“Taking Care” in Intercultural Research

Abstract

Methodological and ethical aspects of intercultural research are frequently discussed in the literature. However, rarely are detailed examples given or practical suggestions offered, particularly in relation to qualitative enquiry. Drawing on global health qualitative research by an international team with indigenous women in Guatemala about access to family planning, this article highlights consequences of different research designs and implementation strategies and practices. We used the constant comparison method for analysis and developed a code for portions of interviews or content which might have been omitted had the research been conducted differently. These applied examples are used to illustrate the gaps and misinterpretations possible in intercultural research and how critical it is to involve a local team early and throughout the study in such stages as: preparing research instruments, recruitment, and conducting interviews; multilingual interviewing, transcription, and team analysis; and reporting and dissemination. International research has been likened to an extractive industry due to the propensity of scholars to conduct research then publish only in English, essentially extracting knowledge in a way that is inaccessible to those in the country where research was conducted. Practical and ethical implications are highlighted for those conducting, funding, and reviewing intercultural research, to ensure that research does not become the latest extractive industry.

With the recent increase in global health research (Glanz, Rimer, & Viswanath, 2008; Glickman et al., 2009; Montaño & Kasprzyk, 2008), it is common for researchers to investigate topics in intercultural settings, both in their home countries and abroad. Although findings from this research abound, there is a relative paucity of literature providing concrete and detailed descriptions of intercultural methods (Shah, 2004; Wong & Poon, 2010). Grounded in our experiences facilitating a global project with indigenous women about family planning in Guatemala, this article discusses the methodological and ethical implications associated with various research designs and methods, particularly in international, intercultural, and inter-linguistic settings.

Indigenous women in Guatemala use family planning far less (28%) than their Ladina peers (54%), who speak Spanish and are of “mixed Spanish and indigenous heritage” (Ishida, Stupp, Turcios-Ruiz, William, & Espinoza, 2012, p. 99; Ministerio de Salud Publica y Asistencia Social, 2009). Our research attempts to better understand the reasons behind these differences in relation to this particular context. Family planning is a taboo topic in Guatemala, making elements of cultural sensitivity even more important (Chirix García, 2003; Pick et al., 2008).

Culture “includes all the characteristic activities and interests of a people” (Eliot, 2010, p. 14). Cross-cultural psychologists use culture as “an overarching label for a set of contextual variables (political, social, historical, ecological, etc.) that are thought by the researcher to be theoretically linked to the development and display of a particular behavior” (Marshall, Walter, & John, 1998, p. 1105). Indigenous researchers have long recognized that failure to understand differences in culture may jeopardize research in terms of both quality and ethics (National Health and Medical Research Council [NHMRC], 2003). Contrary to many research traditions that consider culture as a barrier, this research follows indigenous methodologies in treating culture as an integral part of methodology (Smith, 1999). Culture is therefore treated as a factor to be thought about and considered explicitly and reflexively in research (Smith, 1999).

Critical research paradigms and associated methodologies, such as indigenous methodologies, community-based participatory research (CBPR), and participatory action research, have made significant contributions to understanding how local participation in research from inception to application is important, particularly with marginalized populations (Horowitz, Robinson, & Seifer, 2009; Smith, 1999; Wallerstein & Duran, 2010). Indigenous researchers have aptly criticized the dominance of Western approaches, particularly in research with indigenous people, which has largely been a tool of colonialism (Kovach, 2010; Smith, 1999). Indeed, “having been immersed in the Western academy which claims theory as thoroughly Western…, Indigenous voices have been overwhelmingly silenced” (Smith, 1999, p. 29). Indigenous methodologies have been characterized as congruent with other relational qualitative approaches, such as participatory action research, which value research processes as much as content (Kovach, 2010). CBPR “is an orientation to research that focuses on relationships between academic and community partners, with principles of co-learning, mutual benefit, and long-term commitment and incorporates community theories, participation, and practices into the research efforts” (Wallerstein & Duran, 2006, p. 312). In keeping with CBPR, we worked with marginalized populations to define and carry out the study. Within this article, we consider questions and challenges we encountered around ethical research processes and dissemination.

Existing literature in both quantitative and qualitative research provides lessons and advice about the conduct of intercultural research. On the quantitative side, for example, the burgeoning field of cross-cultural psychology grapples largely with how to test and validate universal psychological constructs and theories (Berry & Dasen, 1974, as quoted in Marshall et al., 1998). The Handbook of Cross-Cultural Psychology features language and interpretation mainly as possible sources of bias, with little practical advice about how to reduce these biases beyond using the committee approach and adapting current instruments (Vijver, 2001). The approaches suggested by cross-cultural psychology to reduce bias have merit in all intercultural research; however, a detailed discussion of how to achieve accuracy in communication, translation, and interpretation is generally lacking. Within this more positivist framing of intercultural research, which ultimately seeks to uncover “universal truths,” the role of the interpreter and translator is reduced to relative invisibility. The interpreter or translator is understood to help achieve equivalence, but their potential role in later analysis and interpretation of data is omitted entirely.

Several, mostly qualitative, researchers have attempted to address this gap by showing the crucial interpretive role that translators play and highlighting special considerations in intercultural research (Berman & Tyyskä, 2011; Björk Brämberg & Dahlberg, 2013; Easterby-Smith & Malina, 1999; Esposito, 2001; Gesink, Rink, Montgomery-Andersen, Mulvad, & Koch, 2010; Lê, 2008; Shah, 2004; Temple, 2002; Wong & Poon, 2010). They assert that translation is not a neutral technical process, rather it involves meaning-making and is influenced by relative power imbalances (McGloin & Carlson, 2013; Orlove, 2002; Stanner, 2011; Sutton, 2001). Decisions about “terms to describe the world” can be “a site of struggle,” and subtle concepts may “suffer badly by translations into our dry and abstract concepts” (McGloin & Carlson, 2013, p. 3; Stanner, 2011, p. 57). Responsible decisions about how best to translate key concepts from an indigenous language may represent painstaking processes of interpretation, trial and error, such as that described by Orlove (2002) in his quest to translate “fisherman” from Quechua to English. However, these few examples from indigenous research and anthropology are exceptional in their practicality and detail about struggles with language. Most qualitative literature about the ever-growing intercultural research focuses mainly on problems encountered rather than strategies for mitigating these complicated issues in translation and interpretation (Björk Brämberg & Dahlberg, 2013; Esposito, 2001, p. 577). Only a few notable exceptions contribute needed practical advice (Gesink et al., 2010; Wong & Poon, 2010).

