This is an open access article distributed under the terms of the Creative Commons
Attribution License (Since 1994, the Rwandan government has attempted to remove the division of the population into the ‘ethnic’ groups (ubwoko) Hutu, Tutsi and Twa and instead encourage people to think of themselves only in terms of a common national Rwandan identity. This approach can be classified as recategorization of social identities – a process where subordinate identities are replaced with a shared or superordinate identity in order to improve intergroup relations. Recategorization means moving from ‘us and them’ to a more inclusive ‘we’ (Gaertner & Dovidio, 2000). In Rwanda this entails encouraging people to think of themselves as Rwandans and to stop thinking of themselves in terms of the ‘ethnic’ identities. How do leaders justify this recategorization approach, which can be seen as a country-wide social psychological experiment?
The conflict between Hutu and Tutsi has proved devastating to Rwanda numerous times since the 1950s, making subordinate identity mobilisation a real – not symbolic – threat. The government refers to this history of conflict when promoting the single superordinate identity, and outlawing the subordinate identities in the public sphere. The Rwandan identity and national unity is carefully constructed through several institutions and programs, for example the government’s main unity organ – The National Unity and Reconciliation Commission; the educational value-building institution Itorero, reintroduced in 2009 to help reinstate social cohesion and Rwandan values in the people; the grass-root justice mechanism Gacaca, where 1.9 million cases were tried by local courts after the genocide (Ingelaere, 2012); the monthly mandatory community work program Umuganda, designed to be “a day of contribution and building the country by citizens themselves” (Rwanda Governance Board, n.d.); the local target and achievement program Imihigo (Ingelaere, 2011), where local governments articulate objectives and strategies to achieve these; and the reintegration camps Ingando where prisoners and former guerrilla soldiers undergo value-training (see Thomson, 2013). Although this response to genocide has understandably received far less attention than the events that triggered them, the programs represent remarkable social scientific phenomena in their own right.
This comprehensive recategorization approach is based on a meta-narrative that essentialises Rwandan unity and de-essentialises ethnic identities, in a context where alternative narratives are not permitted. The 2003 constitution forbids genocide ideology, and a 2008 addition further limits identity discussions. These strict genocide ideology laws ban everything seen as divisionism, implying limited freedom of speech on topics of identity, political power and representation. The genocide ideology law has been criticised for being too broad and used repressively (Waldorf, 2011). It is currently being revised, but despite changes, Human Rights Watch (2013) claim the law is still too vague.
The recategorization is complicated by an official annual one hundred day mourning period (April-July), commemorating the genocide. During this period there are posters throughout Rwanda referring to the genocide as “the genocide against Tutsi.” According to the government, accentuating that the genocide was against Tutsi is necessary to combat revisionism and genocide denial. Other respondents claim it solidifies a categorical image of Tutsi as the victims and Hutu as the perpetrators, which could further contribute to spoiling and stigmatizing the identities, as it makes it difficult to escape the labels (victim, perpetrator) associated with the categories.
Today, Rwandan coexistence has improved greatly, growth levels are high, and levels of intergroup violence are low. However, the government’s general identity approach is widely criticised outside Rwanda as a process built on false unity (Clark, 2010), that glosses over differences rather than dealing with these (Straus & Waldorf, 2011), and in this process ignores interpersonal reconciliation (Ingelaere, 2012).
Recategorization
In social psychology, the most commonly researched approach to intergroup conflict reduction is the contact hypothesis (Allport, 1954/1988), which proposes that relations can be improved through positive intergroup contact. Although there are elements of contact involved in Rwandan government policies (with released perpetrators returned to live in communities next to survivors) the Rwandan government’s recategorization policy involves (psychologically) dissolving the subordinate groups rather than improving relations between those groups. According to the common ingroup identity model (Gaertner & Dovidio, 2000), negative attitudes tend to be reduced when groups are recategorized into a superordinate identity. Several studies have confirmed this (Bilewicz, 2007; for a review see Gaertner & Dovidio, 2012). Others show that recategorization at times leads to an increase rather than decrease in intergroup bias due to identity threat (Hornsey & Hogg, 2000b). Recategorization expands the boundaries of inclusion whereby pro-ingroup bias is extended to former outgroup members (Gaertner & Dovidio, 2000). For example, people help ingroup members more than they help outgroup members (Levine, Prosser, Evans, & Reicher, 2005). Recategorization studies have shown that people cooperate more with outgroups and favour their ingroup less both in terms of rewards and evaluation (Gaertner & Dovidio, 2000), but extending such findings to real-world relations is difficult. According to Gaertner, Dovidio, Anastasio, Bachman, and Rust (1993, p. 20): “In naturalistic settings (…) it would certainly be more difficult than in our laboratories to induce a common ingroup identity,” and probably especially so in intergroup conflict.
