Keywords

It is not like ‘I am who I am because I am Hutu or Tutsi.’ Before the genocide, this was the reality. The little divisionism left in Rwanda comes from the history we have. Our top leadership is planting a sense of unity, and wish to avoid divisionism. Since we are trying to avoid the divisions, we do not talk about the issues that divide us. Also if we talk about the genocide, we have to answer who was it who killed Tutsi? It was Hutu.

Young Tutsi male (11P, 2011).

In 1994, up to a million people were killed in Rwanda. The victims were mostly Tutsi, but also included an estimated 40,000 moderate Hutu who wanted coexistence with Tutsi, rather than extremist measures. The perpetrators were mainly Hutu. The nature of the divide between Hutu (85 % of the population), Tutsi (14 %) and Twa (1 %) is contested (Kiwuwa, 2012). There are stereotypical physical differences, but the groups share language, culture and religion. These subordinate identities have been actively manipulated throughout history. Increasingly, from 1959, the Hutu government claimed the Tutsi were foreigners who should “go back to Ethiopia where they came from”. This anti-Tutsi rhetoric escalated and resulted in the genocide. Many accused the Tutsi-led army of committing atrocities against Hutu after the genocide. Handling these polarised “ethnic” identities was a key challenge for the Rwandan government post-genocide, and they have taken a radical approach: attempting to replace these “ethnic” identities with the national identity . This identity approach is akin to social psychology’s single recategorisation model. This chapter focuses on what the representations of these social identities—the Hutu, Tutsi and Twa —are now, and how respondents justify these different representations. This is key in terms of the identity approach the government has taken. In single recategorisation, the subordinate identities are to be cast off. Abolishing consequential and salient identities is challenging, and particularly so soon after extreme intergroup conflict. The success of the single recategorisation process is likely to depend on the salience (see Crisp, Stone, & Hall, 2006) and the associated definitions of these subordinate identities. Each group has their view of the world, their social construction or social representations of reality serving as the basis for individual thought and interpretation of the world. In Rwanda, however, the question remains: what constitutes the groups, and what representations do people have of these subordinate identities? If these identities are perceived as ethnic, these may in consequence be seen as more static and “essential” compared to other social identity representations (e.g. social class), which may be seen as more malleable. Now, 22 years after the genocide, it is important to look at how Rwandans speak of these subordinate identities in Rwanda. Are they seen as true, false, essential, ethnic or non-ethnic? It is important to unravel what these groups are to people, as this will influence responses to the policy of single recategorisation.

Social Recategorisation

In social recategorisation, as stipulated in the common in-group identity model (Gaertner & Dovidio, 2000), intergroup bias is attempted to be decreased through restructuring group categorisations. Through the inclusion in a common in-group identity, positive in-group bias should be extended to former out-group members. The model has received strong empirical support (Dovidio, Gaertner, & Saguy, 2009; Gaertner & Dovidio, 2000; González & Brown, 2006). Dovidio et al. (1997) found that the more distinct groups felt as a common group, the more the intergroup bias was reduced in evaluations, self-disclosure and helping behaviour. In research on European Portuguese and African Portuguese children, Guerra et al. (2010) found that recategorisation strategies produced positive attitudes toward the out-group children.

However, recategorisation is difficult . Most of the research has been done in laboratories, and proves harder to achieve in real life (Mullen, Brown, & Smith, 1992), especially in highly politicised contexts (see Brewer, 1997) and contexts involving powerful ethnic categorisations (Hewstone, 1996). Dovidio, Gaertner, and Validzic (1998) found resistance to recategorisation even in laboratory studies. As emphasised by Rutchick and Eccleston (2010) : “it is not easy to induce people to set aside who they are” (p. 110).

There are two main recategorisation models: dual, where both subordinate and superordinate identities are retained, and single, in which the subordinate identities are replaced by the superordinate (as in Rwanda). These are closely related to strategies of multiculturalism and assimilation (Dovidio, Saguy, Gaertner, & Thomas, 2012). Preferences for a dual identity are related to preferences for multiculturalism, and preferences for the single identity are related to preferences for assimilation (Dovidio, Gaertner, & Kafati, 2000). Crisp et al. (2006) found that the measured increased bias in high identifiers when subordinate groups were attempted recategorised into a superordinate identity was attenuated when room was given for the subordinate identity to remain salient within the salient superordinate identity. However, the dual approach can make it more difficult to overcome conflict because the conflict categories are retained (Gaertner & Dovidio, 2000), which can make the sense of common identity difficult to sustain, and a return to conflict may thereby be more likely.