In this article, we discuss ways that intercultural local research teams might be formed and employed in line with the conceptualization of interpreters as valued members of the research team. The role of the translator is elevated to have considerable influence and improve reliability, validity, and accuracy. Whereas we acknowledge the sources of bias and misinterpretation of concern to cultural psychologists and more interpretivist qualitative researchers alike, we provide more practical advice about how to approach intercultural research and minimize communication problems, while improving effectiveness and validity even when three languages are involved. The crucial role of the interpreter is acknowledged and described across the research process. The findings presented here were generated through behavioral science global health research with indigenous women in Guatemala about family planning, but the reflections are relevant across disciplines, to those participating in research in a culture other than their own. First, an overview of the study is presented, then the article describes methodological and ethical considerations in the research process, introducing issues in the order they would likely arise in the research process. Finally, further discussion places these findings in the context of the literature and highlights recommendations, limitations, and conclusions.

Overview of Study

This article draws on critical social science qualitative research about family planning with indigenous women in Guatemala to show consequences of different decisions in the practice of intercultural research at various stages. The substantive findings of this behavioral science research, which applies social cognitive theory and a political economy framework, are discussed elsewhere (Richardson, Allison, Gesink & Berry, 2016). This article focuses on the methodological and ethical implications. Ethics approval was obtained for this research at the University of Toronto in Canada (protocol ID 28766) and the University del Valle in Guatemala.

Elicitation interviews were conducted with 16 indigenous married women, aged 20–24, from rural districts of Patzún, Chimaltenango in Highland Guatemala, to better understand the particular barriers to accessing and using family planning in this context. Following methodology consistent with social cognitive theory, a purposive sample ensured that half the participants used family planning while half did not. The sample therefore included married women: from two rural districts in Patzún, Chimaltenango, Guatemala, who were early in their reproductive lives (aged 20–24), and who either used or did not use family planning (Richardson, Allison, Gesink & Berry, 2016). Elicitation interviews ask open-ended questions, so that rich qualitative data may be shared by participants about the barriers they face to the health behavior under study (Middlestadt, Bhattacharyya, Rosenbaum, Fishbein, & Shepherd, 1996; Montaño & Kasprzyk, 2008). The elicitation interview guide asked: general demographic information; ideas about, knowledge of, and use of contraception; as well as diverse open-ended questions to explore impressions of barriers and facilitators to family planning (Richardson, Allison, Gesink & Berry, 2016). Written informed consent was obtained from participants prior to beginning interviews. Care was taken to not inadvertently disclose a woman’s family planning status to her peers or family in the course of the recruitment or interviewing.

During development and implementation of the research study, careful notes were recorded by the first author about key methodological decisions and their possible advantages and disadvantages, including practical and ethical implications. The first author attempted to be reflective about her own possibilities and constraints as a Canadian researcher who is bilingual in Spanish with experience working and living in Guatemala, but whose first language is English. Language considerations were especially important because the first language of participants was also not Spanish, but rather the local Mayan language, Kaqchikel. The first author sought advice from a Guatemalan academic, who specializes in research with various indigenous groups, about interviewing, translation, and transcription considerations in multilingual research (Alburez, 2013). Reflective approaches were also called for because the first author is non-Mayan and nonindigenous and “there is a danger of invalid results due to inaccurate assumptions and interpretation of data from a non-Aboriginal standpoint” (‘Yotti’Kingsley, Phillips, Townsend, & Henderson-Wilson, 2010, pp. 3–4). This article also draws on consultations the first author had with her Canadian-based doctoral committee (who are also nonindigenous), Guatemalan host supervisor, and other local and international scholars of Guatemala.

The constant comparison method was used during analysis, which generates and plausibly suggests “many categories, properties, and hypotheses about general problems” (Glaser & Strauss, 1967, p. 104). The first author read through verbatim transcripts of the interviews and created an initial set of codes that were shared and revised by the local research team in Guatemala, and then later with the doctoral committee in Canada.

Team analysis meetings involved the first author, research coordinator, research assistant, and transcription specialist. These workshops emphasized interpreting Kaqchikel segments of interviews as well as creating, reviewing, and adjusting coding schemes, all in the most participatory and varied way possible. For example, team members listened to portions of audio and read transcribed text from Kaqchikel portions of interviews, then independently recorded their translation/interpretation of participants’ comments in Spanish. Each team member then shared their translations and similarities and differences between interpretations were discussed. At the first workshop, the team organized Kaqchikel quotes into groups of similar potential codes, then named and defined them. This served as an exercise to confirm codes identified by the first author and create new lines of inquiry, such as tracking cases of laughter and interpreting these in light of local cultural context. Moreover, it helped to introduce the methodology of coding to members of the team who had not previously been involved in this aspect of qualitative research. In the second workshop, the team grouped codes into larger themes, in some cases confirming previous groupings envisioned by the first author and in other cases creating new ways to look at the data in terms of interpreting access and barriers to family planning. These team analysis processes led to new insights about data provided in individual interviews which enhanced, validated, and improved reliability of the initial analysis conducted by the first author and are described in select examples below.

A code was developed for portions of interviews or content which might have been omitted had different decisions been taken regarding how the research was conducted. Thus, this article uses these applied examples as much as possible to illustrate the potential gaps and misinterpretations associated with decisions relating to early involvement of a local team in preparing research instruments, recruitment, and conducting interviews; multilingual interviewing, transcription, and team analysis; and inclusive as well as ethical reporting and dissemination.

