“It was Naparama who ended the war,” an elder told me in a village in northern Mozambique.Footnote ¹ For the peasants who experienced the war in the rural areas of Zambézia and Nampula provinces, the community-initiated militia provided much-needed relief. Residents I spoke with perceived Naparama as being crucial to pacify and return their villages to some level of order and stability. As a community resident in Nicoadala stated, “[w]hen [Manuel António] Naparama came, he said that no one will ever [have to] flee again from their home. And indeed, we were able to sleep in our homes.”Footnote ² Not all claimed that the end of the war was due to Naparama, but many granted the militia an important role in reducing violence.

The case of Mozambique shows how important militias are for civilian self-protection and the dynamics of war. But the Naparama militia is not unique. Stories of spontaneous and more organized community initiatives to arm residents and protect communities come up in many civil wars. Scholars and policy makers alike have acknowledged that militias can upend the military balance and impact the dynamics of war and their aftermath and they have studied the causes and consequences of militia formation. Due to the use of different concepts and approaches, however, the field of militia studies has become fragmented and a unified body of research is yet to form.

In this book, I begin to define a research agenda on militias. The book brings conceptual clarity to the phenomenon of militias and explains when, where, and how communities form militias to defend themselves against violence. The book makes two important distinctions: first, between grassroots initiatives (community-initiated militias) and militias that are formed by state agents (state-initiated militias); and second, between militias that exercise their tasks part-time, and those that are committed full-time to their work as militias. The militia’s access to resources such as weapons and training and its level of professionalism has consequences for the type of operations that militias perform and the violence they perpetrate. The focus of this book lies on militias that are formed by communities and initially only complete their tasks during the day or night but otherwise go about their regular activities (community-initiated part-time forces). These are militias with few resources and little professional training. Parts of the book compare these militias to part-time state-initiated militias, which have the advantage of state resources.

The book analyzes how militias form from three perspectives and develops arguments for when, where, and how communities include such forms of self-protection in their repertoire of collective action. Community-initiated militias form when community residents’ lives are in danger and local military stalemates provide windows of opportunity for militia formation. This strategic situation shapes community incentives to form militias rather than flee or rely on the state military to protect them from insurgent violence. In the absence of sufficient support from the national armed forces, the local administrative elite tolerate or even support additional armed groups to fight insurgents.

While this strategic context may explain the timing, it does not fully explain the location of militia emergence. When militias form in neighboring communities and demonstrate their success against insurgents, residents may seek to form community-initiated militias in their own communities. This may lead to forms of collective action diffusing across community boundaries due to ethnic, cultural, or historical bonds between those communities. Such initial diffusion of militias is only transformed into sustained diffusion, however, if relative unity among elites allows the militia’s integration into the local security apparatus, which facilitates broad-based support from the local population, the administration, and the army.

Once a community includes militias in its repertoire of violent collective action, several factors may explain the rapid growth of the organization. When communities appeal to preexisting social conventions and base their creation of innovative institutions on them, they help residents to manage the uncertainty of war. Community-initiated militias attract a large number of recruits when they provide a new and innovative opportunity for self-empowerment, creating a sense of agency and hope among residents to address their hardship during war.

The book develops and illustrates these arguments with a diverse set of evidence from interviews and archival data on the Naparama militia and the Mozambican civil war from thirteen months of fieldwork in central and northern Mozambique. Naparama formed at a time in which local military stalemates between Frelimo and Renamo, characterized by much indiscriminate and collective violence, prevented a solution to the conflict. Community residents in Zambézia and Nampula provinces considered forming Naparama units as their best option in a war about control over people. The local Frelimo administration considered the indirect support for Naparama as their only chance to retain some level of military and political control over district towns in Zambézia and Nampula provinces.

Once Naparama leaders had demonstrated their power over Renamo in the first districts in which they operated, the militia spread from one district to another. Community residents longed for the same stability that neighboring communities were experiencing. However, elite conflicts prevented the militia from forming durable organizations in some districts, as individual elites abused Naparama units to pursue political ambitions.

