Forcing bonds, fuelling violence : combatant socialization and repertoires of violence during the civil war in Sierra Leone

(1)

Forcing Bonds, Fuelling Violence: Combatant Socialization and

Repertoires of Violence during the civil war in Sierra Leone

Nina Pouls – 10645616

Master Thesis Political Science – International Relations Track Civil wars, State-building and Displacement

Supervisor: dr. A. Steele Second reader: dr. J. Krause

22-06-2018 Word Count: 23.376

(2)

Acknowledgements

I would like to express my gratitude to a few people who have helped me with and guided me through writing my thesis. First and foremost, I owe a thank you to the people I interviewed in Sierra Leone. I am incredibly grateful to them for taking the time to share their stories with me, despite the content of my interviews. I am particularly thankful to Alfred, Lahai, Lash and Sayra who have assisted me in finding respondents and conducting the interviews. I also owe a thank you to Miller and Susan, who have been of invaluable help in connecting me to people but who also have provided distraction when needed during my stay in Freetown.

I would not have been able to write this thesis without the help of my supervisor, dr. Abbey Steele. I am thankful for her encouragement to go to Sierra Leone and her support throughout this process. In general, I am grateful for the key role she has played in my academic development the past one and a half year, developing my strong interest in this topic.

Lastly, I would like to thank my friends, roommates and family who have been of great support throughout the past three months. A special thank you to Adam, Andrea, Bo and Ely with whom I was able to share my thoughts in particular stressful moments.

(3)

Abstract

Recent research has increasingly focused on the relationship between combatant socialization and its role in producing violence. This research will focus on the relationship between socialization and repertoires of violence during the civil war in Sierra Leone (1991-2002). I argue that key to understanding differing repertoires of violence among armed groups is the extent to which they have been exposed to formal socialization. Building on Hoover Green’s (2017) work on institutions through which formal socialization takes place, I assess the extent to which combatants from the Sierra Leone Armed Forces and Revolutionary United Front have been formally socialized. Where formal socialization is not extensive, a group will employ informal socialization mechanisms to create a cohesive unit that is willing to fight. I argue that informal socialization mechanisms can account for a broad repertoire of violence because they are inherently violent or motivate violence. Empirically, this research will specify the causal mechanisms between combatant socialization and repertoires of violence for armed groups in Sierra Leone.

Keywords

Armed group institutions, civil war, formal socialization, informal socialization, Sierra Leone, violence

(4)

Abbreviations

ADRA: Adventist Development and Relief Agency AFRC: Armed Forces Revolutionary Council APC: All People’s Congress

DDR: Disarmament, Demobilization and Reintegration CDF: Civil Defence Forces

ECOMOG: Economic Community of West African States Monitoring Group ECOWAS: Economic Community of West African States

EO: Executive Outcomes

NPFL: National Patriotic Front of Liberia

NPRC: National Provisional Revolutionary Council RUF: Revolutionary United Front

SCSL: Special Court for Sierra Leone

SLA: Republic of Sierra Leone Armed Forces SLPP: Sierra Leone People’s Party

TRC: Truth and Reconciliation Commission

(5)

I. Introduction

On the 23rd_{of March 1991, the Revolutionary United Front (RUF) invaded the eastern} provinces of Sierra Leone from neighbouring Liberia. The invasion inaugurated the beginning of the 11-year long civil war, which was characterized by widespread mass atrocities. While in the beginning phase the war was fought semi-conventionally, the lack of control of the Liberian forces under RUF command soon led the RUF to employ a different strategy, relying less on Liberia’s support (Marks, 2013, p. 362-365). Despite the attention the RUF received as a rebel movement, the perpetration of atrocities was not limited to the RUF. Sexual violence, looting, displacement and killings were widespread throughout the conflict and perpetrated by several armed groups. Reports on the war, even long after the end, have emphasized the brutal character of the conflict. Human Rights Watch (HRW) emphasized how “(…) entire families were gunned down in the street” (HRW, 1999) and “children and adults had their limbs hacked off with machetes” (ibid). The New York Times further wrote on child soldiers: “Kidnapped by rebel forces or drawn into the Government’s army, they are forced to become soldiers, human shields, spies, porters and sex slaves” (New York Times, 1999). When reading such horrifying descriptions of the war, it becomes difficult to imagine rationality behind this seemingly indiscriminate violence.

When violence seems so inhumane and cruel that it contradicts with existing explanations of wartime violence, it poses a challenge to research the logic behind such acts. Where ethnic and religious cleavages are absent, as was the case for Sierra Leone, questions on what motivates violence seem to be even more persistent. In an attempt to grasp the logic of violence, this thesis will look into the factors that can explain different repertoires of violence employed by an armed group, posing the research question:

What explains the scope of repertoires of violence employed during the civil war in Sierra Leone?

Initial explanations of wartime violence that go hand in hand with explaining the occurrence of civil wars in the post-Cold War era in general, undermine that there is such a

(7)

innate to human nature (Kaplan, 1994). An entire body of research has already critiqued Kaplan’s work and its applicability to the context of Sierra Leone (Richards, 1996; Hoffman: 2006). Part of this critique is the body of literature seeking a rational explanation for wartime violence in emphasizing the political economy of war (Collier & Hoeffler, 2004). Within this research, violence came to be understood as motivated by greedy rebels, fighting to pursue their private agenda. However, Mitton (2015) argues that these explanations focus too much on rational calculation and “(…) as a result, have failed to explain, or even consider, those instances of atrocity that bear little if any semblance of rational calculation” (Mitton, 2015, p. 1.27). Existing theories falling in the ‘rational versus irrational dichotomy’ trap have proven to be inadequate in explaining variation in violence during civil wars.

These gaps in the literature have paved the way for a new body of research focusing on the role of socialization in producing violence (Wood, 2008; Cohen, 2013; Checkel, 2017). Socialization pays attention to group-level dynamics within an armed group. It considers the “(…) process through which actors adopt the norms and rules of a given community” (Checkel, 2017, p. 592). This thesis proposes that the extent to which combatants are socialized within a group, will affect the scope of repertoires of violence employed by an armed group. This argument will be illustrated by the case of the civil war in Sierra Leone where substantial variation is observed between the different armed groups. In doing so, this research will contribute to our understanding of why armed groups employ different forms of violence, which may initially appear to have no logic at all. In constructing this argument, the thesis will discuss the mechanisms that link combatant socialization to violence.

