Women and the Rwandan gacaca courts: gender, genocide and justice

Abstract This article examines the gacaca trials of women accused of perpetrating the Rwandan genocide, asking whether and how ideas about their gender impacted their defences, testimonies and experiences as defendants. It uses court reports of the trials of 91 accused women; a set of sources that provides novel insights into the role of gender in an African transitional justice system. These sources reveal that ideas about gender – particularly female peacefulness and passivity – were commonly invoked by both accused women and wider trial participants. These gendered ideas not only helped women to achieve acquittals, but they also contributed to the Rwandan state's construction of a ‘truth’ narrative that ordinary Rwandan women are not capable of genocide violence. Additionally, women's trials reveal a further function of the gacaca process: as a political tool that made moral judgements about contemporary Rwandan women's domestic roles and place within the household.

monitoring was part of ASF's wider aims in Rwanda, which were to provide legal services; improve and build the capacity of the legal system, including gacaca; and undertake legal advocacy work (ASF a: ). Except for trial reports from , which are available to view on the organisation's website (ASF ), these remain unpublished and were sent to me in electronic form, to use on the condition that all names mentioned in the reports are changed. It is important to note that these reports do not record every word that was spoken in women's trials, nor every non-verbal action that was undertaken. Observers recorded their original observation notes in Kinyarwanda, before translating them into French for the final reports (ASF : ). This translation process means that there is the potential for translation errors to have occurred, and the reports do not give access to trial participants' original words. Additionally, the precise identities of the ASF observers are unknown. Individuals' decisions regarding what to record during a trial, and how, will have varied. Nevertheless, since these were legally trained observers whose aim was to record the legal aspects of trials, their role was to capture the legalities, arguments and narratives of cases, rather than simply 'soundbites' or 'sensational' moments. The content of the reports clearly reflects the observers' interest in the argumentation of cases, including what defence and accusation narratives were deployed; the arguments of witness testimonies; and how judges spoke to participants and delivered verdicts. The reports contain a high level of detailincluding quotations from testimoniespermitting both close analysis of women's testimonies and a broader thematic analysis of the evidence set. Nobody has conducted research using this set of women's trial reports before. It provides new and important insights into women's complex and varied trials, testimonies and defences, as well as of the narratives that gacaca courts allowed the Rwandan state to construct about women's involvement in the genocide.
The insights generated from these reports speak to broader conversations regarding both women's involvement in periods of mass violence, and postconflict assumptions about women's involvement. It is increasingly argued in feminist scholarship that there is little evidence for women being inherently less violent than men; rather, the social construction of their gender, and the resultant gendered expectations of societies, mean that men more commonly act violently during periods of conflict (Alison : ; Smeulers : ). There has also been growing scholarly recognition that women have indeed played a variety of combatant roles in recent conflicts in places such as Columbia, El Salvador, Eritrea, Ethiopia, Guatemala, Liberia, Nicaragua, Peru, Sierra Leone, Sri Lanka, Uganda and Vietnam (MacKenzie ; Smeulers ; Gentry & Sjoberg ). Furthermore, although the area remains under-researched, some existing work points to how post-conflict justice and resolution processes have failed to comprehend fully women's capacity to commit violence. Gentry & Sjoberg () identify how narratives of women's political violence across the globe tend to portray these women as either 'mothers', 'monsters' or 'whores', who respectively were supporting  W O M E N A N D T H E R W A N D A N G A C A C A C O U R T S and avenging their sons; were extraordinary non-women; or were led astray by the 'evils' of female sexuality. Bernal () points to how women who fought for the Eritrean People's Liberation Front were stigmatised by their communities and seen as undesirable wives upon returning from the conflict. MacKenzie () argues that the de-securitisation programme in Sierra Leone narrativised women combatants as 'wives', 'camp followers' or 'sex slaves' due to assumptions that women were not violent. Rwandan women who participated in the genocide were not gendered anomalies, nor did they represent an isolated case of women's involvement in conflict. The narratives generated by gacaca in its attempts to make sense of women's genocide involvement fit into a wider story of the ways that post-conflict societies have struggled to comprehend fully women's violent agency.
Building on these conversations, this paper also contributes to a growing literature on the challenges and pitfalls of 'transitional justice'. It reacts to growing calls in transitional justice scholarship for analytical critiques of transitional justice's gendered 'blind-spots' towards actors who do not fit the binary male perpetrator : female victim understanding of conflict (Björkdahl & Mannergren Selimovic ; Schulz ). This paper specifically responds to this identified lack of research on accused women perpetrators in transitional justice mechanisms. This paper also contributes to wider discussions regarding African women's agency, voice and 'empowerment' in state-run court systems. Gacaca formed part of a wider and longer story of court systems providing women in African countries with opportunities to gain agency and material benefits through speaking in legal settings (Roberts ; Shadle ; Mujere ). Yet, other scholars have identified that this 'empowerment' has coexisted in tension with some women's aims of exploiting gendered assumptions, such as female dependence, to achieve favourable trial outcomes (Mutongi ; Zimudzi ). The speech acts of women in gacacaparticularly those that emphasised female passivity and subservience, as well as those that were compelled due to women's position as accused individualsfurther problematise assumptions about the relationship between African women's speech in court settings, and their 'empowerment'.