The research processes we reflect on and suggest in this article are consistent with ethical guidelines from the fields of indigenous research and anthropology (American Anthropology Association, 2012; Association of Social Anthropologists of the UK and the Commonwealth, 1999; Canadian Institutes of Health, 2007; Laycock, 2011; NHMRC, 2003, 2010; Shannon, Shibasaki, & Committee, 2005). Indigenous research is pragmatic with the intention of benefiting the community first (Laycock, 2011). Indigenous research is guided by six values: spirit and integrity, reciprocity, respect, equality, survival and protection, and responsibility (Laycock, 2011; NHMRC, 2003). We suggest research processes that align with expressions of these ethical values such as recognizing “the contribution of others” and “minimizing difference blindness” to show respect (NHMRC, 2003, p. 11, p. 15), ensuring members of the research community are employed and develop skills to embody reciprocity (Canadian Institutes of Health, 2007; Laycock, 2011; NHMRC, 2003, 2010), and conducting research that seeks “to advance the elimination of inequalities” as an expression of equality (NHMRC, 2003, p. 14).

Aspects of Data Collection—Local Research Team Incorporated Early

Because of the sensitive nature of family planning in rural, indigenous Guatemala, it was not at all clear at the outset of this research that young, married, indigenous women (aged 20–24) would be interested in sharing their impressions about the taboo topic of family planning in interviews with a Western (Canadian) researcher who is clearly not from their own community. Several factors helped to mitigate this divide, including working with the support of local nongovernmental organizations and hiring a local research team.

Recruitment and Trust: “Getting In”

One important decision was whether to hire a research assistant directly from the rural district where interviews were to be carried out, versus a neighboring rural district. When the research assistant is from the exact community where interviews are planned, ease of recruiting must be weighed against concerns of confidentiality. In this case, emphasis was placed on confidentiality and anonymity during the training of the research assistant, such that even though she would hear responses of women from her own community, she would understand the serious implications of sharing any of their responses outside of the research analysis process. Her lack of previous research experience made it all the more important to carefully explain the aims and methods of the study, to include the research assistant as much as possible in data collection and analysis, and to build local research capacity. Thirteen of the 15 potential participants from the research assistant’s rural district, El Llano, agreed to participate. This relatively high recruitment success allowed us to be more confident that we were capturing the diversity of potential participants from this community.

At certain points in the interviews, the research assistant was able to put participants at ease by setting up a private channel of communication between them. For example, when one woman broke into laughter in the middle of her response about the advantages of family planning, the research assistant then said in Kaqchikel “sina’ij pa qach’ab’älja re’ k’omodo” (you can also respond in our language). In this case, the woman responded with laughter, “we manäqchqa pa qach’ab’äl” (I don’t have the answer in our language either), showing that lack of being able to express herself in Spanish was not the issue in this case. She went on to eventually answer the question in Spanish. The topic of family planning is considered taboo in Guatemala in general and within these indigenous communities in particular, and conversations about sexual relations are even less common than conversations about family planning (Chirix García, 2003; García Chirix, 2009; Pick et al., 2008). It is therefore notable that among only four cases of “having relations” (sexual relations) being mentioned across interviews, one participant switches into Kaqchikel to do so.

Interview Guide and “Getting On”

The interview guide for this research was translated from English to Spanish by the first author, who is fluent in Spanish and has previous work experience living and working in rural Guatemala. However, an important step was to review the Spanish version of the interview guide with the local research team, in this case, the research assistant and a research coordinator. Both these local team members are Kaqchikel women in their 20s from Patzún, Chimaltenango, the municipality where the study was carried out. The local research assistant was 23 years old and had graduated from tercer básico, equivalent to ninth grade. This level of education was above the norm in her community but did not socially distance her unduly from her peers. Because she is from Patzún, Chimaltenango, where the study was planned, she wears the typical traje (long Mayan skirt) of this region and speaks the local Mayan language, Kaqchikel. The research coordinator was a consultant in her final year of an undergraduate degree in social work, with prior experience working with adolescents and young women from rural Guatemala in sexual and reproductive health programming, research, and evaluation. Together, the research assistant and research coordinator reviewed the Spanish interview guide, with particular attention to making the questions as simple as possible, and incorporating local parlance.

An important decision relevant in many multilingual research settings relates to how to proceed with the interviews themselves. Had the first author exclusively carried out the interviews, some participants who were not always comfortable expressing themselves in Spanish would have been inhibited. The opposite option of translating the interview guide to Kaqchikel and having the research assistant conduct interviews in this native language would have the advantage of linguistic and cultural closeness, but the disadvantage that the first author would have been excluded from the interview, including crafting follow-up questions to the semi-structured guide. Even though the research assistant received some training about conducting research, she could not be expected to take the lead on asking follow-up questions, particularly in the initial interviews when she was new to the whole endeavor. The first author solicited advice from a Guatemalan academic (Alburez, personal communication, May 10, 2013) with experience conducting research across Mayan languages, who warned against pursuing the alternative of the local research assistant simultaneously translating all Kaqchikel interview responses to Spanish, as this might cut the flow of the interview too much. A compromise was struck and the interview guide was translated into Kaqchikel, but the initial language of interviews was Spanish, with the research assistant translating into Kaqchikel or back into Spanish when necessary. Determining when to pause the interview to allow for translation, and when to take the lead, depended on the flow of the interview and nonverbal cues, such as eye contact or lack thereof. For example, if the participant looked confused or diverted her eyes while remaining silent, the principal investigator would ask the research assistant to repeat the question in Kaqchikel. The translations of responses from Kaqchikel to Spanish were brief summaries, to facilitate formulation of appropriate follow-up questions without overly sacrificing the flow of the interview, but the richness of original responses in Kaqchikel would be captured through detailed transcription, as discussed below.

