Knowing Violence

When violence emerged as an issue to be taken seriously by anthropologists, it was often in a reaction to the violence that happened in the communities where they were already busy studying other things (Colson 2007: 2–3). In the words of E. Valentine Daniel: “The responsibility of writing an anthropology of violence pierced like a shriek in the dark, my world of other occupations” (Daniel 2000: 334). The strength of these early anthropological works on violence was their focus on the everyday experience of living through/with violence (Das et al. 2000; Feldman 1995; Nordstrom and Robben 1995). These scholars also questioned assumptions about the distinction between war and civil violence and between times of war and peace (Das et al. 2000: 16). These studies have shown that violence is not an end point that needs explaining. Rather, violence is itself generative of meanings and worlds and should be studied on a par with other social relations. Arjun Appadurai's thought provoking article “Dead Certainty” argues that there is even a kind of intimacy between perpetrator and victim in the brutal “vivisectionist” violence of the Rwandan genocide (1999).

There have been a number of attempts to define and categorize different types of violence and how they interact. Nancy Scheper-Hughes's famous ethnography of mothering and child death in Brazil emphasizes everyday violence (1993); the kind of faceless violence that creeps into the everyday lives of the poor. Elizabeth Povinelli works with a similar understanding when she explores suffering in late liberal society as “ordinary, chronic, acute, and cruddy, rather than catastrophic, eventful and sublime” (2008: 511) and she compares the Schmittean lethality of “state killing” with the more amorphous Foucauldian condition of “letting die” (2008: 511). In a retrospective article on his own ethnographic trajectory from the war zones of El Salvador to drug dealers in East Harlem, Philippe Bourgois sketches out the four most common understandings of violence as they are conceptualized in social science; namely political violence, Johan Galtung's structural violence, Pierre Bourdieu's symbolic violence, and Nancy Scheper-Hughes's everyday violence (Bourgois 2001: 7–8). Interestingly, despite the somewhat schematic typology, the main argument of his article disrupts the distinctions, as he realizes that his earlier strong distinction between the political violence of the guerrillas in El Salvador and the everyday, structural violence in the ghetto cannot be so easily sustained.

While these different types of violence interact and overlap, I focus in this article on violent events where physical violence occurs outside of the ordinary similar to what Bourgois terms “political violence.” I am aware that these events are related to structural violence and feed into everyday violence, just as they certainly may be triggered by and in turn trigger symbolic violence. However, such events shake the everydayness of the other types of violence—the uneventful, ongoing slow violence that Povinelli describes, for instance (2008). Furthermore, the Burundians with whom I interacted, also distinguished between such events and other kinds of violence. In emic terms, violence could be “open” or “closed”—the former when people were attacked and killed in broad daylight, the latter when enemies hid their true intentions and “killed in silence.” Similarly, they would often talk about themselves being “blind” to the social injustices and about “opening their eyes” and seeing the real violence. These distinctions between different kinds of violence are important for my analysis, as they reveal the ways those affected by violence explore and explain the violence they experience.

There is a tension in our understanding of violence between the socially embedded and everyday aspects of violence and the world-shattering and exceptional aspects of violence. We witness this tension in emic conceptualizations of violence, when Burundians talk about violence as “closed” and “open” and we witness it in academic understandings of violence. On the one hand, anthropologists argue that we must move away from pathologizing violence as a symptom of something “gone wrong” in the social. On the other hand, we try to come to terms with the fact that violence is somehow out of the ordinary. Concurrently, we can identify the anthropological insistence that we must understand violence as integral to the historical and social context.

We also cannot deny that violence creates trauma—irrespective of context. A way to dig into this is to explore the tension between theories inspired by Freudian, universalist notions of trauma and more constructivist approaches to violence and memory. In their edited volume on memory after violence, Nicolas Argenti and Katharina Schramm criticize the Freudian-inspired and Eurocentric universalist model of PTSD (Posttraumatic stress disorder). “We might not be able to take the aetiology of post-traumatic stress disorder for granted as a universal psycho- neurological syndrome, but rather that this syndrome too must become one of the social phenomena of the culture of modern violence that we analyse” (Argenti and Schramm 2010: 14). Another limitation of the “trauma approach” is that it valorizes narrative remembering as the only “normal” or “healthy” form of memory, while silence by consequence is seen as pathological (Shaw 2010: 255), reproducing what has been called Judeo-Christian truth-telling as a universal model of healing and closure. Meanwhile, anthropologists have shown that violence can be dealt with in many ways in different contexts. It is, for instance, well-known in the African Great Lakes region that silence is a socially constructed and morally accepted way of dealing with large scale violence (Ingelaere 2009; Russell 2019). Argenti and Schramm suggest that “we ought perhaps to pay still more attention to a politics of memory, or in other words to processes of appropriation, conflicting interests and overlapping discourses” (2010: 18, emphasis in original).