This research will draw on fieldwork conducted in May 2018 in Freetown and Waterloo in Sierra Leone. During this fieldwork, I interviewed 13 former combatants from various armed groups, one NGO country director and one person who used to work on the DDR process in various regions in Sierra Leone. Furthermore, I have conducted research in the Truth and Reconciliation Commission (TRC) archives at the former Special Court for Sierra Leone (SCSL) site, reading multiple victims and perpetrator statements made to the TRC. Combining this qualitative data with existing research and reports will enable me to illustrate the relationship between combatant socialization and the different repertoires of violence employed.

This thesis is organized in eight different chapters that are divided into subsections. First of all, a literature review will be provided. This will situate my research within the

(8)

existing literature on combatant socialization and wartime violence. The literature review will provide an overview of the existing research aimed at explaining wartime violence in Sierra Leone. Subsequently, in the theory section, I will identify the gap in the literature that this research aims to bridge. Within the theory section I will also introduce my theory and its underlying mechanisms. The fourth chapter explains the methodological approach to the analysis and my fieldwork process. Within this chapter, I will reflect on the limitations of the methodological decisions made and the constraints on my fieldwork. The fifth chapter will provide background information on the civil war in Sierra Leone and the actors involved. Subsequently, s my analysis for the RUF and SLA will follow. In chapter seven, I will discuss an alternative explanation accounting for the variation observed. I will also reflect on the limitations of my analysis. Lastly, chapter eight concludes my research, answering my research question and formulating recommendations for further research.

(9)

II. Literature Review

This literature review forms the cornerstone of this thesis. The first section focuses on the existing literature researching the widespread violence during Sierra Leone’s civil war and how this paved the way for socialization theory. The second section pays attention to the growing body of literature on linking socialization to different forms of violence in civil wars.

2.1 Explaining Wartime Violence in Sierra Leone

The first scholars attempting to explain violence during civil wars in a post-Cold War era used the so-called New Barbarism theory. This theory seeks to explain the emergence of civil war and wartime violence as a consequence of the general collapse of a state. The loss of a state’s monopoly on violence creates anarchical circumstances in a society, which paves the way for the innately violent human nature to unleash (Kaplan, 1994; Mitton, 2015, p. 2.1-2.4). The use of different forms of violence against civilians by armed groups was a consequence of the removal of restraints on them; they now had the opportunity to act in accordance with their violent human nature. Armed groups were portrayed to have no ideological foundation that would form a brake on this inherently violent human nature (ibid). Violence here is described as opportunistic; where restraints are removed, violent human natures will emerge. This theory fed well into dominant media narratives around the ‘uncivilised dark continent’, emphasizing the primitive and barbaric characteristics of African culture (Baaz & Stern, 2008, p. 58-59). Since the New Barbarism theories have been profoundly criticized within existing research for their racist tendencies and lack of context, it will not be further discussed in this thesis (Hoffman, 2006, p. 11). However, it is essential to mention the theory for two reasons. First of all, it enables us to track down the historical lineage of explaining (repertoires of) violence during civil wars. Secondly, understanding the breakdown theory is fundamental to studying any aspect of violence during civil wars, especially given its persistence up until today in many theories (Mitton, 2015, p. 2.15). The ‘breakdown thesis’ assumed that there was no logic or rationality behind the horrifying violence in civil wars. Resistance against this assumption has provided the basis for much research attempting to account for wartime violence. It has forced academics to be cautious to not fall into the breakdown trap whenever violence seems so horrifying that it gets

(10)

challenging to account for logic behind it. In search for such logic, Mitton has argued that scholars have to move beyond focusing on the removal of restraints and ask themselves what is introduced that enables the employment of violence in the context of (state) breakdown (2015, p. 2.27-2.28).

The response against the New Barbarism theory is divided into two camps. The first response came from academics that emphasized the socio-political logic of wartime violence (Richards, 1996). The second group emphasized economic factors in an attempt to explain wartime violence (Collier & Hoeffler, 2004). Emphasizing the social and political grievances experienced by many RUF rebels in the pre-war period, Richards (1996) argues that violence can be understood in light of those grievances. Many RUF rebels in the initial phase of the conflict joined the rebellion to express their discontent with the dominance of the APC regime under which they have suffered widespread unemployment (ibid). The unemployment mainly affected young urban men, something Gberie (2005) has referred to as ‘lumpen youth’. According to Richards, extreme forms of violence can be understood as both acts of rebellion and as a useful tool to spread fear and control civilians (ibid). Richards also emphasizes the role of violence as a communication tool of one’s significance to the outside world (ibid, p. 60). In later stages of the conflict, violence could be understood as a response to the “(…) revival of patrimonial social forces” (Richards, 1999, p. 437). The revival of the patrimonial forces were, in the eyes of the RUF, the civil defence forces (CDF). The CDF were local militias that had formed to protect themselves against the RUF because, according to them, the Sierra Leone Armed Forces (SLA) failed to do so. However, this last explanation seems somewhat unsatisfying since the increased reliance of the RUF on abduction for recruitment led to a fighting force that seemed to have little connection to the RUF’s political program (Humphreys & Weinstein, 2004). In line with this, Mitton argued that Richards overstated the relationship between violence and political strategy (2015, p. 3. 15). According to Keen, the RUF could not be credited as a political movement because of its “(…) persistent abuses, including extreme atrocities, against the very civilians it claimed to represent” (Keen, 2005, p. 41). Violence as a form of control over civilians also seems to be an overstatement. Civilians were often already very fearful of the RUF, which motivated a willingness to comply with them. Therefore, extreme atrocities cannot adequately be accounted for by a rational actor account emphasizing political grievances

(11)

The second camp within the rational actor approach emphasizes the role which economic agendas and greed have played in fuelling violence. The conflict in Sierra Leone is well known for the role of ‘blood diamonds’, to the extent that it is often portrayed as being solely caused by greedy rebels participating in mass atrocities to pursue their private agendas. Economic motivations have undoubtedly played a role in the initial invasion from Liberia by the NPFL/RUF into the Eastern region of Sierra Leone, but its role has often been overstated (Keen, 2005, p. 48-55). Keen identifies three connections between diamonds and the war. First of all, the financial benefits of diamond extraction were concentrated in the hands of a small group of (often foreign) elites. This bred frustration among large segments of the population (ibid, p. 50). Further feeding this frustration was the weak government infrastructure that proved unable to implement a system of taxation on diamond mining activities (ibid, p. 51). The TRC described the attempts by the APC, the party ruling Sierra Leone since its independence in 1961 up until the civil war, to control and tax diamond mining as ‘a façade’ (TRC, 2004, 3B p. 7). Secondly, diamond extraction created revenue for the RUF to purchase arms to sustain its fight (ibid). Lastly, diamonds formed an incentive for violence in some cases (ibid). The diamond-rich area of Kono and Pujehun had a lot of rebel activity. A miner recalls in his statement to the TRC:

