I
Concepts
[Tjt is to no purpose, it is even against one's best interest, to turn away from the consideration of the affair because the horror of its elements excites repugnance.
Carl von Clausewitz, On War
To understand the living... I found it was necessary to begin with the dead.
Salman Rushdie, The Jaguar Smile
This chapter reviews existing accounts of violence and civil war and clarifies definitional and conceptual issues related to both civil war and violence. I make the case for the analytical autonomy of violence vis-a-vis conflict and introduce three important distinctions: between violence and violent conflict, between violence as an outcome and as a process, and between violence in peace and violence in war.
I.I. CIVIL WAR
Civil war has attracted considerable scholarly attention from various disciplines -though considerably less than interstate war. Important bodies of literature have explicitly or implicitly (as studies of revolution, rebellion, or ethnic conflict) focused on numerous aspects: onset (Fearon and Laitin 2003; Collier et al. 2003; Sambanis and Elbadawi 2002; Gurr 1980), resolution (B. Walter 1997), social bases (Wickham-Crowley 1992; Skocpol 1979), outcome (Leites and Wolf 1970), political and social consequences (Sambanis 2000), and processes of rebuilding, reconciliation, and postwar justice (Bass 2000; Nino 1996). A recent boom in civil war studies has been fueled by the global shift from interstate to intrastate conflict: of the 118 armed conflicts that have taken place between 1989 and 2004, only 7 have been interstate wars (Harbom and Wallensteen 2005).
Until recently, however, civil war enjoyed little conceptual autonomy (Ranzato 1994); the term is still used by analysts and observers in multiple, often contradictory, ways. While historians have used it to describe discrete historical events,
16
Concepts 17
they have generally refrained from analyzing civil war as a phenomenon that transcends particular instances. In historical sociology and political science, civil war was until very recently subsumed under phenomena implicitly deemed more important, such as revolution, peasant rebellion, or ethnic conflict. In everyday language, "civil war" (unlike "revolution") is a term that conveys a sense of violent division, often used as a metaphor for extreme conflict and widespread brutality.
Civil war often refuses to speak its name. Euphemisms abound: one hears of "troubles," "emergency," "situation," or simply "violence." Indeed, civil war is often the object of serious semantic contestation. The very use of the term is part of the conflict itself, conferring or denying legitimacy (or status equality) to the parties in the conflict. The American Civil War was called "The War of Rebellion" and "The Second American Revolution," depending on the favored side. During the war, the term is usually sought out by insurgents in search of legitimacy, and denied by incumbents who label their opponents "bad guys," bandits, criminals, subversives, or terrorists - and describe the war as banditry, terrorism, delinquent subversion, and other cognate terms.1 In fact, the repudiation of the term is common to all incumbent regimes, leftist and rightist, authoritarian and democratic alike (e.g., Horton 1998:11; Pavone 1994). Following a civil war's end, the term is often claimed by the vanquished in their quest for political redemption and inclusion, and denied by the winners who seek the permanent exclusion of the losers from the political, or even national, realm (Bobbio 1992). The spillover effect of this semantic contest has affected research on the topic, as definitions of civil war have tended, until recently at least, to hinge on the war's outcome (Price 2001:33-4).
Civil war is defined here as armed combat within the boundaries of a recognized sovereign entity between parties subject to a common authority at the outset of the hostilities. This definition is a broader and more minimal version of existing definitions (Sambanis 2004); it is agnostic about causes, goals, and motivations. "Internal war" is more precise, but civil war is by far the most familiar term. "Armed combat" (implying a degree of organization on both sides and violence of a certain magnitude) serves political aims when it challenges existing authority, even when also serving additional goals (Chapter 11).
The key intuition is the violent physical division of the sovereign entity into rival armed camps. This entails a de facto territorial division. At the war's outset, the rivals are subjects to a common sovereign or authority (De Lupis 1987:3; C. Schmitt 1976). After 1648 this refers increasingly to state authority, but prior to the spread of modern state sovereignty civil wars took place within entities perceived as sovereign or "quasi-sovereign," from empires down to city-states and kin-based groups.2 In fact, historians use the concept of civil war as an analytical category for the pre-1648 period (e.g., Porter 1994).
1 The German occupation authorities in the Soviet Union made this point explicitly in 1942: "For psychological reasons," the term "partisan" was to be replaced by "bandit"; accordingly, antipartisan operations were to be called "antibandit warfare" and areas of suspected partisan presence were referred to as areas "contaminated with bandit groups" (in Heer 2000:113).
2 Carl Schmitt (1976:32) speaks of "organized units" and Bobbio (in Ranzato 1994:3000) of "autarkic entities." Even when sovereignty was fragmented, decentralized, and overlapping (e.g., in medieval
i8
The Logic of Violence in Civil War
Civil wars have been fought for all kinds of reasons, from "differences of doctrine and intellectual wrangling" (Hobbes, Leviathan, appendix 2:30), to differences of ascription (mainly ethnicity and religion), and to plain power grabbing (Collier and Hoeffler 1999). The parties to the conflict may be united or divided, internationally recognized or isolated and obscure, supported by external actors or relying on local resources, seeking to capture the state or to divide it. However, the conflicts that are constitutive of civil wars can be best described as those related to the effective breakdown of the monopoly of violence by way of armed internal challenge. The armed contestation of sovereignty entails mutually exclusive claims to authority that produce a situation of divided or dual sovereignty (Tilly 1978:191; Trotsky 1965:224) - a concept that can be traced back to Plato, who thought of domestic war or "faction" as the war that arises when "ruling [a city] becomes a thing fought over" (Republic 521a), and Grotius (II, 18:2), who pointed to situations whereby "a people has been divided into parts so nearly equal that it is doubtful which of the two sides possesses sovereignty." Divided sovereignty came to be seen as something unnatural (Rousseau, Social Contract II, 2:3). In the words of a Vietnamese man: "There cannot be two suns and there cannot be two kings for one country" (in Elliott 2003:749).