In the single approach, on the other hand, the subordinate categories are attempted to be replaced with the superordinate identity. In settings where one-group representation is the goal, the single approach has been more effective in reducing intergroup bias than the dual (e.g., studies on corporate mergers and stepfamilies, see Gaertner, Bachman, Dovidio, & Banker, 2001). This is in contrast to two-groups within one representation, for example, Africans and Europeans at the same school . According to Gaertner and Dovidio (2000) , in contexts aiming for one-group representations “the continued existence of the earlier subgroup identities (even simultaneously with a superordinate identity) may be perceived as a sign that the amalgamation process is failing.” (p. 101). However, the dual approach is widely criticised. According to social identity tradition, people strive to be part of positively valued groups, to ensure positive self-imagery. This entails motivation for positive differentiation. Brewer’s optimal distinctiveness theory (1991) postulates a need for balancing distinctiveness and inclusion. The single recategorisation approach hinders distinctiveness (Hornsey & Hogg, 2000). The abolishing of subordinate identities can lead to social identity threat, especially likely to be experienced by high identifying group members (Crisp et al., 2006). Recategorisation can also lead to group members trying to retain their distinct identity, and thereby maintaining a high level of intergroup bias, (Dovidio, Gaertner, Hodson, Houlette, & Johnson, 2005), or experience an increase in intergroup bias as a reaction to the recategorisation (Dovidio et al., 1998).

In general, the dual approach has been more widely supported (Crisp et al., 2006; González & Brown, 2006; Hornsey & Hogg, 2000). However, Dovidio et al. (2005) found that the dual identity approach can have a positive initiation effect, but single recategorisation may work better in the long run even if this may include an initial upsurge in conflict levels. This means that some contexts may be better served with the radical single approach as this offers an “escape” from the conflict categories (see Moss & Vollhardt, 2015 for a discussion of single recategorisation as a potential temporary approach).

Research also finds general differences in preferences between majority and minority group members, where the former prefer the single approach and the latter the dual (Dovidio, Gaertner, & Saguy, 2007). This will vary with context, and what goals the groups have (Guerra et al., 2010). Related are issues of power; in cases where a high power group defines the content and specifics of the superordinate identity, minorities may not be included (Esses, Wagner, Wolf, Preiser, & Wilbur, 2006). These variations and the potential negative ramifications of recategorisation emphasise the need to know more about in what contexts the different approaches will increase or decrease intergroup harmony. The trajectory of the recategorisation will be connected to what representations people have of the identities in question. Exploring these issues in the Rwandan context allows a rare window onto a single recategorisation policy put into life through comprehensive policies. Being able to discuss the “abolished” “ethnic” categories with both political leaders and general population offers access to narratives on current Rwandan social fabric. Such contextually grounded non-Western identity representations seldom emerge within recategorisation research.

Empirical Background

Rwanda has a population of 12.1 million. The group divide has been in place for a long time, but how these groups differ is subject to debate (see Eltringham, 2004; Mamdani, 2001). The differences between the groups have been connected with social class, power and representation, with Tutsi as higher ranking (though, many Tutsi were also poor). In this reading of the categories, Tutsi were identified as being nearer power, and Hutu seen more as subjects (Mamdani, 2001). The groups have also been referred to as ethnic, often referencing the stereotypical physical differences (taller, lighter Tutsi with longer, smaller noses), and the “Hamitic myth” in which Tutsi are said to originate from Ethiopia. In 1933, the Belgian colonisers introduced ethnic identity cards, institutionalising the already existing subordinate group divide, making it likely that Rwandan social solidarity was greater before the Belgians arrived than after (Hintjens, 2001). Tutsi were the ruling class, but from 1959 the situation started changing. The Hutu overtook power during the “Rwandan revolution”, in which thousands of Tutsi were killed, fled and were expelled from jobs and schools through quota programmes (Newbury, 1998). Tutsi in exile were denied re-entry to Rwanda (on accounts of the country being “full”), and anti-Tutsi propaganda increased. The Uganda-based Tutsi-led rebel movement Rwandan Patriotic Front (RPF) invaded Rwanda in October 1990 to topple the Hutu government. The war culminated in the 1994 genocide. RPF, under current President Kagame, won the war, and ended the 100 days genocide.

