1. Introduction
Discussions are increasingly taking place on how to research and assess the impact of transitional justice (Baxter Reference Baxter2002; Van der Merwe et al. 2009). Ethnographic techniques are often used for individual case studies (see for example Baines Reference Baines2007) and community studies (see for example Hamber and Kelly Reference Hamber and Kelly2005). Nationwide opinion and attitude surveys (see for example Gibson Reference Gibson2006)Footnote 1 or cross-national comparisons (see for example Olsen et al. Reference Olsen, Payne and Reiter2010) of a number of variables provide a quantified insight into the nature and impact of transitional justice processes. Seeking to draw on these varied epistemologies, and the value that each can offer to our understanding of transitional justice, this paper explains the reflective and adaptive nature of the research process adopted in the study of the Rwandan gacaca courts and its impact on society.Footnote 2
The modernized gacaca courts were conceived at the end of the 1990s and implemented nationwide between 2005 and 2012 to deal with the legacy of mass violence in Rwandan society. The 1994 genocide was noteworthy, not only because thousands of people were killed – primarily ethnic Tutsi – but especially due to the high involvement of ordinary civilians, mostly ethnic Hutu, in the killings and looting of their neighbors. This was one of the motivations to initiate a grassroots transitional justice process through the gacaca court system. This court system was loosely modelled on a conflict resolution mechanism that had historically existed in Rwandan society.Footnote 3 A key characteristic of the “new” gacaca was the decentralization of the judicial process to the lowest administrative levels of society. Approximately 14,000 courts presided over by lay judges dealt with 1,958,634 cases of alleged involvement in the 1994 genocide. However, such facts of process and procedure can tell us only part of the story of gacaca. A first generation of academic studies on gacaca primarily focuses on this dimension of Rwanda’s grassroots transitional justice process.Footnote 4 The objective of my study was to understand how the gacaca process functioned in practice and how it affected the lives of the Rwandans who practiced gacaca. In doing so, my research objectives were similar to a second generation of gacaca studies.Footnote 5 The actual methodological approach, the research techniques used, and the interpretative process might vary in each of these studies, but the central concern is the actual practice of the modernized gacaca courts.
The objective of this article is not to discuss in detail the nature and outcome of Rwanda’s grassroots transitional justice process. Instead, the following sections describe the main contours of my research approach. In particular, I will discuss two main research principles guiding my research activities: immersion and iteration. In doing so, I will refer to the gradual adjustment of fieldwork activities, research techniques, and the nature of the involvement of Rwandan collaborators and interpreters in order to demonstrate how the navigation of the field of study not only generates knowledge on the topic under study—what some have called metadata—but equally allows for the progressive embedding of the research approach in the very field of study. A recurring theme throughout the discussion is the need to establish and monitor social relationships and increase trust with research participants in an environment of suspicion. Mass violence leaves deep scars on the social tissue. Distrust and suspicion are often pervasive and might be enhanced by the operation of a transitional justice process, as seen in the Rwandan case (King Reference King2009; Begley Reference Begley2009; Fujii Reference Fujii2010; Ingelaere Reference Ingelaere2010, Thomson Reference Thomson2010; Chakravarty Reference Chakravarty2012). Hence prudence needs to guide the research process at all levels: presence in the field, data collection as well as interpretation.
Overall, this paper does not make a claim regarding the methodological superiority of my or anyone’s particular approach in the study of Rwanda’s transitional justice process but highlights how reflexivity results in a research approach and researcher attitude that pragmatically takes into account the characteristics of the research terrain and worldview of its inhabitants. In doing so, the paper moves beyond the Rwandan case by highlighting how prudence and practical knowledge can help to structure research approaches generating knowledge after atrocity. The conclusion elaborates on the worldview underlying such a research approach.