Shared membership to a sovereign entity by all belligerents when the war begins is essential (Bouthoul 1970:447). "The American Revolution was a civil war," Shy (1976:183) reminds us, because "in proportion to population, almost as many Americans were engaged in fighting other Americans during the Revolution as did so during the Civil War." Membership is understood here as reflecting basic obligation rather than citizenship and does not require a subjective perception of belonging.
Reflection on civil war is associated with two intertwined intellectual traditions. On the one hand, the concepts of stasis (faction) and public discord and division preoccupied writers in smaller sovereign entities, such as city-states; on the other hand, the concepts of sedition and rebellion tended to emerge in larger sovereign entities, such as empires.
The ancient Greeks posited a self-evident link between stasis and the polis and understood the concept of stasis to refer to "a polis which is internally divided" (Price 2001:31). Thucydides (3.69-85), Plato (Republic 47oc-b), and Aristotle (Politics Vv-xii) drew a clear distinction between stasis and external war.3 Civil war became the dominant form of war in the late Roman Empire (Brent Shaw 2001) and has been a constant occurrence in Europe since then; these civil wars include factional conflicts of the sort that took place in the Italian medieval republics, as recorded in the writings of Marsilius of Padua, Machiavelli, and others, as well as wars pitting the crown against various corporate entities, such as estates,
Europe), there existed entities with recognized jurisdiction, princes that were "supreme and public persons" (in Hale 1971:8). The concepts of "dominium" and "lordship" describe such quasi sovereignty during the Middle Ages (Davies 2003). 3 There was even a more subtle difference as well: diaphora was a term used to describe civil wars in one's awnpolis, whereas civil wars in a neighboring polis were described as stasis (Price 2001:35).
Concepts
19
religious groups, and cities. Grotius (On the Law of War and Peace II, 19:4) made a clear distinction between civil and "foreign" wars, while Hobbes (Leviathan 13:8) argued that sovereign authority emerges (and is justified) precisely to ward off civil war: men are in a state of war as long as there is no "common power to keep them all in awe" - a point also made by Grotius (I, 4:2). Indeed, restrictions on the right of resistance to a legally constituted authority were justified by Grotius and other authors on the basis of their consequence, namely civil war.
By this definition, most revolutions, sustained peasant insurrections, "revolutionary" or ethnic insurgencies, anticolonial uprisings, and resistance wars against foreign occupiers are civil wars (Malefakis 1996:18; C. Friedrich 1972:37). On the other hand, violent protests, riots, crime, and low-level banditry, all of which leave sovereignty pretty much intact, are excluded from this category.4
1.2. VIOLENCE
Though it may be an intuitive concept, violence is a conceptual minefield. As a multifaceted social phenomenon, it can be defined very broadly and stretched way beyond physical violence (Nordstrom and Martin 1992:8). Some distinguish between violence that preserves the social order ("systemically functional" violence) and violence that destroys it ("dysfunctional" violence) (C. Friedrich 1972; Sorel 1921); others take social and economic oppression (or even competition) to be forms of "structural" violence (Braud 1999; Galtung 1975; Ellul 1969:86). Finally, some think that the range of social acts that qualify as violence is so broad as to include any act that results in mental anguish (Bourdieu 1977:191). This books narrows down the definition of violence to its physical dimension.
At a very basic level, violence is the deliberate infliction of harm on people. Here I further narrow my focus to violence against noncombatants or civilians. This is an ambiguous and contentious category in most civil wars, the object of never-ending legal and philosophical dispute (Nabulsi 1999; Walzer 1997). Because I am interested in intracommunity dynamics, for the purpose of this book I regard as civilians all those who are not full-time members of an armed group, thus including all types of part-timers and collaborators.5 Noncombatant
4 Many studies of occupation and anticolonial insurgencies stress their civil war dimension (e.g., D. Anderson 2005; Bouaziz and Mahe 2004; Pavone 1994; Shy 1976). Civil wars are distinguished from coups when a certain fatality threshold is crossed, entailing significant military operations. Large-scale insurgencies with a predominantly rural basis should not be confused with a class of events described as "peasant rebellions," including spontaneous peasant uprisings, jacqueries, food riots, and the like. These undisciplined, unstable, anarchic, and decentralized processes (Tilly 1978) are not sustained long enough to challenge sovereign authority effectively. Unless harnessed by skilled organizers, jacqueries are usually repressed (Marks 1984:240). Peasant rebellions develop into civil wars (and possibly social revolutions) when spurred and led by organizations (DeNardo 1985; B. Moore 1966:479). Borderline phenomena such as the Chinese Cultural Revolution may be understood as civil wars (L. White 1989:308).
5 I explain how I code noncombatants in Appendix B.
20
The Logic of Violence in Civil War
fatalities in civil wars are not always violent; famine and disease can be highly lethal. Violent fatalities can also be unintentional, the so-called collateral damage. In this book, I account for the violent and intentional victimization of civilians.
