African Studies Quarterly

Ethnicity and Burundi’s Refugees

Burundi: Ethnic Conflict and Genocide. Rene Lemarchand, Cambridge University Press, 1995. Pp. 206.

Purity and Exile: Violence, Memory, and National Cosmology Among Hutu Refugees in Tanzania. Liisa Malkki. Chicago: University of Chicago Press, 1995. Pp. 352.

Fear in Bongoland: Burundi Refugees in Urban Tanzania. Marc Sommers.  New York: Berghahn Books, 2001.  Pp. 224



These three books are about people who share a description of themselves as “Burundians,” or “Hutu,” or “Tutsi,” or any number of other identities from Burundi. But as each book makes clear, the ethnic designation an individual uses is only part of the story because what is also significant is who you consider “not Burundian,” or “not Hutu” or “not Tutsi.” Depending on the group of “Burundians” the question is asked of, you get different answers, often in the form of what Liisa Malkki calls “mythico-histories” about who are “we," and who are “they.” In other words, there is a broad range of response to this question depending on where, of whom, and when the question is asked, despite the fact that each question is likely to be rooted in mythologized claims asserting a “pure” Burundian identity. For example, inside Burundi as described by Lemarchand, the social world has been divided (at different times and contexts) into Hutu, Tutsi, Twa and Ganwa; Rwandan and Burundian; Highlander and Lowlander, and northerner and southerner. Among the Burundian communities in Tanzania, divisions have emerged between Hutu and Ha, Burundian and Tanzanian, and a number of other permutations. For example, in the remote Tanzanian refugee camp where Malkki studied, the answer is that there are Tanzanians, Hutu, and a “remembered” Tutsi, while in Kigoma town nearer Burundi, the “we” is vaguely “Burundian,” and the “they” is remarkably unclear. Finally, Marc Sommers, writing of clandestine Burundian communities in Dar Es Salaam, writes of divisions a putative origin in the Burundian highlands and lowlands.

An important element in all three books is that often there is an explicitly hidden or clandestine story, which Burundians relate to each other and interested outsiders. In three of the locations, (i.e., Dar Es Salaam, Mishamo refugee settlement in Tanzania, and in Burundi itself), the story is in fact “explicitly clandestine,” in the sense that the oral accounts they give of themselves are different form the written public history. Indeed the pattern of the three mythico-histories, presented by Lemarchand, Malkki, and Sommers is consistent. The Hutu protagonists have stories emphasizing how they lead their lives in the context of Tutsi (or Ha, or Tanzanian) dominance. Only one group, described by Malkki, the Burundian refugees in Kigoma town, did not have such a story, and in fact identified with the Tanzanians. This paradox—clandestine identities everywhere except Kigoma—is important and a focus of what follows.


From the perspective of sociology, one way to tie together these meticulous ethnographies/histories is a broader theoretical context for understanding ethnic identity, and ethnic inequality. Since all three books explicitly emphasize the role that storytelling plays in defining group boundaries, I will use this opportunity to place the story of Burundi in the context of what Max Weber wrote about “ethnic communities believ[ing] in blood relationships, and exclud[ing] exogamous marriage and social intercourse. Such a caste situation is part of the phenomenon of ‘pariah’ peoples and is found all over the world.” (1946:189)

Weber emphasized that stories define and re-enforce systems of ethnic stratification, particularly when focused by ascribed occupations such as those found in caste systems. Burundi provides a good example of this: Hutu are believed to be farmers, and the Tutsi are believed to be rulers and cattle-herders. In describing their relative positions, Weber says that such pariah groups emphasize the glories of the coming future while dominant groups emphasize the glories of the past, and both create stories and culture to reproduce their particular view (1946: 189-90). The battle songs of the United States’ Civil War illustrate this point nicely: the Southern anthem “Dixie” emphasizes the glories of the dominant slave-owning planter caste, while the music of the dominated/pariah slaves emphasized the future religious redemption in songs like “Swing Low Sweet Chariot,” or even “The Battle Hymn of the Republic.”  According to Weber, to be maintained as plausible the status groups need to continue being the specific bearer of such conventions associated with that status (1946:191). In other words for the group to persist, the convention must also persist; if it does not the status will disappear and the principles preventing exogamous marriage will ultimately dissipate.