There are two main recategorization approaches: the dual and the single. The dual retains the subordinate identities in addition to the superordinate, whilst the single focuses solely on the superordinate, attempting to discard the subordinate identities. The Rwandan government policy is clearly a single recategorization approach. The dual recategorization approach allows for subordinate categories within a salient superordinate category, but in retaining the conflict categories, the conflict may be difficult to overcome and it may be harder to make the superordinate identity salient (Gaertner & Dovidio, 2000). Accordingly, studies show that when the goal is to build social units (stepfamilies: Banker & Gaertner, 1998; or corporate mergers: Anastasio, Bachman, Gaertner, & Dovidio, 1996) the single approach may be preferable. Here the subordinate categories are replaced by a superordinate identity. However, Gaertner et al. (1993, p. 20) speak of the forsaking of earlier categorizations as “undesirable, impossible, and potentially detrimental” in many contexts, due to salient social identities being threatened in the process. In Rwanda, abolishing identities for which people killed and died for 20 years ago is challenging, and Dovidio et al. (2006, p. 79) warn that if people’s sense of collective identity is threatened by the recategorization process, it may “exacerbate rather than reduce bias.” Hornsey and Hogg (2000a) claim social harmony is jeopardized by weakening subordinate identities. Based on similar concerns over reactance due to identity threat, the dual approach is generally preferred (Gaertner et al., 1993; Haslam, 2004; Pittinsky, 2009). Research does however indicate certain group differences in responses: majority group members generally prefer the single approach, whilst minority groups members – wishing to avoid assimilation into the majority identity – prefer the dual approach (Dovidio, Gaertner, & Saguy, 2007).
Recategorization, perhaps especially in the Rwandan case, is not passive – it has been actively carried out. According to Reicher, Hopkins, Levine, and Rath (2005) leaders wanting to shape social reality need to be entrepreneurs of identity, defining category boundaries and content. In Rwanda – where both groups forego their identity, the main party (the Rwandan Patriotic Front, RPF, widely seen as mainly Tutsi) has more power to define superordinate and subordinate identities and how these identities are used in political discourse (both category boundaries and content). Extensive research demonstrates that leadership can influence public opinion, and Chong (1993, p. 870) describes framing as the “essence of public opinion formation.” Druckman (2001) differentiates between two uses of frames: ‘frames in communication’ and ‘frames in thought.’ The former refers to the communicator’s words and images used to present the information to others, whilst frames in thought entail what the individual is thinking. Both are concerned with alteration in salience. Frames in communication often shape frames in thought, a process referred to as framing effect. The framing effect works by the communicator having audiences focus on certain aspects of an issue over others (Chong & Druckman, 2007). Through such framing leaders can define category boundaries and content, and try to shape both intergroup relations and social reality.
In this paper the aim is to examine how the Rwandan single recategorization approach is justified by the leaders. The presentation and discussion is based on interviews with nine national leaders (politicians and bureaucrats responsible for unity implementation) conducted in Rwanda 2011/2012. Leaders’ framing may or may not influence people’s perceptions that may go along with or reject their narratives, but this is not an analysis of how the population receives the message (for that see Moss & Vollhardt, in preparation). For now, this is an analysis of how leaders justify the single recategorization: the mobilisation of Rwandan identity and demobilisation of ‘ethnic’ identities. This focus allows for an exploration of the elites’ own narratives and framing of the ‘ethnic’ identities. Getting to the leadership reasoning behind the identity policy provides interesting input to the recategorization debate, especially as this is a real-life post-conflict setting, where intergroup tension proved highly consequential.
Based on thematic analysis, I suggest leaders’ identity meta-narrative focuses on a shared Rwandan identity, claiming all Rwandans are one. Three main themes are further suggested for the framing of ‘ethnic’ identities, used to ‘sell’ the meta-narrative: a) identities are an alien construction (imposed by colonialists), b) identities were used to the disadvantage of the people (i. e., used negatively by the former leadership); and c) identities are socially constructed – and can be discarded. The first two frames are partly externally attributing the causes of conflict, whilst the third focuses on the non-essential nature of some social identities. Based on the analysis, the paper finishes off with a discussion of the implications of the leaders’ narratives for Rwandan social harmony.