Intentional and direct physical violence takes several forms, including pillage, robbery, vandalism, arson, forcible displacement, kidnapping, hostage taking, detention, beating, torture, mutilation, rape, and desecration of dead bodies. Although I refer to various forms of violence, my primary focus is on violent death or homicide. As just stated, homicide does not exhaust the range of violence, but is an unambiguous form that can be measured more reliably than other forms (Spierenburg 1996:63; Buoye 1990:255), which is why it is used as the primary indicator of violence in quantitative studies (e.g., Poole 1995; Greer 1935). In addition, there is a general consensus that homicide crosses a line: it "is an irreversible, direct, immediate, and unambiguous method of annihilation" (Straus 2000:7); in this sense, death is "the absolute violence" (Sofsky 1998:53).
1.2.1. Violence, Conflict, War
Violence is typically treated as synonymous with cognate but distinct concepts such as "conflict," "revolution," or "war." Hence most references to, say, "ethnic violence" refer to ethnic conflict or ethnic war rather than the actual violence that takes place within the conflict. However, conflicts, wars, and revolutions are phenomena that cannot be simply reduced to large-scale violence. Conversely, violence, as Hannah Arendt (1970:19) pointed out, is "a phenomenon in its own right" that should not be equated with cognate phenomena. David Horowitz (2001:475) echoes Arendt when he points out that "there is good reason to treat conflict and violence separately." Obviously, war "causes" violence. However, a considerable amount of violence in civil wars lacks conventional military utility and does not take place on the battlefield. If anything, there appears to be an inverse relationship between the magnitude of the conflict, as measured by the size of forces and the sophistication of the weapons used, and the magnitude of violence (Harkavy and Neuman 2001:230). Moreover, areas consumed by the same conflict can exhibit substantial variation in violence. Hence, violence should be analytically decoupled from war, echoing the well-established distinction between jus ad helium (lawful initiation of war) zná jus in hello (lawful conduct of war).
This book places violence at the center of the analysis. The analytical distinction between civil war and violence in civil war is both its premise and major implication. The causes of violence in civil war cannot be subsumed under the causes of civil war; hence a theory of civil wars cannot be a theory of violence in civil wars - and vice versa.6 At the same time, the theory of violence presented herein is compatible with different views of civil war onset: it does not matter whether civil wars begin because of mass grievances or opportunities. Simply put, a civil war is likely to open a Pandora's box of violence.
6 Hence it is incorrect to test theories of civil war onset using an indicator of violence, such as fatalities, as the dependent variable (e.g., Murshed and Gates 2005).
Concepts
1.2.2. Violence as an Outcome and as a Process
The observation that political violence tends to be produced by very small groups of people (Mueller 2004; Valentino 2000:21-5) has led to the conjecture that most people remain uninvolved (Valentino 2000:2); they are an unaware public at best and passive spectators at worst. Likewise, the observation that killers often dehumanize their victims (e.g., Toolis 1997:126) sustains the perception that violence in civil wars is impersonal. Regardless of their empirical accuracy, these conjectures fail to distinguish between violence as an outcome and violence as a process.
Although political scientists and historians tend to subsume violence under violent conflict, many anthropologists, NGO activists, and journalists tend to perceive violence as an outcome rather than as a process, often effectively "black-boxing" it (e.g., Appadurai 1996). The focus is on instances of violence rather than the complex, and often invisible, nonviolent actions and mechanisms that precede and follow them. Often, the description of very recent acts of violence is accompanied by references to ancient historical events, with no reference to the period in between. Like traditional depictions of Balkan feuding, many descriptions of civil war make no effort "to link one episode to another. Each case is treated as isolated in time and space. Nor do these writers attempt to explain the disproportion that so marks what superficially appears to constitute the relationship of cause to effect" (Black-Michaud 1975:34). Furthermore, little or no information is provided about the victims' histories and lives before the advent of violence (Binford 1996:5). Such a view assumes (and further propagates) a dichotomous world populated only by victims and perpetrators, combined with the flawed perception that victimhood and guilt are mutually exclusive categories - hence victims cannot be guilty. Yvon Grenier (1999:2) portrays the literature on Latin American insurgencies as suggesting "a world inhabited by women, children, and the elderly," a point also made about other civil wars (Cenarro 2002:67; Brovkin 1994:5). Typically overlooked is a large "gray zone" populated by those who partake in the process of violence in a variety of ways without, however, being directly involved in its outcome, as either perpetrators or victims. A corollary is that the line between perpetrators and victims is blurred, as yesterday's victims may turn into tomorrow's victimizers and vice versa (Joshi 2003:xiii; Chang 1992:498). Women and children, usually portrayed as victims, are often active and willing participants in all kinds of activities, including combat (Peterson 2000:112). Tzvetan Todorov (i996:xvi-xvii) tells how conducting a close study of a massacre that took place in the French town of Saint-Amand-Montrond in the summer of 1944 forced him to discover the missing sequence of events and revise his understanding of the massacre:
Little by little I realized that the massacre in question had not occurred at that time and place for no reason but was rather the culmination of a series of no less dramatic events that preceded it during that summer. After a short time I was no longer satisfied with having read the few works that told the various episodes in this story. With the help of a friend from the region, I decided to seek out and ask questions of the various contemporaries and witnesses of these incidents. I ran across some unpublished manuscripts. I read both the
22
The Logic of Violence in Civil War
daily and weekly press of the period, and I spent several days undoing the strings around the dusty files in the departmental and national archives. I could no longer tear myself away from the story.... In reading about [the fate of the main actors] I became convinced that, when talking about this period, it was imperative to go beyond both the hagiography of the "victors" (which is nevertheless so fitting for official celebrations) and its reverse image, systematic denigration; the same could be said for the "defeated." Instead of a world of black and white, I discovered a series of distinct situations, of particular acts, each of which called for its own separate evaluation.