Thematic analysis of the ASF report set of  trials, and close reading of individual reports, reveal that gender played a prominent role in the gacaca trials of accused women. Many accused women used expectations about their genderparticularly motherhood and subservienceto defend themselves successfully against charges of genocide. Other trial participants' testimonies also spoke to beliefs about women's incapacity to have a will to commit genocide. Conversely, but similarly linked to a reluctance to consider ordinary women's ability to commit genocide, ideas about femininity worked against women where their violent actions as wives and mothers were judged to have transgressed the boundaries of Rwandan womanhood. Gacaca was thereby a justice process that constructed a state-authorised 'truth' narrative that ordinary Rwandan women were not capable of genocide. Court debates concerning  B E T H B R E W E R women's agency also expose a further function of gacaca: as a political tool that made state-sanctioned moral judgements about contemporary Rwandan women's domestic roles and place within the household.
Before examining the content of the reports, it is necessary to consider the nature of Rwanda's post-genocide justice system; in particular, its role as a truth-generating process for the Rwandan state. This understanding of gacaca, informed by existing scholarship, allows for an analysis of women's trials as processes that were not only impacted by ideas about women's gender, but that also generated state-authorised 'truth' narratives about women's genocide involvement.
After the  genocide, and in line with a growing international movement towards transitional justice in post-conflict scenarios, the newly established Rwandan government aimed to hold all those who had perpetrated the genocide to account in courts of law (Leebaw : ). The UN established the International Criminal Tribunal for Rwanda in , but it only tried the main instigators of genocide (Thomson & Nagy : ). In , the Rwandan government established four categories of genocidal crimes, which in  were reduced to three: category one included instigating genocide, and sexual assault; category two encompassed killing and assault; and category three covered property-related offences (Schabas : -, ). National trials for genocide started in December . However, it became apparent that the existing court system was not sufficient to try all accused perpetrators (Thomson & Nagy : ). To resolve this issue, the government stated that it would revive and adapt what it described as Rwanda's 'traditional' gacaca community justice system to implement justice relating to the genocide (Karekezi et al. : ). A pilot phase of post-genocide gacaca began in , with the court system rolling out nationally in  and concluding its work in .
Post-genocide gacaca was a locally situated, but state-mandated and -controlled, justice system. Its structure reflected the administrative structure that had been in place since Belgian colonial rule, which split the country into sectors and those sectors into cells (Schabas : ). Courts were divided into three levels: there were  cell courts, which were responsible for the pre-trial phase of collecting information as well as trying category three crimes;  sector courts, which tried category two crimes and from  also category one crimes; and  courts of appeal (Nyseth Brehm et al. : -, ). Each gacaca court was scheduled to meet at least one day each week until all accused perpetrators in its jurisdiction had been tried (Ingelaere : -). Seven civilian judges and two substitutes were elected from among the local community to preside over the courts' cases (Bornkamm : ). The state decreed that all adult members of the community should attend gacaca sessions and form the general assembly, although attendance varied across courts and dropped over time (Bornkamm : ; Ingelaere : ).
In the absence of other evidence, spoken testimony played a crucial role in gacaca trials. The precise structure of proceedings varied, but in a typical trial, the accused was summoned before the bench to give their defence statement and answer questions from the judges. Witnesses were then invited to speak, before proceedings were opened to the general assembly for questions and comments (Human Rights Watch : ). The judges then deliberated privately, before returning with their verdict (Clark : -). Every week in these court spaces, members of local communities told competing stories of genocide events with the aim of achieving their desired trial outcome. The communally situated nature of courts meant that local power dynamics, social capital and interpersonal relations impacted who could speak in these spaces and what they felt able to say (Buckley-Zistel ; Brounéus ; Funkeson et al. ; Ephgrave ).
Through these competing stories and resultant verdicts, gacaca courts not only formed judgements about each individual's genocide guilt, but also made broader moral statements about what actions and mentalities were considered 'genocidal' (Doughty : -; Geraghty : -). Gacaca's moral judgements were situated within a context of wider state concerns and laws in Rwanda regarding 'genocide ideology'. This 'genocide ideology' is a state-generated narrative that contends that certain individuals harboured, and continue to harbour, a will to exterminate the Tutsi ethnic group, and that these internal psychologies need to be uncovered and punished by the post-genocide state (Begley : -; Geraghty : ). In this context, gacaca was a state process that made moral judgements not only about individuals' actions, but also their psychologies and will to commit genocide.
Gacaca was a political tool that helped the ruling party, the Rwandan Patriotic Front (RPF), to project state power into local communities and generate further state power in return (Longman ; Thomson ; Loyle ). Chakravarty (: -) argues that each participant's action in the performance of gacaca justice legitimised both the justice administration and the wider right to rule of the RPF regime, thereby increasing the regime's power. The Rwandan state also claimed that gacaca would reveal the 'truth' about genocide events (Waldorf : ; Sosnov : ). This claim was situated within, and appealed to, wider international expectations of transitional justice institutions. Linked to claims about the relations between truth-telling, healing and reconciliation, transitional justice institutions have often been implemented with the aim of revealing truths about past atrocities, without fully problematising what those truths actually are, whether they exist, and what the political and societal consequences might be of creating justiceand state-authorised truth narratives about contested past events (Daly : -; Chapman : -; Clark : -).