In her recent book on violence and memory in Sierra Leone, Mariane Ferme draws on Freud to argue that violence creates traumas that prevent the victim from even recognizing the violence. “(I)t is this temporality of past experience, which is never assimilated consciously, and which returns to haunt the subject, that bears a similar relation to the temporality I observed in wartime Sierra Leone” (Ferme 2018: 2). Violence creates an excess that cannot be assimilated into the present and needs “the work of time” to be recognized (Ferme 2018: 40). She argues that this inability to “know” the violent, traumatic event creates a belatedness, where the violent event might re-emerge and become knowable in a specific context or place (Ferme 2018: 40). Drawing on Reinhart Kosseleck, she uses the concept of “chronotypes” to describe these situations where someone who has not been able to put an experience into words, experiences that a specific place or constellation of material objects brings it back to him/her (Ferme 2018: 13–14).

Perhaps we can reconcile the Freudian argument of researchers such as Ferme with what we may call the more constructivist approach of the majority of anthropologists, mentioned above. By constructivist I simply mean that they argue that the meaning of violence is not universal and needs to be understood in its historical and cultural context. While I agree that there is a risk of universalism in the Freudian approach that can ignore historical context and politics, I also believe that we should accommodate the fact that violent events temporarily destabilize meaning. The creation of memories of violence need not be in the shape of re-enactments of trauma and healing but rather politically and socially contested and context specific. In her study of wartime violence in Sierra Leone, Ferme attempts to remove the trauma from the “pathology of falsehood” as a clinical symptom to a “symptom of history.” In other words, she argues that the trauma is not based in the psyche of the individual but rather it is the impossibility of a history that cannot be fully appropriated by memory and awareness (Ferme 2018: 3). Argenti and Schramm make a similar argument: Despite their constructivist approach, they acknowledge that we should not dismiss the suffering that the trauma approach is attentive to. “(T)here is still a need to question how extreme political violence experienced by individuals, enculturated at the collective level and passed down through generations, and to develop methodologies and theories that can examine this phenomenon without being ethnocentric” (Argenti and Schramm 2010: 19).

In this article, I try to retain this tension between the trauma approach and the constructivist approach to violence. I recognize them both in the emic conceptualizations of violence, and I find the tension productive when studying the temporality of violence and in particular of knowing violence. Thus, wartimes have their own particular temporality and affect the ability to know, witness, and narrate violence. In his article “Mood, Moment and Mind,” Daniel similarly argues that writing about the presence of violence is challenging because it is constantly changing meaning. “Relatively independent of the present, the past and future are easier to fathom because they can be conceptually seized and positioned for a still-life representation, a representation hovered over by the protective shadow of a coherent narrative” (Daniel 2000: 336). Times of violence are best narrated before or after the fact; during the event, they evade narration. In other words, violent events do not necessarily create universal traumas, but they create a presence where the violence is unknowable. My argument is that they break the flow of time, forcing those who are exposed to such events to revise their pasts and their futures. This is what I mean by wartimes making their own temporalities. Through my long-term engagement with the various violent conflicts in Burundi, I explore how violent times are narrated and made knowable, depending on the time of the ethnographic encounter, thereby exploring the temporality of violence.

Revisiting and Revising

My approach to long-term ethnography is inspired by Michael Burawoy's “Revisits: An Outline of a Theory of Reflexive Ethnography” (Burawoy 2003). He argues that we should leave behind the idea of a strict comparative approach; the idea that we can objectively compare visit one, two and three in order to explore the effects of x or y. We can, however, make some kind of diachronic comparison, as long as we do so within a reflexive ethnography, making the ethnographer visible in the equation. That is to say that when we revisit our field site or the field site of another anthropologist, we are comparing not only changes in the field but also in our position and our approach (Burawoy 2003: 650–653). Changes in our relations with the field and changes in the field itself should be taken into account, as we “tack backward and forward” (Geertz 1995 in Burawoy 2003: 645) across our fieldwork over the years.

What I want to add in this article is that studying violence adds a new dimension to this movement back and forth. As mentioned above, the shifting shape of violence forces both the ethnographer and those whom we study to relocate, upsetting the idea of a geographically defined “field.” My movement has thus not just been tacking backward and forward in time but also in and out of the warzone. My research has bounced in and out of the country, at times following individuals, their families, or their networks and at times following themes and ideas among new groups in new settings. This of course reflects a career that has pursued various research interests rather than a planned longitudinal study—but it also reflects the realities of ethnographic fieldwork in violent contexts. Neither my interlocutors nor I were able to remain in the same location for long, as the conflicts evolved, shifted, declined, and escalated in new forms.