I know that in many ways, this point sounds “academic” and “theoretical” but it is an important one for policy-makers seeking to establish peace between dominant and subordinate status groups like Hutu and Tutsi. How refugees come to view themselves, whether as a pariah group or not, matters for what type of future they envision for themselves. Who refugees define as the “other,” and what they define as “home,” affects efforts to have refugees return to their country of origin or resettle as immigrants. Repeatedly in the history of Burundi and Burundian refugees (as well as others), the kaleidescope on these two issues has shifted, as the saliency of one ethnic identity or the other thickens and thins.


From Ganwa King to Colonial Rule

Modern ethnic typologies in both Burundi and Rwanda typically focus on primordial stories defining Hutu and Tutsi as a dichotomy, with an occasional nod to the one percent of the population that is “aboriginal” Twa. A figure indicating that 85 percent or so of the population is Hutu and 14 percent is Tutsi is often used in modern documents. Such simplified typologies admittedly reflect the mythico-histories Burundians tell about themselves and are common in the popular press and policy-making documents. Nevertheless, volumes have been written in response, discrediting this approach. Lemarchand in particular has been among those who aggressively affirm the central point that it is “beliefs” about ancestry that are important, not actual bloodlines.

Rene Lemarchand has been watching Burundi long enough to remember that the “Tutsi” were not always the ruling caste in Burundi. As he writes, until the 1960s at least three identities were relevant in the context of political power: the Tutsi who were powerful in the context of the church and Belgian colonial state, farming Hutu, and the princely “Ganwa” who actually ruled in Burundi. As Weber described, each had a story to tell about itself, justifying its role in the semi-feudal society that existed before Burundi was occupied by the Germans in the 1890s. The stereotypes formed included a princely, tall, and regal Ganwa, who ruled from the royal capital in up-country Gitega; the equally tall and willowy Tutsi who were herders and recent immigrants; and the shorter stockier Hutu clans who were farmers and more ancient arrivals. Powerless, the “pygmoid” Twa were considered to be aboriginal inhabitants specializing in hunting and forest crafts.

Lemarchand tells the story of how each group saw each other without resorting to cliché about primeval origins and notes that the rigid opposition focused on today between Hutu and Tutsi did not emerge in Burundi until the 1960s, irrespective of what modern Hutu or Tutsi nationalists may recall. His point is well-taken; it is the story told which is important for understanding contemporary Burundian society. It is not the positivistic history told by academics, but the stories Burundians tell each other which define and stratify.

From Colonial to Independent Burundi

Burundi has had a violent history since the 19th century, rooted in a combination of stern rule by princely Ganwa, disease epidemics, and the taxation policies of the colonial powers Germany (1890-1916) and Belgium (1916-1961). As is well-known, the Belgians favored the Tutsi, having bought into the “Hamitic myth” that the Tutsi were born to rule, while the Hutu were born to farm. This "history" has been adapted by both Hutu and Tutsi nationalists to suit their own purposes. For example, in 1985-86, Hutu from the refugee settlement in Mishamo in Tanzania explained to Malkki that the Tutsi took advantage of the situation to consolidate their power under the Belgians:

   …At the time of the arrival of the Belgians, the Tutsi employed the malignness…of telling the Belgians that the Hutu are accustomed to cultivating—by that token, therefore they should be taught agriculture…Even today , the Hutu [in Burundi] will tell you that “me, I am accustomed to cultivate. (p. 134)

No matter the salience in Mishamo in 1985-86, from a positivistic viewpoint such mythico-history in fact simplifies colonial history. peasant revolt against combined Belgian/Ganwa taxation regimes were frequent, particularly in the 1920s and 1930s. Differently from the stories told in Mishamo, Lemarchand tells how revolts were rooted in regional identities and not along the Tutsi-Hutu dichotomy as is the current focus; focus was instead on Ganwa in their role as chiefs and princes.

Be that as it may, Lemarchand notes that the development of the stories and fears occurred in the context of Ganwa families, whose authority was gradually eroded by emergent Tutsi clans jockeying for positions in the new colonial bureaucracy, starting about 1930. The result was that while the Ganwa continued to rule both as the king and also as appointed chiefs, the Tutsi began to staff elements of the colonial state.