Despite this state claim, gacaca was not a truth-telling mechanism. The punitive nature of the process meant that some participants were incentivised to lie,  B E T H B R E W E R while others simply did not want to give their full recollection of genocide events in this environment (Buckley-Zistel : ; Rosoux & Shyaka Mugabe : ; Rettig : -). The notion that gacaca could reveal the truth about genocide also assumes the presence of an objective, independent 'truth' of events. It does not consider that multiple individuals might experience a single event in different ways. Even for one individual, it assumes that a person has full knowledge of what they did and what their motivations were. It was unrealistic, and even impossible, for gacaca to have revealed the 'truth' about genocide events.
Nevertheless, the state's claim that gacaca would reveal this 'truth' had a political importance in itself. The post-genocide Rwandan government has a widespread aim of dictating societal 'truths' about domestic politics, foreign affairs and Rwandan history, in the belief that controlling these narratives will allow it to monopolise power (Desrosiers & Thomson : -; Reyntjens : -; Eramian : -; Laws : -). Gacaca sat alongside genocide commemoration events, 're-education camps', laws concerning 'Rwandicity' and 'genocide ideology', and the elimination of political opponents, as one of the mechanisms used by the government to construct state-sanctioned 'truth' narratives about past and contemporary politics (Ingelaere : -). Gacaca's national status as a 'truth-revealing' process meant that the narrative accepted by the bench in its verdict gained a truth status, and became the state-generated knowledge of how the genocide had occurred and who hador had notparticipated in it. By creating 'truths' about genocide events under the guise of revealing them, the gacaca process allowed the regime to control the narrative of the genocide, as well as to affirm its role both nationally and internationally in the punishment of perpetrators, the reconciliation of communities, and the rebuilding of the post-genocide nation. Gacaca was an important tool that legitimised and generated power for the regime that remains in place to this day.
Since gacaca did not reveal the 'truth' of genocide events, the court reports cannot be used as conclusive historical evidence for how accused women had acted during the genocide. Instead, building on this understanding of gacaca as a state institution of truth construction, they can be analysed to ask not only whether and how women's trials were impacted by their gender, but also what 'truths' the state constructed about Rwandan women's genocide culpability.
The reports show that pre-existing conceptions about gender tended to help women defend themselves against charges of genocide. Several women successfully employed ideas about female peacefulness and passivity to argue that, because they were women, they could not possibly have committed these acts of violence. These defences were situated in a wider context of long-standing gendered expectations about Rwandan women's passivity, subservience and behaviour. Before the genocide, a married woman was expected to remain within the household, only leaving home for necessities and to visit her parents (De Lame : ). Men were the legal heads of households, and married women were not able to control household resources, own land, or take on economic work without their husbands' consent (Burnet : ; Mwambari : ; Stern et al. : ). Those women who did undertake public work were expected to behave as 'virtuous wives' or 'virginal daughters' (Jefremovas ). Expectations around women's behaviour continued into the post-genocide period. For example, the narrative of women as peacebuilders played a significant role in the postgenocide national drive to include more women in politics (Uwineza & Pearson : -). These gender norms meant that it was difficult for Rwandan communities to work an understanding of women's genocide culpability into their pre-existing narratives of women's roles and behaviour, especially where women presented and reinforced these narratives when on trial.
As mentioned already, much occurred in and around gacaca courts that the reports do not capture, including individuals' social status and relations. It is therefore difficult to draw direct conclusions about individual judges' motivations for delivering the verdicts that they did in each case. However, for the purposes of considering gacaca's 'truth' generation, it is more pertinent to consider what narratives about women's violence the courts generated through the pronouncement of these verdicts. Regardless of what other factors influenced the judges' decisions, the arguments and testimonies produced in court formed the publicly recorded, and then gacaca-authorised, stories of these women's genocide involvement. The judges' authorisation of these gendered defences in their verdicts shows the construction of a 'truth' narrative by gacaca that ordinary Rwandan women were not capable of committing genocide.
Most common of these gendered defences was the argument that the accused could not have attacked or killed because she was a mother. Virginie was put on trial in August , accused amongst other crimes of having beaten and tortured a male victim. In her defence testimony, the observer recorded that 'L'accusée ajoute qu'elle ne pouvait pas faire du mal à la victime d'autant plus qu'il est son gendre' [The accused adds that she could not have hurt the victim especially since he is her son-in-law] (ASF b: ). Virginie explicitly argued that her role as the victim's mother-in-law prohibited her from committing an act of violence against him.
Another woman, Pauline, was tried in January , accused of abandoning her four children to attackers so that they could be killed. She pleaded not guilty, stating that the man she was on trial alongside was the person who had forced her to leave her children behind and who had alerted the attackers to their presence. Under questioning from the judges about whether she had told anyone that she had left her children because they had 'du sang Tutsi' [Tutsi blood], Pauline responded that  B E T H B R E W E R Je n'ai jamais dit une telle chose. Comment est-ce qu'une personne douée de raison, qui a donné la vie à des êtres humains, peut vouloir leur mort?