As ethnographers, we rarely observe the violence as it happens but explore, instead, how people reconstruct the events retrospectively or how they anticipate violence in the future. There are a number of reasons for this. First, we need to think of our own safety and the safety of our research assistants and our interlocutors. Second, there are practical reasons why we often are out of sync with the events. Conceptualizing the research, applying for funding, and planning fieldwork all take time, meaning that we usually enter the field long after the facts.

While violent events are reconstructed ex post, they may also be anticipated ex ante. Fearing the violence to come is a forceful motivator for action, as people “read” their surroundings for signs of possible violent futures and take their precautions accordingly. They might leave their homes and seek safety elsewhere as a precautionary measure. Or they might turn to violence themselves in order to prevent the violence that they anticipate from others. In other words, what we are studying in our “tacking backward and forward” is itself moving backward and forward over time and space. Our ethnographic visits and revisits might take place at specific times in specific places but often they are not about these specific times and places but about memories of past violence or anticipations of future violence. In this article, I want to explore how long-term ethnographic engagement may contribute to understanding changes in these futures and pasts. Or to put it simply: when we do a single fieldwork without follow up, we can explore the pasts that have been constructed retrospectively, while with revisits we are able to explore how these constructions change and hence also how they are constructed. Likewise, hopes or anxieties about futures also change over time. It is, in other words, not just the present that changes over time, so do pasts and futures. Ethnographic revisits over the years have allowed me to follow such changes in pasts and futures.

Changing Pasts

Violent events call on narratives to suture the symbolic order of society (Argenti and Schramm 2010; Feldman 1995; Malkki 1995; Schmidt and Schröder 2003). Liisa Malkki's study of mythico-histories among Burundian Hutu who had fled the 1972 “selective genocide” (Lemarchand and Martin 1974) is a prime example of this. But what shapes these retrospective narratives about violent events? What role does the positionality of the ethnographer play? And how do we understand the context in which they are told? In the following I make a diachronic comparison between Malkki's fieldwork and my own. When Malkki did fieldwork in Mishamo refugee camp for Burundian Hutu in Tanzania in the 1980s, she was struck by the standardized narratives that she came across again and again, and she famously called them mythico-histories that revealed more about the worldview of the refugees than the events that they had experienced more than a decade earlier—although it was the events that had shaken their worldview and forced them to make up the mythico-histories (Malkki 1995). The mythico-histories linked body-maps of the Tutsi to racialized understandings of Tutsi mentalities and the century-old repression of the Hutu by the Tutsi. When I did fieldwork a decade later in Lukole refugee camp in Tanzania for another generation of Hutu refugees, fleeing a later conflict in another political context, I did not find the same mythico-histories. Not only had the narratives changed, but they were also less standardized and more pragmatic. This did not mean that the refugees did not try to understand ethnicity and the root causes of the conflict. They also had theories about the different international NGOs in the camp, the Tanzanian police, and the Tanzanians living in the neighboring villages. However, these understandings were precarious and always marked by fears that things could change for the worse every day, which resulted in a number of rumors about the international NGOs in the camp (Turner 2004). Rather than standard mythico-histories, shared by all the refugees, I found rumors and conspiracy theories that would come and go and were challenged by other refugees in the camp.

There are different ways to explain the difference between what I was experiencing and what Malkki had encountered. At first, I feared that the fault was mine: I was less of an ethnographer than she, I spent less time in the camp (due to safety regulations, I was not allowed in the camp after dark), and I therefore had not reached the same profound level of rapport. This type of revisit is often used to show that the first ethnographer had been mistaken and that one's own “rapport with the native” is better somehow.¹ I had the opposite sense of being less adept at the art of fieldwork than my predecessor. However, after a while I met a young man, who spilled over with all the stories that Malkki had heard a decade prior to my fieldwork. He talked about ancient myths, about Tutsi kings using the testicles of Hutu for the royal drum and the essential differences between the two races. This made me reconsider. It turned out that the young man had grown up as a child in Mishamo refugee camp where Malkki had done her fieldwork. There must, in other words, be a difference between Mishamo in the 1980s and Lukole in the late 1990s. This made me conclude that “politics matters” and that Malkki had ignored the fact that the radical Hutu party, Palipehutu, was established in Mishamo. When visiting Palipehutu chairman Etienne Karatase at his home in Denmark in 1998, he gave me a publication by Remy Gahutu, the founder of Palipehutu. The publication had been written in Mishamo refugee camp and held exactly the same kind of narratives as the mythico-histories that Malkki had found in the camp. And yet, she saw the mythico-histories as simply emerging from the situation that the refugees were in at the time, without taking the political context of the camp into consideration (Turner 2010). While Mishamo was dominated by one hegemonic narrative, Lukole refugee camp in the 1990s was marked by political rivalry between Palipehutu and CNDD (Conceil National pour la Defénce de la Démocratie), a more moderate party and rebel movement. CNDD was the dominant force in the camp, and whenever I found understandings of the conflict that were similar to the mythico-histories in Mishamo, the narrator would most often be a Palipehutu supporter, as was the case of the young man mentioned above. This political rivalry brought to the light the fact that the narratives were linked to political ideologies and struggles. Without the rivalry, I might have taken the narratives as simply “Hutu” narratives or mythico-histories. In other words, perhaps the difference in theoretical approach of the ethnographer made us see things differently, and perhaps I had an eye for political subjectivities that Malkki did not.

