Buying Them Off or Scaring Them Straight: Explaining Warlord Democrats’ Electoral Rhetoric

Abstract Post–civil war democracies are characterized by intense electoral competition. To ensure continued political relevance, ex-military-turned-politicians, or “warlord democrats” (WDs), can either engage in a rhetoric of fear or ease societal tensions by employing a rhetoric of peace. WDs’ choice of rhetoric can have a profound impact on durable peace by altering societal discourses concerning the legitimacy of using violence. A key question is therefore: Why do some WDs employ a rhetoric of fear, and others a rhetoric of peace, when running for office? We argue that the choice of rhetoric is a function of the patrimonial endowments WDs possess; if WDs lack the resources and social networks needed to distribute patronage, they may instead use a rhetoric of fear to rally voters. To highlight the explanatory value of this proposition, we compare two Liberian WDs who ran for the Senate in 2005—Adolphus Dolo and Prince Johnson.

how elite speech acts can alter broader societal discourses. 9 Once set in place, such discourses define what behavior is socially accepted. 10 Put differently, depending on what type of discourse prevails, it will be more or less easy for spoiling elites to engage in violence. It is particularly vital that benevolent discourses reign during times of elections; studies have shown that due to the stakes involved, postwar elections easily become violent affairs. 11 Hence, at best WDs can help transform the ethos of war permeating society into an ethos of peace. Such societal shifts do not only lay the ground for more inclusive identities but also increase the chance of improving intergroup trust. 12 Due to the role WDs' speech acts have in shaping societal discourses, it is vital to gain a more profound understanding of why ex-military-turnedpoliticians sometimes engage in fearmongering. There is, however, a lack of theories addressing the question at hand. The research question guiding this article is therefore: Why do some WDs employ a rhetoric of fear, and others a rhetoric of peace, when running for office? More specifically, we analyze the effect patrimonial endowments (the combination of economic resources and social networks needed to mobilize clients) have on WDs' choice of electoral rhetoric. This is because access to such forms of endowments plays a key role in shaping the postwar political trajectories of WDs. 13 We argue WDs are most likely to employ a rhetoric of fear when they possess limited patrimonial endowments. Under such circumstances, exmilitary leaders lack the resources and social networks to rally voters via patronage distribution. Consequently, WDs are obliged to develop alternative electoral survival strategies. One such strategy is to recreate wartime anxieties by making aggressive speech acts. Such utterances have the potential to decrease the relative value of patronage in favor of credible promises of future protection. Conversely, when WDs possess significant patrimonial endowments, they have fewer incentives to engage in fearmongering since such utterances risk alienating wartime victims, members of other social groups, or moderate voters. Instead, WDs are likely to use their superior funds and multiple social networks to engage in distribution politics and employ benevolent speech acts to rally voters. In order to assess the explanatory value of this argument, we conduct a structured, focused comparison between two WDs who participated in Liberia's 2005 national elections: Adolphus Dolo (ex-National Patriotic Front of Liberia (NPFL) and Government of Liberia); and Prince Johnson (ex-Independent National Patriotic Front of Liberia (INPFL)). Interestingly, although both WDs contested for Senate seats for the same county (Nimba), they differed in which kind of speech acts they employed. Whereas Johnson systematically used a rhetoric of fear, Dolo shunned away from aggressive statements, instead preferring to stress remorse and forgiveness.
This study makes an important contribution to the literature on post-civil war democratization by underlining the role of the individual actor and stressing the importance of their usage of speech in electoral settings. Exmilitary leaders do not only affect electoral dynamics by engaging in violence, abuse, and fraud but also through the rhetoric they employ. In fact, their utterances help shape communities' sense of security and trust in the new peace order.

Inspiring Peace or Instigating Fear: WDs and the Role of Speech Acts
The relationship between democracy and peace is complex. On the one hand, there is a strong correlation between consolidated democracies and civil peace, where free and fair elections enable societies to more effectively resolve collective disputes. 14 On the other hand, it has been argued that democratization can have a negative effect on stability, as it facilitates elites' ability to play on nationalist sentiments, which in turn can cause conflict along ethnic lines. 15 This is particularly true in post-civil war societies, where electoral mobilization is often carried out along old conflict lines, and former belligerents fear what will happen to them if their opponents prevail. 16 To minimize the risk that elections fuel violence, peacemakers have, over the last 25 years, invested considerable resources in the (re)construction of neutral and strong democratic institutions, as well as viable political parties. The underlying assumption is such transformations can make politics more transparent and broad-based, and minimize governmental abuse. 17 However, in the short term, institutions and political parties tend to be weak and fragmented in war-ridden countries. More often than not, these entities function as a front for the personal ambitions of one or a few Big Men-influential elites who, due to their position in informal and formal economic-political structures, accumulate substantial resources. In these contexts, WDs constitute one of the most influential categories of Big Men, as they often have a competitive advantage over others. 18 First, WDs often accumulate substantial resources during the war and peace processes, allowing them to "outspend" their civilian counterparts. 19 Second, many WDs have a loyal following among communities they protected during the war. During elections, WDs can play on these solidarities to mobilize support. 20 Third, WDs habitually control informal command structures (ties to excommanders/fighters), which can be employed to mobilize followers or intimidate opponents. 21 WDs do not necessarily possess these advantages because they represent a particular party or institution. It is rather a function of their skills as military Big Men: excelling in the art of amassing followers, wealth, and wartime prestige. 22 As such, we argue postwar democratization may have less to do with building institutions and parties than bringing the individual back in and finding ways to transform warlords into peacelords.