2. Immersion
The notion of immersion occupies an important place in the description of ethnographic and anthropological approaches to research (Olivier de Sardan 2008, 51; Emerson et al. Reference Emerson, Fretz and Shaw1995, 2). “Immersion in ethnographic research, then, involves both being with other people to see how they respond to events as they happen and experiencing for oneself these events and the circumstances that give rise to them” (Emerson et al. Reference Emerson, Fretz and Shaw1995, 2). Immersion is closely related to the notion of (participant) observation. However, while the latter is a research technique, the former is, in my interpretation, a particular approach, a guiding principle structuring a research project. Immersion implies a continued and long-term engagement with the research environment in situ, thus, even when not in the process of doing research and when there is no intention to take notes of the experienced or observed events. It is the development of a sort of tacit knowledge, where the “interpretation of a given situation becomes almost a reflex” (Olivier de Sardan n.d.; 2008, 53) that results in a visible difference in the works of a “fieldworker who calls on lived experiences (through immersion) and an armchair researcher working on the basis of data collected by others” (Olivier de Sardan n.d.; 2008, 54). It is evident that this type of knowledge can only be the result of extended and repeated stays in the field.
It is important to stipulate that although the crucial aspect of immersion does characterize my research approach, I have never attempted to become, or considered myself to have become, a Rwandan among Rwandans. I still do not know what it feels like and how it is to have your family exterminated or to be in a Rwandan prison for decades, and I do not know how it feels to personally appear in a gacaca court as a plaintiff, defendant, witness, or judge. In the field, I was aware of the fact that I had an international passport, a credit card, and a plane ticket in my pocket (or at hand). But I attempted and progressively managed to bracket (not erase) these conditions and move closer to these practices and experiences. Indeed, I am confident enough to say that if required, I could now emulate the logics of certain behavioural practices I studied. As a consequence, I would be able to pass “a test that some ethnographers aspire to [is] ‘if you think you understand the X then you should be able to act like the X’” (Olivier de Sardan 2008, 103). Hence the choice of the title of this paper: “Learning ‘To Be’ Kinyarwanda.” In fact, “knowing Kinyarwanda” (kumenya Kinyarwanda) means two things (Nkusi Reference Nkusi1987, 85).Footnote 6 On the one hand, this expression refers to the ability to speak Kinyarwanda, the language of Rwanda, by being familiar with syntax, words, etc. On the other hand, “knowing Kinyarwanda” also connotes familiarity with the local customs, the established practices among Rwandans, and also how language is used, so when and how to speak, remain silent, etc.
A dimension of “knowing Kinyarwanda” (kumenya Kinyarwanda) thus relates to the form of the verbal and non-verbal interactions between people: the way things go on between Rwandans. This type of knowledge, contextual understanding through immersion, allowed me to navigate the field, an issue of the utmost importance considering the nature of the overall research environment. I provide an example of such a navigation exercise.
During one of my research stays in Ntabona, a hill located in the north of the former province of Gitarama, I learned that the former mayor of the area had returned from the Democratic Republic of Congo (DRC) since my last stay on the hill. The man had been mayor for decades before the 1994 genocide. Given the fact that I had been trying to understand the history of the locality in order to situate the ongoing gacaca practice in the longue durée of the locality, I considered it important to meet him. However, I realized that he was without doubt suspicious of any unannounced visit to his place by people unknown to him and eager to discuss sensitive issues such as the unfolding of the local genocide and the ongoing gacaca activities. Maybe he would accept a visit, but that would not necessarily mean he would say anything genuine. The man had lived as a refugee in the DRC for over a decade, which probably meant he had been unable or afraid to return to Rwanda. I had learned that he was not accused in the local gacaca trials. His reluctance to return to Rwanda could have been based on fear of the reigning Rwandan Patriotic Front (RPF), not the fear of being judged for crimes committed. The fact that he returned did not necessarily mean that such a fear had subsided.
This assumption was based on the fact that, while navigating the field, it had come to my attention that feelings and impressions of being under constant surveillance were widespread in the population, especially among people who had occupied important positions in the former Hutu regimes, even at the local level.Footnote 7 To give just one example, in June 2006, I went to visit the former prefect (governor) of a province where I was doing research in order to learn more about the history of the region. The man had occupied the position during the reign of the first Hutu president, Kayibanda, following the so-called “Hutu social revolution” that established independence and abolished the monarchy. After we talked for some time inside his house, he said goodbye and suggested we could meet again in Europe, where he was planning to visit relatives soon. He stated: “It’s better to talk outside Rwanda on these issues because here even the walls have ears.”