However, I would argue that it was the revisit—the diachronic comparison—that helped me refine and develop my theoretical point. Without her study on mythico-histories, I would perhaps just have taken my findings for granted and not probed them further. As it were, by searching for something that was lacking, I became aware of why it was lacking and why something else was in its stead. Both groups of refugees had experienced profound violent events, and both were constructing ways to deal with these experiences. The narratives they told depended not only on the nature of the violence, nor only on their situation in the camps—although this was indeed important—but also on the political situation in Burundi and their position in relation to that.

The revisit called for exploring the larger political and historical context of the camps. Mishamo and Lukole might both be refugee camps for Hutu who had fled ethno-political violence, but the political situation in Burundi had changed dramatically in the meantime—as had the situation in the African Great Lakes. I have explained elsewhere how the changing political landscape in Burundi (an opening of the political landscape) and in the Great Lakes region (the genocide in Rwanda among other things) had made the racial body maps and the ethno-nationalist standard narratives of the 1980s obsolete (Turner 2010), giving way to competing narratives about the nature of the ethnic conflict in Burundi. Liisa Malkki and I did fieldwork among comparable populations, fleeing similar ethnic tensions, but their way of understanding and explaining the violence they had experienced (or that the Hutu as a group had experienced—many of my interlocutors had not actually experienced direct violence) differed considerably.² Ethnographic revisits may uncover these shifting understandings of the root causes of ethnic conflict where national and international politics play a role in the available narratives.

The methodological lesson we may take from this is that my confrontation with Malkki's different findings led me to reflect on my position in the field as well as the theoretical assumptions that we both brought to the field, in this way sharpening my attention to the political ideologies in the camp. It allowed for a diachronic comparison where both national and regional politics turned out to have had an impact on the ways the refugees related to their pasts. Their pasts changed over time, as the nature of the conflict changed.

In terms of conceptualizing the temporality of violence, we see that violent events call for narratives that retrospectively give meaning to the violence. In order to “know” the violence and to testify to it, the violent events must be recreated through memory in socially meaningful ways. The diachronic comparison between Malkki's work and my own allowed me to explore how such memories are produced and how they relate to the political and social contexts of the narrators.

Unstable Presents

Studying violence as it unfolds is challenging for practical reasons, and serendipity plays a crucial role in doing ethnographies of violent conflict. However, violence also disturbs the communities we study, and if we happen to be present in those situations where violent events occur, we might gain insight into the uncertainties that violence produces. In this section I give some examples of how my fieldwork changed in relation to the temporality of violent events. In both cases, I had not anticipated the events that unfolded in front of me, and while they to varying degrees disturbed my plans, they all gave me an opportunity to follow the lived experience of violence. In both cases, unexpected violent events took me and my interlocutors by surprise, allowing me to study how this surprise was transformed into something knowable. In each case, the shift in the relations to the violent event forced me and my interlocutors to re-orient ourselves and our narratives.

The first example of what might be termed a violent event, even though it did not directly involve killing or physically hurting anyone, was the forced rounding up of refugees living in Tanzanian villages by the Tanzanian police. This occurred unexpectedly while I was doing fieldwork in the refugee camp. In early 1998, the Tanzanian authorities decided to remove all non-citizens who were living in the villages close to the Tanzanian border as part of a general clean-up operation. If one could not provide an ID card, one had the choice between being repatriated to a war-torn Burundi or go to the camp (Human Rights Watch 1999). I first met this group of refugees less than two weeks after they had arrived in the camp. Some had lived in Tanzanian villages for decades and had married Tanzanians. Some had moved back and forth across the border in order to cultivate land on one side and have a supplementary income such as masonry on the other. When I talked to them, they seemed in a state of shock and had no idea why they had ended up in the camp. They had had convivial relationships with their Tanzanian neighbors and were surprised that they had been rounded up like this. When I met them a few months later, however, they had formed explanations and had clear ideas of the injustice that they had been victims of. They were angry with their Tanzanian neighbors who had never actually liked them, they said. The Tanzanians had simply pretended to be generous hosts because they wanted the Burundians to cultivate the bush in this part of the country where the population is scarce. It is well-known, they claimed, that the Burundians are hardworking and better farmers than the Tanzanians. So apart from cultivating uncultivated bushland for the Tanzanians, the Tanzanians could also learn some tricks of the trade from the Burundians. Once they had learned these tricks, they made sure that the authorities came and removed the Burundians. In other words, the Tanzanians had a plan to exploit the Burundians all along, and their friendship had been false and part of the scheme.