One of the most fundamental ways WDs shape war-ridden societies is through the speech acts they employ. 23 As argued by Lene Hansen, language is political in the sense that it is "the site for the production and reproduction of particular subjectivities and identities." 24 This is because political action first necessitates the establishment of a legitimizing discourse that contains meanings and rules for what is acceptable behavior. 25 The discursive constructions help shape both collective identities and perceptions of security. The construction of a threat-identity nexus can be linked to the assumptions of securitization. Securitization concerns the intersubjective process in which an actor declares an issue to be a threat to a referent object (for instance, the nation state or ethnic group), requiring extraordinary measures. This so-called securitizing move is directed toward an audience, such as the broader political elite or public, which accepts or rejects the threat image. 26 Oftentimes such moves include awarding negative virtues to "out-groups," branding them as "enemies of the nation." 27 As demonstrated by Jack L. Snyder, Serbian president Slobodan Milo sevi c employed this strategy, as he "tried to discredit his liberal opponents by branding them as tools of a foreign enemy." 28 The making of public threats and securitization of social identities makes it easier for the securitizing agent to suggest extraordinary measures that may otherwise be viewed as unacceptable: for instance, the creation of militias or engagement in ethnic cleansing, or, in other words, the agent can "entice or force [followers] to join in the enterprise of collective 'self-defence. '" 29 The systematic usage of threatening speech acts-here held as a rhetoric of fear-risks polarizing relations between war-affected groups and challenges the new peace order. A rhetoric of fear can manifest itself in several ways. First, WDs can threaten to employ violence if not elected into office. Such speech acts not only risk instilling fears of renewed hostilities but are also likely to polarize relations between the ex-military leader's supporters and opponents. Second, if WDs depict members of other social groups as dangerous, they remind their own constituencies of how vulnerable they are to aggression. In such situations, communities may feel obliged to rally behind a belligerent WD. In addition, the absence of strong and neutral media outlets-a common phenomenon in war-ridden societies-make it difficult for communities to assess the validity of WDs' statements, and the costs of falling outside the protection of Big Men are extremely high. 30 For many WDs, it can be rational to engage in these types of fearmongering. Aggressive speech acts can generate crucial votes during elections; by playing on their wartime credentials, WDs can make credible promises to protect their constituencies should new hostilities erupt. For instance, in Bosnia and Herzegovina's first postwar elections, moderate politicians lost due to the ethnic outbidding of more militant leaders. 31 Although the role of a rhetoric of fear has been addressed in prior studies on, for example, nationalist movements, we argue there is an equally important other side of this coin, namely, the rhetoric of peace. In postconflict settings, the arrival of peace offers an opportunity to deconstruct the aggressive discourses developed during war. A key step toward this is when WDs engage in desecuritization processes; that is, they discard their rhetoric of fear in favor of more benevolent speech acts. A rhetoric of peace calls for an end to hostilities and stresses the economic, social, and political benefits of peace. It can furthermore incorporate aspects of recognition, 27  that is, verbal acknowledgments by the WDs of their-and their followers'-wartime transgressions, as well as calls for reconciliation and acknowledgement of out-groups' suffering. 32 Hence, unlike a rhetoric of fear-which dwells on the past and differences between groups-a rhetoric of peace embraces forward-looking perspectives that underline the commonality of former belligerents. 33 Nelson Mandela-ex-leader of the African National Congress and its military wing-probably constitutes the most lucid example of an ex-military who employed a forward-looking and inclusive rhetoric. During the 1994 elections, Mandela ran a campaign based on a rhetoric of peace that emphasized unity and the commonality of different ethnic groups. 34 Electoral rhetoric is, however, not only about what is said. Of equal importance is how often it is uttered. If an ex-military-turned-politician once lashes out against members of other social groups, it is possible to claim that he was misquoted. The situation is very different if belligerent statements are employed in a systematic manner. Under such circumstances, one can suspect that a WD is actively seeking to incite anxieties. Likewise, for a rhetoric of peace to be seen as authentic, it is necessary for reconciliatory statements to be repeatedly dispersed. At best, a systematic use of conciliatory messages can initiate a process whereby WDs "develop tools for gathering support without extreme nationalist appeals." 35 Finally, it is important to note that it is not uncommon for elites to use "rhetorical ambivalence," switching between belligerent and benevolent speech acts. 36 It is therefore unlikely that WDs only use one framing strategy. This article thus maps rhetorical trends and explains why some WDs employ fearmongering, or peaceful statements, to a greater extent.

Rhetoric of Fear: Mobilization on the Cheap
Why, then, do some WDs employ a rhetoric of fear, while others use a rhetoric of peace? Even if no theories explicitly address this question, it is possible to identify potential explanations in adjoining fields of research. Exploring the role of nationalist persuasion, Snyder has argued that leaders whose power bases begin to fade in times of democratization can use an aggressive rhetoric "to exclude so-called enemies of the nation from enjoying democratic rights … [arguing that] ethnic minorities, the working classes, rival elites, or other political opponents should be excluded from political participation." 37 This scenario is particularly likely under conditions of weak political and civic institutions, an undeveloped free press, and when "the democratizing country is poor, [and] when its citizens lack the skills needed for successful democratic political participation." 38 Meanwhile, scholars examining the political behavior of elites have found leaders with a military background often respond aggressively to external pressures, such as security threats. 39 If linked to the issue examined here, it could be argued that exposure to threats would cause WDs to adopt certain rhetorical strategies.
However, the abovementioned studies' applicability to explain the puzzle probed here is limited. First, the unit of analysis and sociopolitical context tends to be very different. Unlike WDs, these studies' elites are more bound to political parties, an aspect likely to affect political opportunity structures and electoral strategies. In addition, their point of departure is not post-civil war societies, which have proven to be particularly susceptible to intergroup hatreds and armed conflicts. 40 Second, as we compare WDs from the same country (Liberia), country-specific factors-for example, institutional strength, freedom of press, levels of poverty, and democratic norms-cannot explain rhetorical variations between WDs. Finally, Snyder compellingly points out that there is a relationship between elite fears of losing "their parochial economic interests as well as their positions of power in society" 41 and their willingness to employ certain strategies, such as nationalist persuasion. Such an argument may not only explain why elites operating in the same country employ different types of speech acts but also illuminate why individual leaders may switch rhetorical strategies over time. One problem with this argument is there is no benchmarking of when "threats to inflexible elite interests are high." 42 This makes it difficult to apply the proposition to other contexts, such as the electoral maneuverings of WDs in post-civil war societies.
Instead, we argue our research question can be answered by referring to the patrimonial endowments-the economic resources and multiple social networks required to distribute patronage to potential clients-that WDs possess. Resource-rich WDs, who have access to multiple social networks, have incentives to engage in a rhetoric of peace. By distributing patronage, they can attract voters from various social groups and maximize their 37  chances at the polls. This is in sharp contrast to WDs who possess limited patrimonial endowments. Such ex-military leaders are obliged to develop alternative electoral survival strategies. One such strategy is to seek to decrease the value of patronage by using aggressive speech acts to recreate wartime anxieties. This allows them to mobilize voters by offering protection rather than resources. By looking at what may be described as WDs' patrimonial power base, we align ourselves with scholars, such as Snyder, who see elite belligerency as a calculated response to growing political weakness. Before further outlining the causal logic of the argument, it is necessary to gain a deeper understanding of the central role "wealth in people" plays for WDs operating in an electoral context.
In developing countries, where most civil wars take place, elite formation is generally a function of wealth in people 43 -the ability to amass clients who can be used as workers, voters, or fighters. In such societies, power is personalized and social mobility occurs when Big Men incorporate other citizens into his grouping. 44 To amass people, elites must have access to patrimonial endowments. By providing (or promising) money, employment, scholarships, food, and clothes, elites can entice people to provide them with various services (voting, labor, fighting). 45 However, having access to an infrastructure of distribution is just as vital as possessing resources. Barring such structures, Big Men will struggle to allocate the necessary patronage to intended groups. Habitually, elites employ preexisting social networks-ranging from ethnic, religious, and ex-military ties to common membership in political, economic, or cultural organizationswhen distributing benefits. Not only is it more cost-efficient to tap into existing structures, the trust inherent in social ties makes mutual promises more credible. Put differently, the presence of a preexisting bond between Big Men and their followers decreases transaction costs. In order to gain access to as many clients as possible, it is rational for elites to try to tap into multiple social networks. 46 Big Men can do this by attaining numerous leadership roles-ranging from becoming a politician, businessman, and community leader to security provider, religious figure, or head of a sports club. 47 Increasing or safeguarding one's patrimonial endowments is often a key elite objective during civil wars. Aspiring Big Men often see violence as a route to personal enrichment and an opportunity to expand their networks of dependents. 48 Such actors can best be described as "self-made men, entrepreneurs who rely solely on their personal skills as military organizers, political leaders and charismatic orators [who] use war as a resource to benefit from personally, to accumulate prestige and wealth." 