One rebel came with a mask and asked me to produce the diamonds. I told him I had nothing like diamonds in my possession. Failing to produce the said diamonds, I was severely beaten with a stick. He further threatened to kill me. I begged him not to kill me as I am a poor boy working for people. But he seemed not to care and continued beating me until the first stick was broken. He used a second stick until it was finished (TRC, 2004, 3B, 27)

However, the sole focus on economic greed by authors such as Collier (2000) does not provide a satisfactory answer to why the many different forms of atrocities were committed, and it also does not answer the question why violence was perpetrated in areas where no resources were available (Mitton, 2015, p. 3. 72). The overstatement of the role of diamonds fueling the conflict itself also finds its resonance in the Truth and Reconciliation Commission report which states: “Although it is true that the RUF partly financed its war effort through diamond trafficking, diamonds did not yield significant revenues for the movement before 1997” (TRC, 2004, 3B p. 4). In addition to that, the violence in diamond-rich areas in the south and east of Sierra Leone is not merely explained by the presence of

(12)

diamonds but also its proximity to the Liberian border from which the rebels initially entered (Keen, 2005, p. 52).

The rational actor approach has thus left many questions with regards to the employment of broad repertoires of violence unanswered. Assuming that every actor makes a decision based on a calculated logic leads to the mischaracterization of much of the mass atrocities perpetrated in Sierra Leone. Mitton (2015) has contributed a great deal in explaining violence in Sierra Leone, ascribing the commitment of atrocities by the RUF to the psychological factors of shame, disgust and systematic brutalization. His work will be mentioned in the analysis section of this thesis. The gaps within the rational actor approach to account for violence have paved the way for research looking closer into the internal dynamics of armed groups. Theory on the so-called socialization of combatants will now be further discussed.

2.2 Combatant Socialization

The role of socialization of combatants within armed groups has only recently been receiving attention from scholars studying civil wars (Wood, 2008; Autesserre, 2009). Socialization finds its roots in sociology and anthropology. Dawson & Prewitt have described it as “(…) a process of inducting actors into the norms and rules of a given community, the endpoint of which is internalization” (Dawson & Prewitt, 1969 in Checkel, 2013, p. 7). This definition perceives socialization as a process where the desired result is a deeper change in perception of the ‘self’ (Checkel, 2013, p. 7). Socialization then highlights the social processes shaping combatants within an armed group. A central assumption in socialization theory is that being a member of a group means that you are “(…) embedded in social environments, which not only constrain and provide incentives to act, but also reshape interests and identities” (Checkel, 2017, p. 592). Where rational actor explanations have emphasized the external motivations of combatants to commit some forms of violence, socialization theory emphasizes how combatants get internally motivated inside an armed group to engage in particular forms of violence (Hoover Green, 2016, p. 621).

According to Checkel (2017), socialization processes can produce three different types of outcomes that indicate the extent to which someone is socialized within a group. The first one, type 0, “(…) is based on rational calculation of the group member in response to incentives – coercive or not – (…)” (Checkel, 2017, p. 597). This socialization

(13)

that complying with what is told to him/her is more effective than disobeying an order, a rational calculation takes place. It is possible that repeatedly making this decision will eventually result in a form of structural compliance, which can be perceived as internalization and thus a different type of socialization (ibid, p. 596). Type I socialization means that “(…) an individual exhibits pro-group behaviour by learning a role – acquiring the knowledge that enables action in accordance with group expectations” (ibid, p. 597). This does not yet mean that the combatant has entirely taken on the new group identity, but is performing a role in accordance with dominant ideas, norms and values within an armed group. The last type of socialization, type II is the most extensive form. With type II socialization:

An individual accepts group norms as the right thing to do; she adopts the interests or possibly identity of the community of which she is a member. Conscious instrumental calculation and role playing have been replaced by taken for grantedness or full internalization (ibid)

This means that a combatant acts entirely in accordance with the prescribed group norms. Combatants that have undergone this type of socialization pose a challenge to DDR programs, as they need a long-term program to re-socialize (ibid).

Hoover Green (2017) identified four key institutions within an armed group through which formal socialization occurs. She has linked these institutions to the different types of socialization described in the paragraph above. While previous research has perceived military institutions as tools to align commanders’ preferences with those of combatants, Hoover Green highlights how socialization occurs through those institutions (2017, p. 688). The four institutions considered are recruitment, military training, political education and disciplinary regimes (ibid). Following March & Olsen’s (2008, p. 3) definition of an institution, Hoover Green defines an armed group institution as: “(…) the relatively enduring rules and organized practices of armed groups, those that tend to persist despite changes in membership or external circumstances” (2017, p. 689). Recruitment is, in contrast with the other institutions, not necessarily a process but rather a particular moment in time. However, the way someone is recruited has a big influence on one’s attitudes and behaviour within the group and the level of cohesion within a group. Recruitment practices are also the first introduction to the group culture (ibid). Forced recruitment through abduction would be linked to type 0 socialization whereas more extensive, and voluntarily

(14)

recruitment practices may produce type I or type II socialization. Military training aims at creating a unit that is able and willing to employ violence. Hoover Green describes this process as creating a fighter-as killer identity, mainly focused on type I socialization (ibid, p. 690). A more thorough form of socialization can happen through political education (PE). However, as we will see in the analysis section, not every armed group engages in PE. PE focuses on the deepest form of socialization where norms are internalized, namely type II socialization (ibid). PE attempts to communicate a specific ideology and often provides a ground for why the war is fought. The last formal institution, disciplinary regimes, includes: “(…) systems of behavioural rules, rewards, and punishments, and – importantly – the group’s capacity for enforcing them” (ibid). This form of socialization regards the strength of the hierarchy within armed groups and the presence of clear behavioural guidelines and procedures. Disciplinary regimes will mainly lead to type 0 or type I socialization (ibid).