This episode tells us something about the need to find reasons behind sudden violent events. It also shows a sense of betrayal when friends turn out to be enemies. In terms of fieldwork, it tells us that timing is essential for understanding the experience of disruption that violence creates. One might argue that any ethnographic field changes constantly, and that any interluctor may say something radically different to the ethnographer within a timespan of two months, depending on the mood they are in. However, situations of violent events mark a radical uncertainty and unpredictability to a much larger degree, and people are busy suturing this broken social fabric. The sudden violent evictions by the police came without prior notice and disrupted the flow of ordinary time by forcing them to pack-up and leave within a day. This is what characterizes the kind of violent event I am exploring in this article. As opposed to structural violence or everyday harassment by the police, events like these are unexpected and outside of the ordinary, thereby disrupting the predictability of the future. The Burundians, who had lived in the villages and from these experiences had planned an immediate future as self-settled refugees in these villages, were forced to re-assess these plans, leaving them unable to predict their immediate futures. Furthermore, the violent events also forced them to revise their pasts. They believed that they knew their neighbors; a knowledge that they had gained from living with them for years. But the violent evictions made them doubt their own knowledge and revise their own experiences. When I met them immediately after the evictions, their knowledge and their faith in their own experiences with and assessment of their Tanzanian neighbors had been challenged. Only after some months did they recreate a new past and an understanding of the violence. The violence created its own temporality—as a kind of disconnected present—with repercussions backward and forward in time.

At another point in my twenty-year ethnographic engagement with Burundi, serendipity gave me the opportunity to explore the sudden outbreak of violent hostilities in the middle of a peace process. In July 2003, I was in Bujumbura studying the peace process and in particular the role of politicians who had returned from the diaspora, when one Monday morning I woke up at around 4 o'clock in the morning to the sound of gunfire. I had heard sporadic gunfire in the Quartiers Nord (poor neighborhoods with a presence of Hutu rebels) in the evenings every now and then. They normally lasted five minutes and were not seen as a danger to the stability of the city or to the peace. However, this time was different, as the sound came from the opposite direction and the shots went on for much longer—joined by heavier artillery. The situation was new to us and hence not easily explained. Could it be a coup d'état? There were several factors that could substantiate this theory. The sound was from the direction of the presidential palace. Also, at this early stage in the transition, the majority of army officers were Tutsi, while the interim president was a Hutu, and the last Hutu president had after all been abducted and killed by Tutsi officers ten years earlier. We grabbed our phones to call friends who lived close to the presidential palace, but the phones were dead. We later learned that the government had ordered the mobile phone companies to close all the cell phone towers to interrupt rebel information channels. Fortunately, I was staying with a friend working for an international NGO who had a radio in the house, and we were able to communicate and gather information.

We later learned that it was rebels from one of the two remaining rebel groups—FNL (Forces Nationales de Libژration)—who were attacking the city from the surrounding hills. Bujumbura lies in a bowl with Lake Tanganyika to one side and surrounded by hills to the other sides. These hills were the stronghold of the FNL. Most of the fighting was taking place in the southern neighborhoods like Musaga but grenades were also launched into the city center, causing a few casualties and damage to random buildings.

What did this experience mean to my ethnographic engagement with the conflict? Most importantly, I learned the unpredictability of violence when it occurs through my own experience. I did not experience physical violence on my own body, but it was an embodied experience, nevertheless. For weeks after my return from fieldwork, I would feel a physical unease at the sound of helicopters or fireworks. Even a lawnmower could trigger this affect. I experienced the need to make sense of the violence in order to know what to do. This is again a concrete, bodily experience, as it had implications for whether I could meet interlocutors, go shopping, and access an Internet cafژ to tell my wife that I was okay. To start with I made sense of it by likening it to the gunfire I had heard earlier in Quartiers Nord. Later I made sense of it by seeing it as a coup d'état. These short-lived explanations were preliminary attempts to understand and “know” the violence, constantly changing as new information was added.

When violent events shatter the symbolic order (Feldman 1995) and unmake worlds (Nordstrom and Robben 1995: 131), rumors may fill the void (Simons 1995) as free-floating interpretations of the situation—what Feldman has called “collective wide-awake dreaming” (Feldman 1995: 230). Rumors also point toward a future and hence are prognostic not in terms of prediction but in terms of a sense of possibility (Feldman 1995: 230).