The current Rwandan government’s identity approach is comprehensive . There are several programmes and policies in place to mobilise the Rwandan identity and demobilise the ethnic identities. These include an extensive value-training programme (Itorero), re-education camps to prepare and reintroduce prisoners and former rebel group members to society (Ingando), the grassroots justice mechanism (Gacaca), and monthly mandatory community work (Umuganda; see Straus & Waldorf, 2011 for several of these programmes and policies). The government has also replaced the flag, the national anthem, rewritten curricula and laws, and changed the second official language from French to English (demonstrating, as discussed above, the power an elite has to influence what the official, practised identity content should be). Further, there is a comprehensive government narrative on Rwandan identities, presenting the Rwandan identity as the true identity, and the “ethnic” identities as alien constructions, used to the disadvantage of the people. The essential Rwandan identity thus stands in contrast to the non-essential “ethnic” social constructs (Moss, 2014; see also Vandeginste, 2014). Strict anti-divisionism laws hinder alternative public narratives, where the concepts of Hutu and Tutsi are only accepted when discussing the genocide (Amnesty International, 2010). Several researchers argue the identity policy is a cover for a “tutsisation” of power (Reyntjens, 2013; see also Gready, 2010), and that voicing concern over increased Tutsi power would be seen as divisionism and thus illegal. Either way, this comprehensive approach mobilises the Rwandan identity, and demobilises the “ethnic” identities.

Every year, however, during the 100 days mourning period commemorating the 1994 genocide, the categories are again emphasised as posters go up all over the country: “X years since the genocide against Tutsi”. In the commemoration ceremonies tensions are at times exacerbated, as some Hutu say that they feel collective guilt is wrongly ascribed to all Hutu (Amnesty International, 2010). This concern is echoed by Eltringham (2004) saying that the polarisation of Rwandans into survivors (Tutsi) and perpetrators (Hutu) leaves no room for moderate Hutu, as this narrative assumes all moderate Hutu were killed during the genocide.

Some researchers claim the smooth surface in Rwanda hides strong ethnic bias (Reyntjens, 2013) that the identity model is counterproductive (Ingelaere, 2010), and that the ethnic differences should be discussed rather than silenced (Clark, 2010). Several of these echo concerns voiced by social psychologists against single recategorisation in general. Research finds that these subordinate groups are still highly salient, and that Rwandans express a strong desire to be able to categorise people based on these group memberships on a continuous basis (Hilker, 2009). Buckley-Zistel (2006) found intergroup antagonism between respondents of the two groups. The Rwandan government on the other hand says the reconciliation is on track, and that unity is improving (NURC, 2010).

Present Research

The Rwanda-focused research within psychology is still scarce (though see for example McGarty, 2014), and is less focused on the identity approach (exceptions include Moss & Vollhardt, 2015). The aim of this present research is to look specifically at how people speak of the subordinate group identities of Hutu, Tutsi and Twa , and what representations people have of these. Further, the chapter explores how these representations go together with the single recategorisation, as managing to replace the ethnic identities with the national identity depends on the representations of these identities.

Methods: Discussing Abolished, but Highly Salient, Identities

Semi-structured interviews, with a convenience sample from the local population (N = 56) and political leadership (N = 9), were carried out during 4 months in the field, covering the periods June–July 2011; June 2012Footnote 1 and November 2012. These time slots included tense periods, with grenade attacks within Rwanda and spillover effects from the heightened conflict in DR Congo.