There are two factors that current scholars working on the role of socialization in producing violence have to pay attention to. First of all, early work on socialization processes has neglected the agency of the targets of those processes. Regularly, targets are perceived to be clean slates onto which particular norms, ideas and values are inscribed (Checkel, 2017, p. 594). Targets of socialization processes can resist such attempts as well. A caveat has to be made about child combatants who usually tend to be easier targets for socialization (Vermeij, 2014 in Checkel, 2017, p. 597). Child combatants were widely employed during Sierra Leone’s civil war (Humphreys & Weinstein, 2004). Another interesting study that has explored the limits of socialization was conducted by Devorah Manekin (2017). She argues that soldiers from the Israel Defense Forces came to resist orders, and thus exercise a form of agency when they encountered situations in their deployment that did not reconcile with the training they have received. Theory should also acknowledge that prior to joining an armed group, combatants have already been part of a social context shaping them, which will likely influence socialization processes within the new group (ibid, p. 595). Furthermore, commanders are not only agents of socialization but are also targets of socialization processes (Wood & Toppelberg, 2017). Wood & Toppelberg have demonstrated how commanders’ socialization into the masculine structure of the army contributed to the persistence of sexual assault within the US army.

(15)

operates through institutions such as boot camp or ideological training. The insight that socialization processes not necessarily produce violence but can be inherently violent will be central to understanding the scope of repertoires of violence in Sierra Leone (Checkel, 2017, p. 595).

(16)

III. Theory

In this section, I will identify the gap in the literature and formulate the hypothesis, which will be tested in the analytical section. I will also describe the causal mechanisms underlying my hypothesis.

3.1 Gap in the Literature

Substantial variation in the forms of violence used by a particular armed group within and between civil wars can be observed. No research up to now has focussed on explaining variation in the scope of repertoires of violence during conflict. Within one specific civil war, one can often observe armed groups with different repertoires of violence. More structural factors (such as natural resources) contributing the different forms of violence employed to circumstances that are equal for all the parties involved cannot account for this variation. Since the frequent employment of a lot of different forms of violence can be costly to an armed groups’ legitimacy, it is even more pressing to understand why an armed group would engage in such costly behaviour. It is thus important to develop an answer to what can account for the different repertoires of violence within civil wars.

Building from the literature review and the existing research on the relationship between socialization and violence, my hypothesis will focus on the relation between formal and informal socialization processes and the scope of repertoires of violence. In doing so, this research will contribute to highlighting the specific mechanism between socialization processes and repertoires of violence. I will differentiate between formal socialization processes that happen through the institutions outlined in the literature review and informal socialization processes since I believe both forms of socialization have a different outcome on the repertoires of violence. Hoover Green (2017) has researched the role of formal institutions on socialization processes but, as she acknowledges in her introduction, she has not taken into account the role of more informal socialization processes. I emphasize the importance of informal socialization processes in my theory because it can play a key in understanding violence. Cohen (2013) has provided an exploration of the role of informal socialization on violence. She has argued that rape can serve as a socialization tool in groups where cohesion is low.

(17)

formal socialization happens) should display repertoires of violence that are narrower than groups that do not have political education, I hypothesize the following:

H1: Armed groups with strong formal socialization will on average display a narrow repertoire of violence when compared to armed groups that rely on informal socialization mechanisms.

Formal socialization here takes place as a top-down process and refers to “social processes that seek to implement official policies and values (including traditions, exercises, and rituals)” (Wood & Toppelberg, 2017, p. 626). Informal socialization refers to “social processes (often prohibited) that undermine official policies and values (including traditions, exercises and rituals)” (ibid).

If we observe strong formal socialization institutions within an armed group, I expect the armed group to, on average, display a narrow repertoire of violence. If the armed group has weak formal socialization mechanisms, I expect to see a broad repertoire of violence. Empirically, this would mean that when a group has received, for example, no political education, little formal training and there is a weak discipline within the ranks that I expect that this armed group to engage in sexual violence, amputation, torture, displacement, mass killings etcetera. The causal mechanism to this relationship will be explained in the subsequent section.

Furthermore, it is important to notice that if full internalization of group norms has taken place (type II socialization), the content of the group ideology and the commands given to combatants also matter in deciding whether socialization will lead to a general restraint on violence (Hoover Green, 2016). If an armed group has managed to reach type II socialization for most of its combatants, but the group ideology clearly justifies violence against group X, then one can expect the employment of various forms of violence against group X. In this hypothetical case, strong socialization will not lead to narrow repertoires of violence. Thus one scope condition to which my argument applies is that the commanders within an armed group do not approve of extra proportional violence against particular groups. In this thesis, I will test my hypothesis for the case of Sierra Leone. During the civil war in Sierra Leone, every armed group claimed to fight for the protection of civilians and principally disapproved of violence against them. This contrasts with, for example, Rwanda where violence against Tutsis was clearly ordered.

(18)

Figure 1: Hypothesized effects of formal socialization on repertoires of violence.

3.2 Causal mechanisms

It is important to outline why one would expect armed groups with extensive formal socialization to show, on average, narrower repertoires of violence in comparison to groups that rely on informal socialization. To do so, I will use existing literature on how violence can serve as a socialization tool among combatants to identify two causal pathways that clarify the relationship between my variables. My dependent variable is repertoires of violence, and my independent variable is formal socialization. Prior to identifying my causal mechanisms, two assumptions underlying my argument, have to be clarified.

First of all, violence against civilians is a costly activity for armed groups because civilians’ support can be instrumental to an armed group to achieve their goals (Wood, 2010, p. 602). Therefore, we would expect that violence against civilians is generally discouraged

(19)

and disapproved of by commanders.1_{Secondly, an armed group is always confronted with} the challenge to create a cohesive unit that is willing to fight, the so-called creation of a ‘fighter-as killer-identity’ (Gates, 2002; Hoover Green, 2017). The creation of this identity is inherently a group-based process. Morris (1996) further demonstrates the importance of this process. She has researched how increasing social ties and thus commitment among combatants becomes the main motivational factor for combatants’ willingness to fight. This means that socialization, in whatever form, will always take place to a particular extent to create a force willing to fight. These two assumptions underlay the causal pathways that I will now identify.

When a combatant becomes part of an armed group, he/she will be subjected to socialization efforts through the formal institutions identified in previous parts of this thesis. The institutional capacity of an armed group heavily influences this process. While one may expect that a state army has the resources to provide extensive training to their combatants, insurgent groups may not have such well-established institutions through which military training takes place. The capacity of institutions to socialize a combatant differs thus per armed group.