Approaching violence as a dynamic process allows an investigation of the sequence of decisions and events that intersect to produce violence, as well as the study of otherwise invisible actors who partake in this process and shape it in fundamental ways.
1.2.3. Violence in Peace and in War
As studies of civil war have tended to overlook violence, studies of "political violence," a broad and imprecise concept that covers everything from campus demonstrations to genocide, have tended to disassociate it from civil war. This body of research often intersects with research on social movements - particularly studies of "contentious politics," a term that also includes phenomena ranging from nonviolent collective action to sporadic violence (Tarrow 1994)- This work tends to treat violence "either as the unproblematic extension of ordinary social movement processes, or conversely, as a pathological effect of competition or decline within social movements" (Seideman 2001:2).
Conflating violence in the context of contentious action with civil war violence suggests a failure to recognize that war and peace are radically different contexts that induce and constrain violence in very different ways. To be sure, these contexts share many mechanisms (Tilly 2003); however, the way in which these mechanisms are activated as well as their effects' diverges. Most obviously, forming and expressing political preferences are fundamentally different in times of peace and during a war. At the very least, the stakes are much higher in wartime.
The difference between violence in peace and violence in war is clearly one of degree. The total number of deaths in all reported episodes and campaigns of protest is negligible compared with the total number of deaths in all reported rebellions (Gurr 1986:52). Even terrorism involves violence onamuch lowerscale than civil war (Guelke 1995). Sri Lanka, a country with the misfortune of having experienced both peacetime riots and civil war has experienced significantly more fatalities due to the latter.
More important, the difference between violence in peace and violence in war is a difference in kind. As Vladimir Brovkin (1994:419) reminds us about Russia,
"the civil war routinized the unthinkable____It substituted for politics as usual the
politics of war." War structures choices and selects actors in radically different ways than peace - even violent peace. As a former Greek insurgent remarks, "an armed confrontation is not like a [workers'] strike. You can be defeated in a strike once and twice and three times, and still survive. When you opt for an armed rebellion you bet everything you have" (Papakonstantinou 1986 1:583).
Concepts
23
Contentious action represents a challenge to the government in place in a context characterized by an undeniable monopoly of violence by the state. In contrast, the defining characteristic of civil war is the absence of such monopoly. Contentious action in democratic settings is causally different from rebellion: whereas the former thrives in the presence of political opportunities, the latter is likely in situations where such opportunities are absent (Goodwin 1999); in ethnically heterogeneous societies at least, the dynamics of riots and demonstrations are the exact opposite to those of rebellion (Bates 1999). Unlike civil wars, riots tend to be a predominantly urban phenomenon (Varshney 2002:10; C. Friedrich 1972:70), lacking significant retaliation (Horowitz 2001:224), heavily influenced by institutional (often electoral) incentives (Wilkinson 2004), and facilitated by crowd anonymity; the ratio of perpetrators to victims tends to be inverse in riots and civil war: in the former participation is public and the victims are an unlucky few, whereas in the latter a few participate directly in victimizing an unlucky public. In Sri Lanka, ethnic riots declined and almost ceased after the civil war began, and there were no riots in Indian Punjab during the Sikh insurgency of 1984-94 (Horowitz 2001:482-5). Varshney (2002:11) is thus right to argue that a theory of civil wars must be "analytically distinguished" from a theory of riots. This is true, even when riots and pogroms take place in the context of war (Petersen 2002). The situation can be compared with the occurrence of genocide and war: although the two almost always intersect, the study of each phenomenon is usually distinct.
I.3. SCOPE CONDITIONS
Available conceptualizations of political violence as an object of research vary depending on the criteria employed: the scale of violence (mass killing; mass crimes; massacres) (Verwimp 2003; Valentino 2004; Sémelin 2000; Levene 1999), its mode and technique (riots, pogroms, reprisals) (Wilkinson 2004; Varshney 2002; Geyer 2000), the motivations of perpetrators (Straus 2000; Fein 1993), or the specific historical and social context of a particular instance (Browning 1998). A careful delineation of scope conditions is, thus, necessary. The intersection of two key features of violence defines the domain of analysis in this book: the aims and the production of violence.
1.3.1. The Aims of Violence
Political actors use violence to achieve multiple, overlapping, and sometimes mutually contradictory goals. Various literatures detail more than twenty uses for violence, including intimidation, demoralization, polarization, demonstration, radicalization of the public, publicity, the improvement of group morale, the enforcement or disruption of control, the mobilization of forces and resources, financing, the elimination of opposing forces, the sanction of cooperation with the enemy, and the provocation of countermeasures and repression (Hovil and Werker 2005; Schmid 1983:97-9, Mallin 1966:59, Molnar 1965:169). Further, violence may be used with no goal in mind, and war may generate violence that is
24
The Logic of Violence in Civil War
completely independent from the intentions of the main actors and materializes as a by-product of their action, such as looting or certain forms of revenge. Such a profusion of diverse aims can paralyze the analysis.
It is necessary, first, to address the issue of violence that serves no instrumental purpose. Such violence is said to be expressive when its use is restricted to the "strictly consummatory rewards of inflicting pain on one's enemies or destroying a hated symbol" (Rule 1988:190). Often described as anomic or nihilistic, expressive violence is often combined with "identity" or "sectarian" violence, that is, violence directed against persons exclusively on the basis of who they are. This understanding of violence dominates popular accounts that emphasize the madness of violence (e.g., Rosenberg 1991) and is present in many scholarly works stressing the discursive, symbolic, ritualistic, and generally noninstrumental character of violence.7 Interpretations of violence as expressive motivation can also be found in victims' testimonies: "They killed for killing's sake - like mad dogs going after their prey" (in Tarnopolsky 1999:52).