Community residents were convinced that the Naparama militia provided a solution to their hardship because Naparama’s social and cultural references resonated with them and the militia provided an innovative institution that seemed more promising than preexisting ones. Residents felt empowered to not only protect themselves from violence, but also to regain agency over their life course and provide for themselves and their families.

Analyzing the origins of militias in Mozambique helps us understand wartime civilian agency and violent resistance, as well as how the rise of third actors affects the dynamics of civil war. As a result, this study sheds light on similar movements beyond Mozambique.

8.1 Civilian Agency and Violent Resistance

When Naparama emerged in Mozambique in the late 1980s, it was treated as a story of surprising civilian resilience. An international journalist who reported on the militia’s formation at the time spoke of a “rag-tag army” confronting the rebels and a surprising “triumph of spears over guns.”Footnote ³ Armed with spears and machetes, the grassroots movement fought back Renamo’s military organization of about 20,000 men and pacified many areas in northern and central Mozambique, an achievement the army had been unable to accomplish alone. This was a surprising story as civilians are often depicted as passive victims of war, and even if they try to organize, they often have the challenge of limited resources to defend themselves. Paying attention to the various ways in which civilians can claim agency helps us to understand how and why they would want to protect themselves and to recognize the challenges that come with such violent displays of agency.

The stories from Zambézia and Nampula show the ways in which the Naparama militia offered civilians agency. The militia empowered civilians to respond to their hardships and protect themselves. For militia members, becoming a Naparama was an opportunity to overcome the passivity that they felt was forced upon them, and to regain the ability to secure their own livelihoods:

But for those who did not formally join the militia, Naparama’s activities also had an empowering effect. As a community leader explained to me, when Naparama arrived, “the people began to experience the war [first-hand] and face the enemy, [and the war] became a popular war. Before, when it was just Frelimo, it was the army that engaged in war and the people fled.”Footnote ⁵ As with the Naparama combatant above, regained agency meant not only to survive but also to provide for one’s livelihoods. In the same conversation, the community leader explained that before Naparama arrived, “people were put into protected zones (villages). They received food and clothes in these villages. When Naparama arrived and vaccinated people, the community was able to return to the fields.”Footnote ⁶ Thus, the agency that community residents were longing for was to secure their survival and to ensure that they could provide for themselves.

The type of community empowerment we see in Mozambique fits with a new research agenda on civilian agency and civilian self-protection in civil war (Jose and Medie Reference Jose and Medie2015, Reference Jose and Medie2016; Kaplan Reference Kaplan2017; Krause Reference Krause2018), and elucidates what civilian self-protection can encompass and how it emerges. Concepts and typologies of civilian self-protection often include violent means, but in-depth analyses mainly refer to nonviolent individual or collective means to protect noncombatants. The book shows that the analysis of agency and self-protection can and should also encompass violent means of self-protection. I consider the type of militia under study in this book, community-initiated part-time militias, to be a form of civilian self-protection.

While civilians may seek protection from government or rebel forces by joining or supporting them, this book contributes to our understanding of why community residents opt for a “third” or neutral strategy in response to war – in other words, when they engage in “noncooperation” with armed groups (Masullo 2021). The military situation civilians find themselves in – indiscriminate or collective forms of violence – requires civilians to come up with a new, innovative response that allows for self-protection. My analysis of the Mozambican civil war, therefore, confirms Masullo’s (Reference Masullo2015, 47) argument that conditions of indiscriminate violence are unlikely to lead individuals to join and rely on insurgents or state forces as “indiscriminate targeting makes cooperation in exchange for protection ineffective.” In such violent and uncertain environments, community residents look for innovative ways to respond to the violence, and adopting violent or nonviolent means of resistance is such a response. What remains to be refined, however, is our understanding of the conditions under which communities and individuals opt for violent as opposed to nonviolent forms of resistance. Attention to civilian tactical choices will allow researchers, policy makers, and practitioners take into account the challenges that come along with the rise of third actors in civil war.