When type I or type II socialization are the outcome of the socialization process, then I would expect a narrower repertoire of violence. First of all, the (partial) internalization of group norms in combination with the group cohesion created in the formal socialization process makes these groups less reliant on violence as a socialization tool (Cohen, 2013). Secondly, the capacity of commanders to restrain violence to the extent that they judge necessary for their goals, increases in a well-disciplined unit (Hoover Green, 2016). This is not to assume that once restraints are removed that combatants become uncontrollably violent and do not have any moral considerations anymore. However, it does imply that individuals who are part of an armed group are more predisposed to commit violence through their military training, a predisposition that gets reinforced by their combat experiences (ibid, p. 622). This first pathway to a narrow repertoire of violence is thus characterized by restraint and a sufficient level of group identification, which decreases the role informal socialization has to play in the creation of a group identity.

When formal socialization efforts lack or are insufficient, which can already start when a combatant has been forcefully recruited into the armed force, a group has to rely on other methods to socialize combatants. This socialization is needed to create a cohesive unit

1_{See Reed Wood (2010) and Kalyvas (2006) for circumstances under which violence against civilians becomes} part of the strategy of an armed group.

(20)

willing to fight. Socialization, I argue, is not necessarily a top-down process, as demonstrated by Cohen (2013). She argues that groups that rely on forced recruitment, which causes low social cohesion, can employ gang rape as a socialization tool. These horizontal forms of informal socialization, undermining group policies will become important when formal socialization efforts are insufficient. Informal socialization processes can both be inherently violent or produce violence. In the case of gang rape, the creation of cohesion is violent in itself. It can produce violence when an armed group needs to demarcate group boundaries to create a group identity. A way to do this is to dehumanize the ‘outside world’. This process leads a group to perceive the others as not, or less human, than themselves, and this makes them easier targets for violence. Keen (2005, p. 76) has argued that dehumanization removes a feeling of moral shame from the perpetrator to carry out a particular form of violence. The use of violence to socialize fighters gradually leads to a systematic brutalization of combatants where the perpetration of violence becomes rewarding in itself (Mitton, 2015). Several informal socialization mechanisms will be identified in the analysis part.

(21)

IV. Methodology

In the methodology section, I will introduce the methods used for my research. I will explain why I have chosen a particular method and the implications of this decision for my research. I will start off describing my methodological approach and explain my case selection. I will continue with the operationalization of my variables. Then I will explain the data gathering process during my fieldwork in Sierra Leone and the ethical considerations taken into account during my time there. Lastly, I will discuss the limitations of my data.

4.1 Methodological Approach

This research is a qualitative case study of armed groups active during the civil war in Sierra Leone. I have used a case study since it would allow me to examine the causal mechanisms my hypothesis is concerned with more in-depth. Gerring (2004) has defined a case study as “(…) an intensive study of a single unit for the purpose of understanding a larger class of (similar) units” (p. 342). The context of this research is the civil war in Sierra Leone, and the cases are the RUF and SLA. The variation across and within cases both over time and space is needed to observe whether cause and effect co-vary and thus to see if there is a relationship between the variables. An in-depth case study will best reveal the mechanisms potentially linking combatant socialization to a certain scope of a repertoire of violence. A case study is not the ideal method to establish a general causal effect, but that is also beyond the aim of this thesis. Juxtaposing cases that differ on the independent variable enables one to observe the influence of this on the value of the dependent variable. When a relationship between the two variables is observed, variation within an armed group will help us highlight the mechanism linking the two variables researched here and thus how combatant socialization might explain the scope of a repertoire of violence. More generally, an advantage of conducting a case study is that it allows to “(…) peer into the box of causality to the intermediate causes lying between some cause and its purported effect” (Gerring, 2004, p. 348). The causal argument developed in this thesis will be probabilistic. This means that the cause, the extent of formal socialization, has an increasing or decreasing effect on the scope of the outcome, which is in this case, the repertoires of violence employed by an armed group (ibid, p. 349).

(22)

The purpose of my research is to contribute to the growing literature on the relationship between combatant socialization and violence. Having developed a hypothesis based on the existing literature, I have been able to test it during my fieldwork in the Western Area of Sierra Leone. Existing research and reports were used where necessary to complement my analysis. To test my hypothesis and to theorize the causal mechanisms underpinning my hypothesis, I have conducted interviews and archival research. I have also gained insights from interviews and surveys conducted by other researchers with former combatants in Sierra Leone (Mitton, 2015; Humphreys & Weinstein, 2004).

While my approach can provide an in-depth insight into the relation between combatant socialization and repertoires of violence and its underlying mechanisms, one has to be aware of the limitations of this approach. Conducting a case study of armed groups within one particular civil war makes it difficult to generalize findings to a broader level. It is difficult to differentiate between case specific and structural factors when just one civil war is studied. Conducting a case study and generalizing its findings, risks drawing generalizations from a case that may be idiosyncratic. The particularities about the Sierra Leone civil war and the process of case selection will be discussed further in subsequent sections of this thesis.

4.2 Case Study Selection

This thesis aims to research how combatant socialization influences the scope of repertoires of violence employed by different armed groups. To do so, variation on the independent variable needed to be present in the case I would choose. Inspired by Cohen’s (2013) theory on rape as socialization tool, I became interested in Sierra Leone where a wide variety in recruitment patterns could be observed. This made me wonder how, and if, the larger understanding of combatant socialization influences repertoires of violence.

While clear lines between armed groups are sometimes blurry (Keen, 2005, p. 132-159), this thesis will look into two armed groups specifically: the Sierra Leone Armed Forces (SLA) and the Revolutionary United Front (RUF). Both armed groups have engaged in human rights violations, but both to different extents (TRC, 2004, 3A p. 524-564). The RUF and the Armed Forces Revolutionary Council (AFRC) are sometimes empirically difficult to distinguish. Following the TRC report, this thesis will consider the RUF as a distinct group from the beginning of the conflict up until the 27th_{of May 1997 when the RUF joined the}

(23)

After removal from power, the AFRC and RUF largely split up again. However, people did remain to contribute atrocities to the ‘rebels’ since it proved difficult to trace back a perpetrator to his/her institutional affiliation (ibid, p. 525). After May 1997 the number of violations contributed to the RUF might thus have deficiencies. Background information on these armed groups will be provided in subsequent parts of this thesis; however, I will stress here why I have chosen these armed groups and the larger context of the Sierra Leone civil war for closer examination.