Individual motivations of violence can be, and often are, expressive (Petersen 2002; Horowitz 2001:123). Greek tragedy is a treasure trove of expressive violence, with orge (anger), eris (discord), or pthonos (envy) driving violent acts (Bernand 1999). Criminological research recognizes the importance of expressive motivations, because a large part of nonpredatory murders are not premeditated or driven by instrumental means-end motivations and are conducted with an indifference to consequences (J. Katz 1988). Many descriptions of violence in civil wars are apparently devoid of any instrumental significance and fit the expressive template very well. Consider the following parallel recollections, from the Spanish and Lebanese civil wars:
Later they shot Saturnino along with thirty-six others in reprisal for a civil guard's son who was killed at the front.... When the father heard the news of his son's death, he went to the Toro jail and began saying, "This one, this one, this one!" without knowing who they were. Thirty-six were shot. (Sender Barayon 1989:162-3)
We're heading straight for the slaughterhouse.... it's just a couple of blocks behind your house. You know the empty lot there. That's where Halabi, the Moslem butcher whose son was kidnapped, is collecting Christian Maronites. He wants his revenge, that man! We'd better stay out of that area. (Tabbara 1979:64-5)
An overriding emphasis on expressive motivation, however, runs into problems. In general, it is extremely difficult to uncover with an acceptable level of accuracy the individual motives behind violent acts (Tilly 1975:512). Deducing motive from behavior is a bad idea, as is replacing evidence with politically motivated classifications, as in the case of "hate crime" (Rothstein 2oo5:E3): the problem of observational equivalence is common since a particular act may be consistent with several motives. Furthermore, motives are typically subject to (strategic or unselfconscious) reinterpretation and ex post rationalization by the subjects. Even when fully revealed, intentions often turn out to be mixed or even contradictory. For example, individual motivations of violence may mix hatred (of many sorts), peer pressure (Browning 1992), obedience (Milgram 1974), honor,
7 E.g., Mahmood (2000:74-81); Geyer (2000:201); Crouzet (1990); Zemon Davis (1973).
Concepts
25
rituals, and collective imaginaries (Nahoum-Grappe 1996; Zemon Davis 1973), greed (Paul and Demarest 1988), revenge (Frijda 1994), or sadistic impulses; they may also result from the consumption of alcohol (Tishkov 2004:139; G. Jones 1989:124) or the use of drugs (Aussaresses 2001; Peters and Richards 1998). Complicating things is the prevalence of correspondence bias - the tendency of observers to draw inferences about enduring individual dispositions from behavior that can be explained by the situation in which it occurs (Gilbert and Malone
!995)-
Obviously, these problems apply to all types of motives, instrumental and noninstrumental alike. However, many observers tend to be biased toward interpretations that stress expressive motives. For instance, while several observers were quick to attribute the violence between the Dayaks and the Madurese of West Kalimantan in Indonesia to the ritual reenactment of headhunting, others remarked that violence was strategically deployed in the course of the conflict (Peluso and Harwell 2001). Consider the following remark by Mario Vargas Llosa (1994:428): "I was scribbling the speech... [when] the news of the assassination
of our leader in Ayacucho, Julian Huamani Yauli, reached me____His murder was
a good example of the irrationality and stupid cruelty of the terrorist strategy, since it was not intended to punish any violence, exploitation, or abuse committed by the extremely modest and previously apolitical Julian Huamani, but simply to terrify through the crime those who believed that elections could change things in Peru." In a misleading, though extremely common way, Vargas Llosa overlooked the clearly instrumental nature of this murder, which he himself acknowledges, to describe it as an irrational act.
Seneca observed that "no one proceeds to shed human blood for its own sake, or at any rate only few do so" (in Grotius II, 2 2 Indeed, expressive motivations may be less widespread than is often assumed. People involved in the production of political violence appear to lack the kind of "extreme" personality features that tend to correlate with expressive violence. A number of studies of perpetrators of violence have failed to uncover pathological traits (Kakar 1996; Delia Porta 1995), whereas others have pointed out that the ritualization of violence often serves instrumental purposes (Richards i996:xx; Schroeder 1996:432).
The stress on expressive violence may result from a double confusion: between individual and collective motivations and between descriptive and causal accounts. Arguments about the expressive and symbolic aspects of violence claim to address the motivations of collective actors (e.g., why the Hutu attacked the Tutsi) when in fact they only describe the way in which individuals perform violence (e.g., how some Hutu attacked some Tutsi). For example, in discussing the incidents of cannibalism perpetrated in the Liberian Civil War, Ellis (1995:193) points out that "the observation that there is a 'cultic' element to violence of this type does not imply that the militias fight primarily as a form of ritual behaviour." Inge Brinkman (2000:2, 14) first notes that her informants, Angolan refugees in Namibia, interpret the violence of the civil war as primarily senseless and absurd, "deemed to be beyond comprehension"; still, she then reports that her informants were also sharply aware that these practices were used to instill a paralyzing and incapacitating fear: "they do it," she was told, "to frighten people." Likewise, violence in Mozambique was often sadistic and reinforced by drug use, but there
26
The Logic of Violence in Civil War
is also substantial evidence that it "was co-ordinated and systematic rather than spontaneous" (Vincent 1994:87).