49 Meanwhile, entrenched elites fight to defend their and their followers' more privileged positions. 50 Irrespective of whether it is the former or latter, armed conflicts provide elites with an opportunity to reach out to new followers. During the war, elites habitually sponsor nonstate armed groups such as rebel movements or militias. Through the command structures of such outfits, Big Men gain access to loyal cadres of fighters. In addition, in areas military leaders' armed factions control, community members can be incorporated into the formers' networks of dependents. 51 Elites can sustain wartime structures in various ways, ranging from coercion and appeals to collective identities to the provision of security and rudimentary services. Oftentimes, wartime leaders complement such actions with the dispersal of economic benefits to key followers. 52 With the arrival of peace, elites habitually seek to "convert their gains made during the war into material security and social status." 53 The most efficient manner to do this is to ensure a continued wealth in people. Oftentimes ex-military leaders put significant emphasis on cultivating old wartime relations. This can entail befriending ex-combatants also after armed groups have been dissolved and unofficially continuing to provide security to traumatized communities. Such conservative strategies are not without risk. As security improves, wartime capital depletes. It is therefore crucial for Big Men to transform their wartime capital into other forms of social capital that can attract new constituencies. This may necessitate taking on new peacetime leadership roles. It is not uncommon for ex-military leaders, who have invested in business enterprises during the war, to fashion themselves as businessmen. 54 Another notable strategy is to become religious leaders. In certain religious contexts-such as evangelical churches-it is not necessarily an impediment to have committed gross human rights violations. On the contrary, the miracle of conversion becomes much grander when those with blood on their hands repent. 55 However, for most ex-military leaders, becoming "democrats" is the most essential form of shape-shifting. 56 This is because it offers the best prospect of continued political relevance and access to state resources. By engaging in similar forms of shape-shifting, wartime elites can gain access to new social groups to which they have few wartime ties. It is, however, not enough for ex-militaries to champion themselves as the head of numerous social groups. They must also provide benefits to transform network members into loyal clients. To some extent, this can be done by continuing to provide security to clients. 57 However, as peace processes consolidate, the provision of economic resources usually becomes more important. This is particularly true for WDs. Studies have shown that running for office in democratizing and developing countries is an extremely costly affair. 58 Not only are aspirants expected to distribute large amounts of money, food, and jobs to potential voters, they also need to exhibit their prosperity by displaying expensive goods and organizing festive events. The latter may be particularly essential since it "somehow reassures the followers of a Big Man about his capacity to supply and satisfy the network of dependents." 59 Commonly, economic inducements are dispersed via formal or informal community leaders-chiefs, ex-commanders, religious leaders, as well as heads of youth or women's organizations-who claim to represent various social groups. 60 Due to the benefits of mobilizing clients from multiple social networks, WDs are more likely to employ a rhetoric of peace if they possess substantial patrimonial endowments. In fact, by making aggressive utterances, exmilitary-turned-politicians risk alienating certain groups, such as wartime victims, business groups, and moderate voters, who may be reluctant to support aggressive politics. A safer strategy to ensure broad-based mobilization is therefore to distribute patronage-a coveted resource among most groups-via multiple social networks. What happens when WDs do not possess the patrimonial endowments needed to credibly compete in elections? Under such circumstances, WDs are obliged to develop alternative, and more risky, electoral survival strategies to attract clients that can be mobilized as voters. We argue that the most cost-efficient way to do so is to engage in a rhetoric of fear. Studies in adjoining fields of research have shown how Big Men have used various forms of aggression-such as fearmongering-to salvage depleting networks of clients. 61 One reason why elite belligerency has the potential to pull back fledging followers is that it increases the costs of falling outside the protection of a leader. As perceptions of insecurity increase, people tend to attach themselves to like-minded individuals who share similar traits. 62 In war-ridden societies, this usually entails professing loyalty to old wartime identities. Under such circumstances "leaders who have the most violent past may make the most convincing claim that a vote for them is a vote for peace." 63 Hence, by instilling fear among the electorateirrespective of whether it is through threatening to unleash violence, depicting other groups as dangerous, or proclaiming the in-group is in jeopardy-WDs can decrease the value of patrimonial endowments. In this sense, "rhetorical deployments can have structural effects" that "change the game's structure." 64 A rhetoric of fear can therefore best be described as mobilization on the cheap, which allows WDs with restricted resources and limited social standing to shift how the electoral game is played. Put differently, "references to violence in campaign speeches alone can change realities. Politicians use interpretations of the reality, which the electorate can easily relate to and in doing so shape political discourses on violence, peace, and politics in their favor." 65 Hence, in this context, the audience is potential clients-in search of protection-who can be mobilized as voters.

Methodological Considerations
This study speaks to the wider population of cases of WDs in post-civil war countries characterized by electoral politics, weak state institutions, and systems of patronage. Similar cases can be found in Sub-Saharan Africa (for example, Burundi, Côte d'Ivoire, DRC, Guinea-Bissau, Mozambique, Liberia, Republic of Congo, Sierra Leone, South Sudan, Uganda) and a number of Eurasian countries (for example, Afghanistan, Indonesia (Aceh), Kosovo, Tajikistan, Timor-Leste). 66 For the purposes of this study, we compare two WDs from Liberia. There are two reasons for this. First, Liberia's two civil wars empowered a large number of (ex-)military actors, and Liberian politics have traditionally been dominated by Big Men rather than strong political parties. 67 In conjuncture, these dynamics mean WDs have constituted key electoral agents in the two elections-1997 and 2005-meant to consolidate peace. As such, Liberia offers a large pool of ex-military-turned-politicians from which to choose. Second, there are also benefits of selecting WDs from one country, as it allows us to hold a number of country-specific factors constant, such as institutional/electoral design and strength of state institutions, transitional justice mechanisms, foreign intervention, relatively strong peacekeeping troops, and international investments in peacebuilding.
Although several WDs have run for office in Liberia, 68 we have selected two-Adolphus Dolo and Prince Johnson-who both ran for the Senate in the 2005 national elections. There are several reasons for focusing on these two ex-military-turned-politicians. First, both were key military leaders during the hostilities; Johnson headed his own armed group, INPFL, and Dolo was one of the most influential generals of ex-rebel-turned-president Charles Taylor. Second, both Dolo and Johnson were aspiring senatorial candidates for the same county-Nimba (two seats were assigned to the county). It is thereby possible to hold several local factors constant, such as ethnic composition, economic conditions, and wartime experiences. There is, finally, an interesting variation in the outcome of interest: whereas Johnson systematically employed a rhetoric of fear, Dolo preferred to stress remorse and forgiveness.
To collect data on the WDs' rhetoric of fear and peace, we use a twopronged strategy. First, we collected Dolo's and Johnson's electoral statements in national and international newspapers. In this process, we read every article in which our selected WDs appeared, from the date their supporters petitioned them to run for office 69 to 9 November 2005-one day after the completion of the 2005 elections. 70 To identify relevant articles we employed the FACTIVA and AllAfrica search engines, using the WDs' names. This resulted in many articles where the WDs figured, including both local newspapers (for example, the Liberian Observer, the Analyst, and the Inquirer) and international news agencies (for example, Integrated  70 We use this cut-off point to include speech acts made in newspapers on the day of the final rounds of the elections. Regional Information Networks (IRIN), 71 Reuters, and Voice of America). This material was complemented with archival research at National Archives of Liberia. This allowed us to identify additional articles (in Liberian newspapers) not picked up by FACTIVA and AllAfrica. 72 In total, we collected 126 articles (85 for Johnson and 41 for Dolo). 73 Of these, 80 came from local and 46 from international newspapers. Second, to capture rhetorical patterns at political events not covered by media, we interviewed individuals-ranging from ex-combatants/commanders and politicians to journalists and ordinary citizens-who had witnessed rallies the WDs organized. In total, we collected testimonies from 17 individuals who participated in a total of 60 rallies. We asked the informants about what the WDs said during the rallies, how the latter spoke about the war and peace processes, and interethnic relations.
To analyze the collected texts and examine the WDs' rhetoric, we conduct a structured discourse analysis (SDA). Most types of discourse analysis are post-structuralist inductive analyses that take a critical stance in order to detect hegemonic structures of domination and subordination. 74 By using SDA, this study instead follows a methodological tradition in which key theoretical concepts guide the analysis, and the discursive method is used mainly as a tool to chisel out what is being said, how it is said, the underlying meaning of the statements, and how different statements are bound together in a web of intertextuality. 75 Based on the theoretical discussion above, we have identified some key elements that make up the coding frame for each binary value of our dependent variable. Rhetoric of fear is operationalized as when WDs repeatedly employ statements that risk polarizing relations between war-affected groups. Expressions coded in this category are, for instance, threats to employ violence, depictions of other groups as dangerous, or claims that the in-group is in peril. This includes statements such as "this will unleash violence," "the peace may stop," or "the war will come back," and they often concern what will happen to the in-group if they do not vote for a given WD.