Historical Background
According to the 1991 census, the Rwandan population (then at 7.15 million), consisted of 90.4% Hutu, 8.2% Tutsi, and 0.4% Twa (Sellström & Wohlgemuth, 1996). Hutu and Tutsi share the same language, religion, customs, and culture, but are said to differ in terms of their historic social economic status (Newbury, 1988) and political representation (Ingelaere, 2010). Physical stereotypes accompany the groups (taller, lighter Tutsi with longer noses). These physical stereotypes are crude and often misleading, and Rwandans today seldom treat these as absolutes, but rather as “an indicator that a difference, however fuzzy, did indeed exist between Hutu and Tutsi” (Fujii, 2009, p. 112). Fujii (2009, p. 61) emphasises that the meanings of Hutu and Tutsi “were social, not ethnic, in nature, referring to social origin, status, or place of birth,” and according to Gravel (1968, p. 21) Tutsi used to refer to “a noble” and Hutu to “a commoner”, thus to socio-economic groups (castes or classes) rather than to ethnicities. Mamdani (2001) and Eltringham (2004) question how this socio-economic explanation of the group divisions can be reconciled with the large percentage of poor Tutsi. Mamdani (p. 75) offers a middle ground: “to be a Tutsi was to be in power, near power, or simply identified with power just as to be a Hutu was more to be a subject.” The German (1894-1919) and the Belgian (1919-1962) colonial rulers for many years co-opted the lighter-skinned, taller Tutsi to assist with the administration of the colonies. The Belgian introduction of ethnic identity cards in 1933, and their differentiated treatment of the groups, contributed to making these formerly social identities into ethnicities (Clark, 2010). The 1959 ‘social revolution’ replaced Tutsi power with Hutu power. Tens of thousands of Tutsi were forced into exile by intergroup violence. In 1962 Rwanda became independent under a Hutu-led government. Ethnic quotas, violence and anti-Tutsi propaganda followed, and the Hutu regime came to frame the situation as one where Hutu were under existential threat from the Tutsi (Clark, 2010). From October 1990 a civil war ensued between the Tutsi-led Rwandan Patriotic Front (RPF) invading from Uganda and the Hutu regime. From 7 April 1994, the (Hutu) regime in Rwanda orchestrated the killing of up to one million Tutsi and moderate Hutu in Rwanda. The violence was initially political more than ethnic (Reyntjens, 2004), with the main fault line between moderates and hardliners, and further influenced by regional north-south dimensions. In July 1994, the RPF ended the genocide, and won the war against the Hutu-regime in Kigali. Around two million Hutu refugees fled to neighbouring countries in fear of Tutsi retribution, and several sources attribute mass-killings of Hutu in the DRC to the RPF (Emizet, 2000; Lemarchand, 2009). Rebel fractions wanting to overthrow the current Rwandan government still operate in DRC. Presidential elections were held in 2003 and 2010, with President Kagame and RPF taking 95 and 93% of the votes. Kagame is to step down for new presidential elections in 2017.
Methods and Material
This paper is based on in-depth interviews with nine Rwandan leaders. June-July 2011 and November 2012, I conducted 65 semi-structured interviews in Rwanda with members of the general public, academics, civil society organisations, and national leaders (politicians as well as implementation agencies). Throughout this paper, due to the focus, I draw exclusively on the leader interviews.
I used purposive sampling, based on a list of leaders I wanted to interview. These were high-level, national representatives from the three largest political parties (making up the government) and the most influential unity implementation agencies. Official letters asking for interviews with relevant high-level representatives were addressed to the secretaries general or executive secretaries of these offices. In these letters, the research project was presented along with the main interview questions, and the coexistence levels in Rwanda were praised. Three of the interviews were conducted in 2011 (marked below), and the six others in November 2012. Doing the last six interviews after having done 65 interviews in Rwanda was advantageous, as these provided additional contextual knowledge of the identity processes.