A corollary enterprise related to this was the translation of the interview guide into Kaqchikel. The research team considered the merits of hiring a professional translator, but the local research coordinator explained that the Kaqchikel spoken in research communities was actually a hybrid using many local expressions and a mix of Spanish words: Participants might not understand a translation into pure formal Kaqchikel. Instead, the local research assistant produced a Kaqchikel version of the interview guide based on how the language was spoken in her community, which was adapted after back translation into Spanish by the research coordinator, to ensure original meanings were captured.

Beyond “getting in” to the interview, which is concerned more with “gaining access” to begin the interview, there are tacit elements necessary for “getting on” or achieving “social access” in the interview, only some of which can be prepared for in advance (Shah, 2004). The social closeness of the research assistant helped to “get on” in interviews, not only translating between languages but also helping to interpret responses in a fluid way to keep the interview on track and not miss important concepts. For example, the first time a participant mentioned the concept of “being a woman,” the first author took this statement too literally. The research assistant was able to provide a more accurate interpretation which facilitated understanding of this important concept of “being a woman” as related to having and maintaining many children, an important theme across interviews:

Participant:

My sister in law well…she plans [uses family planning]; she knows how this is done and because she starts to tell others, people say that she is not a woman, because she can’t, she can’t take care, that she can’t have children, and like that.

First author:

And she’s not a woman?

Research assistant:

Let’s say, she’s not a woman because she doesn’t know how to maintain many children.

The research assistant’s presence was also critical for building trust throughout the interview, which proved essential as many women were initially vague about the situations they faced. For example, one woman revealed only slowly that her husband was in fact not in favor of using family planning. When first asked if her husband supported family planning, she said “el está de acuerdo, está de acuerdo” (He agrees, he agrees). Later, she reveals that her husband “casi no está de acuerdo” (almost isn’t in agreement; about using family planning). Finally, toward the end of the interview, the woman acknowledges, “no va a estar de acuerdo” (He won’t agree; to using family planning). An outsider might cut or end the interview at the wrong time, not allowing enough time for trust to be developed and for women to slowly reveal the obstacles they face in accessing and using family planning. Or, without the presence of a local peer in the interview, sufficient trust might never be established for women to share more in detail about their situations.

One participant specifically commented at the end of the interview about how she appreciated the research assistant’s presence and contrasted this with a previous interview experience:

Because you [the research assistant] are here, if we don’t understand in Spanish, you can tell us in Kaqchikel. In contrast, [in the case when she was previously interviewed by two Canadian researchers], one could not understand their Spanish and didn’t understand what they said. Maybe they felt upset because one only looked at them. One does not understand nor do they understand. It’s hard.

This example not only shows the utility of having interviewers fluent in Spanish and Kaqchikel but also highlights ethical concerns with a less linguistically proficient team, whereby participants might be left feeling frustrated or inadequate.

Aspects of Analysis—Multilingual Transcription

Many participants availed themselves of the opportunity to respond in Kaqchikel, so it was important to handle analysis of the data in a way that respected the richness of their responses, whether in Spanish or in their native Mayan language. An important consequence of this related to transcription. Often in multilingual research, interviews are transcribed by professional transcribers directly from the native language of participants to a language familiar to the researcher, and the roles of the translator and transcriber are kept in the background (Temple, 2002). In this case, this would have meant transcription by a professional who understands Kaqchikel and is able to translate into Spanish, and he or she listening to audio of interviews in Spanish and Kaqchikel but transcribing only in Spanish. However, such a process precludes effectively looking for patterns in Kaqchikel, which might be lost in translation (Alburez, personal communication, May 10, 2013). Although more involved, transcription for this research was carried out by a professional linguist who is fluent in Kaqchikel and Spanish. He transcribed the interviews verbatim, including all parties’ comments in their original languages, then provided his detailed translation into Spanish of any original responses in Kaqchikel and advice about language patterns to consider during analysis.

Team Analysis

A team analysis methodology was developed to help with interpretation of the interviews, with emphasis on segments in Kaqchikel, the native Mayan language, as the first author is fluent in Spanish but does not speak Kaqchikel. This approach was consistent with CBPR and the philosophy guiding this research, including involving all members of the research team as extensively and interactively as possible in the analysis process. As discussed above, all three members of the local research team were bilingual in Kaqchikel and Spanish, and each brought a unique perspective to the analysis process. For example, the transcriber’s professional background as a linguist, with considerable experience in social science research, helped with the identification of important language patterns in participants’ responses. The research coordinator was able to look for particular themes in the data, representative of accumulated knowledge from working in issues of sexual and reproductive health with young, indigenous Guatemalan women over several years. The research assistant, being from the rural district where research was carried out, was able to confirm the local meaning of specific words and felt a responsibility to capture the experience of participants conveyed through the interviews where she had been present. The first author framed areas of exploration based on knowledge of existing literature and theoretical frameworks as well as previous experience in critical social science research.

Multiple versus singular meanings—“Taking care” to interpret appropriately

The word “cuidar” or “take care” came up often in the interviews with various meanings, including using natural family planning, using a modern form of family planning, and in reference to looking after oneself and one’s children. During team analysis of one of these latter cases, we examined the comment of a participant in Kaqchikel about the disadvantages of having many young children: “man yajosq’ij ta, man naya’ ta ri atención chke rije’ y na’an chqa disfrutar rat ke re’ at te’ej…” This had been translated by the transcriber into Spanish: “No los mantiene limpios, no les pone atención y disfrutas ser madre,” meaning, in English “You don’t keep them clean, you don’t pay them attention or enjoy being a mother.” However, when we discussed the translations and interpretations of the members of the analysis team, it was explained that Kaqchikel term “yajosq’ij ta” has many meanings in Spanish. The closest meaning in English is “take care of,” but the word in fact has multiple meanings in Spanish including cuidar, limpiar, atender, haciar, proteger, dar de comer, and bañar (take care of, clean, attend to, wash, protect, feed, and bathe). Therefore, when participants comment in Kaqchikel that having many children close together means they cannot “care” for their children, this should be understood as cannot “care/clean/attend to/wash/protect/feed/bathe their children.”