Statements coded as rhetoric of peace include utterances that can ease tensions between war-affected groups: for instance, calls for reconciliation, declarations of remorse for wartime abuses, or appeals to peace. A peaceful 71 IRIN is now known as the New Humanitarian. 72 For the studied period, the National Archives of Liberia do not possess every issue of the newspapers they collect. This is, unfortunately, also true for the newspapers' own archives, which are inadequately stocked. 73 The difference in the number of articles between Johnson and Dolo is rather misleading. In many of the rhetoric can include statements such as "forgive me," "we are not enemies," or "turn toward peace and reconciliation." It is important to stress that WDs' electoral statements are not always so clear-cut. Studies have shown that in post-civil war contexts, leaders often make veiled threats that are indirect and culturally embedded. 76 We, therefore, take great care to analyze statements according to the historical and cultural context in which they are made. 77 When it comes to operationalizing WDs' patrimonial endowments, we focus on its two core components-economic resources and social networks. More specifically, WDs are only seen as possessing substantial patrimonial endowments when they both control significant resources and head multiple social networks. WDs have significant resources when they, for instance, own businesses or important tracks of land, have access to large amounts of cash, or head (il)legal enterprises exploiting valuable natural resources. Meanwhile, we define WDs as heading multiple social networks when they share ties with, and claim to represent, several social groups (such as communities based on ethnic, religious, gender, ex-military, political, business, cultural, age, ethnic, or sport identities). It is vital to stress that there may be variations concerning how extensive networks are. This concerns what portion of a given constituency (for example, ex-combatants) a WD has integrated into his network. For this reason, we only include those instances when WDs' networks encompass a significant portion of a given social group.
To establish the strength of WDs' patrimonial endowments, and gain access to rally testimonies, we conducted in-depth interviews in Liberia with key informants during November 2016 and May and August 2017. To minimize the risk of selection bias, and gain multiple perspectives on our selected WDs, we engaged with informants from various social circles-politicians, journalists, nongovernmental organization (NGO) representatives, academicians, and civilians close to the WDs, as well as rally witnesses and ex-members of the WDs' armed groups. These interviewees were either people who were, or had been, close to the two WDs, had taken part in/ witnessed rallies, or were political insiders with key knowledge of events that took place during 2003-5. Some interviewees were used both to gain information about the WDs' patrimonial endowments and utterances at local rallies. This especially concerned individuals such as journalists, campaign managers, and workers, as well as clients of the WDs (both ex-combatant and civilian), who had attended political assemblies. Care was taken 76 Sj€ ostedt, S€ oderberg Kovacs, and Themn er, "Demagogues of Hate or Shepherds of Peace?" 77 Efforts to contextualize statements made by the WDs, and detect possible veiled threats, were facillitated by the fact that one of the authors has conducted reseach on Liberia since 2010. Our collaborating partners at the University of Liberia and Network for Empowerment and Progressive Initiatives also assisted us in these interpretations.
to also identify individuals-besides journalists-who had no connection to Dolo or Johnson but who had witnessed their rallies (that is, ordinary citizens). All informants were identified with the help of four local assistantsone from Network for Empowerment and Progressive Initiatives and three from University of Liberia. The assistants used their own networks to identify initial informants, whereafter they employed snowballing to gain access to additional interviewees. By cooperating with diverse facilitators-who worked independently-it was possible to gain access to informants who were independent of each other and to make extensive use of source triangulation. In total, forty-eight interviews were conducted. We only refer to specific interviews when we employ direct quotes.

War and Peace in Liberia and Nimba County
The origins of the Liberian Civil War can be traced back to the inability of President Samuel Doe (1980-90), largely supported by the Krahn and Mandingo ethnic groups, to uphold his patronage networks and ensure support from key elites. Consequently, Doe increasingly used repressive means to fend off dissent. His administration predominantly directed these abusive policies toward the Gio and Mano communities in Nimba. It was, therefore, no coincidence that when NPFL-led by Charles Taylor-first invaded Liberia in December 1989, they did so via Nimba. This allowed them to tap into a large pool of disgruntled recruits. The dynamics of the war shifted in September 1990, when Prince Johnson-an NPFL general who had split from Taylor to form INPFLcaptured and executed Doe. The political vacuum that followed saw a proliferation of armed groups. Despite numerous efforts to broker peace, it was only in August 1996 that the parties agreed to a peace accord, which among other things stipulated the organization of national elections. The 1997 elections resulted in a victory for Taylor and his newly formed National Patriotic Party (NPP).
In 1998, Taylor launched a crackdown against his former opponents. After regrouping in Guinea, parts of the opposition-who eventually became known as Liberians United for Reconciliation and Democracy (LURD)-began launching armed attacks into Liberia. In early 2003, a new armed faction joined the fray-Movement for Democracy in Liberiaadvancing from bases in Côte d'Ivoire. In April, the regime suffered another blow when LURD invaded Nimba. Even though the attackers were eventually repelled, large parts of Ganta-the county's main town-were destroyed. After the rebels laid siege to Monrovia in mid-2003, Taylor sued for peace. It was, however, not until the latter agreed to go into exile in Nigeria in August that an agreement was reached.
The Comprehensive Peace Accord called for the creation of an interim power-sharing government, the stationing of United Nations (UN) peacekeepers (UN Mission in Liberia; UNMIL), the disarmament, demobilization and reintegration of ex-combatants and new elections. The 2005 presidential elections were won by Ellen Johnson Sirleaf of the Unity Party (UP), who defeated the Congress for Democratic Change standard-bearer George Weah in the second round. Even if the top two presidential contenders were not WDs, some dozen ex-military leaders did run in the presidential race or for seats in the Senate and House of Representatives-such as Sekou Conneh, Adolphus Dolo, Saah Gbollie, Alhaji Kromah, Prince Johnson, and Isaac Nyenabo.
The peace process constituted both a challenge and opportunity for Nimba. Widespread fears existed over what would happen to Gios and Manos after Taylor's fall from power. The two latter communities were particularly wary of revenge attacks by Krahns and Mandingos. However, an entrenched peace also offered the possibility of revitalizing Nimba's historically vibrant economy based on iron-ore mining and regional trade with Côte d'Ivoire and Guinea. These dual sentiments allowed political entrepreneurs-vying for one of the two Senate seats allotted to Nimba-to either employ a rhetoric that emphasized peace and economic development or a backward-looking political platform focusing on wartime abuses and experiences. Prince Johnson and Adophus Dolo were two of the contenders seeking to be elected as senators for Nimba. 78 being run over by Taylor loyalists. By the end of the year, Johnson accepted an offer by Nigeria to go into exile. 80 During his stay in Nigeria, Johnson became a "born-again" Christian and declared he had shed himself of his military past. In fact, not only did Johnson attend a theological seminar, he also began preaching. 81 A central pillar of Johnson's metamorphosis was to present himself as an agent of reconciliation. In this spirit, Johnson came to terms with Doe's family and called on Liberians to reconcile themselves. 82 Even if Johnson had declared his intention to run for Senate already upon his return to Liberia in March 2004, it was only that October that he more formally committed himself to the electoral process. At that time, supporters organized a petitioning ceremony, asking Johnson to be their senatorial candidate-a request Johnson accepted. 83 In the October 2005 elections, Johnson ran as an independent candidate, where he won one of the Senate seats allotted to Nimba (with 33.8 percent of the votes, Johnson came in the first place). Meanwhile, during the second round of the presidential race, Johnson supported George Weah's bid to become head of state. During his election campaign, a systematic effort to employ a rhetoric of fear characterized Johnson's speech acts. We argue this was a calculated response to his limited access to patrimonial endowments.