[I never said any such thing. How is it that a person endowed with reason, who gave life to human beings, can want their death?]. (ASF b: -) Not only did Pauline argue that she personally could not have killed her own children, but by using this rhetorical question she appealed to the idea of a general truth that no sane mother could possibly want her children to die. She asked the judges to refer to their 'knowledge' that the state of motherhood was incompatible with child-killing.
Both women were successful at invoking their motherhood to defend themselves and were acquitted of the charges against them (ASF b: ; b: ). Since gacaca did not reveal the 'truth' of genocide events, it is impossible to know from the reports alone whether either woman had taken part in attacks and killings, as well as whether these motherhood defences reflected genuine beliefs on either of their parts that this identity inherently prevented them from committing such violence. Regardless, their use of such defences in gacaca shows that these women believed that a story of motherhood being incompatible with these accusations would help them to achieve an acquittal in this justice system. Their acquittals meant that, although the possibility of mothers being killers had been discussed in these trials, the judges authorised these lines of defence. Their verdicts constituted a state-generated societal 'truth' that sane and ordinary mothers were not capable of wanting their children to die, and so had not taken part in these genocide killings.
Other accused women aimed to use the assumption of their subservience to their male heads of households to deny their agency in the perpetration of genocide violence. The reports provide examples of accused women drawing on their expected female position within the household to argue that they did not have authority over what took place inside their home, and therefore had no responsibility for any crimes that took place there.
In this context, we return to the trial of Agnès, who stood before an appeals court in April . She was appealing her conviction for the crimes of refusal to testify about her knowledge of genocide events, and intimidation of witnesses. The observer recorded that Invitée à réagir sur ces allégations, elle affirme qu'elles sont fausses et sans fondement, parce que pendant le génocide, son mari n'était pas à la maison.
[Invited to react to these allegations, she affirms that they are false and unfounded, because during the genocide, her husband was not at home]. (ASF a: ) In this defence statement to the judges, Agnès presented the story that she could not have committed genocide crimes without the presence of her husband in their home and his consent. Her story was supported by a defence witness, who said that she was taking refuge in Agnès's house during the genocide. This witness did not testify that Agnès did not kill; she only confirmed that Agnès's husband was away, having left at Easter for his parents' house, and that the couple was only reunited after the killings (ASF a: ). Together, these testimonies presented a story that Agnès was a good and 'traditional' Rwandan wife who had stayed at home, and who could not have had involvement in the genocide because her husband had not been there to facilitate it. No other witness testimonies were presented, and the bench acquitted Agnès (ASF a: ). It cannot be known for certain that the argument in these testimonies was what convinced the judges to find Agnès not guilty. However, in delivering a verdict that followed this argument, in a court where no other argumentation was presented, the judges generated a court record, and a state-authorised narrative, that this gendered story had led to Agnès being found innocent.
Béatrice also relied upon her assumed position as deferential to her male head of household in her defence, but unlike Agnès, her story of innocence relied upon the presence of her head of household, not his absence. Béatrice told a story that decisions taken about Tutsis hiding in her house during the genocide were made by her father, not her. She had been detained in prison as a suspected perpetrator in January , before appearing in gacaca in August  accused of complicity in the killing of two Tutsis. Béatrice pleaded not guilty, and the observer recorded that in her defence statement [She makes it known that the victims did indeed come to her house and that her parents took them into the house and showed them a hiding place. Some minutes later, M and P came and threatened her father that they would kill all his family with a grenade if he did not hand over to them the Tutsis who were hiding in his house. Panic-stricken, her father showed them the victims' hiding place and they took them away to kill them]. (ASF c: -) By telling this story, Béatrice emphasised the agency of her parents in this genocide event. She argued that it was her parents who took the decision to hide the Tutsis in their house, removing herself even from this alleged act of rescuing. Above all, Béatrice argued that it was her father who actedalbeit under duressto reveal the victims' hiding place and contribute to their eventual deaths. She said that it was to her father that the attackers spoke, presenting a story that these attackers addressed the head of the household by threatening the family he supposedly protected. In her story, she herself took no action, and had no responsibility over either the house being a hiding place or the victims being given to attackers. Instead, she argued that the house in which she lived was a location where her parents, but most of all her father, had authority. Nobody present in court challenged this story. The judges did not question Béatrice about whether she had taken any action on this day of  violence, and the only witness who spoke said that she was hiding and so had no knowledge of these events. Béatrice was found not guilty (ASF c: ). The court thereby authorised her story that her father was responsible for decisions about what occurred inside the house, while she was simply a passive observer to this genocide killing.
Agnès and Béatrice provide examples of women using their female roles within the household to argue before gacaca that they were deferential to their male head of household's decision-making. Agnès contended that without her husband's presence in their home, she could not possibly have taken part in genocide events. Béatrice argued that decisions about what happened within her home, including whether or not to hide Tutsis, were taken by her father, relying on his presence to absolve her of responsibility in this violence. Both women were acquitted, with the courts accepting two contrary, but compatible, points stemming from the belief of women's subservience to their male heads of households: that Rwandan women could not take decisions without their husband or father, but that when women were with these men, they had to obey them. For these cases, the verdicts established 'truths' that Rwandan women's subservience to their male heads of households absolved them of genocidal culpability.