First of all, all cases vary substantially in socialization practices. For example, looking at recruitment practices, the RUF is well known for its recruitment through forced abduction of (child) soldiers. Despite the RUF’s notoriety for this, the CDF and ARFC have also engaged in abduction to fill its ranks (ibid, 470). The SLA would usually draw from voluntary recruits. Furthermore, the length and depth of military training between armed groups vary substantially too. While the RUF combatants would generally receive violent forms of training, the SLA training was more formal and institutionalized (ibid). This variation on the independent variable is needed to observe if the variables of my hypothesis are covariational.

Secondly, reliable data needs to be available on the perpetration of violence per armed group. To establish the repertoire of violence of an armed group, I heavily rely on the well-documented statistics within the TRC reports on Sierra Leone. However, violations by some smaller armed groups such as the West Side Boys or external forces present during parts of the conflict, are less well documented. This does not neglect their importance for closer examination, but a narrowing down of the armed groups researched was needed.

Lastly, there were some practical considerations that I was forced to take into account. When I decided that I want to go out to conduct research in the field I had to go to a country a) with a recent past of civil war, b) where I could manage without a translator and c) where the situation right now is relatively safe to travel by myself. Given that most people in Freetown do speak English and the relative safety of the city, Sierra Leone felt like a good place to be introduced to conducting fieldwork.

4.3 Operationalization of the Variables

The variables in my research are formal socialization (independent variable) and repertoires of violence (dependent variable). As previously discussed, socialization can be both formal and informal. However, a reliance on informal socialization will occur when formal

(24)

socialization has not been extensive. I will discuss the indicators for measurement of formal socialization and the repertoires of violence.

4.3.1 Formal socialization

Wood & Toppelberg define formal socialization as “(…) the traditions, exercises, rituals, and other social processes that seek to implement from ‘above’ the organization’s policies as stated in its regulations, training documents, and other official sources” (2017, p. 626). Formal socialization, as mentioned in the literature review, happens through four key institutions (Hoover Green, 2017). First of all, I will look at recruitment patterns; were combatants abducted or did they join voluntarily? Secondly, I will look at the duration of their military training to measure the extent of socialization through training. I will also look at the content of their training; what do they recall learning? Where military training was simply focused on ‘how to shoot a gun’, I classify it as not extensive. For political education, I will observe whether combatants have received political education and what they were told about the purpose of the war. If combatants could reconcile the situations they encounter on the battlefield with the purpose of the war; then political education counts towards type II socialization. Another indicator is the extent to which they still identified with the armed groups’ political agenda after the war was over. Lastly, for disciplinary regimes, I will look at whether combatants were punished when they did not follow orders. Wood’s (2018) theory on explaining the persistence of rape during conflict has shown that disciplinary regimes are not just characterized by one’s willingness to listen to a commander but also by what is tolerated by a commander. Therefore, to fully capture the concept of disciplinary regimes, ‘order’ in this context does not merely refer to what is being directly ordered to a combatant but also the ability to enforce the rules and regulations within an armed group. If rape is formally disapproved of within an armed group, but not punished when perpetrated, then this will count towards having weak disciplinary regimes.

Informal socialization processes are somewhat more difficult to measure but necessary to depart from traditional principal-agent models to explain violence. Informal socialization is defined as “(…) the rituals, traditions, exercises, and other processes between peers that undermine those policies” (Wood & Toppelberg, 2017, p. 626). ‘Those policies’ in this definition refers to policies within an armed group aimed at regulating

(25)

commanders, who are subjected to socialization efforts as well, can also be agents of socialization processes that undermine official policies. In the analysis section, I will identify mechanisms observed within the RUF that served to socialize its recruits while simultaneously undermining group policies.

4.3.2 Repertoires of Violence

The dependent variable is the scope of a repertoire of violence. An armed group’s repertoire of violence is “(…) the subset of all possible forms of violence against civilians in which it regularly engages” (Gutiérrez-Sanín & Wood, 2017). Regularly is added within this definition because when only one combatant has engaged in, for example, mutilation of civilians, then it is more likely that this very rare occurrence can be ascribed to individual characteristics rather than a group level phenomenon. A repertoire of violence will be established in the analysis section. The frequency of a particular form of violence will be based upon the statements given to the TRC and the Benetech Sierra Leone War Crimes Documentation Survey. A form of violence will be considered part of a group’s repertoire when that form of violence by one armed group constitutes at least one per cent of the total amount of violations reported to be committed by the RUF and SLA.2_{The TRC report has} acknowledged that it has inadequate data about forced drugging and forced cannibalism. Furthermore, I have not included ‘abduction’ and ‘forced recruitment’ as a violation in the repertoires. Abduction was often used as a recruitment practice, which is an indicator of my independent variable. Including both violations as well in the dependent variable would then lead to incorrect inferences. Hence, these violations will be left out of the analysis. If the TRC data can be complemented with other data sources on, for example, sexual violence during conflict (Cohen & Nordås, 2014) then this will be done. Many survivors of sexual violence may be fearful to report their experiences to the TRC because of the stigmatization they face, thus having reason to believe that sexual violence was underreported.

A repertoire of violence will be classified as a narrow repertoire when an armed group has employed five or fewer forms of violence. It will be classified as a broad repertoire when more than five forms of violence have been employed. This categorization is my own, to make comparisons among armed groups easier. Forms of violence that could

2_{The TRC has recorded 40242 violations in total. 24353 of them have been perpetrated by the RUF. The RUF} and SLA combined are responsible for 22.725 violations that got reported, excluding abduction and forced recruitment. (TRC, 2004, Appendix 1).

(26)

make up a repertoire could include but are not limited to: killings, torture, forced displacement, forced cannibalism and different forms of sexual violence.

Hypothesis: Armed groups with extensive formal socialization processes will on average display a narrow repertoire of violence when compared to armed groups that rely on informal socialization mechanisms

Variables Indicator

IV: Formal Socialization

1) Recruitment practices

2) Length of and experience with military

training

3) Received political education? 4) Disciplinary regimes: was violence

against civilians punished or tolerated? How was discipline maintained? DV: Repertoire of Violence Forms of violence regularly employed

-Regularly: it has been reported to the TRC more than 230 times.