In fact, individual motivations alone are unlikely to result in large-scale violence over a long period of time. The Nazi policy of reprisals across occupied Europe was developed centrally even though it was often implemented by semi-rogue and openly sadistic junior officers (Heer 2ooo;Mazower 1993). It is, indeed, possible to overlay instrumental action on expressive action by imputing strategic behavior to leaders and expressive behavior to followers (May 1991:253; Coleman 1990:483). Unlike riots, civil wars are contexts that place considerable premium on organization, hence reinforcing interpretations of violence as instrumental.
Violence can be used to exterminate a group or to control it (Semelin 2000; E. Walter 1969:14). This book focuses on the latter type, also known as coercive violence. Although the methods used to achieve compliance and physical destruction may be similar, these objectives differ. A way to distinguish between the two is to ask whether at least one political actor intends to govern the population it targets for violence; an empirical indicator of this intention is whether the targets of violence have the option to surrender. In many civil wars amnesty programs encourage insurgent defection and spare or even reward civilians who defect and collaborate with them, whereas in genocides the surrender of victims does not prevent their murder but expedites it (Fein 1993:21). Analytically akin to physical destruction is mass deportation, sometimes referred to as "ethnic cleansing."
When violence is primarily used to control a population, it becomes a resource rather than the final product (Gambetta 1993:2). This type of violence entails an analytical distinction between the victims and the targets of violence (E. Walter 1969:9). If someone tortures a child in order to get her to reveal where somebody else can be found, the child is simultaneously a victim and a target. But if the same child is tortured in order to get her father to reveal somebody else's whereabouts of which the child knows nothing, then it is the father who is the target although it is the child who suffers the violence; the father can comply or refuse to comply, whereas the child can do neither (O. O'Neill 1991:172-3). In short, violence is intended to shape the behavior of a targeted audience by altering the expected value of particular actions. Put otherwise, violence performs a communicative function with a clear deterrent dimension - consistent with the description of civil wars as times of fear and eras of terror (Senaratne 1997:145). As Trotsky (1961:88) put it, "the revolution... kills individuals and intimidates thousands" - an insight also expressed in the Chinese proverb "kill just one and frighten ten thousand others." Mao Zedong called for "blows to the traitors and collaborators who undermine the army and the people" (in Heilbrunn 1967:145) and Che Guevara (1998:91) justified "assaults on persons" as a means of preventing information leaks. In Grossman's (1995:207) emphatic formulation, "One of the most obvious and blatant benefits of atrocity is that it quite simply scares the hell out of people. The raw horror and savagery of those who murder and abuse cause people to flee, hide, and defend themselves feebly, and often their victims respond with mute passivity." Note, however, that coercive violence is not necessarily massive. In fact, successful terror implies low levels of violence, since violence is "off
Concepts
27
the equilibrium path." Coercion fails if it merely destroys the subject whose compliance is sought.
Coercive violence may be strategic and tactical at the same time. Targeting a person to eliminate a particular risk (e.g., information leaks) is tactical, but using this act of violence so as to deter others from engaging in similar behavior is strategic. The counterrevolutionary rebels in western France directed their violence against people accused of informing the republican soldiers about their movements; they abandoned the mutilated corpses near republican-held towns and hung a tag around the informer's neck with his name and those of the victims who were avenged by his death; in this way they sought "to make examples in order to deter similar vocations" (Dupuy 1997:161). Martyn Latsis, a Communist leader during the Russian Civil War, asserted that "one must not only destroy the forces of the enemy, but also demonstrate that whoever raises the sword against the existing order of class, will perish by the sword" (Werth 1998:85). In Colombia, summary execution of suspected collaborators is the rule: "An assassin, dispatched day or night, ends any potential for collaboration and closes the case irrevocably, while also sending a crystal-clear message to the local population that the armed group will not tolerate such activities" (Fichtl 2004:5). Consider the following description of an IRA assassination: "Flood had become an RUC military asset who had to die to protect the IRA and deter other would-be informers" (Toolis 1997:202). In a different formulation, coercive violence tends to be both retrospective in its intention to punish an action that has already taken place and prospective in its goal to deter a similar future action by someone else.8
Even a cursory reading of descriptive accounts suggests the widespread strategic character of violence in civil war contexts. Consider the following examples. A Zimbabwean peasant explained the murder of a government collaborator by guerrillas by saying that "they only wanted to show the [masses] they had the power to do anything and instill fear so that none would repeat the mistake" (Kriger 1992:156). In Peru, "from the beginning, even without an infrastructure of war weaponry, Shining Path sought to terrorize and paralyze opposition, to inspire fear by displaying overwhelming force that demolished the enemy" (Del Pino 1998:168). Jeffrey Race (1973:135) was told that "the Vietcong use terrorism to instill fear. In a hamlet they will pick out a couple of people who they
say cooperate with the Americans, and shoot them, to set an example____After
they kill a few people, the whole hamlet is afraid and the Vietcong can force them to cooperate." A British agent in German-occupied Greece in 1944 stressed the same aspect to describe the violence used by the partisans: they "were masters of the psychology of the 'exemplary atrocity.'... They seem to specialise in picking on the one man whose death or disappearance would cause a whole area to continue its more or less docile support for their cause."9
8 Obviously, this logic is part of justice systems everywhere.
9 "Report by Cpl Buhayar," PRO, HS 5/698. See also Toolis (1997:81) on Northern Ireland, Senaratne (1997:121) on Sri Lanka, Kheng (1983:180) on Malaya, Ortiz Sarmiento (1990:190) on Colombia.