Women's choices to use gendered defences took advantage of pre-existing ideas in Rwanda about female peacefulness, agency, and domestic roles. The gacaca process undoubtedly forced communities to discuss the possibility of women's violence for those women who were accused. Yet, the success of these defences shows not only that the post-genocide justice process was impacted by these gendered assumptions, but also that gacaca was an institution that generated new state-authorised 'truths' of ordinary female roles being incompatible with genocide violence.

R A T I O N A L I S I N G W O M E N ' S V I O L E N C E
Arguments for women's gendered peacefulness and inability to commit genocide did not just come from women in their own stories, but also from other participants in their trials. Witness statements and judges' verdicts show the emergence during some women's trials of a disbelief that women could want to commit genocide. As well as those who denied women's participation, other witnesses and audience members admitted trying to rationalise women's violence by providing other, non-genocidal, explanations for their violent actions. These participants acknowledged women's violence, but presented women as acting violently in isolated, personal and understandable incidents related to intermarital problems, rather than acting violently as a result of 'genocide ideology'. These attempts to explain women's violence exposed pre-existing ideas about women's psychologies and violent capabilities. The acceptance of these stories in judges' verdicts then turned them into a state-generated 'truth' that ordinary Rwandan women were not capable of thinking and acting with genocidal intent.

W O M E N A N D T H E R W A N D A N G A C A C A C O U R T S
Gladys was put on trial in June . She was charged with false testimony, in relation to a statement that she had given as a witness in a previous trial about how two victims had stayed at, and left, her home before being killed (ASF b: ). She had claimed, and stated again in this trial, that the victims had come to her home briefly while her husband was out but then left of their own accord to go to her father-in-law's house, and that as a result she did not know anything about their deaths. She presented herself as a wife left at home, with no active agency in the victims either leaving or being killed. However, her father-in-law contested this narrativewhich implicated him as the last known person to have seen the victims before their attackclaiming instead that the victims had never come to his house (ASF b: , ). These competing narratives were dangerous for Gladys and her father-in-law: betraying the hiding place of Tutsis or leading them to their killers were treated by gacaca law as acts of complicity in killing (Hogg : ).
The judges' questioning and the wider discussion from witnesses and audience members revolved around the question of whether the victims had left Gladys's house out of choice, as she claimed, or whether Gladys had made them leave. The court thereby aimed to establish whether Gladys was lying and had in fact acted in a violent manner to drive the victims out of their hiding place and lead them to their attackers. During their questioning of Gladys, one judge said that it was 'incompréhensible que les victimes soient sorties d'elles-mêmes' [incomprehensible that the victims left of their own accord]. Other participants, including her father-in-law, testified that Gladys was lying (ASF b: -). A narrative was beginning to be formed that Gladys must have driven the Tutsis out since it was unrealistic for them to have wanted to leave the place where they were hiding from the violence of genocide.
Significantly, however, what emerged during this court discussion was not just the question of whether Gladys had driven the victims out, but also what could have motivated her to have done so. One participant in particular, a judge, aimed to lay claim to Gladys's motivation for this violent act. The observer recorded that Un juge déclare que lors de la collecte d'informations, [F] avait déclaré qu'elle s'était réfugiée chez l'accusée et que cette dernière l'avait chassée malgré l'opposition de son mari; que l'accusée avait insisté en disant qu'elle ne pouvait pas vivre avec une autre femme dans sa maison.
[A judge declares that during the information collection session, F had declared that she was taking refuge at the accused's home and that the accused had driven her out despite the opposition of her husband; that the accused had insisted by saying that she could not live with another woman in her home]. (ASF b: ) With the authority of a person who held a role of court power, this judge recalled what another witness had said in the initial cell-court trial phase and affirmed their explanation for why Gladys had committed this violent act. He presented a story that Gladys had acted against her husband's wishes and driven the victims out because she was emotionally unable to cope with a female rival for her husband's affection in her house.

B E T H B R E W E R
Ultimately, through its verdict, the bench accepted the story that Gladys had given false testimony related to the victims leaving her house and their subsequent deaths, and it sentenced her to three months' imprisonment (ASF b: ). By implication, the bench authorised the narrative that emerged during the trial that she had indeed driven the victims out because she could not live with another woman in her home. This verdict raises questions about the moral judgement that the court gave to Gladys's actions during the genocide. Firstly, the court judged that, unlike the other women discussed, Gladys was capable of exerting agency within her home against her husband's wishes. Secondly, Gladys was convicted of false testimony, meaning the court only sanctioned her for lying. The court did not sanction her for the action about which they determined she had lied: driving the victims out of their hiding place. Nor did the bench in this trial determine that she should be tried separately for these alleged actions, implying that it was not a genocide crime worth being investigated in gacaca. The motivations for such a judgement cannot be known for certain from this evidence alone. Nevertheless, the judgement raises the question of whether the court had rationalised and explained away her actions such that they were no longer deemed 'genocidal'. The court made a moral judgement that, rather than committing this act out of genocidal intent against the victims due to their Tutsi ethnicity, Gladys was a wife who had acted out of jealousy and without her husband's consent to drive a female rival out of her marital home.