Dwindling Patrimonial Endowments
One reason why Johnson lacked patrimonial endowments at the end of the war was that he had struggled to sustain himself and his relations from Nigeria. During his exile, Johnson did not have any business activities to support him. Instead, Nigerian authorities largely supported Johnson, providing him and his family with a house and car. He received further economic assistance from a number of wealthy supporters in the Liberian diaspora and donations from Nigerian congregations that invited him to preach. Johnson's economic difficulties meant he was largely unable to send money to assist his clients in Liberia. The terms set by his hosts constituted another obstacle. The Nigerian government conditioned their hospitality on Johnson refraining from intervening in Liberian politics and are said to have monitored Johnson's communications. It was, therefore, not only difficult, but also dangerous, for Johnson to interact with followers in Liberia. Making things worse, most INPFL fighters had been compelled to join Taylor's forces after Johnson left the country. As a bitter enemy of Johnson, Taylor punished anyone caught communicating with him. For these reasons, Johnson was, as one anonymous politician expressed, "practically cut off from Liberia" during his time in the diaspora. 84 Johnson's situation was, therefore, precarious when he returned to Liberia. That he lacked influence in Nimba was a particular problem. After Taylor departed for Nigeria in August 2003, he had left generals-such as Dolo and Roland Duo-in charge of upholding security in the county. The influence these generals had over the fighters continued also after the completion of the disarmament process in November 2004. The resilience of these ties meant Johnson could not easily tap into the ex-INPFL/NPFL combatant community for support. Even if Johnson possessed a mythical aura-due to his role in "liberating" Gios and Manos from Doe's rule-few Nimbadians had any personal relationship with him. Besides a short spell in 1990, Johnson had not spent any time in the county since the mid-1980s. Subsequently, Johnson lacked inroads to other social networks-be they youth, religious, student, or business-which could be employed for political mobilization. Despite his religious credentials as a pastor, most Christian leaders and communities did not take Johnson's conversion seriously. Making things worse, Liberian elites showed little interest in Johnson after his return. During the war, a number of factional delegations had visited Johnson in Nigeria to try to convince him to join the fray. However, Johnson was not invited to take part in the 2003 peace talks, and once the fighting ceased, most elites saw Johnson as a spent force.
To overcome his isolated position, Johnson reached out to a number of ex-INPFL commanders-such as Samuel Varney-in an effort to rebuild his relations with Nimbadians. These ex-militaries functioned as brokers, who sought to "reintroduce" Johnson to central actors in the county. This initiative had limited success for two reasons. First, ex-Taylor loyalists still commanded a lot of sway in Nimba; not only did ex-NPFL generals act as informal security providers, most mayors were Taylor appointees. These individuals tended to be suspicious of Johnson and his intentions. It was, therefore, difficult for Johnson to openly build up a following. Second, since Johnson was not a wealthy man when he returned in 2004, he struggled to reestablish old networks, and develop new ones, by distributing resources.

Scaring Nimbadians Straight
Even if Johnson employed a rhetoric of fear during the run-up to the 2005 elections, he initially used alternative strategies. Between Johnson's petitioning ceremony in October 2004 and July 2005, Johnson did not make any aggressive statements in the media. Instead, his utterances were characterized by messages of regret and reconciliation. For instance, in July 2005 Johnson declared, "We just don't need leadership but people who can promote the process of reconciliation amongst Liberians." 85 Why did Johnson first employ a rhetoric of peace? It could be argued this was due to an initial desire to become integrated into the mainstream political system. When Johnson returned to Liberia, he first declared his intention to run on the ticket of Liberian Action Party (LAP). Johnson later switched allegiance to UP, and in July 2005 he contested in the latter's Nimba primaries. By attaching himself to key political parties Johnson could gain access to economic assistance and preexisting networks of clients. For instance, LAP had strong historical roots in Nimba, and its chairman, Gyude Bryant, was the interim president (2003-6). Meanwhile, UP-headed by Johnson Sirleaf-had attracted much support from the business community. Both parties had a strong civilian and somewhat pan-ethnic appeal, wherefore aggressive rhetoric may not have been beneficial.
Johnson failed to win the July 2005 UP primaries. The defeat seriously distraught the ex-military leader, who alleged Johnson Sirleaf had actively manipulated the process. 86 Without the backing of a party, and with limited access to resources and social networks, Johnson was unable to base his campaign on "distribution politics." 87 In fact, during his campaignwhich officially began in mid-August of the same year-Johnson did not, like other candidates, systematically provide cash, rice, or t-shirts during his rallies. This astounded commentators, including an anonymous academician who observed two political gatherings. According to him, Johnson's "campaign was unbelievable in that the money that he campaigned with was very different from that of other candidates." 88 The only patronage Johnson is said to have distributed was ten cows, sacrificing one each in ten communities.
Without patrimonial endowments-and with the failed experience of extolling the virtues of peace fresh in mind-Johnson seems to have drawn the conclusion that his political survival required him to embrace a rhetoric of fear. This was particularly pronounced at local rallies-which were frequented by women, church groups, elders, chiefs, ex-combatants, students, and marginalized youths-organized in Nimba during August and the Gio and Mano electorate about their continued vulnerability, especially in relation to their Krahn neighbors. Johnson's fearmongering was facilitated by ex-military leaders from "opposing" ethnic groups-in particular, Krahns-having declared their intention to run for the Senate and House of Representatives (for example, Isaac Nyenabo, ex-LURD). According to Johnson, electing him was the only way to counter the threat posed by the likes of Nyenabo. On several occasions, Johnson went so far as to say that the war would come back if he was not voted into power. To give further credence to his image as a "protector," Johnson repeatedly reminded his audiences of his role in freeing them from the oppressive rule of Doe. Johnson's strategy to vilify the Krahns is lucidly captured by an anonymous ex-journalist who witnessed several Johnson rallies. According to him, "Johnson is always, you know, in the business of saying things that will make Nimbadians [Gio and Manos] feel that they do not have any good relationship with the Grand Gedens [Krahns]. So that message is not anything hidden. It's a message … 'I killed Doe! I killed Doe!'" 89 By engaging in fearmongering, Johnson made use of the only advantage he enjoyed-his war aura-or, as developed by one Christian leader: Johnson was not a businessman. He did not give rice or distribute things. Just after the war what the Nimban people wanted and demanded was protection and security and Prince Johnson used this to his advantage in his campaign and presented himself as the one who could give them that security. 90 In this sense, Johnson was both the pyromaniac and firefighter; 91 he first fanned the anxieties of Gio and Manos, and then offered a solution to their predicament-by electing him, they would be safe. For the economically restrained Johnson, employing a rhetoric of fear made sense, since it, according to one anonymous politician, "was the easiest way to get people behind him." 92 A former youth activist expressed similar sentiments, who argued it was vital for Johnson to employ aggressive statements since he "had just returned [to Liberia]. He needed to identify some strategy to mobilize people." 93 Rather cleverly, Johnson even made his inability to distribute patronage into a virtue. Johnson habitually told crowds that "I won't spend a dime on you. I have already liberated you, you owe me!" 94 Hence, by reminding people of his past sacrifices, Johnson could "reconnect" with those in Nimba. This strategy was particularly efficient in mobilizing support among elders, who were old enough to remember the abusive rule of Doe. Elders' fears were instrumental when reaching out to other groups, such as youth, women, Christian congregations, secret societies, and ex-combatants.