By observing Rwandans, I had learned that strangers would look for “linkages” to assess and qualify their social relationship. Therefore, I looked for a go-between: someone who trusted me and who had a good relationship with the former mayor. Through key informants in the area, relationships that had been built over the years during several visits, I learned whom the man frequently interacted with, and whom he apparently had warm relationships with. I was given a few names. One of them turned out to be one of the priests of the parish where I was lodging during my stay on the hill. Since I had returned several times to these locations, I had developed a good relationship with the priests of the parish. One night, during a dinner with the priests, I mentioned I had learned that this former mayor had returned. I framed my interest in the man in the overall and rather neutral discourse I used for my research and explained that I would find it interesting to come into contact with the former mayor. The priest who frequently visited the latter suggested that he might mention my presence at the parish and in the sector during an upcoming visit. He would mention that I had stayed many times in the parish and had become familiar with the local inhabitants of the area. I had no doubt he would to give a favourable impression of my character and doings. Indeed, the next week, I was told to pay the former mayor a visit. I did so together with a translator. The interaction did indeed provide me with some insightful information. Does it mean that my interlocutor interacted with me without hesitation? I cannot be absolutely sure. But I have no doubt it would have been different if I had not been first introduced by a “trusted friend.”
And, in any case, leaving aside what the former mayor had actually told me, this example demonstrates that part of the knowledge of the nature of social relations and experience with transitional justice processes resides in the mere experience of navigating the field. It was by observing Rwandans construct linkages among trusted nodal points in the web of social relations and reflecting on how I started to emulate these practices that I became more clearly aware of the nature of the social landscape, of how relationships were being mended or not in the aftermath of mass violence. Other researchers, also in the Rwandan case, have referred to such a process as the generation of metadata that constitute valuable knowledge complementary to the actual data gathering (King Reference King2009; Fujii Reference Fujii2010). However, a researcher taking immersion as a structuring principle of his or her research approach will consider these insights as data as such, essential rather than accidental to the research process. Moreover, my argument here is that such an understanding contributes to the methodological operationalization of the research activities.
As the previous example demonstrates, I was accompanied by Rwandan field assistants and translators. My collaborators were chosen based primarily on their ability to interact with the rural population and their skill in the navigation of “the field,” as explained above. I valued these skills as well as the capacity to reside in rural communities over any formal education. They were even more important with respect to issues of translation and interpretation. This brings me to the other meaning of “knowing Kinyarwanda” (kumenya Kinyarwanda): the use of language. I cannot say that I speak Kinyarwanda fluently, in the linguistic or grammatical sense, although I am familiar with common expressions and words. I am also able to identify the topic of a conversation, especially when it is related to the research theme under investigation. I have only a basic understanding of Kinyarwanda, since I have been working on a sequence of research projects, each of limited duration and with no guarantees of future funding. For these very practical reasons, I was always of the opinion that it was not useful to invest in an in-depth study of the language. More importantly, I realized that Kinyarwanda is such a complicated language, both grammatically and as used, that I would need (a) translator(s) anyway to guarantee the quality of my understanding of my interlocutors. The latter aspect made me reluctant to invest more in my understanding of the grammar of the language. This allowed me the opportunity to focus more on that other dimension of speaking Kinyarwanda, discussed above: how things go on among Rwandans, as captured in the previous example of how to access former dignitaries.