As the gunfire lulled later that morning, we assumed that the rebels had been beaten, but the shooting started again in the afternoon—this time accompanied by helicopters from the Burundian air force. After two days, the pattern became almost regular, I knew roughly where and when to expect gunshots and helicopter fire, and I could sleep at night and would venture into the city center to buy groceries, check emails, and even get a decent cup of coffee—ignoring the risk of the inefficient grenades that landed in the city center in irregular intervals.

I also learned from friends who were caught in the crossfire what the process of fleeing looks like, as it unfolds. When the phones were working again, I called Didier (a pseudonym) who lived in Musaga. He told me about the shooting in the streets and the heavy presence of government soldiers, and that he was stuck in the house with his wife and their three small children. Naively, I asked why they did not leave to a safer part of town. “We have all our belongings (Nous avons tous nos biens), and they will be looted if we just leave the house,” he replied. Didier was not a wealthy man and he had no stable income. However, he had managed to accumulate some valuable items such as a television, a car, and a refrigerator, and it would take him a long time to accumulate such wealth again. Furthermore, the fighting was not going to last more than one day; it never did, he assured me. In this situation, Didier was trying to read the situation by drawing on similar experiences of violent events in the area, hoping that this would not be different. “We'll wait until tomorrow,” he said. Tomorrow came and the fighting had not ceased. I called him again. This time the children were hiding under the table. “Tomorrow we'll leave, I promise.” When I called him on the third day, he had bribed a soldier to let his wife and children leave. He was in the house with the “houseboy.” I left the country on the Saturday and the fighting ended on the Sunday.

While one might see my experience in Bujumbura as more authentic than interviewing people about their past experiences because I had the rare opportunity to experience violence in its “raw” form in real-time (and even then, I did not actually see anyone get hurt), the experience in many ways only gave limited access to knowledge about the violent event and its consequences. Mariane Ferme talks about belatedness in understanding—or even just accessing—violence (Ferme 2018). Similarly, it was only after a while that the refugees who had been rounded up from the villages to the camp, started to be able to make sense of their traumatic experience. In Bujumbura, violence simply emerged and imposed itself upon us—forcing us to act, despite not knowing fully what was going on.

In that week in July 2003, none of us knew what was going on and even less what would happen in the immediate future—and Didier had to make decisions that weighed the risk of life and death against the risk of losing his possessions and hence his livelihood in this situation of radical uncertainty. I have interviewed numerous Burundians about their flight from conflict. Some had fled general insecurity; others had held positions that meant that they felt particularly targeted, while yet others had fled due to general lack of opportunities. Their stories were, however, always constructed ex post, and however much I dug for the facts in my interviews with these people, they were always told from a place where the narrator knew the outcome both of the violence and flight and of their own actions. Talking to Didier on the phone as violence developed around him—and around me to a lesser degree, as I did not know the direction of the violence either—gave a different picture of the relationship between violent events and the decision to flee. Didier's future plans had been disrupted by the unpredicted violence, leaving him at odds as to how to act in the present. Not only had the violence emerged unexpectedly, but it was also not possible to predict how it would develop. Would it fade away tomorrow? Or was this just the beginning of something much worse? One might say that neither Didier nor I could understand or know what kind of violence it was. Was it the last convulsions of a dying rebel group or was it the beginning of genocidal violence? Over the years, Didier had managed to accumulate an amount of material goods that could provide him and his family access to the middle-class status that he aspired to. Some of these investments—such as the taxi—were not just investments in middle-class lifestyle but also in a future income. The sudden violence in July 2003 threatened this investment in the future. The violent events also threatened the lives and well-being of his children in whom he had also invested love and good education so that they might have better futures than he. In those days, hiding under the table, he was constantly juggling between these potential futures and how to keep them safe while not actually knowing what the immediate future might bring. Rebecca Bryant proposes the term “uncanny present” to refer “to a particular sense of present-ness produced by futures that cannot be anticipated” (2016: 20). In such moments, the present becomes “heavy” and “visceral.” It is as if the present moment is weightier and the decisions we make are more decisive in such situations of crisis where the future cannot be anticipated, she argues. Didier certainly experienced the weight of his decisions while the unknown violence surrounded his house.