The age range spanned from 21 to 80 (mode around 35 years), and occupations included farmers, shopkeepers, students, business people, nurses and carpenters. The political leaders included politicians (representing the three largest parties) and high-level representatives of unity implementation agencies (e.g. National Unity and Reconciliation Commission; Itorero). Participants were recruited in rural and urban areas in three of five districts (South, North and Kigali), mostly in the capital, Kigali. Asking about group membership is not possible in Rwanda; however, 34 participants volunteered this information during the interview (Hutu: 16; Tutsi: 16; Twa: 1; mixed: 1). Four men declined the invitation to be interviewed. Women who were approached frequently said they were busy with work, and only 13 women were interviewed.

The interview guide was discussed with both assistants and a Rwandan researcher, and thereafter adjusted. The interviews were mostly individual, but, when participants preferred, the interviews were done in small groups (seven interviews, 2–4 respondents). Here respondents corrected, objected and debated each other’s statements. To enhance trust and openness, several respondents were interviewed twice or more. Interviews were conducted in offices, at cafés, hotels or in peoples’ homes, depending on the respondents’ wishes, and were held in Kinyarwanda, English and French, with the help of research assistants. Both a Hutu and a Tutsi research assistant were present in the rural interviews (compared to only a Tutsi assistant in Kigali). Most interviews lasted approximately 45 min. As only ten respondents allowed the interview to be recorded, extensive note taking was relied on for the rest.Footnote 2 The study was described as examining Rwandan identities, and questions addressed how participants felt about the current peace, the potential transition from group identities to more focus on national identity, whether or not people identified more now as Rwandans, and what factors they thought could explain any changes in identification. Through these questions, it usually came up what respondents meant that these groups are (ethnic groups, non-ethnic groups, social groups, fake groups etc.). If not, I asked directly. “Ethnic identity” was only used in the interviews when the respondents themselves characterised these groups as such, if not these were referred to as “the group identities” .

Analysis: Representations of Subordinate Identities Post-Genocide

The material was analysed using thematic analysis (Braun & Clarke, 2006), where respondents’ understandings were treated as valid—but subjective—constructions. After iterative readings of the transcripts, the material was ordered under thematic headings and subheadings. From this thematic mapping , I identified two main dimensions pertaining to the respondents’ representations of the subordinate identities in Rwanda: ethnic and non-ethnic.

Three themes were identified for representations of the groups as ethnic: ethnicity is not the problem and should be retained; history and physical traits demonstrate that these groups are ethnic groups; and lastly, social classes having become ethnic groups. For the non-ethnic dimension, three themes were identified: these groups are social classes, not ethnic groups; the shared characteristics across the groups “prove” their non-ethnic nature; and as does the shared memberships in the same ancestral clans (see thematic map, Fig. 13.1). In the following, the chosen extracts are marked with respondent number, year of interview, and either P for general population respondents or L for leaders.

Fig. 13.1
figure 1

Thematic map of interview responses

Full size image

Ethnic Groups

Many respondents refer to these identities as ethnic identities. In a society where sentiments are strictly regulated, respondents were surprisingly open about these sensitive issues. Many voiced opinions that directly opposed government narratives (which emphasise that the groups are social classes). Three main themes were identified.

Ethnic Groups Not the Problem

Some respondents explain that the groups are ethnic groups, but that this is not and never has been the problem. A farmer said:

For me, ever since we have had ethnic groups it never caused any problem. I just wonder, why did we utilise these ethnic groups to kill each other? Why did they become tools of hatred? We should stay the way we were before, with the ethnic groups, but focusing more on unity (51P, 2012).

She openly discussed the groups as ethnic, and she wanted these to be retained, but not used negatively. She continued:

I think now we no longer have ethnic groups. We are supposed to be just Rwandan. Perhaps people have this in their hearts, but they can’t show it. Before I say anything else, I need to emphasise that now, under this set up, these ethnic groups are no more (51P, 2012).

In the interview she was clearly frustrated that the government had removed these ethnic groups, pretending the groups were non-existent. She says they are supposed to be just Rwandan, and that people have to hide what identities they have in their hearts . That the groups are not a problem was reiterated—interestingly—by the head of a unity implementation agency:

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. (…) Hutu and Tutsi has never been a problem. We should not make it a problem now. Even now I think these three things [Hutu, Tutsi, Twa] are not a problem. The problem is in one place—the stomach, where our interests are. This [pointing at his stomach] is a bad ethnic group (38L, 2012).