I interviewed three senior elected members of government parties (politicians P1, P2, P3) and six senior officials of national unity agencies (officials O1 through O6). Due to the identity policy, the leaders’ ethnicity was not available to me. In her work in Rwanda, Fujii (2009) emphasises that not knowing the ethnic make-up of her sample helped unmask assumptions, as questions were more open. The general ethnic make-up of the leadership in the unity government is said to be rather balanced, but many claim there has been a Tutsification of power (see Gready, 2010), where especially Anglophone Tutsi (those who were in exile in Uganda) hold the most powerful positions (Reyntjens, 2011). I offered to send each interviewee the interview transcript, giving them full rights to edit the material (which established rapport during the interviews and improved the accuracy of the transcripts). When transcripts were corrected, these were only minor adjustments. The interviews were conducted in English, French or Kinyarwanda (with the help of a research assistant). Some meaning was probably lost in translation, but the presence of the research assistant added vital contextual input. Each of the nine interviews was conducted in the representative’s own office, with only the respondent, me, and the research assistant present, and lasted approximately 60 minutes. I recorded six of the nine interviews (with the respondents’ approval), and took extensive notes for all. The Rwandan government has often been criticised for its identity policies, and some defensiveness could be expected when discussing these. It was therefore important to establish rapport and a calm atmosphere by conveying a stance of being curious and questioning rather than critical. Having praised the level of coexistence in Rwanda in the initial letter asking for interviews; working within psychology; and being Norwegian (a country with no significant legacy of negative involvement in the region) may have been advantageous. During the interviews, I presented the research project, and reminded respondents they could terminate the interview at any point.
Interview topics included whether and how Rwandan social relations have changed; reasoning behind the recategorization process; responses to the critique of the single approach; thoughts on whether discarding ‘ethnic’ identities is possible; whether this is in fact happening; and general reflections on the impact and challenges of these identity policies. Perceived lack of freedom to articulate own opinions may have influenced answers, and official roles may have interacted with personal sentiments. The analysis here draws on Fujii’s (2009) focus on meaning-making rather than differentiating between factual and non-factual statements, overlapping with the constructivist stance that meaning is socially produced and reproduced. Traumas may also have affected leaders’ navigation of the topics. When respondents hinted at difficult experiences, I gave ample room for them to retreat or share (see Uvin, 2009), overlapping with Malkki’s (1995) advice of leaving stones unturned.
After transcribing the material, I commenced with a constructivist thematic analysis, based on Braun and Clarke (2006). The thematic analysis was conducted in four steps. After initial open coding, sections of data were structured into recurrent themes. Several themes were particularly relevant to the identity process, and in the third step I focused further analysis on the single recategorization, demobilisation, and mobilisation. There were topics with very little disagreement across the interviews (the need to demobilize the ‘ethnic’ identities), whilst other contained more diversity in sentiments (the degree to which these identities need to disappear completely). One overarching theme – the meta-narrative – was identified, along with three sub-themes constituting the frames used to “sell” this narrative. In the fourth step, the material was structured using Nvivo 10, where queries were run on the themes to refine the specifics, name these, and help select the quotes to illustrate the data constituting the different themes. Quotes are marked with the month of interview.
Analysis and Discussion
Through the thematic analysis, I found that the leaders justify the single recategorization model using an identity meta-narrative, focusing on the shared Rwandan identity. Three frames were found in use to “sell” this narrative. These rhetorical justifications refer to these identities as being alien; used for negative purposes; and essential social constructs. Together these speak to the communicated non-legitimacy of these identities, mobilising instead the shared Rwandan identity. Below I first map out the meta-narrative, and look at how the government has made this the only available narrative, after which I go through the three frames providing supporting evidence from the interviews.
The Meta Narrative
The Rwandan government’s identity meta-narrative ‘we are all Rwandan’ essentialises Rwandan identity and de-essentialises the ‘ethnic’ identities. Several leaders made historic references to the salience of ‘Rwandanity’ in the past to ‘prove’ the legitimacy of this superordinate identity. A political leader (P1) (11.12) explained:
In those old times, the main purpose of the leadership was always to unite Rwandans, as Rwandans, not as individual social categories and classes; to unite them behind that idea of ‘Rwandanity’ (…) So a Rwandan used to feel that he is a Rwandan before his membership of any other social class or category.