Team analysis revealed another participant’s Kaqchikel comment to also have several meanings. In this case, the participant referred to her mother being supportive of her using family planning, and expressing that previously these kinds of methods were not available to her: “As my mother says…‘I wanted to take care of myself but before there weren’t those things…there was no way to help yourself.” The translations and interpretations of each team member revealed that the Kaqchikel term “manaq modo nato’ awi” means more than “helping” oneself; in fact, this term simultaneously means helping, taking care of, and protecting oneself. The intercultural team analysis process thus helped to unpack language used by participants.

In contrast to the above cases where team analysis highlighted multiple meanings in Kaqchikel, at other times, it was important for confirming the singular meaning of participants’ comments. For example, one participant spoke about a breach in her privacy which occurred when she sought family planning from the local extension health services. The English translation of what she said in Kaqchikel is: “I planned with my daughter, but my sister-in-law saw me one time when I was being injected. She started to talk about me.” When probed later in the interview the participant said her sister-in-law had not in fact seen her directly while she was being injected, but the community health worker who carried out the injection told her sister-in-law. It was therefore critical to return to the transcripts and audio from the interview with the research team to confirm she initially said she had been “seen” by her sister-in-law. All members of the team confirmed that “xirutz’ët” unequivocally means “me vió” or “saw me,” a phrase she repeated 4 times. Although this team analysis was useful for confirming the literal meaning of the participant reporting being seen receiving her injection, it is not possible to be sure whether she had simply misspoken or did not want to impugn the health worker. Both interpretations had to be considered, so we needed to go down two analytical paths with these distinct possibilities always in mind. If she had literally been seen by her sister-in-law, the breach in privacy was more direct. However, even if her sister-in-law was later informed about her using family planning by the community health worker, the participant’s wording and revelation shows the extent of the breach: so directly was this woman’s privacy violated that she expressed being seen by others who were informed of her family planning status. Either way, we were able to draw conclusions about the importance of guarding indigenous women’s privacy and confidentiality in community-level service provision of family planning methods.

Over the course of the two team analysis workshops, each lasting about 5 hr, there were 10 discrepancies in interpretation that arose when examining the transcripts in Kaqchikel and Spanish. In 8 of the 10 of these cases, initial discrepancies in translation and interpretation were resolved through group discussion. This above example was one of the two cases where team analysis could not fully resolve discrepancies in interpretation that were related to language and meaning. The second case came from the next passage in the same interview, where the participant described how others knowing about her family planning status brought on more criticism. The team could not resolve from reading the transcripts if she described this as a reflection she made to herself, or one her husband made to her. We decided to revisit the same passage in the second workshop by listening to the audio transcripts. The pauses in the participant’s comments made it possible for the team to come to a consensus that she was most likely making this reflection to herself. This process shows it is beneficial to conduct analysis in intercultural teams of both multilingual written transcripts and audio recordings of interviews, especially in cases where discrepancies in analysis are not easily resolved.

More accurate meanings in cultural context

Even though transcription was carried out in Kaqchikel and Spanish by a bilingual, professional linguist, the participation of the local research coordinator and research assistant helped to validate his translations into Spanish and, in some cases, make important adjustments, based on their age, gender, cultural, and geographical proximity to participants (the transcriber was also Kaqchikel but from a more distant part of Chimaltenango). For example, the transcription of one participant’s Kaqchikel comment was: “Ri ruventaja k’a ri’, xe ri ta’ij che ri, manlo nk’oje’ chik jun abebé, ja ri’ ri ruventaja ri nuk’äm pe…,” which translated into English means, “Well the advantage [of using family planning] is not having another child; this is the advantage of it brings.” The research coordinator had a different interpretation, which hinged on the Kaqchikel word “manlo.” She explained the woman was likely to have instead said “malon,” which means “delays.” The research assistant was able to confirm through her knowledge of idioms local to her rural district, and her participation in the interview itself, that the participant meant, “The advantage is that the child is delayed.” This enhanced interpretation is critical, particularly because the distinction between unmet need for contraception for spacing and for limiting ¹ depends on this difference between wanting to wait before having more children or not wanting to have any more children at all.

The research assistant provided a subtle yet important alternative translation of another participant’s comment about critiques in the community of women who use family planning. The transcriber’s initial translation from Kaqchikel was: “What I’ve heard is that those who use that [family planning] are not willing to have children.” The research assistant corrected this translation to read: “…are not willing to receive children.” This subtle difference has important connotations in this context where people often feel children are blessings that are received from God. The transcriber’s initial translation made having children seem more of a neutral decision by the individual couple; the interpretation provided by the local research assistant reflects the perception of many participants that children are “received,” and to deny this is to deny God’s blessing. In another case, the research assistant corrected the transcriber’s translation of a participant’s comment about her aunt who was a bit “fat” because of injections rather than a bit “deaf.”

New lines of analysis

Joint analysis of transcriptions and audio from interviews helped to identify unusual interactions that might not have been noticed by the first author who comes from outside this cultural context. For example, one participant commented that her mother had been instrumental in speaking with her husband and convincing him about the merits of family planning. During the first team analysis workshop, the transcriber, a man in his 40s from the same Kaqchikel cultural group, noted how this interaction was exceptional in a culture where most direct communication is along gendered and family lines. The fact that a mother-in-law spoke with someone of opposite gender, who was not a blood relative, caught his attention. This observation led the first author to pursue a new line of analysis, where the patterns of communication about family planning were tracked across interviews, ultimately confirming the exceptionality of this case where the participant’s mother spoke to her son-in-law about family planning.