8.2 The Rise of Third Actors and Order in Civil War

Ignoring realities on the ground, the national and provincial Frelimo government adamantly tried to suppress civilian mobilization into community-initiated militias. Already weakened from the civil war against the rebels, the government feared that if it had to fight an additional armed group, an end to the war would be out of sight. What the Mozambican government realized – and conflict research has confirmed – is that multiple armed actors complicate the solution to civil wars. Though militias often support the government, they can turn on their sponsors and target the population they are supposed to protect. More systematic attention to third actors in civil war is needed to specify the ways in which such actors influence levels and types of violence, and the length and outcomes of civil wars.

This book shows that if local military stalemates are conducive to forming community-initiated militias, we should see more militias in prolonged irregular wars, or civil wars in which the adversaries are from the outset relatively equal in their military capabilities, such as symmetric nonconventional wars (Kalyvas and Balcells Reference Kalyvas and Balcells2010). In this sense, although the book’s theoretical arguments apply primarily to wars in which an asymmetry exists between state and rebel forces, it can also apply to wars with a range of armed groups of equal military capability that exercise high levels of indiscriminate violence. Violent military stalemates do not only prevent incumbents and insurgents from winning the war, they also make settlements between them more difficult, thereby contributing to the length of civil wars. In such situations, militias proliferate. The state may attempt to gain a military advantage over the insurgent by multiplying its forces and delegating tasks to militias. The incumbent may also increase pressure on the insurgent by outsourcing violence against civilians under the insurgent’s control to militias. Communities may form militias in prolonged civil wars, as they do not see any other option to protect themselves. War-weary communities may want to bring an end to the war themselves and so form militias. What is difficult to ascertain here, however, and what needs more research, is the precise causal direction; though I argue that prolonged civil war leads communities to adopt militias, there is some cross-national evidence that militias themselves make wars last longer (Aliyev Reference Allen and Vlassenroot2020a).

What is important to recognize is that militias are not just any additional armed group. They do have distinct characteristics that distinguish them from other armed groups, at least in irregular wars.Footnote ⁷ For example, a key distinction between community-initiated militias and insurgent groups appears to be the locally rooted character of militias. Militia recruitment is generally more localized than rebel recruitment, which enables reliance – and innovation – upon previously existing forms of organization. Such use of social conventions may help manage wartime uncertainty and explain the rapid growth of some community-initiated militias. It is therefore likely that in some contexts, counterinsurgent recruitment is less ideologically determined and more security-related, as community residents may participate in militias to signal their loyalty to the government to receive protection (Schubiger Reference Schubiger2021). This focus on security represents militia members’ short-term motivations. For example, in contrast to the supporters of the insurgents in El Salvador, on which Wood’s (Reference Wood2003) concept of pleasure of agency is based, Naparama members did not have a vision of social justice and did not think they were making history. The objective to bring back family members and reclaim land resulted from the concern for the family’s short-term survival, and did not include any long-term political goals. However, as this book shows, even short-term goals may only be met with participation in an armed group that allows for sufficient empowerment, which raises the sense of purpose and expectation of success. Such short-term vision also points to a source of armed group cohesion that is different from ideology or socialization, and therefore worthy of further study.

The importance of social conventions for attracting militia members also has implications for state-initiated militias. When states or occupying forces rely on preexisting forms of organization or social conventions, they are more likely to create sustainable militia organizations. Policy makers have recognized how important familiarity of new organizational structures is for mobilizing new militia members. For example, the US military in Afghanistan built on previously existing forms of community policing to create militias that were later united into the new Afghan Local Police (Goodhand and Hakimi Reference Goodhand and Hakimi2014). As with community-initiated militias, the institutional context of communities influences whether militia organizations form for the long term. In the Khan Abad district in Afghanistan’s Kunduz province, for example, militias quickly fell apart due to elite conflicts that turned newly formed militias into private armies.Footnote ⁸ Scholars have begun to analyze the consequences of the “familiar” character of militias, showing that recruitment based on ethnicity is associated with less lethal wars (Aliyev and Souleimanov Reference Aliyev2019). These examples show, however, that we need to go beyond ethnicity as a proxy for familiarity. To fully understand how and why familiar structures create sustainable militia organizations, we need to trace their precise origins and analyze how social conventions structure their relation to the population.