An agency official (O3) also talked of this pre-colonial harmony (11.12): “Before the colonialists came here to Rwanda, Rwandans were unified, were one. They used to live in harmony, there were no differences between Hutu, Twa, and Tutsi – no one would mind and think about ethnic groups.” This pre-colonial harmony image overlaps with the Rwandan official history account (see Republic of Rwanda, n.d.): “The (Tutsi) king was supreme but the rest of the population, Bahutu, Batutsi and Batwa lived in symbiotic harmony.” This portrays the latter of two versions of the social relations in pre-colonial times: one that emphasises the imbalance in power and means caused by aspects such as clientship and forced labour (where the rich Tutsi ruled the rest of the people, mainly the Hutu majority), and the other that does not (Ingelaere, 2012). The first narrative was supported by the former Hutu regime. The potential contributing role of Tutsi kings and chiefs in dividing society along Hutu and Tutsi dimensions before and during colonial rule (Fujii, 2009, p. 45; Newbury, 1988, p. 11) is not recognized by the interviewed leaders.
‘Rwandanity’ is portrayed as the most salient identity before, which is used to legitimize the return to this single Rwandan identity today. Uniting under this single identity is not an easy process, as recognized by an agency official (O6) (11.12):
We know our situation is unique; from a genocide landscape we want to create a unified community and we want even to go beyond the differences and the genocide identities and transform them to being unified within one, and that is extremely challenging, even for ourselves, even for the people.
Through this identity meta-narrative the Rwandan government tries to control the definition of the category boundaries. According to Reicher et al. (2005) such definitions will determine category content, extent – and direction of the relevant mobilization. In the promotion of chosen category definitions, the presentation of identities as legitimate, spoilt, or harmful is not a given but a rhetorical construct of identity entrepreneurship, making leaders’ motivations in these choices interesting. According to a political leader (P1) (11.12), the goal of the Rwandan government has long been to make the superordinate Rwandan identity salient: “We started saying why not just put into minds that Rwandanity was more important than Hutu, Twa, Tutsi.” However, spoken goal and leadership motives may be far apart. For example, both Gready (2010) and Reyntjens (2011) claim there is a Tutsification of power in Rwanda. If so, a primarily Tutsi leadership would be well served with abolishing ethnic identification, to protect themselves against allegations of Tutsi rule, especially as Tutsi constitute a numerical minority. A Tutsi-biased version of history (of pre-colonial intergroup harmony, and exclusive victim consciousness; see Vollhardt, 2012) could improve their position.
In mobilising the ‘we are all Rwandan’ meta-narrative, alternative narratives could be a threat. According to Haslam, Reicher, and Platow (2011) leaders need to analyse resistance to their own project and organisation of the social world, and have strategies for overcoming such alternatives. To solidify their meta-narrative, the Rwandan leaders repeat it and have disallowed all disputing narratives, making their version the only available (Reyntjens, 2004). The before-mentioned genocide ideology laws effectively limit the discussion of ethnicity, as emphasized by an agency official (O2) (07.11): “You can’t expect to operate within politics in this country based on ethnicity. You have to be within the allowed structures here, and here in Rwanda you cannot talk about these things openly.” A political leader (P3) (11.12) said it was necessary: “For me the fact that there are laws that punish and reprimand all sorts of divisionism is something very positive.” The 2001 genocide ideology law made ethnic self-identification criminal (with severe ramifications: up to 30 years in prison, and fines of up to US$8800; Thomson, 2009). Through such measures, alternative narratives are silenced and demobilised. The government legitimizes these laws with reference to internal stability, to hinder ‘ethnic’ mobilisation, but also to counter claims of dual genocide, of the genocide in fact being a civil war, and more broadly, of former and current Tutsi oppression of Hutu. Such claims are effectively outlawed in Rwanda (again playing into the point above on the category defining and narrative defining power of a subgroup), and are thereby limited to the diaspora. With reference to this ‘outside’ ideology, an agency official (O1) (11.12) emphasised: “I think we still have to put strength in combating this outside ideology, because they influence people too much.” This discourse is thus externalised, made into something the Rwandan people needs protection against. Rwanda has repeatedly been accused of human rights violations and denial of freedom of speech in maintaining their identity narrative. Thomson (2013) found that many farmers see the main effect of the unity and reconciliation mechanisms in the exclusion of groups from political organisation. Similarly, Reyntjens (2011, p. 30) criticises the Rwandan identity narrative that there are no Hutu and Tutsi, and calls it a “concealment of domination by Tutsi.”