Table 1: Variables and indicators

4.4 Data

I have gathered data for this research from multiple sources. Existing interviews with former combatants conducted by Kieran Mitton (2015) and survey research conducted by Humphreys & Weinstein (2004) proved to be very valuable for my analysis. Furthermore, the reports by the Special Court for Sierra Leone and the TRC reports have also formed an important source of information for the analysis. Data on violence, perpetrators and victims was published in the final report of the TRC and the Benetech Human Rights Data Analysis Group. This data, however, is based on statements given by people who decided to accept the invitation of the TRC to give a statement. The data is not necessarily generalizable to the entire population. One can, for example, imagine that survivors of sexual violence were less likely to give a statement because of stigmatization. I have still chosen to use this data because it is the only dataset where the segregated data needed for establishing the repertoire of violence, is available. I have also conducted fieldwork in the Western Area of Sierra Leone from April 29 to May 18, 2018. In Sierra Leone I have conducted twelve interviews in the capital city Freetown and three interviews in a smaller city near Freetown.

(27)

4.4.1 Fieldwork interviews

Most of the data used in the analysis has been collected during my fieldwork through interviews and archival research. I was motivated to conduct fieldwork to gather first-hand information. Furthermore, this opportunity would familiarize me with conducting fieldwork and allow me to move beyond literature research. My gathered data consists out of fifteen interviews (Appendix 1). One interview was with a policy officer involved in the reintegration of former combatants during the DDR and another interview was conducted with the country director of ADRA, a Christian NGO that was involved in the DDR process. The other thirteen interviews were conducted with former combatants. Seven interviews were conducted with former RUF combatants; four interviews were with conducted with former Sierra Leone Armed Forces soldiers; one with a former Kamajor (CDF) and one with a combatant from the West Side Boys. Triangulation of the data received through interviews with combatants is done where possible in this research to increase the reliability of the data.

Throughout my stay in Freetown, I have met several people who were very willing to help me find respondents. Despite some significant obstacles that I faced with getting concrete appointments with people, three people have proven to be very valuable in setting me up with former combatants. Given the difficulties to find people who are still willing to talk about the war, something I will discuss a little further in the next section, I would interview any former combatant active during the war that was willing to speak to me.

The interviews I conducted were semi-structured so that it would allow me to ask further questions where I heard something relevant for my research. After my first four interviews, I have adjusted and elaborated my questionnaire somewhat to provoke more in-depth answers. After asking for consent from the respondent for both the interview and recording it, I have been able to record all 15 interviews. I guaranteed anonymity to all combatants and asked the other two respondents whether they would allow me to use their name, which they agreed upon. The interviews with combatants usually took place in their neighbourhood, close to their homes. The locations were not ideal; often several people were present and there was a lot of background noise from kids, workspaces or just people who came to observe what was happening. Furthermore, a lot of respondents did not speak English sufficiently to the extent that they would have been able to answer my questions. In these cases, the person who had set me up with the interviews would translate the content of the answer. The use of a translator made it difficult to engage in a conversation style interview with the interviewees, which, given the sensitivity of the topic for some, would have enabled me to build up more repertoire with the interviewee’s. Using a translator also

(28)

risks that some answers did get lost in translation. After the interview, I would listen back to my recordings and label relevant quotes under categories linked to my variables.

4.4.2 Archival Research

The Peace Museum at the former site of the Special Court for Sierra Leone contains the archives where all the SCSL documents and TRC statements are filed. Before travelling to Freetown I was not aware of these archives. By paying a visit to the museum I became aware of the option to access these documents. After having paid an access fee through the Sierra Leone Human Rights Commission, I was allowed entrance to the archives where I could look into all but the confidential statements given to the TRC. Both victims and perpetrators have given statements to the TRC, but in the 150 statements I read, I have only come across two perpetrator statements. The victim’s statements, however, did sometimes reveal information about timing and space of particular forms of violence and were therefore very useful.

4.4.3 Challenges and Ethical Considerations

As briefly mentioned earlier, the biggest challenge was to find people who were willing to talk about a war that officially ended sixteen years ago. I was often asked why ‘we’ were still coming to conduct research and what was the purpose of all this research conducted by people from ‘overseas’. I would be honest about my research intentions at all times and would also tell my respondents about how research can contribute to an understanding of violence that hopefully in the future may help prevent such atrocities. At the same time, I acknowledged that I was not going to be able to make a difference in the lives of my respondents with my research. It was challenging to go into an area, subtract the information I needed and then leave again being aware of the sometimes very harsh circumstances respondents were in.

Linked to the reality that the war has been over for a while now, it was also challenging to get a story from a respondent since many could not exactly recall the details of their involvement in the war anymore. Despite specific questions, answers always remained a little broad. Of course this also had to do with the little repertoire that I was able to build up with my respondents since I was only there for a short time. My interviews are relatively short, but I received a lot of background information on the context of the war in

(29)

A third challenge was the circumstances in which the interviews were conducted. Most of my respondents lived in slum areas, which were not ideal to conduct interviews at. However, their comfort was very important to me since I realized the sensitivity of my topic. I recall one interview where a fight was going on in the background, which was a very distracting factor to our interview. Furthermore, during many interviews, people kept walking in and out to see what was going on. The nuisances that were present during the interviews regularly formed a distraction to my respondents.

A fourth challenge was time and infrastructural restraints. During the limited time I spent in Sierra Leone, I could not travel all over the country, thus a slight urban bias is introduced in the data I collected. A limitation of having conducted fifteen interviews is also the generalization of my results. To partly deal with this limitation and bias, I have consulted other research that has interviewed former combatants on violence during Sierra Leone’s civil war as well. This allows me to see whether my findings were in line with the experiences of other combatants and thus to check whether my respondents were not idiosyncratic cases.

The last challenge I faced with my interviews was my interpretation of the information provided by my respondents. What if an answer fully contradicted findings from the TRC? In one case, I recall a former RUF combatant denying the large-scale mistreatment of civilians. The respondent ascribed our ‘misinformation’ about how civilians were treated by the RUF to reporting bias. While such answers can definitely be telling about one’s ties to the RUF, it does pose challenges to the credibility of other answers the respondent gave.

I was well aware that I was dealing with human subjects and thus had to take into account ethical considerations. I would ask verbal consent from every respondent after explaining to them the goal and purpose of my research. I would also reassure them that they could stop at any time and if they felt uncomfortable, they should let me know. I did not want my respondent to feel any pressure to answer my questions. Most people did not seem to be afraid to talk to me or would not question me about my intentions, but I still guaranteed them anonymity, as doing so would hopefully contribute to the creation of an atmosphere where they felt they could freely talk. With regards to the questions asked, I also took research ethics into account. I did not at any time ask my respondents about the violence they might have perpetrated. I was aware that I am still a student and thus still practicing the art of interviewing.