More generally, militias represent informal institutions of security governance that emerge during civil wars to create alternative forms of political and social order. Debates both on how rebels “rule,” build proto-states, and provide public goods and social order to civilian communities, as well as how warlords contribute to “multiple layers of authority” in war-torn countries have dominated the discussion of governance in civil wars (Mampilly Reference Mampilly2011; Arjona Reference Arjona2016; Malejacq Reference Malejacq2019). Scholars usually assume that governments are “unitary” actors. Together with the emerging research agenda on (pro-government) militias in civil wars, this book challenges this assumption as it shows that states are fragmented and rely on multiple armed groups to respond effectively and efficiently to rebel threats. Theories on the dynamics of war should therefore take into account the potentially fragmented nature of the state.

8.3 Beyond Mozambique

Grassroots initiatives, which seek to protect the local population from insurgent violence, have influenced and continue to influence the dynamics of war in many other cases. A prominent case is the rise of militias in Sierra Leone’s civil war from 1991 to 2002, which provides a fitting example of the militia mobilization process as theorized in this book, albeit in a different kind of civil war. A variety of community-initiated militias influenced the dynamics of war in Sierra Leone: Kamajors, Tamaboros, Donsos, Kapra, Gbethis, and the Organized Body of Hunting Societies (OBHS), whose membership was largely defined by ethnicity and geographic origin (Wlodarczyk Reference Wlodarczyk2009, 62). Similar to Naparama, Sierra Leone’s militias relied on the belief in a bullet-proof medicine that every new member would receive during an initiation ceremony.

The case of Sierra Leone helps to explore the scope of my arguments. In contrast to the militias in Mozambique that formed toward the end of the war, many of the groups in Sierra Leone formed in the early years of war in order to protect communities against insurgent violence and support or even substitute the army (Ferme and Hoffman Reference Ferme and Hoffman2004; Muana Reference Muana1997). This could be explained by reference to the type of war; the war in Sierra Leone is considered a symmetric nonconventional war as the rebels challenged a relatively weak state (Kalyvas and Balcells Reference Kalyvas and Balcells2010). In wars that are characterized by relative symmetry among the armed groups, community-empowering stalemates can emerge much earlier than in irregular wars, which would explain the rise of militias in the early stages of war. A second important difference to the case of Naparama is that militias in Sierra Leone evolved into more autonomous forces over time, eventually replacing the state army. Recognizing the militias’ power and its own military weakness, the newly elected government of the Sierra Leone People’s Party (SLPP) officially recognized the militias in 1996/97 and united them into the Civilian Defense Forces (CDF). This aligns with the expectation that militias in wars characterized by more symmetry between armed groups gain more autonomy (Kalyvas Reference Kalyvas2006, 107n44). The focus of this book on militia formation provides a limited perspective, however, on how and why militias in Sierra Leone evolved, professionalized, became co-opted, and perpetrated violence against those they were supposed to protect. Sierra Leone’s President Kabbah formed the CDF in order to unify the different hunter militias in the country. Yet the government’s intervention to unite all militias diluted leadership, slackening the once-strict selection and initiation procedure that every young man had to undergo and contributing to indiscipline and increased levels of violence against civilians (SL TRC 2005; Forney Reference Forney2012). That the militias lost some of their original structures and processes through the government’s interference can explain parts of these developments, but this is an important question to take up in further research (Hoffman 2011).