In demobilising the subordinate identities, most leaders seem motivated to suppress public displays of ‘ethnic’ identity that provide a rallying point for the organisation of hatred, not on private expressions, a domain distinction involving material threat trumping symbolic threat. As a political leader (P1) (11.12) said regarding subordinate identities:
It may be your choice, you may hold it as you want, you may even make a flag and put it in your home, but let it be in your house, not in the street, because it will hurt me.
Thus controlling alternative narratives is mainly about controlling the public identity discourses. In addition to limiting the talk of identities through legal ramifications, the alternative narratives are also habitually made illegitimate with references to traitors and self-interest, as demonstrated by an agency official (O5) (07.11): “Those people who wish to keep these ethnic groups as they say, were the ones who were in the machinery of the genocide. They wish to come to power, not for the benefit of the people, but for themselves.” Through this, the leaders attempt to manipulate emotions in the public (see De Castella & McGarty, 2011), making alternative narratives associated with genocidaires, trying to elicit both fear and anger in people.
The meta-narrative emphasising the superordinate Rwandan identity is portraying pre-colonial intergroup relations as harmonious. Returning to this identity and overcoming ‘ethnic’ identities is the core project (though the explanations of the motivations for this project range from unity to the cover up of Tutsification). Disallowing competing narratives demonstrates the Rwandan leadership’s identity entrepreneurship. The meta-narrative is further ‘sold’ and justified through the use of three frames.
Framing
Studies have shown that framing effectively influences attitudes (Chong & Druckman, 2007), and it is one of the primary means leadership has to influence its population’s opinions (Druckman & Nelson, 2003). By influencing the language used to navigate identities through the frames, leaders can influence related attitudes (see, e.g., Landau, Sullivan, & Greenberg, 2009). Individuals will often base their own perceptions on a combination of the leadership narrative and relevant citizens’ take on the situation (Druckman & Nelson, 2003), but the population may also resist the leadership narratives and subscribe to alternative versions of events (see Bamberg & Andrews, 2004), making it vital to sell the meta-narrative as convincingly as possible. Based on the thematic analysis of the Rwandan leadership narratives, the subsets of considerations the leaders are emphasizing in terms of ‘ethnic’ identities in Rwanda are threefold, and are presented below.
a) ‘Ethnic’ Identities Are Alien Constructions
The leaders portray Hutu, Tutsi, and Twa as social, not ethnic groups. An agency official (O1) (11.12) said: “It really is a social group, but it has been deformed and it has been created in a conflict way.” These social categories were, according to this frame, made into institutionalized ethnic categories by the Belgian colonisers. According to an agency official (O4) (07.11) the Belgians created the intergroup problem:
We faced this bad period because of foreigners’ ideas, which were strongly influenced by divisionism. The Belgians had these divisions themselves and since they had this in Europe, they came here looking for similar divisions. (…) From 1959 the colonialists synthesized one ethnic group to kill the other – most of them were killed, were murdered, their houses were destroyed, and some of them fled the country.
According to a political leader (P1) (11.12), colonialists destroyed unity: “They simply broke that kind of cohesiveness that was in the society.” Belgians partly divided the population based on the group’s stereotypical physiological traits (the tall, thin, lighter-skinned Tutsi with long, small noses). Several of the leaders ridiculed these arbitrary divisions drawn up by the Belgians. An agency official (O6) (11.12) said:
If you take the definition of noses by the Belgians (…) where do you draw the line? What qualifies as a big nose and a small one? Either you are going to be short, or you are going to be tall. But where does the shortness stop? You can also have someone with a small nose, but who is also short – which category does this person fit into?
Ethnic groups are commonly seen as more durable, long-lasting, and static than most other social identities (Verkuyten, 2012), thus removing the perception of these as ethnic could be useful. To ‘prove’ the non-ethnic nature of these groups, the shared membership in clans, where people are believed to come from the same ancestors, is habitually referred to. A political leader (P1) (11.12) said: “In those three classes you find different clans.” Leaders additionally emphasise that nothing substantial divides the groups. The government did not construct this contestation: the nature of these identities is not settled, and different meta-narratives have existed throughout time (see Kiwuwa, 2012, p. 66). An agency official (O6) (11.12) referred to his own research asking people about the differences between the groups: “A Hutu is what a Tutsi is not, a Tutsi is what a Hutu is not – that’s the only definition they will give you, and with that you can’t get anywhere.”
b) ‘Ethnic’ Identities Were Used to the Disadvantage of the People
Most of the interviewed leaders emphasised that the ‘ethnic’ identities were adopted and used negatively by the Hutu leadership that came after 1959, as outlined by an agency official (O5) (07.11): “How the colonial forces put in place ethnic mechanisms is very vital, especially since the local leadership continued to follow these mechanisms. (…) After 1959, Hutu leadership began to use ethnic differences, which were really not differences.” An agency official (O3) (11.12) reiterated this:
From 1959 up to 1994 (…) there was leadership that educated people in divisionism, whereby everywhere, even in the kitchen, whatever anybody would do, you could smell divisions and hatred. So from that you could not expect anything but genocide and killings.
