (Ed.), Critical issues in qualitative research methods (pp.263-279). Thousand Oaks, CA: Sage.
Twine, Frances Winddance (2000). Racial ideologies and racial methodologies. In Frances Twine & Jonathan Warren (Eds.), Racing research, researching race: Methodological dilemmas in critical race studies (pp.1-34). New York: New York, University Press.
Van den Berg, Harry (2005). Reanalyzing qualitative interviews from different angles: The risk of decontextualisation and other problems of sharing qualitative data. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(1), Art. 30, http://www.qualitative-research.net/fqs-texte/1-5/05-1-30-e.htm.
Ventuti, Lawrence (1995). The translator's invisibility: A history of translation. London: Routledge.
Venuti, Lawrence (1998). The scandals of translation: Towards an ethics of difference. London: Routledge.
Venuti, Lawrence (2000/1993). The translation studies reader. London: Routledge.
Wadensjo, Cecilia (1998). Interpreting as interaction. London: Longman.
Bogusia TEMPLE is Professor of Health and Social Care Research in the Department of Social Work at the University of Central Lancashire. Her interests include narrative approaches to research and methodological issues in cross language research.
Department of Social Work
University of Central Lancashire
Preston PR1 2HE, UK
Rosalind EDWARDS is Professor in Social Policy and Director of the Families & Social Capital ESRC Research Group at London South Bank University. Her research interests mainly focus on family policy and family life, especially in relation to gender, class and ethnicity, and she also has an interest in methodology.
London South Bank University
103 Borough Road
London SE1 0AA, UK
Claire ALEXANDER is Senior Lecturer in the Department of Sociology at the London School of Economics. Her research interests are on race, ethnicity and identity in Britain, particularly in relation to masculinity and youth. As an ethnographer, she also has a research and teaching interest in ethnographic methodology, particularly in relation to race.
Department of Sociology
London School of Economics
London WC2A 2AE, UK
Temple, Bogusia; Edwards, Rosalind & Alexander, Claire (2006). Grasping at Context: Cross Language Qualitative Research as Secondary Qualitative Data Analysis [46 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 7(4), Art. 10, http://nbn-resolving.de/urn:nbn:de:0114-fqs0604107.