As for the mobilization process, we see interesting similarities among the two cases that point to the importance of innovation when civilians face a security crisis. As in Mozambique, militias in Sierra Leone mobilized members by creating innovative institutions that relied on preexisting social conventions, which resonated with the relevant communities. The Tamaboro in the north among the Kuranko ethnic group were among the first militias to form. In 1992–93, they worked as guides to the Gola Forrest to support the National Provisional Ruling Council (NPRC), a military junta that ousted President Momoh in a coup in April 1992 (Fithen and Richards Reference Fithen, Richards and Richards2005, 127). The NPRC valued the hunters’ knowledge of the local terrain and their reliability, which exceeded that of the regular army. The NPRC decided to mobilize militia members through Poro societies, the traditional initiation societies for adolescents. The Kamajor militia in the south of the country was modeled on the Tamaboro in the north. A hunter in the Jong chiefdom (Bonthe district) among the Mende ethnic group began training a local militia based on hunting rituals (Fithen and Richards Reference Fithen, Richards and Richards2005, 127). The example of the Tamaboro provided a crucial innovation for the southern communities: the hunter (kamajô) in the south had much more individualist connotations and was less connected to war traditions than in the north (Ferme Reference Ferme2001, 121). The first initiators brought together craft hunting and organizational modalities from youth initiations to transform the “mysterious individual hunter” tradition into a mass movement: “The hunter civil defence might thus best be understood not as a preexisting institution but a syncretic institutional response to the security threat posed by the RUF” (Fithen and Richards Reference Fithen, Richards and Richards2005, 128). This innovation successfully mobilized a large number of militia combatants. By 1996, almost all chiefdoms in the south and east had Kamajor societies and mobilization mechanisms (Wlodarczyk Reference Wlodarczyk2009, 64). The community-initiated militias empowered the people in Sierra Leone to respond to the increasing threat posed by the RUF – the Revolutionary United Front. Militia mobilization “took place in the context of the national army’s failure to protect villagers in the east and south of the country against the initial excesses of RUF ‘special forces’” (Fithen and Richards Reference Fithen, Richards and Richards2005, 127).

Both cases, Mozambique and Sierra Leone, demonstrate the power of storytelling for mobilization processes. In Mozambique, stories of Naparama’s “magic” – their invulnerability to bullets and António’s alleged powers – resonated with community experiences of spirit mediums and traditional medicine. In Benford and Snow’s (Reference Benford and Snow2000) terms, Naparama’s framing had “narrative fidelity” and resonated culturally. The heroic and magical stories about the capabilities and success of both Naparama and Kamajor were crucial in creating hope and motivated many to join. The importance of such framing can be found across civil wars in Africa. As Ellis (Reference Ellis1999) has shown in the context of the civil war in Liberia, existing spiritual practices may serve as a source of invention to mobilize and prepare fighters for war. For Sierra Leone, Hoffman (Reference Hoffman2011, 76) argues that storytelling that referenced social conventions was more important in the Kamajor civil defense movement’s growth than preexisting kinship groups and social networks. It shows that social embeddedness of armed groups needs to be understood in much broader terms – beyond social and ethnic ties to also include social conventions and organizational repertoires that may account for how armed groups form and mobilize their members.

While this study has focused on a particular case of community-initiated militia formation during Mozambique’s civil war, the theoretical and empirical implications illuminate similar processes in other types of wars and contribute to our understanding of political violence, civilian agency, and collective action more generally. This book deals with the important questions of when, where, and how community-initiated militias form. Yet, other relevant themes need to be explored. For example, the social embeddedness of community-initiated militias gives rise to various questions: Are community-initiated militias better at protecting civilians than state-initiated ones? What about the trajectory of community-initiated militias? Under what conditions can they retain their relative autonomy from the state? Under what conditions do they turn on the people they are supposed to protect? And (when) should we expect such militias to provide more than just security to communities and establish more complex forms of governance parallel to the state?

The limitations of my theory point to important future avenues of research. Understanding why community-initiated militias emerge is just a first step. To evaluate their impact, we need to study how they evolve over time, when they ally with other (armed) actors, and why they sometimes violently target civilians they are supposed to protect.