Worldview of interpreter affects analysis

It is important to note that although the research assistant, research coordinator, and the professional linguist and transcriber share Kaqchikel culture, their interpretations were still coloured by other aspects of their background, beliefs, and epistemological positioning. During the second analysis workshop, an illuminating discussion transpired about a participant’s comment in relation to the “wantedness” of her child. The participant had said in Kaqchikel that her baby had come “choj k’ate,” meaning “de repente” in Spanish or “all of a sudden” in English. We discussed how according to definitions in national surveys, this response might trigger the participant to be included in the group with an “unmet need for family planning,” as her child could be construed as unwanted or wanted later. However, the professional linguist and transcriber believed these are Western categories that do not organically exist in his Kaqchikel culture and that all children are in fact “wanted,” probably reflecting a view widely held by middle-aged Kaqchikel men. The research coordinator then contested the implications of his reasoning: that all children are “wanted.” She maintained that if all children are “wanted” in the Kaqchikel community, then there is no such thing as an “unwanted” child. However, she felt she had seen cases where children of Kaqchikel families were clearly “unwanted.”

This example also highlights where this dichotomous framework of “wanted” or “unwanted” children is problematic (Ampofo, 2004). In fact, there are many shades of gray between and the inclusion of an intermediary category might help to reduce an exaggerated assessment of “unwantedness.” One might know one should “want” children because this is what the culture dictates, but struggling against this are aspects of the personal or family situation.

Timing of Translation to English

Because of the many layers of language involved in this research, it was not obvious at which point the first author should translate into her native language of English. This was necessary to receive feedback from some coauthors who do not speak Spanish. However, translating to English too early would take the analysis farther away from Spanish, the language in which the majority of most interviews was conducted and could preclude the identification of language patterns visible only in Spanish. A balance was struck where initial codes were developed in Spanish and validated with the Guatemalan team. The coding scheme was translated to English for the purpose of sharing with coauthors, along with key illustrative quotes. Original quotes in Spanish were referenced in initial article drafts such that bilingual English–Spanish speaking coauthors could suggest further improvements in translation and/or interpretation in both languages. Advantages were therefore garnered by working and analyzing in the original language of the interviews for as long as possible.

Ethical Aspects of Reporting and Dissemination

Following through on responsible and inclusive cultural and language considerations has implications for reporting and dissemination of research results (Canadian Institutes of Health, 2007; Shannon et al., 2005). While attending the Guatemalan Scholars Network Conference in July 2013, the first author heard concerns by many “Guatemalanists,” both Guatemalan and international, about academia becoming the latest extractive industry. This analogy to mining alludes to the propensity, for international scholars in particular, to conduct research in Guatemala, extract knowledge, then publish only in English, without adequately considering the resulting inaccessibility of findings to participants and colleagues who could most benefit from the research in the country of origin. This extractive industry analogy runs parallel and is consistent with the indigenous critique of research as colonialism in general (Kovach, 2010; Smith, 1999). After this reflection, the first author’s resolve was cemented to ensure at least one of the publications resulting from this research was in the Spanish language, and thus more accessible to Guatemalan scholars and policy makers.

Authorship of articles is a related concern. To recognize the contributions of the local research team and partners, it is important to integrate coauthors as much as possible. Although community research assistants might not have academic affiliations, this should not serve as an impediment (American Anthropology Association, 2012). This article, for example, names the research assistant, research coordinator, and transcription professional as coauthors, recognizing and making more visible their significant contribution to the data collection and analysis processes.

A further question of ethical salience is how to ensure participants of research with marginalized populations are sufficiently informed of research results (NHMRC, 2003), particularly when they would not normally have access to academic publications. A short, accessibly written summary of results was prepared for interview participants and these reports were personally delivered by the research assistant.

Further Discussion

This article contributes to a small but growing literature that probes critical issues in intercultural research across several languages. Several authors point out deficiencies in the standard practice of intercultural research, such as treating interpreters as mere translators between languages, kept in relative invisibility, and not critically examining the role and implications of the outside researcher (Björk Brämberg & Dahlberg, 2013; Lê, 2008; Shah, 2004; Temple, 2002; Wong & Poon, 2010). We confirm the importance of creating intercultural teams in international health research and providing space for the co-construction of interpretation, but we go one step further. By drawing on concrete examples from research about family planning with rural indigenous women in Guatemala, this article adds empirical evidence about the importance of active roles by diverse team members in setting up, conducting, analyzing, and reporting qualitative research.

Adequate translation of interview guides is especially important when particular psychosocial constructs are being probed in intercultural settings (Vijver, 2001). Word-for-word translation and back-translation techniques have been deficient when formal translations neglect to employ local idioms and syntax (Esposito, 2001; Vijver, 2001). Using local informants and “the committee approach” are two strategies suggested for improving comprehensibility of interview guides (Esposito, 2001; Vijver, 2001). For example, Gesink (2010) employed four trilingual Greenlandic translators to translate and back-translate her sexual health survey iteratively and in teams. The spirit of this “committee approach” was achieved in this Guatemalan behavioral science research through early involvement of a local research coordinator and research assistant. Each made suggestions on how to make words and meanings in the interview guide more appropriate and accessible for women in their culture.

The essential participation of the local research assistant in recruitment, establishing rapport, and “getting on” in interviews cannot be overstated, and important decisions were made about how to manage translation during the interview. Aligning gender, ethnicity, age, and background between interpreter and participant has been shown to have significant influence on the research process, not only for gaining access to participants but also for establishing trust and maneuvering during the interview (Björk Brämberg & Dahlberg, 2013; Shah, 2004; Temple, 2002). In a study of Vietnamese participants in Australia, Lê (2008) shows how “trust is difficult to build when the cultural gap between researchers and participants is wide” (p. 7). By having the local research assistant take the lead in recruitment, and participate in all interviews, the cultural divide between the first author and participants was effectively bridged.