This led to the paradoxical situation that as independence drew near in 1961, the one truly popular national figure was the Ganwa son of the King, Prince Louis Rwagasore. Rwagasore was a charismatic figure, able to mobilize the forces for independence including the Hutu masses against the Belgian rulers. Indeed, with the support of the Hutu masses, he won election as Prime Minister shortly before independence in 1961. However, Rwagasore was assassinated in a plot organized by a Belgian-supported rival to the throne a few months later. As Lemarchand writes, it was only after this event that what was a three-way focus on power shifted to the two-way ethnic competition that still characterizes political maneuvering in Burundi today (Lemarchand 1996:53-56).

Independent Burundi, 1961-72

Ethnic segregation in post-colonial Burundi rigidified during the 1960s as members of Tutsi clans came to dominate provincial leadership, pushing aside both Ganwa and the few Hutu who held authority. Lemarchand documents this process well in a series of tables that illustrate how Tutsi clans came to dominate administrative posts across a period of about 40 years from the 1920s through the 1960s.

After independence, discrete events punctuated the shift to Tutsi domination over the levers of power offered by the new post-colonial state. After Rwagasore’s assassination in 1961, an unsuccessful coup organized by Hutu officials in 1965 led to the assassination of the prime minister and some tens of thousands of Hutu refugees fleeing to neighboring countries. The coup plotters and many higher level Hutu were executed, and most high level Hutu were excluded from the government. In July 1966, the Ganwa king abdicated in favor of his son, who in turn was deposed and exiled in a republican coup led by Tutsi officers that November.

The Tutsi-dominated military faction from the southern province of Bururi then seized control of the cabinet at the expense of not only Hutu and Ganwa, but also the powerful Tutsi faction from northern Muramvya. In 1969, accusations of coup-plotting were leveled and the Hutu elite were again targeted with 100 being executed and all Hutu eliminated from the military. By 1972, the Bururi faction had turned its attention to the Muramvya Tutsi, who in a series of show-trials received severe sentences, contributing to the fears of the government.

In the context of the increasing competition between Tutsi groups, in April 1972, expatriate Hutu invaded from Tanzania and killed some 10,000 Tutsi along the shore of Lake Tanganyika in the southern tip of the country. The Ganwa King, who had briefly returned from exile, was also killed by the Tutsi-dominated military. The response by the Tutsi-dominated army was brutal enough that Lemarchand considers it the first genocide in the Great Lakes Region. Over 100,000 Hutu were massacred and several hundred thousand fled to neighboring countries, primarily Tanzania, between 1972 and 1974.


Eastern movement from Burundi into Tanzania’s Kigoma Region was common in the past; beginning in the 1600s, most of Kigoma Region was settled by migrants from what is today Burundi in a punctuated series of population movements. Linguistically, Kigoma’s Ha language is mutually intelligible with modern Kirundi and Buha was traditionally ruled by a Tusi dynasty calling itself Ntare, the same name the kings of Burundi used. However, in 1972-74, the Tanzania-Burundi border took on new dimensions as approximately 150,000 Burundians fled to Tanzania, heading toward areas along the forested border, the lakeshore of Lake Tanganyika, and into Kigoma Township. Over the next months tens of thousands of those who survived, often with the assistance of Tanzanian villagers, were resettled in two inland refugee settlements, Ulyankulu and Katumba, with financial assistance from the United Nations High Commissioner for Refugees (UNHCR). Tens of thousands of others remained in both rural areas of Kigoma Region and the urban areas of Kigoma.

In 1978-79, 30,000 of the Hutu refugees moved to a sparsely-settled corner of western Tanzania, Mishamo, where the UNHCR arranged for three years of food rations, hoes, and land with which to re-build their lives as farmers. This they did. But as Liisa Malkki found while living there in 1985-86, they also re-built an ethnic identity. This identity focused on a defiantly revanchist “Hutu” identity (rather than “Burundian”), rooted in claims of forced subordination to both the absent Tutsi and the Tanzanian officialdom who distributed passes and retained political power in Mishamo. The Hutu of Mishamo dreamed of an eventual return to Burundi, rooted in the future-focused belief that “the last will be first,” a situation taken advantage of by a political party formed in Mishamo, PALIPEHUTU, whose goal was the expulsion of the Tutsi from Burundi to be replaced by an explicitly Hutu government. An important element sustaining such belief was the view that any Hutu returning to Burundi would be immediately killed by the Tutsi government.