The first two sentences of the extract are in line with the official narrative: precolonial intergroup harmony. He then ventures off “official script”, stating that the groups are ethnic, but that different ethnic groups per se is not the problem. An elderly Tutsi woman in a group interview had similar sentiments. Asking whether the categories of Hutu, Tutsi and Twa were disappearing, she said: “I am not excluding ethnicity. (…) Ethnic groups and class were never a hindrance to unity, but we should not focus so much on ethnicities. We should focus on unity” (52P, 2012). The respondents cited here agree the focus should be on unity, but emphasise that ethnicity does not need to be removed to attain this.

History and Traits

Some discussed factors such as history, family and physical traits as “proof” of the ethnic nature of these groups. An academic in Kigali said:

After the genocide, the ethnic labelling was removed from the identities. But still people know themselves, because I am affiliated with a family, I have a father and a grandfather. (…) Even if Israel was destroyed, you can’t delete the Israelis. Similarly, you can’t delete Hutus, you can’t delete Tutsis (30P, 2012).

He thereby discusses ancestry and history embedded in perceptions of ethnicity . He further referred to the stereotypical physical differences, saying it was easy to differentiate between people based on looks. A Tutsi also referred to these stereotypes, saying it is crucial for people to establish what groups others belong to:

Here it is a crime to ask people or accuse people of being one or the other group. But we still find out. There are many ways we use to figure it out. The first thing we look at is looks. For some people it is easy to establish what you are based on looks. Secondly we look at the behaviour. (…) It is easy to make a mistake when trying to judge what people are. (…)You need to live in this society to know exactly how this works. You need to just observe, sit back and look, and then you will know what group the person in question belongs to (11P, 2011).

This speaks to the continued need to still be able to categorise people into these groups. The existence of stereotypical physical group traits may make the groups seem more static and may make it harder to abolish these identities (see Gaertner & Dovidio, 2000).

Social Classes Having Become Ethnic Groups

Some respondents spoke of the identities as having become ethnic, and that the problem is that these identities now are ethnic groups. For example , a young man discussed that former social classes had come to be ethnic groups:

A Tutsi was someone rich; a Hutu was poor. If that had still been what they meant now when using these categories that would be fine because people can become both rich and poor. The problem is that now it means ethnic groups, and then talking about it, calling yourself Hutu, Tutsi and Twa, hinders the government policy of unity (20P, 2012).

This respondent claims the groups now are ethnic, emphasising the process of moving from social classes to ethnic groups (attesting to identity malleability). Interestingly, and in contrast to the extracts above, he uses the fact that the groups now are ethnic groups as an argument for the government policy of silencing the identities, not against. A student in Kigali put forward a similar argument:

The division between Hutu and Tutsi went from being the difference between social classes to ethnicity. Violence brought about this change. Belgium came in, and they imported their divisionism. They made us move from economic classes to tribal groups, and they shaped it well, so it grew strong roots (11P, 2011).

He here attests to the strong current representation of these identities as ethnic, even though he sees these as having been social classes before the Belgians.

Not Ethnic Groups

Several respondents focused on what these groups are not: the non-ethnic nature of these groups. Respondents used different “proof” to back up this identity representation.

Social Classes

One of the more common responses (and importantly, government endorsed) is that the groups were not ethnic groups, but social classes. The head of a unity implementation agency (45L, 2012) said: “Many signs and evidence show they are social groups. One could pass from one group to another one, because of number of cows or his richness. It really is a social group”. This was backed up by a student:

Everyone who could acquire wealth was called Tutsi. The word does not mean anything. And the poor class was called Hutu. The Tutsi were generally the people who had cows, and Hutu were in agriculture and hunting. In that kingdom, Tutsi were the ones ruling. Kings came from this group (11P, 2011).

According to this man, the groups were social classes, and membership hinged on wealth. In a group interview with a Hutu and a Tutsi, the Tutsi said: “For me, the groups are social groups. Expressions like “I am not your servant”, call servants Hutu. This proves it is social classes and not an ethnic group” (47P, 2012).