28
The Logic of Violence in Civil War
Perpetrators and victims often acknowledge the deterrent character of violence. The following entry from the diary of an Algerian man is telling: "November 29,1956: Each time a traitor or so-called traitor is executed, anguish seizes the survivors. Nobody is sure of anything anymore. People are really terrified. Terrified of the soldiers, terrified of the outlaws" (Feraoun 2000:155). A Nicaraguan Liberal writing in 1928 about Conservative violence pointed out that "All of the above delinquencies have been committed by Conservative bandits, and per the general opinion to put fear into the Liberals." Michael Schroeder (2000:38) who quotes him, concludes that "the fundamental objective of all political groups" was to instill fear (para infundir terror). This dimension is also consistent with ways of killing that sometimes border on the baroque, such as abandoning corpses in public spaces or stuffing the victim's mouth with banknotes to warn against accepting cash payments from rivals (Dalrymple 1997:123; Crozier 1960:163). Mutilation often serves the same purpose: it is a "walking example" (Leites and Wolf 1970:106). In Sierra Leone, the "cutting off of villagers' hands and fingers inscribe, on the landscape and in the bodies of village people, a set of political messages rather more firmly than if they had been spoken over the radio" (Richards 1996:6). In fact, instances of harrowing and seemingly absurd violence often reflect strategic calculations. Paul Richards (1996:181), an anthropologist who studied the civil war in Sierra Leone, argues that such an analysis makes sense of "patterns of otherwise apparently senseless violence by the RUE" William Finnegan (1992:58), an American journalist, likewise argues that many massacres perpetrated by the insurgents in Mozambique (and which were interpreted as gratuitous by ill-informed observers) were intended to send specific messages to the local population.
Although the underlying logic of coercive violence is similar across contexts, its form varies depending on aims and local cultures. Targets can be prominent local personalities or weak and marginal people, and the intensity of their victimization varies widely. A Northern Irish petty thief recalls how his defiance in the face of the IRA demand that he leave the country led to his abuse (in Smyth and Fay 2000:124): "So, the IRA was blaming me and saying it was my fault because if I hadn't stood up to them, no one would have stood up. There was nobody stood up to them before me."
In short, although violence in civil war may fulfill a variety of functions, the instrumental use of coercive violence to generate compliance constitutes a central aspect of the phenomenon. This is not to say that it is the only one. However, given the present level of theoretical development it makes sense to focus on it.
1.3.2. The Production of Violence
Violence can be produced unilaterally (by one actor, usually the state) or bilat-erally/multilaterally (by two or more competing actors). The main difference between unilateral and multilateral settings is that strategic interaction is more critical in the latter. When the population has the option to join or assist existing rival actors, its reaction to violence must be factored in because it is consequential
Concepts
table i.i. A Typology of Mass Political Violence
29
Aims of Violence: Political Actor Intends to Govern the Population Targeted
Production of Violence
Yes
No
Unilateral
Bilateral (or multilateral)
State terror Civil war violence
Genocide and mass deportation Reciprocal extermination
for the outcome of the conflict. Because the villagers of Due Lap in South Vietnam were mistreated by the government soldiers assigned to protect them, they welcomed the Vietcong who got "them off their backs" (Ellsberg 2003:131). As aMozambicanmanputit (inNordstrom 1997:9), "Youknow, sometimes when there is only one force, they can do anything they please. There are problems with this, they can begin to throw around their power, make people do things they don't want to do, use violence against people to get what they want. When you have two forces, people now have an option. Each force has to be more responsible. People can say, 'Hey you can't treat us this way, there are others to protect us.'"
The intersection of aims and production of violence generates four ideal types of mass political violence: state terror, genocide and mass deportation, civil war violence, and a type that may be referred to, for lack of a better term, as "reciprocal extermination" (Table 1.1). These categories are not intended to capture the entire spectrum of real-world variation. They provide, instead, a useful way to specify the scope conditions of the book.10
Coercion is present in standard definitions of state terror (Mitchell et al. 1986:5). As a Spanish inquisitor put it in 1578: "We must remember that the main purpose of the trial and execution is not to save the soul of the accused but
10 State terror may be delivered in a quasi-multilateral fashion by competing state agencies; the goals of compliance and extermination may coexist in the purposive elimination of one group so as to terrorize others; counterinsurgency campaigns launched with the intention of reestablishing government control over rebel-held areas may degenerate into genocidal violence; and governments, such as the Argentine junta, may justify repression by claiming that they are fighting a civil war. For example, Margolin (1999) argues that the Indonesian government's purpose in killing thousands of Communists was intimidation rather than extermination; Gurr (1986:47) disagrees. Diaz-Balart and Rojas Friend (1997:15) describe the violence exercised by Nationalists in the Spanish Civil War as intended to fulfill both intimidation and "often" extermination purposes. Likewise, Ranzato (1988) and de la Cueva (1998) show that the persecution of Catholic clergy by the Republicans during the same war reveals both an intention to scare predominantly Catholic Nationalists into compliance and a desire to exterminate as many priests as possible, simply because they were priests. The White Terror of the Russian Civil War included violence to exterminate the Jews (Figes 1996), while the Red Terror demanded the extermination "to the last man" of certain groups, such as the Cossacks (Brovkin 1994:103). The Nazi occupation of Poland aimed both at the extermination of the Polish elites and the exploitation and control of the masses (Jan Gross 1979:76). Still, it is possible to use this typology to sort out different processes taking place in the same location and time: the Nazis used different methods against partisans and Jews in the Ukraine, and, although both the Tutsi and Hutu were killed in Rwanda, the violence against the two groups followed distinctive patterns (Verwimp 2003).