Shared experience, such as this alignment between the research assistant and participants with regard to geography, gender, ethnicity, age, and background, is part of what contributes to a common worldview and a more nuanced analysis (Brown & Strega, 2005; Haraway, 1988). Feminist scholars have built the case for situated and embodied knowledges: “Feminist objectivity is about limited location and situated knowledge, not about transcendence and splitting of subject and object” (Haraway, 1988, p. 583). Research must therefore be understood as a process of accessing multiple, complex, contested, and partial perspectives rather than one correct, objective version of events (Brown & Strega, 2005; Haraway, 1988). While the first author’s situated vantage-point as a White, Canadian, middle-class, 37-year-old researcher framed her design of the research process and interpretation of this research data, the cultural divide with young Mayan women participants was somewhat bridged through local research assistance as well as previous work and research by the first author in Guatemala and personal connections including being married to a Guatemalan. The aspiration was to contribute to “partial, locatable, critical knowledges sustaining the possibility of webs of connections called solidarity in politics and shared conversations in epistemology” (Haraway, 1988, p. 584). The first author hoped, therefore, to contribute to the work of social justice researchers who consider representation and collaboration as vitally important, do not gloss over the complexity of working within and between power differentials, and pull back from the margins to the central questions of positionality and interpretation (Brown & Strega, 2005).

Recognized as being advantageous (Easterby-Smith & Malina, 1999; Shah, 2004), the research assistant’s personal contacts no doubt helped to improve recruitment in her own community, where, of the 15 eligible participants who were married women between the ages of 20 and 24, 13 agreed to participate. These women might have felt more comfortable participating in the study because they knew one of the team members already. The way the local research assistant put participants at ease and established a privileged line of communication in their Mayan language created another avenue for establishing trust and expressing themselves about uncomfortable topics. Even during Spanish parts of the interviews, a language in which the first author is fluent and most participants are comfortable speaking, the research assistant was able to periodically provide critical interpretations, because of her cultural closeness with participants. For example, the research assistant helped to interpret and draw attention to the important concept of “being a woman,” assuring we did not miss conceptualizing and asking follow up questions about how the identities of these indigenous women are so often tied closely with having many children.

In regard to how to organize translation from Kaqchikel during interviews, we were faced with the trade-off between real-time complete translation, whereby the first author is “no longer trapped in the past tense…[and] can hear the dialogue as it occurs and intervene to clarify or explain a different area of interest,” and the challenge this can pose to immediacy and the flow of the interview (Alburez, personal communication, May 10, 2013; Björk Brämberg & Dahlberg, 2013; Esposito, 2001, p. 577). We adopted one suggestion for mitigating this loss of openness and immediacy by sitting with the interviewer and informant facing each other, and with the interpreter sitting to one side, “which strengthens the interpreter’s role as a conduit” (Björk Brämberg & Dahlberg, 2013, p. 245). Furthermore, we settled on a compromise, whereby the interpreter provided summaries of women’s responses in Kaqchikel, guarding fluidity more than if she had provided a complete translation, but allowing the more experienced researcher to take the lead in crafting follow-up questions and probes.

Our transcription and analysis process helped to address the concern that “interviews are rarely transcribed in the original language, and possible differences in the meanings of words or concepts across languages vanish into the space between spoken otherness and written sameness” (Temple, 2002). A professional Spanish–Kaqchikel linguist transcribed all interviews verbatim, including original comments in Kaqchikel, accompanied by his translation into Spanish. We ensured economic compensation for all members of the research team, and the transcriptionist in particular, was commensurate with their experience (i.e., “fair return for assistance”; Commonwealth, 1999, p. 4). Potential pitfalls relating to “monopolies” on interpretation, when one person alone translates and interprets data, were mitigated through team analysis processes, including the “articulation of all ideas and subtle meanings that might cause misunderstandings” (Shklarov, 2007, p. 535). By sharing the analysis and interpretation process both with the local research team in Guatemala, and the doctoral committee in Toronto, Canada, “cultural and contextual interpretation” were sought, subtle meaning differences debated (Lê, 2008; Shklarov, 2007, p. 531) and misunderstandings arising from “lack of sharing a frame of reference” (Adelman in Shah, 2004, p. 12) were mitigated. This process was also consistent with guidelines that advise indigenous people to help interpret the meaning of research findings and encourage everyone undertaking indigenous research to make sure the Aboriginal community participates in the interpretation of data to improve accuracy and cultural sensitivity (Canadian Institutes of Health, 2007; article 14; NHMRC, 2010).

Active involvement of the research assistant and research coordinator in interpretation and analysis, with each reading transcriptions of women’s original accounts in Kaqchikel, helped identify several cases where initial translations by the professional linguist did not completely reflect the meaning conveyed by participants, illustrating how their even greater closeness with participants, in gender and age for example, allows them to better understand nuances in their responses (Shah, 2004). This process allowed for detailed analysis by the research team of all aspects of the interview and respected the richness of data provided in the native language of participants. Furthermore, when discrepancies in interpretation arose, the research assistant and translator could be consulted about her interpretation based on being present in the interview itself. This allowed her to help transmit what women said in the interviews, a desire she herself expressed during a team analysis workshop. Because of an awareness of the potential negative implications of power differences in the research team (Easterby-Smith & Malina, 1999), the team analysis meetings were organized to be the least hierarchical possible, emphasizing that each member of the team brought a unique perspective as a function of their relationship to the research question, the participants and their epistemological and cultural backgrounds. There is no doubt the first author’s identity as outside researcher and employer in this case would have coloured relationships with the research coordinator, research assistant, and transcriptionist. For example, my identity as outside researcher and doctoral candidate would inevitably shape what I presented as initial analysis ideas. As the first author of the articles from the research, I must recognize that my version of the analysis, albeit revised, was the one that would eventually be written. I did, however, build into the analysis process as much scrutiny and as many opportunities for feedback as possible with the local research team. Collaborative approaches were encouraged both in formal discussions about the data and in informal meal sharing and breaks during the analysis workshop days. Minimizing and resisting hierarchical relations is consistent with the value of respect integral to indigenous research approaches, as is the recognition of contributions for all members of the research team (NHMRC, 2003).