Related to the social classes, several respondents said one could move between the groups, as explained by a student: “Hutu could become rich and move from Hutu, as in the saying Umuhutu y’ihutuye, which means “he has come out of the class of Hutu and now he is in the class of Tutsi”, or “he is now Hutu-free” (11P, 2011). A political leader referred to this as “proof” against the ethnic nature of these groups:

Well, you could migrate from one position to another, so it means there is no identity there; if you can migrate from one position—if you are a Bahutu [Hutu, plural] you can go to Batutsi [Tutsi, plural], and it is valid, it is recognized by the society; it means that you are actually one people (35L, 2012).

Many respondents similarly defined the groups as former social classes. However, other statements indicate that this may not be as shared a sentiment, and that many people see the groups as ethnic. An academic in Kigali said: “In the strict sense, there are no ethnic groups in Rwanda. Many people will not agree with me, but it was social classes” (31P, 2012). The complexity of these representations is clear when comparing this theme to the third theme under the Ethnic representation as discussed above, where the focus was on that the groups had been social classes but now had become ethnic.

Shared Characteristics

Many referred to shared language, culture and religion as “proof” of the groups’ non-ethnic nature, as this civil society representative in Kigali: “For me, for us, we are not Hutu, Tutsi or Twa, we are not separate ethnic groups—we have the same language, traditions and culture. In Uganda and Kenya, the separate tribes have their own languages, traditions” (5P, 2011). This man further asks: without such differences, how can these be different ethnic groups? This was reiterated by an academic: “I will not say they are ethnic groups, we have the same language, same understanding of culture all over the country.” (31P, 2012). Similarly, a high level political leader said: “We [Rwandans] are the same at the level at culture, at the level of our language” (34L, 2012).

Clan Memberships

Several respondents also referred to clan memberships. Rwandan clans cut across the three subordinate groups, and clans commonly contain both Hutu and Tutsi, as explained by a musician in Kigali:

They told us we have three tribes, but among these, people come from the same clans. The clans come from one shared ancestor each, and we have many, many different clans, but most or all contain all three tribal groups. So it does not make sense to talk about it [ethnic groups] like this (12P, 2011).

These shared memberships are used as “proof” against the ethnic nature of these groups, as this political representative explained:

If you look at our societal composition in Rwanda today, we have what you commonly call ethnic groups, which has been described as ethnic groups but which are actually not; Batwa, Bahutu, and Batutsi—but when you go to the real clans (…) you will find that in those three classes you find different clans. I belong to the Basindi—you have a Mutwa [Twa, singular] who is a Musindi, you have a Tutsi who is a Musindi, you have a Hutu who is a Musindi. (…) The Western world and the Europeans who came here failed to understand this, and they took a machete, cut into the wood, carved out three groups (35L, 2012).

Using the clans and the associated shared ancestory, the respondent thereby challenges the logic of the ethnic labels applied to Hutu, Tutsi and Twa.

Within these two dimensions—ethnic and non-ethnic—various representations of the social identities are evident, even within individuals, who at times argue for both dimensions. Respondents clearly engage in meaning-making and construction of identity narratives within a complex reality.

Discussion

The above discussions of the Rwandan subordinate identities as ethnic and non-ethnic entail a story with many players, from colonial powers, to different groups in power, through extreme violence, to something on the other side. Now, on the other side—having gotten through the most difficult of the post-genocide period —the nature of these group identities is still unclear. For many respondents this distinction seemed to overlap to some extent with an “essential” and “non-essential” quality of the categories, where these should or should not be abolished (without any of the respondents necessarily prescribing to primordial views of ethnicity). Elcheroth, Doise, and Reicher (2011) argue that social representations critically influence how people act, and the way they see the world. Further, these are shared meta-representations, influencing people more powerfully than personal beliefs, and these shared beliefs influence reality. Several aspects make this unclear in Rwanda: the public metanarrative; the forbidden alternatives and the private narratives resisting or partly/fully agreeing with the metanarrative. In the following, the two dimensions—sentiments describing the categories as ethnic and as non-ethnic—are discussed up against the two core aspects of single recategorisation theory: inducing a common identity and abolishing the subordinate identities.