Did the fact that I had only a basic knowledge of the local language constitute an obstacle? Not necessarily: Svensker Finnström (2003, 33–34) regrets having limited knowledge of the local language of the people he studied. He, nevertheless, produced an insightful and widely praised ethnography evoking local understandings and practices of dealing with past violence. A research approach characterized by immersion, and with the objective of unpacking local understandings, does, however, need to pay sufficient attention to language, especially in the Rwandan context. Rwandans tend to speak in proverbs and images. Moreover, phrases in Kinyarwanda can have multiple meanings. Therefore, I attached particular importance to the translation process. For example, often, and always in group discussions, I used two Rwandan translators during interviews. One of them would translate in order for me to understand responses in situ, and the other would take notes in Kinyarwanda or French, depending on the type of interview. In order to avoid suspicion and a reservation in response, I did not use tape recorders. A third collaborator, based in the capital Kigali, would then later type up, annotate, and, if needed, translate the interviews. Phrases with multiple meanings were then jointly discussed. I will return to the role of translators and collaborators in the following section.
3. An Iterative Research Process
Iteration is another principle underlying my study of the Rwandan gacaca courts. I use the notion in three ways. First, it refers to a successive movement to and from the physical setting of the field, namely Rwanda and the Rwandan hills. Second, it means rotating through multiple locations in the field. And third, iteration refers to a psychological movement: the intellectual reflection on the “field,” namely the topic under study and how it is being studied.Footnote 8 Both dynamics, physical and reflective, had the objective—through the very nature of the process of iteration—to progressively embed the research strategy (as well as the researcher) in the field of study.
First, regarding the physical dimension of iteration, my study of the gacaca courts is based on approximately thirty-two months that I have spent in Rwanda since 2004. I made over ten return visits to Rwanda. The bulk of that time was spent in rural areas. When in Rwanda, often for extended periods of six months, I also moved frequently between these rural localities. Danielle de Lame (2005, 25), studying one hill in the period preceding the genocide, asserts that in-depth research will enable the researcher to understand the specificities of the site under investigation, but that a nearby site, another hill, will always remain strange and unapprehended. Iteration, therefore, also refers to the physical movement between different research locations.
By iterating through different research locations inside Rwanda, the objective was to understand both the breadth and the depth of processes (Barron et al. 2004). Any researcher is faced with two extremes in methodological approaches. On the one hand, large n-studies, typical nationwide surveys, can establish the breadth of a study. By collecting data through survey questionnaires distributed to randomly selected respondents in randomly selected communities, one can infer statistically sound conclusions based on a significant part of the population that are, therefore, representative of the population as a whole. Every social setting is marked by idiosyncrasies. The use of large-scale surveys avoids those idiosyncrasies by reducing the complexity of reality and producing universally valid predictions and statements. But predictions are not explanations, and although predictions and statements are valid for a large population, the data are not rich in detail because they are mostly quantified and collected on the basis of concepts drawn up beforehand.
On the other hand, ethnographic research generates information that is very rich in detail and emic conceptions, and gives insights into why and how events happen and processes take place. Ethnographic approaches are able to identify underlying patterns and themes that will not easily surface by using questionnaires: they are well suited to understanding issues of process. This approach can also identify social categories that remain invisible and themes that “fall through” the tight grid of preconceived questions and already-coded answering possibilities that are common in survey research. However, the question of representativeness remains. Why should findings gathered in one place, albeit rich in detail, be valid for a larger population or even for another, similar, nearby place?
These two approaches—broadly summarized as survey versus ethnographic research—do not have to be as mutually exclusive as they are presented to be above. Integration of both viewpoints is possible and results in a better understanding of the topic in question. The idea is to use the best of both worlds and avoid the weaknesses of each of them. Iterating through a multitude of research locations inside Rwanda helped to establish both the breadth and the depth of the study.
Figure 1 below gives an overview of the interlocking aspects of the depth and the breadth of my research approach. As highlighted in the graph below, I used a sort of pyramidal approach in the oscillation between breadth and depth. I have an ethnographic understanding of four sites since I lived there for longer periods; they were studied in-depth on the basis of longstanding presence in the field. In these four sites I used multiple research techniques and followed all the gacaca activities.Footnote 9 In addition, however, I collected a substantial amount of data, including two rounds of life-story interviewing with a representative sample of the population and repeated gacaca observations. Sporadic gacaca observations took place in a wider range of “sites” (eleven additional locations in different regions of Rwanda). In these sites I used multiple research techniques ranging from surveys to formal/informal interviews, (focus) group discussions (FGD), and observations.