As Malkki elegantly describes with “story panels,” this new revanchist Hutu identity, radicalized in an isolated situation where the Hutu-Tutsi dichotomy included the element of subordination, was remembered, reproduced, and rigidified, even though there were few if any Tutsi present. The stories she heard were Br'er Rabbit-style, focusing on eventual redemption of a surbordinated group more clever than their powerful opponents. The protagonists Malkki describes though were not only the absent Tutsi, but the Tanzanian officials who dominated the camp and controlled movements in and out. The accounts Malkki collected from the Hutu reflected this persistent subordination in both Burundi and Tanzania. According to her informants:

[In Burundi] There were two categories of school—one for the children of the Tutsi chiefs, another for the Hutu children. The Hutu were taught—are preapred for agriculture, and the Tutsi are prepared to govern. It is like this today in Burundi. The Tanzanians also have schools for themselves where they learn—how is it called?—the social rights—for example, the politics of the country. For the Hutu it is the métiers like mechanics, construction, and carpentry. These are their chosen schools for us, the technical schools…. (p. 133)

Flight and Urban Resettlement

At an extreme opposite of the politicized Hutu identity Malkki found in remote Mishamo is the apolitical “Burundian” identity she found in Kigoma town. Arriving in Kigoma after her time in Mishamo, she expected to find a similarly politicized group of Hutu terrified of a forced return to Tutsi-dominated Burundi, focused on a military re-conquest of the homeland, and sympathetic to PALIPEHUTU. After all, they were all presumably “Hutu” with similar grievances when they left Burundi in 1972-73.

To Malkki’s surprise, over the previous 12 years in Kigoma town a different identity, a more general “Burundian” one had emerged. In the urban setting, there were no mythico-historical tales around which to tell the story of the “Hutu.”  Instead, she found people seeking assimilation as Tanzanian by acquiring markers as Muslims, speaking Swahili, cards as members in the Tanzanian CCM party, marrying Tanzanians, and self-identifying as Tanzanian Ha, rather than Burundian. A few were even able to obtain official papers to legitimize their status in Tanzania. Others even took the overnight boat to Burundi to visit relatives there, a trip considered impossibly dangerous by their cousins living in Mishamo.

Thus, Malkki draws an important contrast: in Mishamo, people describe themselves as militantly Hutu and refugees, believing it impossible to return to Burundi except by force. In Kigoma, their cousins, brothers, and sisters, have yet another attitude and are attempting to blend into the Tanzanian milieu. The contrast is stark: refugees in a difficult urban milieu, “melt” into a local background, while refugees re-established in a remote (and also difficult) forest settlement pull inward socially and create a revanchist movement seeking a violent return to Burundi.

The strength of Malkki’s chapters about the Hutu refugees in Mishamo and how they viewed their status as refugees is in documenting the consistent narrative they tell about themselves. The absence of such a story in Kigoma made the application of such a technique difficult and for this reason her descriptions of the “refugees” in Kigoma are brief, as would be expected from a group for which there is no narrative about identity, hidden or otherwise. But it is also for this reason that the story from Kigoma is important; it shows how Burundians in different circumstances respond differently to their status as refugees.


Understanding the pulls and pushes of Burundian urban life is most intimately seen in Marc Sommers’ ethnography of Burundian refugees in Dar Es Salaam. Sommers has a talent for moving beyond the politics of the refugees and telling the story of individuals, in this case young men who work and live in two tailor shops. In doing this, he does not confront the reader with the politicized narrative that Malkki found in Mishamo, but the stories that individual tell about what they think and do. As Sommers’ subtitle indicates, his focus is on “fear."  In the case of the refugees in Dar Es Salaam in the early 1990s, this fear reflects the dangers of being “illegal immigrants” in Dar Es Salaam; most Burundians living in Dar Es Salaam were in violation of Tanzanian immigration laws. The Burundian refugees that Sommers describes had moved out of the settlements of Katumba and Ulyankulu, and re-established themselves in the rapidly growing Dar Es Salaam of the 1990s.

In Sommers' book, Dar Es Salaam is nicknamed the Swahili “Bongoland” (brain-land) due to the claims of the refugees Sommers interviewed, who said that they need to always think in order to survive. Dar Es Salaam, like many African cities, is populated by millions, including the Burundian refugees he interviewed, living semi-legally in interstitial areas. Sommers describes individuals with personalities and dreams who are members of overlapping family groups and friendship networks, which can loosely be described as “ethnic.” But their relationships also reflect the normal shift and flow of very human relationships and friendships (Sommers’ story about teaching a young Burundian man to drive is particularly interesting, though not directly relevant to this review).