Not only in studying statements made at local rallies is it possible to detect a rhetorical shift from before and after the UP primaries in July 2005. It was only after Johnson launched his campaign as an independent candidate that he began employing fearmongering in the media. These statements did, however, differ from those used at rallies, as they tended to be more veiled. For instance, in late September Johnson declared he had fought the war because "a group of people had declared war on Nimba." 95 By assigning blame to a "group of people," rather than AFL as an organization, Johnson seems to have alluded to it being the Krahns, as an ethnic group, who had started the armed conflict. The ex-military expressed similar anti-Krahn sentiments in early October. In an interview, Johnson stated that "we [Gio] have apologized because we want peace, but the Krahn people have never apologized for what they did to the Gio." 96 The underlying gist of Johnson's message was that by refusing to apologize for their transgressions, the Krahns still wanted war. Such covert messages were presumably directed toward older Nimbadians, who had personally experienced the abusive rule of President Doe and could more easily pick up cues about the belligerence of the Krahns.

Adolphus Dolo: the Peacelord
When the first civil war broke out, Dolo was a student at the University of Liberia. Although Dolo initially joined NPFL, he followed Johnson when the latter broke off to form INPFL. Like many other INPFL fighters, Dolo was absorbed back into NPFL after Johnson's departure for Nigeria. While with Taylor, Dolo became a key military leader and, by the end of the first civil war, had become a commanding general.
After Taylor was elected president in 1997, Dolo was integrated into the reconstructed AFL. With the outbreak of the LURD rebellion, Dolo was ordered to form his own militia, called Executive Strike Force. Dolo attained national fame in April 2003 when the rebels launched a surprise offensive in Nimba. To salvage the situation, Taylor made Dolo chief of staff of the county. Dolo organized a counteroffensive that drove LURD out of Ganta. Since Dolo's actions saved numerous lives and properties, he received war-hero status.
Dolo was one of the generals who continued to provide security in Nimba between the end of the war and the finalization of the disarmament process in November 2004. In February of the next year, Dolo made his political intentions public by accepting a petition from community members to be their senatorial candidate. Running on LAP's ticket, Dolo won one of the Senate seats (second place with 17.4 percent of the votes) allotted to Nimba and rallied behind Johnson Sirleaf during the second round of the presidential elections. Despite his military background, Dolo largely used a rhetoric of peace during his election campaign. We argue that this can be explained by his possession of substantial patrimonial endowments.

Bourgeoning Patrimonial Endowments
Dolo's access to patrimonial endowments was partly a function of the economic assets at his disposal and partly a result of the ties that bound him to various social groups. Dolo had established himself as a businessman prior to the war; during his time at university, Dolo owned a popular club in Monrovia. His sense of business came in handy during the conflict, which created opportunities for crafty entrepreneurs. Not only did the general found a prosperous security company (North Star Security), he also acquired a hotel. Above all, Dolo made a fortune selling plundered scrap metal, and captured mining and forestry equipment. Dolo's good fortunes did not cease after Taylor's exile in 2003. Besides the abovementioned assets, Dolo also controlled numerous additional enterprises in Nimba (a rubber plantation, several provisional shops and petrol stations, as well as a club with a hotel) and Monrovia (a second hotel) after the end of the second war. Of equal importance were Dolo's close ties to key elites during the interim period. Officially, Dolo sought to distance himself from Taylor by joining LAP. This gave him access to economic and political support from LAP and the latter's presidential candidate, Varney Sherman. However, Dolo covertly retained his ties to Taylor, and the latter helped to finance his campaign.
Dolo was not only economically well off after the war; he was, according to an anonymous politician, also "assimilated into normal society. He was more than a warlord and war hero." 97 Another anonymous commentator (a former politician), expressed similar sentiments when he declared that Dolo possessed "a larger social network and could relate to and think outside his little military clique. Dolo's relationship went beyond his military faction and tribe." 98 Dolo had, thanks to this "assimilation," access to an array of networks from which to draw potential clients. First, Dolo had substantial support within the business community, both because he himself was a successful entrepreneur and also due to his role in protecting the businesses' properties during the war. Second, the ex-general also had a strong following among students. When interacting with them, Dolo could point to his own background as a university student. Third, Dolo hailed from a well-known Nimbadian family. Dolo's father had been the principal of several schools, and a number of his relatives had been mayors and held other government-appointed positions in the county. The Dolo name thus commanded respect. Fourth, in 2005 Dolo founded a football team, which recruited local talent. This helped him gain support from local youths and sport enthusiasts. Finally, Dolo still commanded the loyalty of many ex-NPFL combatants. Until the completion of the disarmament process in late 2004, Dolo controlled the largest grouping of NPFL combatants in the county. This influence continued even after the official dissolution of his militia, as ex-combatants made it a habit to pay homage to their former general.
Thanks to the economic assets that Dolo controlled, he was able to invest in transforming members of these networks into loyal clients. According to one former NGO representative: After the civil war many people, they were homeless.
[Dolo] bought zinc [roofs] and other things for them, provided household utensils for them … Some people who were not financially equipped, he helped them, provided scholarships for them to go to school … In addition to that he established a vocational institution. 99 Dolo was particularly keen to assist youth gain access to education. Besides offering scholarships and founding a vocational training institute in 2004, Dolo financed a high school in Nimba run by his family. Ex-combatants and business owners were two additional communities in which Dolo invested. He was well known for assisting ex-fighters in various ways, ranging from offering them food and clothes to paying their school fees and employing them in his North Star Security firm. Meanwhile, Dolo was able to strengthen his ties to local entrepreneurs by investing in their businesses and upholding law and order when, for instance, ex-combatant riots occurred.

Affording to be Conciliatory
Dolo's strong standing among various social groups allowed him to engage in the art of "being all things to all people" during the election campaign. 100 Dolo had a canny ability to "attach himself to other backgrounds" than his military one. 101 This allowed him to make use of multiple leadership roles during the election campaign. According to his campaign manager, Sabato Gonlepa, "Dolo was in student politics before [the war] so Dolo knew how to get around the student community … Dolo would visit various university and college campuses." 102 Furthermore, the ex-general covertly exploited the fears of ex-NPFL fighters regarding what would happen to them after Taylor had left the country. Dolo used the ex-combatants-especially those employed at North Star Security-during political rallies to get "numbers" on the streets. He was, however, keen that the exfighters did not visualize their military background, so as not to intimidate local communities. In addition, Dolo often stressed his role as a successful businessman, arguing these virtues would make him a good senator. Such rhetoric was particularly useful in drawing marketwomen into his campaign. As such, Dolo's electoral strategy was not so much about reaching out to new constituencies as it was employing his networks to cement control over those he already held sway.
Irrespective of which group Dolo targeted, distribution constituted a key campaign strategy, or as expressed by one anonymous scholar, "Dolo was distributing during the election campaign. He gave money to people, made promises of scholarships, paid student fees." 103 When Dolo entered a village, he would commonly give a sum of money and food to community leaders, who would then allocate the gifts to potential voters. At the actual rallies-which were attended by members of various social categories (businesspeople, elders, chiefs, women, youths, students, and ex-fighters)t-shirts and caps, as well rice, were given to the participants. As a consequence, community members often gave Dolo a warm welcome, or as expressed by one journalist, "The elderly, the blind, the crippled and others swarmed him like bees." 104 To capture his generosity, Dolo used the slogan, "Let me butter your bread" during the election campaign. This phrase was particularly clever considering his nom de guerre had been General Peanut Butter.