Figure 1 Study Breadth and Depth
Source: Adapted from Barron et al. (2004)
These research locations were purposively chosen, but this choice aimed at maximizing variance. Maximizing variance (Rao and Woolcock 2003) on specific variables (regional, conflict dynamic, historical bases of power, etc.) was the criterion for the selection of provinces and communities: it helped to sharpen patterns, made recurring themes emerge, and established findings significant for a wide range of environments.
By going back and forth between Rwanda and Belgium as well as by going back and forth between Kigali and the rural research locations under study, I attempted to deepen my understanding of the topic under investigation through a continued follow-up on the changing dynamics. In addition, I also attempted to increase trust between myself and the inhabitants living in these locations. I returned many times to the same research locations and to the same people to conduct interviews, have conversations, or simply spend some time. In doing so, I was deepening my relationship with these locations and these people. Multiple return visits to the same location showed them that there was no harm in sharing thoughts and time with me. However, I never considered this process established or taken for granted: I was always prudent, careful not to disturb my relationship with these localities and its inhabitants.
I attempted to increase trust not only through iteration. Whenever possible, I resided on the spot or very near to the hill, most often in parishes. While residing on site or nearby I took part in local life to the extent it was possible. Most importantly: I continuously walked around, interacting and engaging in small talk. I also went to the market, frequented local bars (cabarets), played soccer and volleyball with young people in the local playgrounds, etc. I brought volleyballs or footballs with me several times. I did so in an effort to give something to “the community,” since I was always clear that I would not give any reward, financial or otherwise, to individuals. Giving to individuals could potentially create envy in the population and pollute the voluntary character of the interactions. In order to give back to the community as a whole, I started a small library on one of the hills where I lived for several months. I collected French and English books at home and shipped them to Rwanda during my next return visit. A library service was initiated together with the local schoolteachers.
Iteration not only refers to the physical movement of continuous return and rotations within the field, it also evokes the reflective and adaptive nature of the research process. The research process was divided into several phases, of which some were more open-ended and exploratory. During these phases, I operated in what can be called an “unstructured” research mode. Other phases were more focused and characterized by the use of more structured research techniques.
At times, I alternated periods of focused qualitative/unstructured research (mainly informal/formal interviews and observations in the field) with periods featuring more structured/quantitative techniques such as survey questionnaires. Finally, I also quantified (coded) qualitative data to bring more structure to apparently unstructured data with the objective of allowing more systematic analysis. The treatment of the transcripts of gacaca observations is a case in point. I highlight some important features, and in doing so, I return to the theme of practical knowledge and prudence as well as a discussion of the involvement of Rwandan collaborators in the research process.
From 2005 onwards and with the introduction of the gacaca court system nationwide, a total of 1,917 trials were observed dealing with allegations against 2,573 individuals. When I was residing in Rwanda, I made observations with the assistance of a translator, so that I would be able to follow the proceedings. Another field assistant would record the proceedings verbatim in Kinyarwanda. They were later translated and electronically captured by a third assistant. As with interviews, the third assistant would also highlight specific phrases in Kinyarwanda that were subject to multiple interpretations. These were later discussed with several collaborators to identify the intended meaning in the context of the trial. This type of procedure was, again, informed by my previous navigation of the field and awareness of the need to proceed carefully in the translation process as highlighted previously.
When I was not present in Rwanda, a Rwandan field assistant went to the research locations on the day the gacaca trials were taking place. These assistants resided in Kigali but travelled to the areas under observation when gacaca activities were scheduled. They wrote down every word spoken during the trials in Kinyarwanda. In addition, they took note of important non-verbal interactions during the trials. They also added a field report to every observation of a gacaca session that detailed relevant information on events, rumours, and social dynamics observed or established through informal interactions with the inhabitants in the community during that particular day of observation.