(30)

V. Sierra Leone’s Civil War

Before analyzing the civil war in Sierra Leone, it is necessary to provide background information on the war, armed groups and their characteristics. I will start off describing the social, political and economic circumstances in Sierra Leone that paved the way for the rebellion by the RUF/NPFL. Afterwards, some background information on the main actors will be given. Last, I will describe the civil war in Sierra Leone by highlighting its key events

5.1 Discontent in a Weak State

Sierra Leone gained independence from Great Britain on 27 April 1961. The British had left behind a state fragmented between the urban and more developed ‘Colony’ (Freetown and surroundings) and the underdeveloped ‘Protectorate’ (the hinterland of Sierra Leone). This division faded a little through the emergence of party politics dominated by the All People’s Congress (APC) and the Sierra Leone People’s Party (SLPP). Both parties would respectively draw support from the Temne ethnicity from the north and the Mende’s from the south and east of the country (Keen, 2005, p. 14).

In 1967, the APC led by Siaka Stevens won the elections. Under the APC, ethnic Temne’s replaced many government and military officials from Mende descent. By distributing positions of power to his inside circles, Stevens created a system of extensive patronage upon which he heavily relied. Simultaneously, widespread corruption and the export of natural resources and raw agricultural products for low prices led to the economic decline of Sierra Leone (Keen, 2005, p. 16-35). In the 1980’s, the import of rice, the most consumed food commodity in Sierra Leone, made up more than half of all export incomes (ibid, p. 26). Under pressure of the IMF and World Bank, austerity measures were implemented as a condition for financial aid. The austerity measurements contributed to the economic and social crisis of the late ‘80s (ibid). Social services, health care and education were unavailable to large segments of the population; teachers went unpaid for months and (youth) unemployment was widespread (ibid). The failure of the government to effectively subtract taxes from diamond mining, largely controlled by foreign actors, also created resentment among the population (ibid). The aggravated frustrations with the unequal

(31)

inequality provided a fertile ground for conflict is further demonstrated by Richards et al., (1996). Their interviews with combatants have shown that it was the large groups of marginalized youth with little education that made up the largest part of the irregular combatants (Richards et al., in Keen, 2005, p. 28). External factors also play in role in the conflict in Sierra Leone. Two years prior to the RUF/NPFL invasion in Sierra Leone, Charles Taylor’s NPFL started an uprising against Samuel Doe’s regime in neighbouring Liberia.

5.2 Main Actors

The civil war started off with the Revolutionary United Front launching a rebellion against the government of Sierra Leone, whose military is referred to as the SLA. Foday Sankoh led the RUF up until his arrest in 1997. After the arrest, the RUF was led by Sam Bockarie alias ‘Mosquito’. While the political program and ideology of the RUF were not very clear (Peters, 2011), they did publish a manifesto in 1995 explaining the purpose of the war. Their purpose seemed to be the creation of a democratic and more equal Sierra Leone:

We are fighting for a new Sierra Leone. A new Sierra Leone of freedom, justice and equal opportunity for all. We are fighting for democracy and by democracy we mean equal opportunity and access to power to create wealth through free trade, commerce, agriculture, industry, science and technology (RUF, 1995)

However, the portrayal of the RUF as a political movement seems to be misplaced given the widespread violations against the same people it claimed to fight for. The RUF is responsible for most atrocities committed during the civil war (TRC, 2004, appendix 1). In the beginning phase of the conflict and around 1997 when Taylor became president of Liberia, the RUF received military support from the NPFL. The SLA was at their turn backed by ULIMO forces from Liberia.

During the war, the SLA has been criticized for cooperation with the rebels, failing to protect civilians (Keen, 2005, p. 132-159). The lack of protection of civilians by the SLA contributed to the emergence of many smaller defence groups that would later collectively fight under the umbrella organization of the Civil Defence Forces (CDF). Danny Hoffman (2007) has argued that the CDF should not be perceived as one military organization but rather “(…) as the militarization of a web of social relations” (Hoffman, 2007, p. 640). The

(32)

main civil defence force was the Kamajors under the command of Samuel Hinga Norman. The Kamajors were usually Mendes while the Gbethis and Kapras defence militias usually descended from the Temnes. The military government in power from 1992-1996 perceived the CDF as undermining their authority (Keen, 2005, p. 7). This perceived threat coupled with the increasing issue of soldiers cooperating with rebels, led the Sierra Leonean government to introduce another actor into the conflict. The Executive Outcomes, a private military from South Africa arrived in 1995 in an attempt to restore order within the SLA and fight the RUF (ibid, p. 151).

During the war, the SLA fractionalized leading to the creation of the Armed Forces Revolutionary Council (AFRC) and the West Side Boys. Johnny Paul Koroma, a former SLA commander, led the AFRC. Together with the RUF, they formed the ‘The People’s Army’ alliance from 1997-1998. Their alliance was characterized by mutual distrust and little cooperation (Marks, 2013, p. 365). The West Side Boys emerged early 1999 as a splinter group from the AFRC. They were stationed in the Occra Hills east from Freetown, and their activity remained limited to this area.

5.3 Sierra Leone Civil War (1991-2002)

On 23 March 1991, a group of RUF rebels crossed from Liberian territory held by the National Patriotic Front of Liberia (NPFL), into Kailahun, a district in eastern Sierra Leone. Two years prior to the invasion, the civil war in Liberia started with the NPFL led by Charles Taylor, fighting Charles Doe’s regime. Taylor allegedly supported the rebellion into Sierra Leone out of economic interest in the resource-rich country (Human Rights Watch, 2005). The TRC has categorized the war in Sierra Leone in three phases, which will be used here to describe the sequence of the war.

The first phase is what the TRC refers to as ‘conventional target warfare’. This period lasted from the outbreak of the conflict until November 1993 (TRC, 2004, 3A p. 88). This period was characterized by battles between government forces (SLA) and the RUF over land and resource-rich areas. During this time, the RUF mainly operated at the border areas of the southern and eastern districts. They had parts of Kailahun district and Kono district under their control (Keen, 2005, p. 37). During the first phase of the conflict, the RUF received military support from the NPFL and Burkina Faso President Blaise Compaore

Forcing bonds, fuelling violence : combatant socialization and repertoires of violence during the civil war in Sierra Leone