Gladys was not the only woman whose alleged actions during the genocide were explained by court testifiers as having been motivated by jealousy. Grace was tried in November , accused of alerting the assailants who killed a victim. She pleaded not guilty and said that she and her husband were hiding Tutsis when attackers came and took the victim. Grace's accuser, the victim's brother, testified that Grace's husband had hidden the victim pour en faire sa femme … Etant donné que l'accusée n'était plus nourrie par son mari, elle est allée alerter des assaillants qui ont emmené les victimes.
[to make her his wife … Given that the accused was no longer being provided for by her husband, she went to alert the assailants who took the victims away]. (ASF c: -) As with those who tried to rationalise Gladys's actions, the victim's brother argued that Grace had acted violently towards this victim because the victim constituted a rival to her position as a wife. Her accuser might have been relaying these events as he understood them, or he might have told this story in a way that he thought would make her violent actions believable to the judges. Regardless of his intentions, his decision to present this narrative of Grace as a jealous wife speaks to a desire, even from an accuser, to provide a more understandable and palatable explanation for women's violence than a will to commit genocide.
The emergence of these narratives in gacaca shows that, even in trials where women's violence was recognised by many to have occurred, court participants struggled to accept that women could be motivated by a will to commit genocide against the Tutsi ethnic group. They preferred instead to believe that women's violent actions stemmed from domestic disputes. There is, of course, a tension between this narrative and the one created in courts where women successfully emphasised their female domestic role: this second narrative deemed that women overcome by jealousy were capable of exerting a certain type of agency in their homes without their husbands' consent. These courts subsequently punished women for exerting this domestic power. Ultimately, this narrative meant that courts debated and judged whether women were jealous and violent wives, above whether they were genocide perpetrators. Verdicts from these trials had wider societal consequences as these non-genocidal reasons for women's violence, and most importantly women's identities as jealous wives rather than genocide perpetrators, became the state-established 'truth'.

N A R R A T I V E S O F G E N D E R E D T R A N S G R E S S I O N S
Gacaca courts did, however, confront some women's genocidal intentions. In comparison to those women for whom gendered defences and narratives helped them to achieve favourable outcomes in gacaca, the report evidence suggests that there was stigma in trials towards those women whose genocide actions were deemed to have transgressed gendered expectations of female submissiveness, especially in relation to their domestic roles. Women faced punishment for accusations of exerting power over their male relatives and inciting them to commit violence. Although the nature of these charges fits with wider research that women often played 'supporting' roles to men in the perpetration of genocide violence rather than participating in attacks and killings themselves (Jones : ; Hogg : ; Nyseth Brehm et al. : -; Brown : ), the debates in gacaca suggest that these women were not seen by court participants as 'supportive' actors in this violence, but instead as instigators who exerted power over their male relatives. The recorded evidence of testifiers' words cannot reveal for certain the motivations behind each individual's decision to accuse or judge women for these crimes. Nevertheless, whether individual testifiers' allegations were deliberate or unconscious attempts to punish women for transgressing gender roles; whether they represented testifiers' understandings of how women had acted in the perpetration of genocide; or whether they stemmed from other reasons, these allegations all took place in the context of a widespread expectation that virtuous Rwandan women should stay deferential to their male relatives (De Lame : -; Hogg : -). They also took place in a country where communities often stigmatised women who transgressed gender norms. For example, journalists in the s and s portrayed women politicians as controlling their husbands, and ridiculed them in a sexualised manner (Holmes : -; Watkins & Jesse : -). Jefremovas () points to how businesswomen in the s risked being labelled as 'loose'. Such a stigma continued in the post-genocide period, with female street hawkers and sex workers facing arrest (Berry : -). Also in the post-genocide  period, it was common for families and communities to ostracise women who reported their abusive husbands to the police (Kagaba : ). Jessee () furthermore points to how suspected women genocide perpetrators faced heightened stigma due to their gender. Her interviewees reported being avoided in the street, losing business, not being visited by family members in prison, and no longer being seen as a woman. In this context, the evidence suggests that communities and the state judged these women in gacaca for two interconnected transgressions that fed into one another: firstly, for genocide violence, and secondly, for transgressing their gendered domestic identity. It appears that the transgression of their gendered domestic identity made allegations of their genocide violence more believable and comprehensible. Court attitudes and stigma towards these women did not sit in opposition to the reluctance to comprehend fully women's capacity for violence, but rather fed into and created a state 'truth' that those women who had displayed a will to commit genocide had deviated from their natural gendered submissive and peaceful states, and were not 'ordinary women'. As discussed, many gacaca courts debated accused women's domestic agency and roles. Building on this analysis, the gacaca trials explored in this section highlight most strikingly a further function of gacaca courts: as a political tool that made state-sanctioned moral judgements about contemporary Rwandan women's domestic roles and place within the household.
Ruth's trial, in May , is one such case where a woman was punished for transgressing her gendered domestic role. Her trial centred around two competing narratives of womanhood. Her narrative was that she was a submissive wife incapable of agency in her husband's absence, while her accusers' narrative was that she was a controlling wife and mother who had exerted power over her male relatives in the perpetration of genocide. Ruth stood accused of having taken part in a killing and a looting, and of denouncing a victim who had taken refuge in her house. Ruth pleaded not guilty and claimed that when attackers came to her door to find the hiding victim, she told them that 'je ne pouvais leurs permettre de défoncer la porte sans la présence de mon mari' [I could not allow them to break down the door without the presence of my husband], and that the attackers then left to fetch him (ASF a: ). Like women already discussed, Ruth presented herself as a woman who was not capable of involvement in this genocide act, since she could not even let the attackers into her home without the consent of her male head of household.