In addition to the two Rwandan field assistants that travelled each time to one of the research sites, I had personally contacted one or more inhabitant(s) of the research sites to request extensive summaries of the ongoing gacaca activities. Following the immersion and iteration principles discussed previously, I had already spent several months in each of the research locations before the systematic observations started. I was thus able to identify trusted and able assistants belonging to the community. Most were young educated people and thus not implicated in the genocide. None were members of the local administration. They wrote observation and research reports on the gacaca activities and provided information on the dynamics surrounding the gacaca trials. None of these people were personally known to the research assistants travelling to the research sites from Kigali. On the one hand, this established a control mechanism through which the observation activities of the research assistants could be verified. On the other hand, it was a security measure to guarantee the quality of the observations when I was not present on site. Also, by using an “insider” and an “outsider” to the community, the research reports could provide complementary information, at least with respect to the social dynamics surrounding the gacaca activities.
4. Conclusion
Overall and in conclusion, the research approach discussed in this article calls attention to the importance of a “phronesis-like” approach when generating knowledge after atrocity. Immersion and iteration help to develop something similar to what Aristotle calls phronesis in his philosophical works. Phronesis is similar to practical wisdom but also has the connotation of prudence. In the aftermath of mass violence and in times of transitional justice, the “field” is riddled with a range of obstacles that demand a practical understanding of navigating the terrain. Rwandans behave with prudence due to the experience of mass violence and atrocity in their midst. This is no different in other situations after large-scale violence and abuse. In fact, I learned to behave in such a way by observing Rwandans’ behaviour. Therefore, this article calls attention to the fact that such an understanding creates knowledge, not only regarding the topic under study but equally regarding the research process as such: the nature of the study of that topic.
Bent Flyvbjerg (Reference Flyvbjerg2001) elaborated on the Aristotelian concept of phronesis in a broader argument on the characteristics of the social sciences and in an attempt to make social science matter (again).Footnote 10 Important in a phronesis-like approach is an intimate familiarity with contextualized settings. Generally speaking, the awareness of being situated in an environment of suspicion as well as knowledge-producing power relations (and power-producing knowledge relations) in the field, informed the phronesis-like stance adopted and the research principles and techniques that flowed from it. It solicited an overall “heightened awareness in data collection” (Flyvbjerg Reference Flyvbjerg2001, 158). This article has highlighted the importance of such a reflective process through which the mere navigation of the field not only generates knowledge on the topic under study but equally allows the researcher to pragmatically embed the research approach in the field of study.
Adopting such an approach evidently comes with a particular worldview, a worldview that is not shared by all researchers and differs from other approaches. By worldview I mean “how we view the world and, thus, go about conducting research” (Creswell and Clark Reference Creswell and Plano Clark2007, 21). This worldview refers to a set of beliefs or assumptions about the reality of phenomena and how they can be explored and understood. In fact, the phronesis-like character of the research process stresses a pragmatic stance—which I use in the common-sense meaning—in that respect.Footnote 11 A pragmatic stance or worldview places the research question at the centre of the study and values both subjective and objective knowledge. It considers reality (ontology) to be both singular and plural and results in a “pragmatically governed interpretation of studied practices” (Flyvbjerg Reference Flyvbjerg2001, 140). It implies an anti-foundational as well as an anti-relativist position, with regards both to the position of the researcher in the research process as well as to the “nature” of the object under study. The situational position of the researcher is inevitable and should not be denied (positivist attitude), nor should it be exalted (subjectivist attitude) (Olivier de Sardan 2008, 97).
Furthermore, the knowledge and validity claims resulting from this phronesis-like approach imply that the findings and interpretations emerging from such a study are “no final truth” (Flyvbjerg Reference Flyvbjerg2001, 139). This applies to my study of the Rwandan gacaca courts as well. However, this does not mean that the interpretation emerging from such an approach—for instance, my study of the gacaca courts—is only as valid as any other interpretation. That would imply a relativist stance that is equally rejected. I consider the insights flowing from a reflective and adaptive research process, as discussed here, necessary pieces of the puzzle in the study of how people deal with past violence. I do so since such a research process results in the adoption of a basic characteristic of the field of study as its structuring device: prudence as practical wisdom.