The inexact correspondence of expressions between languages complicates translation in intercultural research (Lê, 2008). Team analysis in this research evidenced that Kaqchikel expressions used by participants sometimes had several simultaneous meanings in Spanish and English. Furthermore, discerning each team member’s “position in relation to the research” or “intellectual biographies” (Temple, 2002, p. 847) aided in understanding how divergent interpretations of women’s accounts were reached. For example, worldviews of research team members affected whether or not they think the concept of unmet need for family planning was valid, their interpretation of interview data and the conclusions they drew. The linguist and transcriber, based on his understanding of his own Mayan culture, and as a middle-aged man, gave his opinion that the concepts of unmet need for family planning did not apply in his culture. The research coordinator, a younger professional woman from the same Kaqchikel culture, challenged this opinion based on her own experience. Without an approach that elevates the role and “intellectual biographies” (Temple, 2002, p. 847) of team members, these multiple meanings would have been lost, interpretation too narrow, and cognizance of multiple interpretations negligible.

Most critical literature about intercultural research, while extending from methodological to analytical considerations, falls short on following the implications through to dissemination and reporting. Researchers are often therefore complicit in the ethically questionable practice where “there is often no end product in any other language than English and so no visible sign of the interpreter’s presence” (Temple, 2002, p. 844). Such research practices have been likened to an extractive industry, such as mining, which exports valuable resources from a country and leaves behind little benefit. We suggest the minimum standard of communication be providing an accessible summary of results for participants and publishing at least one article in the national language where research is conducted. This recommendation is consistent with the value of reciprocity from indigenous research, which “implies inclusion and means recognising partners’ contributions” (NHMRC, 2003, p. 10). Results would thus be more available to researchers and practitioners in the country where the research took place, rather than only in the principal researcher’s country. The principal researcher would be therefore less guilty of participating in research as the latest extractive industry.

Recommendations

In our research, we adopted several strategies that contributed to a more effective and ethical research practice and produced a more representative analysis than most other approaches to intercultural research. Regarding setting up the research, we recommend: working with local organizations; hiring a research assistant from the specific community where most of the study will be conducted; having locals review final versions of the interview guide to ensure simplicity of questions using local parlance; and conducting interviews in the language most comfortable for participants, allowing for but minimizing interruptions for simultaneous translation. At the analysis stage we recommend: transcribing and looking for patterns in the original languages of interviews; having intercultural team members independently review transcripts and audio recordings from interviews, then discuss multiple and single meanings of words in key passages; and encouraging suggestions for new lines of analysis in cultural context, while remaining mindful that the worldview of team members affects interpretation. As part of research dissemination, we believe it is important to publish at least one article in the local language of the country where research was conducted, include community researchers as coauthors, and provide research participants with a simple, accessible summary of results. Finally, funders and academics institutions should provide additional funds and allow more time for intercultural teams to engage more thoroughly in each of these research processes.

These recommendations are consistent with existing guidelines though they indicate that legal compliance alone has not historically been enough to avoid “lack of communication and infringement of deeply held values arising from cross-cultural insensitivity” in indigenous research (NHMRC, 2003, p. 4). Our findings help show: how guidelines may be implemented in practice, difficult decisions that need to be considered along the way, and examples of oversights and misinterpretations that might otherwise occur.

Limitations

One limitation of this study is that a member of the research team, the research assistant, was local to the community. This may have introduced bias in the responses of participants, although there is no evidence of this and the benefits of having a local research assistant appear to outweigh any potential bias. Another main limitation of this study is that advantages and disadvantages of the research design and methodological decisions are discussed based on hypothetical as opposed to empirical comparisons. Mixed with issues of intercultural and interlinguistic research is the inevitable oversimplicity of some questions and the possibility that some respondents might want to keep some things to themselves. In single culture studies, many of these same issues arise and might be resolved through reinterviewing; however, these standard issues interact with intercultural ones in complicated ways. Whereas this article represents an empirical advancement compared to commentaries about problems in intercultural research which do not discuss potential solutions (Temple, 2002), there is room for more systematic inquiry into these issues. Future research could contrast, for example, results from the same study using different approaches, with various levels of integration and participation of intercultural team members in preparing, conducting, analyzing, and disseminating results. Furthermore, intercultural research with larger budgets could extend the findings presented here and present new suggestions for intercultural analysis, such as employing a local research analyst to participate in coding in the local language, thus allowing for measurement of inter-rater (and to some extent intercultural) reliability. This study was also time-limited, preventing more reciprocal reporting and dissemination. While a Spanish version of research findings is being prepared for publication, it must be recognized that most university-educated people in Guatemala are nonindigenous. Furthermore, the written summary that was shared with participants may not have been the most meaningful way to communicate research results. We would therefore recommend that intercultural researchers engage in consultations about the most appropriate and useful research products for the communities involved in such studies. As recommended in indigenous research ethical guidelines, “researchers should not make the publication of research findings a greater priority than feedback to the community in an appropriate or understandable way” (NHMRC, 2003, p. 12).

Conclusion

An overview of important findings from this research has been provided, along with implications for researchers, ethics boards, funding bodies, and publishers. Translators and interpreters are important members of intercultural research teams. Recognizing and funding them, along with allowing them sufficient time (and credit) for their role in recruitment, interviewing, and analysis, will strengthen the quality of intercultural research. Careful and pragmatic multilingual translation, transcription, and team analysis can improve the reliability and validity of research results. Thus, funders, ethics boards, and publishers need to take into account, and require, important details concerning translation (Esposito, 2001) and intercultural interpretation. The impact of research will be strengthened if all parties consider the ethical advantages of publishing in the national language of the country where research is conducted. Taking care in intercultural research, therefore, has multidimensional implications across the research process and for multiple stakeholders.