However, in Ruth's case, this defence of gendered passivity did not go unchallenged. As well as witnesses claiming that she had denounced the hiding victim to attackers, further allegations emerged during her trial about her incitement of her son to commit genocide. These accusations were not listed as charges in her dossier, but during her trial a member of the audience stood up to say that he was a judge in the cell court and that lors de la collecte d'informations au niveau de la juridiction de cellule, plusieurs personnes ont témoigné en disant que l'accusée aurait dit à un de ses fils ceci: «si tu ne tues pas [L], n'ose plus remettre tes pieds chez moi».


[during the information collection at the level of the cell court, several people testified saying that the accused apparently said this to one of her sons: 'if you do not kill L, do not dare to set foot in my home again']. (ASF a: ) This same allegation was reiterated by a later witness, who testified that Un autre jour, je me rappelle que l'accusée a demandé à son fils qui vivait encore sous le toit parental d'aller tuer [L], en ajoutant que s'il ne le faisait pas, il n'allait plus remettre les pieds dans sa maison.
[Another day, I remember that the accused asked her son who still lived under his parents' roof to go and kill L, adding that if he did not do it, he was no longer going to set foot in her house]. (ASF a: ) These testifiers entered a further allegation against Ruth, linked to a killing carried out by her son. Although her son had allegedly killed the victim, the testifiers laid a significant proportion of the blame for this murder on Ruth, as the mother who had exerted power over her son to commit this act of violence. They did not see her as supporting his genocide violence; rather, they saw her as instigating it, using her authority over him and their home to ensure that the victim was killed. Compared with the 'peaceful mother' defences used by many women in gacaca to claim that they could not possibly have harboured a desire to kill, the allegation that emerged during this trial was of a manipulative, controlling, powerful mother who had forced her son to kill by threatening him with no longer being allowed in the family home. This crime was not one with which she was initially charged, but the judges' verdict stated that having 'Incité un de ses fils à tuer' [Incited one of her sons to kill] was one of the charges of which she was convicted, along with denouncing the hiding victim to attackers so that he could be killed (ASF a: ). The court therefore deemed that her alleged actions towards her son constituted the perpetration of genocide.
Ruth's trial dealt with the fundamental questions of whether women could exert power in their homes and over their male relatives, and what should happen to those women who had done so. Compared with other women, Ruth's defence of being powerless in her role as a wife under the direction of her husband was not successful: the judges deemed that she had the power to determine what occurred in her home and that she had made active decisions leading to the hiding victim's death. Similarly, they judged that she had used her motherhood and the resultant power she had over her son for genocidal means. The gacaca court sanctioned Ruth significantly for these acts, sentencing her to  years' imprisonment (ASF a: ), a sentence corresponding to a conviction of killing, or of injuring with intent to kill (Hola & Nyseth Brehm : ). The narrative that emerged during this trial portrayed Ruth as a woman who had transgressed her natural peaceful and submissive role as a wife and mother to exert both violent power and power over her home and male relatives. This finding builds on wider research that identifies Rwandan men's perceptions that women's gains in political, economic and social rights in the post-genocide period have led to their own wives and female relatives  challenging their household authority (Burnet : -; Stern et al. : -). This wider research also points to how such male perceptions have led in some instances to views that the institution of marriage is being eroded, as well as to increased marital conflict and episodes of domestic violence (Burnet : ; Carlson & Randell : ; Slegh et al. : ).
In the context of this post-genocide reaction of many men to what they see as women stepping outside their natural subservient domestic roles, the recorded evidence shows that gacaca became a state institution that both heard and made moral judgements about what should happen to those women who had exerted power within their household and over their male relatives.
Similar narratives of controlling women exerting unacceptable power over their male relatives emerged in other trials. During Anne's trial in October , a witness spoke to say that they had croisé l'accusée en chemin et que celle-ci menaçait son mari de tuer [D]. Celle-ci a été tuée quelques minutes après. … L'accusée a également demandé à son fils du nom de [H ] de venir lui montrer le corps de [R].
[passed the accused on the way and that the accused was threatening her husband to kill D. D was killed some minutes after. … The accused also asked her son called H to come and show her the body of R]. (ASF : ) As was the case in Ruth's trial, an allegation that was not a listed charge emerged that Anne had exerted power over her husband and son in a way that was both threatening and unnatural for a Rwandan woman. This allegation, which was corroborated by other witnesses and formed a prominent narrative generated in this court, also implied that Anne had responsibility for these two men's genocide actions. Despite Anne's protests that she could not be held culpable for crimes committed by her husband, she was found guilty and sentenced to life imprisonment for a genocide perpetration that included her alleged power over her male relatives' violence (ASF : , ).