3°
The Logic of Violence in Civil War
to achieve the public good and put fear into others" (in Kamen 1998:174)." The key feature of state terror is that it is exercised against a population that lacks organized alternatives; this may account for the arbitrary character it sometimes takes. Chang (1992:218) describes how the Maoist "Anti-Rightist" Campaign of 1957 produced categories of "rightists" described in everyday language as "lots-drawing rightists" (people who drew lots to decide who should be named as rightists), "toilet rightists" (people who found they had been accused of being rightists in their absence after they could not restrain themselves from going to the toilet during the long meetings), and the rightists "who had poison but had not released it" (those who were named as rightists without having said anything against anyone). Przeworski (1991:47) cites a Soviet joke: "Three men meet in a gulag. One asks another, "What are you here for?' 'I was against Radek,' he says. 'And you?' 'I was for Radek.' They turn to the third man, thus far silent. 'I am Radek,' he says."12
Genocide is premeditated, purposive, and centrally planned; it aims toward extermination rather than coercion. At its core is "intentional group annihilation" (Straus 2000:2). From this perspective, genocide is neither mere continuation of severe repression through other means nor just mass killing, but a phenomenon of an altogether different kind (Straus 2000; Chalk and Jonassohn 1990). The violent, purposeful, and permanent deportation of populations, usually in the pursuit of nationally pure space ("ethnic cleansing"), is also driven by the logic of group elimination, though the elimination is spatial rather than physical (Snyder 2003).13
Reciprocal extermination is a type of violence that emerges in multilateral, interstate or intrastate contexts where neither political actor intends to govern the
11 Kamen (1998:174) adds that "The coming of the Inquisition to a town was, in principle, designed
to cause fear____The public activity of the Holy Office was thus based on a premise, common to
all policing systems, that fear was the most useful deterrent."
12 Arendt (1973:305) argued that whereas "tyrannical terror" ends once it has paralyzed individuals, and "revolutionary terror" ends when the opposition is destroyed, "totalitarian terror" begins only after the opposition has been destroyed. In this situation, "terror is no longer a means to an end; it is the very essence of government." Violence in totalitarian dictatorships can turn not only against the regime's alleged enemies but against completely innocent people (Gillespie 1995:244) or even its friends and supporters (Arendt 1970:55). McAuley (1992:50) describes Stalinist terror as a completely arbitrary system where it was impossible to know how to avoid arrest and where the most committed supporter of the regime could be arrested and the most apathetic left untouched; the Russian writer Ilya Ehrenburg recalled about this period that "the fate of men was not like a game of chess, but like a lottery" (in Schmid 1983:175). In such extreme environments, violence can easily become a goal in itself. For example, toward the end of 1977, one of the most notorious detention centers of the Argentine military, finding that the "natural" supply of "subversives" was drying up, had been calling factory managers to inquire whether they had any "troublemakers" (Gillespie 1995:244).
13 Mass population movements should be distinguished from mass deportation when they are the unintended by-product of war or the intended but temporary product (as in "forcible evacuation"). The possibility of returning to one's place once the war has ended provides an acid test for distinguishing between mass deportation and population movements. Mass dislocation that is unrelated to armed conflict is, obviously, a different issue; an estimated 40 to 80 million people have been physically displaced worldwide by the construction of dams (Rajagopal 2001).
Concepts
31
population it targets for violence; put differently, political actors hold symmetrical intentions to exterminate each other's "civilian basis." Such intent often entails mass deportation. Often, this type of violence is associated with circumstances of state collapse and a type of warfare I label "symmetric nonconventional" (Chapter 4). Examples include the Balkan Wars (1912-13), the partisan war between Poles and Ukrainians during the Second World War, the partition of India, and the Serb-Croat War. In general, however, exterminatory violence tends to be unilateral rather than reciprocal. In fact, the unilateral nature of genocide appears to be such an empirical regularity as to be part of many definitions (e.g., Chalk and Jonassohn 1990:2 3).14
This book focuses on the final category, civil war violence. Unlike state repression and genocide, it is not unilateral: it is produced by at least two political actors who enjoy partial and/or overlapping monopolies of violence. Unlike the unilateral production of violence, targeted individuals often have the possibility of shifting their support and resources to competing actors; this is possible because at least one actor intends to govern the population it targets rather than to exterminate or deport it. This feature turns violence into a process with obvious strategic implications. First, political actors need to anticipate their opponents' strategy and the likely effects of their violence on civilians. Second, violence is not merely state terror multiplied by two; whereas the violence under unilateral provision is more or less a direct expression of the intentions of the actor initiating it, in civil wars it reflects the strategic interaction of at least two actors that are simultaneously present on the same territory.
I.4. CONCLUSION
This chapter has clarified the conceptual terrain. I have supplied working definitions of civil war and violence, discussed their parameters, established a set of crucial conceptual distinctions, and specified the scope conditions of the analysis. The phenomenon under investigation is intentional physical violence against noncombatants that takes the form of homicide, in a context where at least one actor seeks to control the population. The particular subset I examine is coercive violence, which is used to obtain popular compliance - a type of violence that tends to be strategic.
14 The distinction between civil war violence and "reciprocal extermination" does not overlap with that between the violence of ethnic and nonethnic civil wars. In the majority of ethnic civil wars at least one actor (usually the incumbents) intends to rule over the population that constitutes the ethnic basis of its opponent. Reciprocal extermination appears to be a sub-type of either ethnic or "ideological" civil wars (Kalyvas 2002).