Anxious Futures

As the case of Didier shows, violent conflict not only disrupts pasts, but may also disrupt futures and how people imagine their potential futures. In this section, I give two examples of how futures may be imagined in relation to violent events. The first case explores how futures may be imagined in the wake of violence, while the second case explores how violence itself may be an imagined future. In Nairobi in 2004, I followed a group of young men and women who had left the refugee camps in Tanzania in order to seek their fortunes in the city. As refugees, they were concerned that their lives had been set on “standby”; they had been forced to stop their schooling and were worried that their future options were shrinking as they sat in a refugee camp. While the camp provided food, health, and some security and hence predictability, it did not—in their minds—provide opportunities for a better future. The city on the other hand, meant precariousness and uncertainty, but it also provided hope for a better future. In the city they tried to remain unstuck in the present in order to prepare for a better future elsewhere—back home in Burundi or in Europe or North America (Turner 2016). I have been inspired by an increasing body of anthropological inquiry into the concept of hope to understand how these young Burundians positioned themselves in relation to potential futures (Appadurai 2007; Crapanzano 2003; Kleist and Jansen 2016; Miyazaki, 2005; Pedersen 2012). Inspired by Ernst Bloch's ideas of hope as future-oriented and hence indeterminate (Bloch 1986), I see hope as a means to understand and explore how individuals orient themselves toward unknown futures rather than simply build on their pasts. The kind of intransitive hope, as hopefulness or affect (Jansen 2014), that I found in Nairobi was closely linked to being “born again Christians.” The Burundians I met in Nairobi were living a life of extreme poverty and precariousness, exposed to the brutality of local police officers and in their own words “living off miracles” (Turner 2015). In this Christian view, one must not hope for something in particular but simply be hopeful. They saw Nairobi as uncertain and dangerous but also “a bit developed,” and they would suffer in the present, hoping that the future would reward them.

More than ten years later, Burundi was again experiencing violent conflict and new groups were fleeing the country. This time around, many people actually fled Burundi before the violence because they anticipated violence to come, in a sense inverting the sequence of the usual understanding of forced migration, where violence leads to and precedes displacement. Based on past experiences of mass violence, they tried to predict the future outcomes of the present tensions and act accordingly, thus moving back and forth in time between past experiences of ethnocide and potentially violent futures in order to act in the very volatile and unpredictable present.

My recent fieldwork among urban, middle class Tutsi, who fled to Kigali in 2015, tried to capture and grasp how they dealt with, and perceived of, a life that had been interrupted by violence. When I first visited them in Kigali in August 2015, they had only been in the country for a few months and did not see their situation as permanent in any way. Often, nuclear families were split, so that one parent remained in Burundi, earning an income and sending money and food to the other parent who would be in Kigali with their children. They saw their futures back home in the near future, once things had settled down. However, these hopes were accompanied by anxieties, as they also feared that the conflict might spiral into the kind of genocidal violence that they had experienced in 1993 or even worse what had happened to the Tutsi in Rwanda in 1994. This caused me to revise my previous understanding of futures being made through hope. Perhaps potential futures were also assessed through anxiety? Anxiety had, after all, triggered their decision to leave the country in the first place. However, while we may intuitively believe anxiety to be the opposite of hope –seeing negative and positive potentialities respectively—they also have a lot in common and may be seen as each other's flipsides (Turner 2020). Both are oriented toward a future, looking for signs of what the future might bring. While the hopeful looks for signs of improvement, however precarious the present might be, the anxious will fear that it may turn sour, and look for signs of this negative future. In his discussions of anxiety (Angest), Søren Kierkegaard argues that anxiety is a productive emotion that is linked to “freedom's possibility” (Kierkegaard [1855] 1981). In other words, both emotions are the result of an uncertain present and an unknown—and unknowable—future. And while anxiety points to possible negative futures, it also leaves open the possibility that other futures may be possible, and most often the Burundians in Kigali would be constantly oscillating between hope and anxiety. In other words, violence created uncertainty that disrupted futures. However, although violence disrupts the predictability of futures, it also creates cracks and openings that open up for other futures. In these cracks and openings those who have been affected by violent events imagine possible futures through hope and anxiety. In short, the temporality of violence disrupts the flow of time but also creates positive and negative potentialities for alternative futures.

When I visited the same people a year later, some had indeed returned to Burundi, while the majority remained in Kigali where they now felt stuck. Often the spouse in Burundi had joined them because it was too dangerous to live in Burundi with relatives abroad. When I asked them about their hopes for their futures, they often responded that as human beings, there is no hope, while as Christians they must have hope (Turner 2020, 727). While in 2015 they feared that the violence might turn genocidal, they were not certain, leaving a shimmer of hope that the East African Community, the African Union, or the UN might intervene and end the violence. In 2016, they realized that this was not going to happen, and although the violence had not turned into a genocide, there were no signs that it was going to improve either. In this situation, uncertainty had given way to certainty, and anxiety had given way to despair. While both anxiety and hope are about anticipating possible futures in situations marked by uncertainty, despair is the emotion that is left when one cannot imagine a future. These are the situations that Bryant refers to as the “uncanny present” where futures cannot be anticipated (2016:20).

Violent events produce a void in the present, challenging our abilities to understand or even know the present. It also affects our abilities to imagine futures, creating anxieties but also hope as unstable means to propel oneself toward the negative and positive potentialities of unknown futures. My longitudinal ethnography allowed me to explore the ways futures changed in conflict and crisis. In situations of precarity and uncertainty, futures were imagined through hope and anxiety. But when there seemed to be no future, only despair was left. It was my revisit to the field after a year that let me sense this change in “mood” among the Burundians in Kigali, pushing me to re-evaluate my concepts.