An agency official (O5) (07.11) also underlined the spreading of divisionism during this period:
The speeches from the [then] president were really hate speeches. (…) People learned to live in hate. All institutions were constructed on this hatred. To explain all problems – social, economic or whatever – the answer to them all was: ‘It is because there are Tutsi in our country.’
Taken together, the first two frames stress the alien and harmful nature of the ‘ethnic’ groups. Through this focus, the government in Rwanda is partly attributing the genocide to external factors, where the divisions are products of colonialism and bad leadership. Both of these are not present anymore, and that may make coexistence easier. This could increase the salience of the Rwandan identity, as it becomes a road to coexistence, security and a way of escaping spoilt identities (see Goffman, 1963). However, as discussed above, single recategorization may lead to identity threat, and for many these ‘ethnic’ identities are still salient and legitimate. As people may be unwilling to adopt the first two frames that challenge the legitimacy of these identities, a further frame is necessary.
c) Identities Are Socially Constructed – And Can Be Discarded
This third frame stresses that some social identities are social constructions without substantial or essential meaning today, and therefore these identities can be rejected. Leaders in Rwanda point to clans and regionalism as examples of social categorizations that were salient but that are not salient anymore, and thus argue that the ‘ethnic’ identities can also be discarded. An agency official (O1) (11.12) referred to regionalism:
Before genocide, regional groups and regional identity was more pronounced and was becoming more important than ethnic identity. If we had continued some 10-15 years with those divisions, we would now have had two ethnic groups, one of the north and another of the south, with the denomination of Abakiga and Abanyanduga.
Another agency official (O6) had similar arguments regarding the clans (11.12): “Nobody cares about his clan, but 50-60 years ago that was the most important political identity.” He further used this to demonstrate the need for strict single recategorization (11.12):
Have you heard anyone say: ‘Ah, this minister is from this or that clan’? There are zero such statements, because when Hutu-Tutsi identities were coming up strongly they did not manage their clan identities simultaneously (…) no, they said ‘forget about it.’ That is why these categories disappeared. So if today we put in policies that create that kind of ambiguity – making room for groups and national level identity – we won’t make it.
In emphasising that identities are social constructions, the government is saying that even though an identity is salient (as the ‘ethnic’ identities clearly still are; Buckley-Zistel, 2006; Clark, 2010), it is natural for social identities to be changed, thus discarding these is acceptable. The framing of these identities as social constructions opposes common perceptions of ethnicity. According to Verkuyten (2012, p. 75), even though ethnicity is “dynamic, changeable, and socially constructed,” ethnicity often entails a perception of “common descent and continuity with the past” – that is, as discussed above, a more static identity than many other social category memberships. Some leaders try to combat this lasting ethnic nature further by emphasising the limited awareness of ethnicity in the early 1990s, and that children were frequently told in school to go home and ask their parents which group they belonged to.
Through these three frames, these leaders try to justify their single recategorization model. They claim the ‘ethnic’ identities are illegitimate (alien and harmful), that it is insecure (akin to the social identity notion that there are cognitive alternatives), and that it is against shared Rwandan interests. The ‘ethnic’ identities are thereby attempted to be demobilised, whilst the superordinate Rwandan identity is mobilised. How is this single recategorization approach influencing Rwandan social harmony?
Consequences for Rwandan Social Harmony
In Straus and Waldorf’s (2011) edited volume on remaking Rwanda, most contributors express concern about the oppressive Rwandan identity approach, asking how long this will last and how deep it goes. One consequence of criminalizing talk of ethnic groups (except with reference to the genocide) is that it severely limits the possibility of knowing people’s responses to the single recategorization.