By targeting numerous social groups and engaging in distribution politics, Dolo had no need to employ a rhetoric of fear. Such belligerency risked intimidating key supporters, such as the business community. In addition, for many Nimbadians Dolo's wartime deeds were still vivid in their minds, and potential voters knew Dolo was a powerful general with a substantial ex-combatant following. In this sense, Dolo did not need to flaunt his power. This sentiment is lucidly captured in the following statement by an anonymous head of an NGO: "From Nimba, 95% of people supported him [Dolo] before the elections. People were so grateful to him for protecting the county against LURD. So he did not have to talk war." 105 A former journalist followed the same line of thinking, arguing that Dolo "had a huge following of ex-fighters and was a businessman and had money, so probably he did not need to use threat talk." 106 It was, in other words, more rational for Dolo to evoke more positive, nonviolent images when engaging with the public. From his petition ceremony in February 2005 until the finalization of the election process, Dolo more or less consistently used a rhetoric of peace. A reoccurring theme in his media appearances was to call on Gios, Manos, and Mandingos to reconcile and unite. 107 In February 2005 the ex-military vowed to reconcile and unite Nimbadians and declared that "we should now rally together to bring the desired peace to make our parents and people proud." 108 Dolo also put significant emphasis on using media outlets to explain why he had become a fighter. A key narrative was: "I did not start this war; the war chose me." 109 Even if Dolo projected an image of having been obliged to take part in the fighting, he did not shun away from taking responsibility for wartime transgressions. In a radio interview in February 2005, he asked listeners to "please forgive me for my past misdeeds." 110 Dolo only made one utterance that can be interpreted as aggressive. In April 2005, he reacted to a comment made by a media advisor at the UNMIL Electoral Division. The latter had declared individuals who were associated with the war and under a UN travel ban, such as Dolo, would not be permitted to take part in the elections. Dolo responded by warning that such utterances had the "propensity to spark confusion." 111 It is difficult to determine what Dolo meant by "confusion," but, in a Liberian context, confusion sometimes means violence. Irrespective of what he meant, the intended audience was arguably international peacemakers rather than potential voters. Benevolent speech acts can be detected when analyzing Dolo's utterances at local rallies during July-October 2005. The ex-general was particularly keen to present himself as a forward-looking businessman who could bring development to the region. Another reoccurring theme was the need for forgiveness. According to an anonymous teacher, who witnessed a number of political gatherings, Dolo often stated, "If I have offended anyone in the war, please forgive me." 112 Dolo's rally declarations did, however, differ from his media statements in one regard: when speaking directly to local communities Dolo was more willing to play up his war-hero status. He would, for instance, declare that "during the war, I [was] the one that help[ed] you," so vote for me. 113 Meanwhile, while taking part in a church service, Dolo is reported to have stood up, removed his shirt, and shown the bullet holes on his back, visualizing his sacrifice for the county. Dolo was, however, careful not to imply that it was for security reasons they needed his leadership, but to bring development.

Comparative Analysis
When comparing the electoral speech acts of Dolo and Johnson, there is a clear variation in rhetoric. A cornerstone of Johnson's campaign was to employ a rhetoric of fear, constantly reminding Gios and Manos about their vulnerability in relation to their Krahn neighbors-stating that "if you do not vote for me, the war will come back." He also habitually reminded his audience that he had liberated them from the oppressive rule of Doe and would-if elected-also protect them in the future. Dolo's approach was markedly different. By systematically employing a rhetoric of peacewhich included declarations of remorse and calls for reconciliation-Dolo put greater emphasis on looking toward the future. Here a central theme was to point to the prospect of economic development. Interestingly, even though Dolo and Johnson employed fundamentally different rhetorical styles, both succeeded in winning a Senate seat. This indicates WDs can carve out postwar political spaces for themselves using various electoral strategies.
However, two important nuances appear regarding rhetoric. As noted above, Johnson initially had a much more benevolent approach, commonly stressing the need for reconciliation. It was only later that he discarded this rhetoric of peace for a more aggressive framing strategy. Furthermore, both Dolo and Johnson appear to have been acutely aware of the type of audiences to which they were speaking. They commonly adjusted their message depending on whether they were interviewed in media or talking at local rallies. For instance, even if Johnson made belligerent statements in media, these tended to be more veiled than utterances at local gatherings. Dolo was less willing to depict himself as a war hero when being interviewed by journalists. The observation that WDs vary rhetoric over time and space highlights the necessity of viewing rhetoric of fear and peace as broad, malleable narrative clusters rather than narrow and mutually exclusive categories.
How can we understand the abovementioned shifts in rhetoric? One explanation could be what Hansen calls the "silent security dilemma." 114 In contexts outside Western democracies, actors often do not publicly express security concerns due to fears of repercussions from political opponents or the international community. In Liberia, one such group of actors was UNMIL peacekeepers. Fearful of being branded as spoilers, and subsequently excluded from the political process, Dolo and Johnson may have preferred to moderate their rhetoric when speaking "on the record." The benefit of employing more veiled messages in societies permeated by fear is the intended audience usually comprehends the underlying meaning since they belong to the same cultural-political setting. 115 Another explanation could be that both WDs were already eyeing the second round of the presidential elections (held approximately one month after the senatorial race). By endorsing a winning candidate, Dolo and Johnson could acquire economic and political concessions. However, if they were portrayed as aggressive or too boastful of their wartime deeds in the national media, they may hurt their respective presidential candidates' chances of mobilizing support from non-Gios and Manos. In this sense, Dolo and Johnson had incentives to moderate their speech acts when talking to a more national audience. Put differently, the objective may have been to maintain their support base rather than attract additional support.
Why was it Johnson, and not Dolo, who employed a rhetoric of fear? As discussed above, studies in adjoining fields of research have, to some extent, examined why political contenders employ different rhetorical strategies. It might, therefore, be appropriate to consider whether the difference in speech acts can be explained by other factors than patrimonial endowments. Could it, for instance, be that the question at hand can be explained by Johnson being more exposed to threats than Dolo? 116 The empirical evidence suggests this was not the case. Of the two WDs, only Dolo received any form of public intimidations. In early October 2005, a group calling themselves the Yanna Boys circulated a leaflet threatening to kill key supporters of Johnson Sirleaf-such as Dolo-unless a supposed threat against Weah was eliminated. 117 This declaration did, however, not generate any form of verbal belligerence from Dolo's side.
Even if there were no similar public threats made against Johnson, he did, in early November 2005, claim there was an UP plot to assassinate him. 118 There is, however, no evidence supporting Johnson's claim, and several observers hold that the accusations were fabricated to rally supporters. Such a strategy could have worked to Johnson's advantage-if he was the only person who could ensure the safety of Gios and Manos, then an attack against Johnson was an attack against them. Even if the allegation was true, it cannot explain Johnson's initial decision to employ a rhetoric of fear. He made his first aggressive utterances three months earlier, just after he had lost the UP Nimba primaries.
Our evidence suggests the diverging rhetorical strategies can instead be explained by the patrimonial endowments-in the form of resources and social networks-each WD possessed. During his Nigerian exile, Johnson was cut off from his followers and increasingly seen as a spent force amongst Liberian elites. Johnson was consequently a marginal figure in Liberia's networked society when he returned in March 2004. Making things worse, he largely lacked the economic means to attract old and new clients. To overcome this disadvantage, Johnson engaged in fearmongering. By reminding Gios and Manos of the persistent threat emanating from their Krahn neighbors, Johnson once again made his wartime credentials relevant. Dolo's situation was markedly different. After the end of the hostilities, Dolo was one of Nimba's most influential Big Men. Not only did he have inroads into multiple social networks, he was also a wealthy man. As such, Dolo did not need to use a rhetoric of fear to mobilize electoral support. Such statements may even have scared off some clients, such as members of the business community. A better strategy was therefore to make peaceful utterances, coupled with systematic distribution of economic resources through the various networks he controlled. By "buttering" Nimbadians' bread and "being all things to all people," Dolo was able to expand his clientele base and maximize his chances at the polls.