Conclusion

Doing ethnography in situations of violent conflict is impossible to plan, as the situation is volatile and may change radically in a short while. How violence is remembered, recounted, explained or anticipated by those who live with it, largely depends on time, and thus for the ethnographer, on timing. Therefore, it is important to work with serendipity and adjust one's plans accordingly. What I have focused on in this article has been the important roles of memories and imagined futures in situations of violence, and how long-term ethnographic engagements may explore how pasts and futures also change over time in these situations.

My ethnography of the refugee camp in Tanzania a decade after Malkki's ethnography of a similar camp—what we might call a revisit albeit not by the same researcher—challenged my preconceived ideas of how the Burundian Hutu refugees would remember the conflict. Rather than the dogmatic, beautiful and coherent “mythico-histories” that she had so vividly recounted (Malkki 1995), I found several competing, incoherent and pragmatic versions of the Burundian past and the causes of the conflict. This comparison between her work and mine was fruitful as it forced me to abandon my initial assumptions and explore the political context of the camp, ultimately revealing the lacunae in Malkki's analysis, while sharpening my own attentiveness to the lack of mythico-histories that put my work into contrast.

Just as violence shakes up pasts, as I showed in the first sections of this article, violence and the uncertainty that it creates, also shake up potential futures. My ethnography of Burundian Hutu, living clandestinely in Nairobi, showed how they strategically used the unpredictability of the city as a means to hope for better futures and act accordingly. Through perseverance and faith, they kept their hope for a better future alive. It was this hope that kept them going despite their dire situation. My recent fieldwork made me revisit my approach to future-making, as anxiety seemed at first to have taken the place of hope. My revisit to the same field a year later made me aware that anxiety might be giving way to despair and a situation where refugees were anxious about their futures and actually could not imagine a future of any kind; a situation where the present became uncanny.

Through long-term ethnography and ethnographic revisits, I have been able to grasp not only changing futures, but also changes in how futures are imagined. They were caused by the changing character of the violent conflict and the changing position of those in exile. Such changes in what Burawoy calls the forces beyond the field have also made me adjust my approaches to and understanding of future-making in precarious situations. Thus, I had to adjust my framework when I discovered that the refugees in Kigali seemed to be more affected by anxiety than hope. And more radically, I found that anxiety gave way to despair, making me try to conceptualize situations where no future is imaginable.

The idea of “cycles of violence,” so often heard in commentaries on the African Great Lakes region, is analytically flawed because it assumes an “essence” of violence—often related to ethnic hatred. Similarly, we cannot see the conflicts in the region as simply “waves” of retaliation, as one group takes power from the other. However, conflicts do happen to come and go in the region. They escalate and de-escalate; they morph and transform—from political to ethnic, from urban to rural, from warfare to genocide. Having a longitudinal engagement with the region has permitted me to explore these changes over time—while they took place. Furthermore, those who live through these conflicts see them as waves and cycles where they understand new events through the lens of previous violence. Hence—as a kind of self-fulfilling prophecy—the conflicts feed into one another and act as fractals of previous conflicts. Perhaps the local understanding of conflicts in Burundi describes this the best: “The conflict is always there. But sometimes it's closed and other times it's open.”

When we ethnographically analyze violence, we are often exploring how violence is remembered and explained retrospectively, or we explore how future violence is imagined and anticipated. Often the memories of past violence feed into the imagined and anticipated violence of the future, folding the past into the future, and both into the present. I started this article with a tension between perceiving violence as socially and culturally embedded in a local context and perceiving violence as events that cause universal trauma. This has consequences for how those who experience violence make sense of this violence and how we as ethnographers make sense of their sense-making (as Geertz would have it [1993]). My long-term engagement has led me to see that violent events do indeed have dramatic effects that change the ways Burundians imagine their pasts and futures. Furthermore, I found that their attempts to understand the violence and to anticipate the future were unstable at the moment of the violent events. However, when I claim that violence radically upsets and destabilizes meaning, I still maintain that the memories, rumors, hopes, anxieties, and despair that emerge are culturally and socially specific. I have argued that violence shatters the social and symbolic order and that people through rumors and other narratives need to re-assemble meaning. Such making sense of violence can take place as memories of past violence, and it can take place as anticipation of future violence, while during the event, meaning is fluid and ungraspable.

Violence creates its own temporality, a temporality that renders past experiences irrelevant and imaginaries of the future impossible. In these situations, the present becomes heavy and visceral, making it difficult to make sense and make decisions. However, the disruptive forces of violent events and their ability to break the flow of time also create cracks and openings where pasts can be remade and futures re-imagined.