According to Brewer’s (1991) optimal distinctiveness theory, the needs for distinction and for inclusion must be balanced. Regarding the need for distinction, single recategorization is likely to threaten social identities. As we saw above, the forsaking of earlier categorizations is detrimental and impossible in many contexts. The problem in Rwanda, Clark (2010, p. 144) emphasises, is not the attempt to build a new superordinate identity, but that this identity “demands and depends upon the suppression of ethnicity.” The dual recategorization on the other hand would likely not elicit the same level of threat, as the subordinate categories are retained – making this the generally supported model. Studies have demonstrated that the level of identification with the ingroup moderates the effect of recategorization. Crisp, Stone, and Hall (2006) found that maintaining subordinate identities within the superordinate identity (the dual approach) seemed to buffer against the distinctiveness threat a single approach caused for high identifiers. However, as the subordinate categories in Rwanda are associated with extreme intergroup conflict, the dual recategorization approach could be problematic. Salient intergroup boundaries are often characterised by intergroup anxiety and distrust, and when the group lines are retained the risk of a return to conflict remains high (Gaertner & Dovidio, 2000). Also, in extreme cases such as Rwanda, perhaps the need for inclusiveness is more acute than the need for distinctiveness?
The need for inclusion may be facilitated by the single recategorization (if this is in fact inclusive). As a result of the conflict, many respondents indicate these subordinate identities have become at least partly spoilt and stigmatized. The spoilt identities of Tutsi (victim, security concerns) and Hutu (perpetrator, guilt) gives people an opportunity to adopt the shared identity (Rwandan) as a way out of stigmatized identities. The single recategorization could offer the opportunity to draw positive self-esteem from group membership (see Tajfel & Turner, 1979), and avoid collective guilt (see Branscombe & Doosje, 2004). When the aim is building a social unit, the single approach has showed to result in more harmony (see discussion in Gaertner & Dovidio, 2000). Even if a single recategorization approach may cause initial conflict escalation due to identity threat, it may reduce tension in the long run (Gaertner & Dovidio, 2009). This may have made the single recategorization seem like the only viable option, as emphasised by a political leader (P2) (11.12):
I think that the tragedy was so deep, in a sense it has helped the leadership to let people understand that we cannot go further on this negative path, so instead of dividing or fighting between us, we should be united and build a future for our kids.
Some of the local population respondents talked of the preferred model as one in stages, where single recategorization was spoken of as important for establishing peaceful coexistence, and then dual recategorization could come after, as a necessary second step towards achieving lasting peace. Gaertner and Dovidio (2000) suggest that different social categorisation approaches may be beneficial at different stages of intergroup conflict reduction. Buckley-Zistel (2006) supports this, and speaks of chosen amnesia in Rwanda, where community cohesion is made possible by the avoidance of certain memories. This, she says, is positive for local coexistence, but not beneficial in the long run as it impedes processes of social transformation removing social inequalities and avoiding a return to intergroup violence. Could it be that the single recategorization was a necessary first step to build coexistence and stability in Rwanda, but that to move on and establish lasting peace, room for subordinate identities within the superordinate identity is necessary? Asking the interviewed leaders about a dual approach, none of them hesitate. An agency official (O6) (11.12) said:
We need to be firm on national identities, that is when those other identities (…) will go, we make them irrelevant, and they go. But if we make them a little bit relevant, they won’t go, and they will then not let the new one emerge.
Final Remarks
In this paper, I examined how the Rwandan single recategorization model is justified by nine national leaders. Their single recategorization approach is based on a meta-narrative essentialising Rwandan unity and de-essentialising ethnic identities, which is strengthened by a lack of alternative narratives. The leaders’ framing to legitimise the discarding of these ‘ethnic’ identities is threefold: a) identities are an alien construction (imposed by colonialists), b) identities were used to the disadvantage of the people (used negatively by the former leadership); and c) identities are socially constructed – and can be discarded. The Rwandan identity on the other hand is presented as essential and legitimate, and offers a potential way for people to escape spoilt and stigmatized identities. Several researchers ask whether it is feasible to discard meaningful and salient subordinate identities in real life (Hewstone, 1996; Hornsey & Hogg, 2000b), and only time will tell whether or not the Rwandan government’s single recategorization approach yields social harmony or a return to intergroup conflict. In the meantime Rwandan leadership insist that Rwandans are all one, as emphasised by a political leader (P2) (11.12): “I do not see the reason why anyone should be sitting there always claiming that you are Hutu or Tutsi as if those people are not the same.”