When tracing Johnson's electoral campaign, it also becomes apparent that it is not only the actual possession of networks and financial assets that influence WDs' rhetorical strategies. Just as important is the prospect of gaining access to such endowments in the near future. As long as Johnson was seeking to attach himself to LAP and UP-and acquire the perks associated with doing so-the ex-military leader shunned aggressive speech acts. It was only when this possibility closed, and Johnson was obliged to run as an independent, that he commenced his rhetoric of fear.
Johnson's sudden shift from benevolence to belligerency also highlights that WDs' choice of rhetoric is not a priori defined; as access to patrimonial endowments alter, so does ex-militaries' willingness to incite fear.

Extending the Analysis
This article has proposed an explanation of WDs' choice of political rhetoric in postwar electoral settings. A relevant question is, however, what broader security effects Johnson and Dolo's utterances had on the Liberian peace processes. Even if this question can only be properly addressed by conducting a more in-depth detailed analysis of post-2005 political dynamics, it is at least possible here to point toward some illustrative patterns. Despite fears that the 2005 elections would trigger new violence, the presence of UNMIL peacekeepers ensured the elections were carried out in a calm fashion. However, evidence suggests Johnson and Dolo's electoral speech acts had a long-term effect on broader societal discourses and ethnic relations. Johnson's efforts to vilify the Krahn community further entrenched the deep-seated polarization between Gio/Manos and Krahns. 119 These tensions constituted a serious problem during the work of the country's Truth and Reconciliation Commission (TRC; 2006-10). To intimidate the TRC, Johnson painted a daunting scenario of what would happen if he was obliged to testify, and by extension, charged with war crimes. In these depictions, hostile relations between Gio/Manos and Krahns played a key part. In February 2008, Johnson declared that "my people, the Nimba people, will resist any attempt by the TRC to forcibly have me appeared before it [TRC] to explain circumstances in connection with Doe's death," and he continued by saying his testimony could trigger a new war between Gio/Manos and Krahns. 120 Such threats carried some weight; by electing Johnson as senator, a large segment of the Gio and Mano communities had accepted the proposition that they needed Johnson's protection against the Krahns. As such, any move against Johnson was a move against them. The threat mongering carried out by Johnson, and other ex-military leaders, not only generated a debate about the merits of following the TRC's 2009 recommendation to set up a war crimes court 121 but arguably constituted the gravest threat to the peace process since the completion of the disarmament process in 2004.
Unlike Johnson, Dolo did not have the same history of battling Krahnbased armed groups. While based in Nimba, Dolo had predominantly led his fighters against the Mandingo-dominated LURD. His statements during the 2005 elections reflected this: Mandingos figured much more frequently than Krahns. As previously discussed, a cornerstone of Dolo's electoral rhetoric was to call on Gio/Manos and Mandingos to reconcile. Even if some unresolved issues existed between Gio/Manos and Mandingos after the 2005 elections-such as unsettled land disputes 122 -relations between the two communities were never politicized to the same extent as Gio/ Mano-Krahn relations. Even if it is difficult to ascertain that this was merely due to Dolo's benevolent rhetoric, at a minimum his utterances did not further entrench ethnic tensions. Johnson's political maneuverings during and after the 2005 election underline that the ultimate goal of fearmongering is not necessarily a return to war. In fact, Johnson reaped the benefits of polarizing ethnic relations-being elected into the Senate and undermining the work of the TRC-without firing a single shot. Studies have shown that postwar leaders who have demobilized their factions are often reluctant to use arms. 123 This is, according to Alex de Waal, "because most members of the elite are content with their personal stakes in the status quo, they rarely use violence among themselves." 124 Oftentimes the rationale for engaging in fearmongering is rather to produce a sense of crisis, which can increase WDs' bargaining range; after escalating tensions, they can offer their services to contain the situation.
However, there are reasons to believe a rhetoric of fear can, at times, trigger an escalation process eventually leading to violence. For instance, in Mozambique, the former rebel leader of Mozambican National Resistance, Afonso Dhlakama, increasingly employed aggressive speech acts-such as threatening to use arms-during the 2000s. One reason for this shift in behavior was Dhlakama struggled to distribute resources to his clients and subsequently risked being politically isolated. When his rhetoric of fear failed to rally enough voters to win the presidency-and pressure the government into making concessions-Dhlakama had few options but to finally deliver on his promises. Subsequently, the ex-rebel leader launched a new uprising in 2013. 125 Another problem is that once identities have been polarized, they often take on a life of their own. Stuart J. Kaufman, for instance, has shown that mobilized ethnic communities can pressure reluctant elites to engage in warfare. 126 There are indications this article's main findings are applicable to WDs outside of Liberia. In the abovementioned case of Dhlakama, it was only after the ex-rebel leader struggled to uphold his patronage networks that he began to employ a rhetoric of fear. Such dynamics are not only reserved for oppositional WDs. For instance, Ugandan president Yoweri Museveniformer rebel leader of the National Resistance Movement-has varied considerably when it comes to the usage of electoral aggression. Whereas there was a relatively limited amount government intimidation and violence during the 2011 presidential elections, things deteriorated during the subsequent 2016 elections. 127 During the latter, Museveni ran adverts containing images of skulls of civilians killed in the previous civil war. The gist of the ads was: vote for me, or risk another war. 128 One reason for the increased belligerency of Museveni was that promises of kickbacks from expected oil revenues were less credible in 2016 than 2011 due to delays in the exploitation of Ugandan oil fields. 129 With less patronage to offer, Musevni's networks of clients began to falter. In response, Museveni chose to use fearmongering and violence to get the votes he needed. Having said this, it is vital to stress that there is a need for further comparative studies of WDs operating in various postwar environments before we can truly assess how far the patrimonial-endowments argument travels.

Shape-Shifters in Multiple Contexts
In countries transitioning from war to peace, WDs often have a profound impact on political dynamics. This is particularly true during elections, when WDs' speech acts can help to either ease or fan tensions between war-affected groups. There is, however, a conspicuous lack of studies investigating why WDs differ so in their electoral rhetoric. There is a particular need to gain a more profound understanding of the role patrimonial endowments-generally seen as a key political resource in postwar societies-have in shaping WDs' choice of electoral rhetoric. This article has sought to begin to fill this lacuna. Experiences from Liberia show WDs make calculated decisions about whether, when, and in front of which audience to engage in fearmongering. Ex-military-turned-politicians appear to be acutely aware of the audience they are speaking to and often adjust their message depending on whether they are engaging an international or national crowd, or local communities. More specifically, we find WDs are particularly prone to employ a rhetoric of fear when they lack patrimonial endowments. Without the resources and social networks needed to engage in distribution politics, WDs need to develop alternative electoral survival strategies. One such strategy is to decrease the relative value of economic patronage by recreating wartime anxieties and providing credible promises of future protection. How well does our theoretical perspective travel outside of Liberia? The argument is probably most applicable to post-civil war societies, like Liberia, that are characterized by electoral patronage politics and weak state institutions. This applies to countries in large parts of Sub-Saharan Africa, as well as some Eurasian states, such as Afghanistan, Indonesia (Aceh), Kosovo, Tajikistan, and Timor-Leste.
The findings presented here can provide peacemakers with valuable information about when there is a need to engage with postwar societies in general and WDs in particular. Due to the central role that wealth in people plays in elite formation, diminishing patrimonial endowments can be seen as an early warning for when WDs are likely to escalate their belligerency and subsequently provide cues for when there is an increased need for peacemaking.