8.4 Demobilized at Last?

The 1992 Rome agreement ended the Mozambican civil war and ushered in a long peace process that was challenged by renewed violence. Tensions between the rebel-turned-political party Renamo and the government escalated twenty-one years after the war’s end, in 2013, and led to violent confrontations between Renamo-affiliated armed forces and government forces. It took several rounds of peace talks and two new peace agreements to bring that renewed conflict to an end in August 2019 (Darch Reference Darch2015; Pearce Reference Pearce2020). The long-time Renamo leader Afonso Dhlakama died in May 2018, which helped to bring about a new phase in Renamo’s transformation into a political party. In addition to tensions with their long-time political rival, the Mozambican government has faced a new insurgency in the far north of the country since October 2017 (Habibe, Forquilha, and Pereira Reference Habibe, Forquilha and Pereira2019; Morier-Genoud 2020). The Islamist movement, which pledged allegiance to the Islamic State of Iraq and the Levant (ISIL, known as ISIS) in July 2019, grew over two years to a serious armed challenge and was able to capture and occupy territory in early 2020.

Contrary to what we might expect, the political violence of recent years has not reactivated Naparama. But this is mostly due to Mozambique’s political geography. The resurgence of Renamo violence took place in the center of the country, in the Sofala province, where the rebels maintained their main headquarters during the civil war and loyal supporters remained living there throughout the postwar period. The Islamist insurgency is mostly confined to the northernmost province of Cabo Delgado, thousands of miles away from where Naparama had operated.

After the war had ended, the Naparama militia disintegrated in all the districts I worked in, except Nicoadala in Zambézia province, where the militia’s headquarters were located during the war. Nicoadala was also among the districts where former Naparama combatants (and soldiers) revolted after the war. Although no official relationship between the Frelimo government and Naparama forces existed, Naparama combatants in Namapa (Nampula) and Nicoadala districts demanded inclusion into the demobilization scheme to receive pensions and other benefits (see Dinerman Reference Dinerman2006).Footnote ⁹ Naparama units in these districts sought recognition from the government, as the demobilization process conducted by ONUMOZ did not include combatants for the state-initiated and community-initiated militias.

In 2011, these conflicts resurfaced when Hermínio dos Santos, leader of one of the associations for demobilized soldiers (the Forum for the Demobilized of the War [Fórum dos Desmobilizados da Guerra]), staged demonstrations to demand that the government raise demobilized soldiers’ pensions and include state- and community-initiated militias into such pension schemes. The protests were a reaction to a new law that redefined the status of demobilized soldiers, which previously only applied to combatants from the independence war. The new statute for former combatants sought to include combatants from the civil war under the umbrella category of “former combatants.” While other (Frelimo-loyal) associations for demobilized soldiers supported the new law, dos Santos criticized the statute for not including former members of the state security agency, the police forces, and (state- and community-initiated) militias.Footnote ¹⁰

The Naparama leadership in Nicoadala welcomed dos Santos’s support, as efforts to collect signatures from former Naparama members and other documents had not brought them any closer to official recognition by the provincial government. In my interviews with them, former Naparama combatants repeatedly complained that Frelimo had never officially recognized that Naparama had contributed to the war effort. “We worked, too” was a common expression with which former militia members demanded access to benefits similar to those that officially demobilized soldiers currently receive. But the renewed mobilization did not gain any traction. Dos Santos died in 2017, so the former militia members lost an important spokesperson and, therefore, political relevance.

The theme of incomplete demobilization is one that has dominated the continued tensions between the Frelimo and Renamo parties after the 1992 peace accord, as Renamo kept a small armed force throughout the years, which facilitated the return to violence in 2013. Incomplete demobilization represents a broader theme of unequal access to economic and political opportunities. The Frelimo party has long marginalized certain social groups and regions and more recently turned to repression of the opposition. While the precise origins of the recent insurgent violence in Cabo Delgado are difficult to ascertain, it appears that the violence was fueled by a government that did not sufficiently respond to citizen concerns, and early repression led to an escalation that escaped the army’s control. Mozambique’s transition to democracy and peace will remain a struggle for citizens’ political, social, and economic equality. A luta continua.