Inducing a Common Identity

In almost all the interviews people spoke of the importance of a shared Rwandan identity, and none argued against the need for this. Only a few respondents said directly that there was little improvement in levels of unity (e.g. “I think it is not that much change, as people have different problems” (46P, 2012); “People still stick on this issue [ethnic groups], and keep to this” (41P, 2012). Most respondents advocated for unity, and many referred to the salience of the Rwandan identity—both before the genocide, and now with the improvements in intergroup coexistence. Instead the main controversy was with regard to the second key aspect of single recategorisation.

Abolishing the Subordinate Identity

The different representations of the social identities vary in their fit with the single recategorisation model. Ethnic groups are often perceived as static (Verkuyten, 2012)—more so than non-ethnic categories such as social classes, which can change with alterations in economic status. If so, framing the group identities as social classes should—if successful—make these identities easier to abolish as they will seem less static. Those arguing for the identities as non-ethnic are not necessarily thereby claiming these should be removed, but may be more prone to support the single recategorisation model.

Several of those arguing explicitly for these groups as ethnic however, say in the same sentences that this is not where the problem lies and that the ethnic identities can be there, but that the focus should be on unity (e.g. 51P, 2012; 52P, 2012). These respondents do not argue against recategorisation, but indicate the need for a dual rather than a single approach where subordinate identities should be retained, nested within the superordinate Rwandan identity.

When it comes to group patterns, this study does not show ethnic differences in the representations of the subordinate groups (though only 34 of the respondents disclosed their “ethnic” group membership, so such claims cannot be made with certainty). The apparent lack of group differences in accounts of Rwandan identity may be due to several factors: the single recategorisation may have helped people to focus on the shared Rwandan identity instead; abolishing the contested “ethnic” groups may have made the public expression of aspects pertaining to these so difficult that such sentiments are not voiced; or these groups simply do not have separate social representations regarding this topic.

The material does however show that the interviewed Rwandan leaders—compared to the general population—are firmer on the need for the current identity approach (though with some variation, where one high level leader advocates directly against the ethnic groups being the problem, see sub-theme one under the Ethnic representation). The leaders also generally offer more elaborate reasoning to back up their arguments for the non-ethnic nature of these groups, and seem more aware of their choice of words, and what it means to call these groups ethnic. Such differences between the leaders and the general population are to be expected, as many of these leaders are directly involved in implementing the unity measures.

Limitations

The sample is small and non-representative, thus standing in the way of empirical generalisations. This was however never the intention. In getting to representations of social identities, the focus was on diversity and depth. Despite attempts to recruit both genders, few women were interviewed. Women approached often declined the invitation to be interviewed saying they had to work or that they were not comfortable discussing politics. This sample bias was detrimental for the study, and somewhat surprising in a country holding the highest number of women in parliament in the world. This increased gender balancing is still recent however, and has not necessarily translated down to lower levels.

Access to personal narratives and not mere reiterations of government propaganda is a challenge in Rwanda. There is a cultural and government imposed tradition of silence on sensitive issues, and people are sceptical as to whom they can trust. Research permits demonstrating permission to conduct interviews on this topic were shown to respondents in each interview. Repeated interviews with the same people were useful for establishing rapport, as were personal introductions and snowballing. The use of two research assistants seemed to facilitate openness, potentially as respondents had an “ally” when navigating histories of intergroup antagonism. All in all, the level of openness was surprising, and respondents frequently ventured far beyond the official narrative. The context is however sensitive, and awareness of the danger of re-traumatisation—both of respondents and research staff—was critical. Participants were given room to—but not pushed to—follow up on painful issues.

Conclusion

The categories of Hutu, Tutsi and partly also Twa have been actively placed at the core of this century’s conflicts in Rwanda. Asking Rwandans about their representations of these subordinate categories is key to find out more about how these formerly extremely polarised identities are spoken of now, but also to put these perceptions and representations in connection with the government imposed identity policy. The interlink between contextually grounded identity representations , leadership’s use of social identities and general population responses to such identity approaches is an area within peace psychology that requires further attention.