As in Burundi and Mishamo, the refugees in Dar Es Salaam fear the clever manipulations of an unseen “other.” By the time the refugees reach Dar Es Salaam, they are fearing not only the Tutsi, who none of the young men have personal experience with, but also a range of others viewed as outsiders and whom can be avoided only through guile. Within the Burundian community, there is a wariness between highland Hutu (Banyaraguru) and lakeside Hutu (Imbo) and most importantly, there is a wariness of the Tanzanian authorities. As with many residents of urban Africa, the Burundians are subject to periodic round-ups and are always ready to slip a bribe in anticipation of deportation. They believe that it is only by being more clever than powerful Tanzanian officials, and by carefully observing signs that they are permitted to survive as watu wa chinichini (little people) in “bongoland.”

A particular strength of Sommers’ book is the focus he places on the role that the Pentecostal Church plays in the reproduction of Burundian identity in Dar Es Salaam. Pentecostalism is an important social force in central Africa, particularly among Burundian refugees (as well as Rwandans), who respond to its prophetic messages of eventual triumph of the poor and rejected. As Malkki notes in passing, it was important in Mishamo and I have seen its importance among Burundians living in Kigoma. Sommers persistently pursues this issue and evaluates the meanings that the Pentecostal revival with its emphasis on future redemption of the powerless has on the traumatized populations of Burundi, whether in Burundi itself or in Tanzania.


All three of these books about Burundians discuss clandestine narratives which Burundians use to describe the situation in their own country and in Tanzania, and for this reason liberally cite James Scott’s books about “hidden transcripts.”  Hidden in the three books however, is also a broader story about ethnic change in the context of a caste system, which is best seen while stringing the books together and especially by paying attention to Malkki’s example of Kigoma, where there was no strong mythico-history, clandestine, or otherwise.

Thus, at times, ethnicity is “thick” or “rigidified” in a manner which can lead to further revanchist movements and ethnic confrontation. Ethnic Hutu nationalists took advantage of this situation in Mishamo, with one result being that that population was radicalized enough to provide the core element for PALIPEHUTU and violent militias, which helped organized the invasion of Burundi in the 1990s. This is not a product of particular events, but of the circumstances in which Hutu refugees established mythico-histories while exiled to a remote forested area.

But at other times however, the same ethnicity is thin, malleable, and even dissipates. Thus, despite the fact that it was more or less random who became a “Mishamo Hutu refugee” and who remained in Kigoma to become a reluctant Burundian within 13 years after flight, two distinct identities had emerged. Along the same lines, the division between lowland and upland Hutu, which Sommers reported from Dar Es Salaam, also achieved a saliency not found in Kigoma. Alternatively, other salient identities, even those for which people fought and died, can dissipate and are no longer relevant in the context of modern Burundi. The Ganwa, who have seemingly disappeared from today’s narrative about Burundian ethnic competition, are of course the most obvious example of this although it is worth noting that the Hutu of Kigoma, discussed at length by Malkki, must have passed through a similar process of “thinning” ethnicity during the brief period they were separated from the parts of their family sent to Mishamo. For that matter, the more obscure (to us today) regional identities, which focused violent mobilization by Burundians in the 1920s and 1930s in response to colonial taxation policies, have also lost their saliency.

Resettlement affects the contours of such imagined communities and with it the mythico-histories. For this reason, the stories that people tell about themselves and use to define the group are important to listen to. Who does the group define as the “other?”  How do they describe their own position, vis a vis, this other?  And most importantly, what groups and people are left out in the self-definition? 

The sum of these mythico-histories, reflects what Weber describes as “ethnic coexistences  [which] condition a mutual repulsion and disdain but allow each ethnic community to consider its own honor as the highest one.”  But what the examples from Burundi indicate is that this disdain is neither inherent, nor necessarily persistent. It responds to resettlement policies and circumstances, in patterned ways.


Weber, Max (1948) From Max Weber, ed. By H. H. Gerth and C Wright Mills. New York: Oxford University Press.

Tony Waters
California State University, Chico