The allegations that emerged in these trials must be seen in the context of wider societal ideas about both female peacefulness and female domestic identities. Where a woman had transgressed one gendered identity, it was easier to comprehend that she could also have transgressed another. The punishment of Ruth and Anne for acting violently and for exerting power over their male relatives did not sit in opposition to the reluctance to comprehend fully women's capacity for genocide. Rather, it helped to produce the court-and stateauthorised 'truth' that those few women who had wanted to participate in the extermination of the Tutsi ethnic group had transgressed women's natural state and were gendered anomalies rather than 'ordinary women'. These testimonies, and especially the gacaca courts' final verdicts, also sent a broader message that was not directly related to the genocide: that good Rwandan women should not exert power over their male relatives. In this way, gacaca courts not only judged what had occurred during the genocide. They were also state tools that produced moral statements about how women should behave in contemporary, post-genocide, Rwandan society, with gacaca trials becoming one of various ways that Rwandan communities stigmatised women who transgressed gender norms.

C O N C L U S I O N : A G E N D E R E D P O S T -G E N O C I D E J U S T I C E
Gacaca did not reveal the 'truth' of genocide events or women's involvement in them. Yet, the state claimed that it did, and this political function meant that the narratives accepted by courts in their verdicts gained a particular state-authorised 'truth' status, regardless of what had happened in . Talking about the genocide in these court spaces created a set of knowledge and 'truths' about how the genocide had occurred, and who hador had notperpetrated it.
In many respects, the gacaca process expanded the boundaries of discussion about women's genocide agency. This agency had largely been ignored in Rwandan society in the decade since the genocide, and gacaca courts forced local communities to debate women's capacity for violence for those women who had been accused. This paper does not claim that no gacaca court comprehended that ordinary women were capable of committing genocide. Gendered narratives did not explicitly enter all trials, and for many women who were convicted, their guilt was seemingly accepted by the community as well as established in court and state records.
Yet, gendered defences, arguments and verdicts were a theme across women's gacaca trials, and thus contributed to creating at least one of the state's 'truth' narratives about the genocide and who its perpetrators were. The content of women's trial reports shows that many women's gacaca trials were impacted by ideas about their gender; most notably, the peaceful and passive nature of Rwandan womanhood. Many accused women employed expectations about motherhood, peacefulness and domestic identities to argue successfully that they were not capable of committing genocide. These narratives of women's inability to act and think with a will to commit genocide also emerged from other court participants, who argued that where women's violent actions had occurred, they resulted from episodes of marital jealousy rather than a desire to commit genocide against the Tutsi ethnic minority. Where these narratives were accepted by judges in their verdicts, as they often were, they not only reaffirmed these gendered ideas but gave them a state-authorised 'truth' status in post-genocide Rwandan society.
The evidence also suggests that some women faced heightened stigma in courts for allegations of transgressing their gendered domestic identity and exerting power over their male relatives in the incitement of genocide violence. While these women's genocide actions and intent were accepted by the courts, they were accepted in the context of beliefs that these were women who had exited their natural female state and acted outside the confines of Rwandan womanhood. Verdicts in these cases fed into court narratives and state-sanctioned moral judgements that ordinary Rwandan women were not capable of wanting to commit genocide, and that those who had committed violence were either acting for non-genocidal reasons or were gendered anomalies.
It is important to recognise that the details of these verdicts and 'truth' narratives were not always consistent across gacaca courts. In fact, some of these inconsistencies in women's trials reveal a further function of the gacaca courts when making moral judgements about women. Internal tensions in narratives about women's violence appear most evidently in judgements about women's ability to exert agency within their own homes. Some women successfully argued that they were subservient to their male heads of households and were not responsible for what occurred in their domestic space. Conversely, other courts accepted and established narratives that jealous wives were successfully able to drive out women taking refuge in their homes, without their husbands' consent. Some courts also sanctioned women for exerting power and control over their male relatives and their domestic space. These debates and inconsistencies reveal that gacaca courts were not just contending with women's violent agency, but also with what their agency and power was, and should be, within their domestic lives. Gacaca courts had a secondary function as a political tool that made state-sanctioned moral judgements about contemporary Rwandan women's domestic roles and place within the household. These findings thereby further an understanding of the present-day Rwandan state's attempts to control narratives about the country's past and simultaneously exert power over its population.
Ultimately, the set of court reports shows that ideas about gender played a role in helping many women to defend themselves against charges of genocide in gacaca. This finding has wider implications beyond Rwanda. Firstly, it adds to literature arguing that African women's agency in court systems has been simultaneously 'empowering' for certain individuals whilst often reliant upon the employment of particular gendered narratives (e.g. Mutongi ; Zimudzi ). This paper identifies tensions between individual women's success in using speech acts to achieve favourable trial outcomes, and women's involvement in generating narratives of female passivity and subservience. It further adds to this literature by identifying that Rwandan women's agency in court coexisted in tension with their forced participation in a system that both produced authority for the state and acted to control women's domestic behaviour. These tensions further complicate assumptions about the relationship between African women's voices in court spaces, and their 'empowerment'. Secondly, this paper responds to calls for considerations of the ways that global transitional justice mechanisms address actors who do not fit the framework of men-as-perpetrators and women-as-victims. Given the prevalence of women's involvement in global conflicts, this research on women's gacaca trials argues for further critical reflection not only of how cultural gendered beliefs impact the post-conflict trials of suspected women combatants and perpetrators, but also of the narratives that transitional justice institutions generate about the involvement or otherwise of ordinary women in the perpetration of violence, both within Africa and globally.