[2]
The Symbolic Politics of Ethnic War
Explaining ethnic war requires answering two different but related questions. The first is the question of ethnicity. What exactly is an ethnic group? Why do ethnic groups stick together in politics when individual interests often cut across ethnic lines? Why did ethnic groups become especially important politically only in recent times? The second question concerns the causes of war. How do ethnic wars get started, and why do they happen? Why are people willing to fight and die for their ethnic group?
To be convincing, a theory needs to answer both questions in compatible ways—showing the ways in which the nature of ethnic groups explains how and why they fight. For example, if the theory assumes that ethnic groups act like economic interest groups, it should explain why economic interests motivate people to stick with their ethnic groups and fight. Additionally, the best theory will be supported by evidence at every stage. Thus, lor an ancient hatreds account to be convincing, it should show that ethnic groups are ancient, that the hatred is ancient, that their hatred is the motivation for them to fight ethnic wars, and that variations in the amount of ancient hatred explain why ethnic wars break out in some places but not others. The task of this chapter is to assemble these combined theories, pinpoint their weaknesses, and explain how a symbolic politics approach offers a stronger alternative.
Clarifying Definitions
The terminology used in discussions about ethnic conflict is so confused, and confusing, that it is important to sort out the meanings of key terms before beginning the analysis. First, ethnic group and nation are separate but
[15]
iiliin Hut ink
m ri lapping concepts. An ethnic group, in Anthony Smith's definition, is a group sharing five key traits: a group name, a believed common descent, common historical memories, elements of shared culture such as language or religion, and attachment (even if only historical or sentimental) to a specific territory. These elements are all tied together by a "myth-symbol complex" (defined below).1 A nation, by contrast, is a socially mobilized group that wants political self-determination.2 Thus, not all nations are ethnic groups (some, like the American nation, are ethnically heterogeneous); and not all ethnic groups are nations (many do not aspire to political autonomy). Cases of ethnic war, however, are always cases of competition for political dominance, so they all involve ethnic nations on at least one side. Strictly speaking, therefore, this book is about "ethno-nationalist" wars, but I will stick to the shorter formulation for the sake of simplicity.
Another set of terms relevant to the discussion is nationalism, chauvinism, and hostility. Nationalism is the belief that one's own group should be politically autonomous—the belief, that is, that one's own nation should take its rightful place among the nations of the world. Chauvinism is the belief that one's own group is better than others, and therefore has the right to dominate or displace them.3 Hostility means relating to another group as to an enemy. The phenomena are separate: nationalists may seek equality for their group, and avoid being chauvinists; chauvinists may be content with their nation's status and see others paternally as "younger brothers" rather than enemies. Frequently, however, extreme nationalists tend to be chauvinists, and chauvinists tend to be hostile to nationalists of other groups.
Central to understanding ethnicity are the terms myth and symbol. According to Murray Edelman, on whose theory of symbolic politics I build, a myth is "a belief held in common by a large group of people that gives events and actions a particular meaning."4 The truth or falsity of the myth is irrelevant; its purpose is to help a person understand what a set of events means to him or her. For this reason, it is appropriate to talk about the "myth" of Kosovo or the Armenian Genocide: while these are real events, what affects politics now is less the events themselves than the mythologies that have grown up around them.
A symbol is an emotionally charged shorthand reference to a myth. In Serbian mythology, for example, the meaning of the Battle of Kosovo Field is the martyrdom of the Serbian nation in defense of Serbian honor and of Christendom against the Turks. "Kosovo" therefore is a symbol referring to this myth of Serbian martyrdom; the point of invoking the symbol is usually to express, to communicate, or to evoke among Serbian listeners the emotions, such as pride or a sense of national grievance, associated with the myth. The web of myths and related symbols like these that collectively define what it means to be a Serb forms the Serbs' "myth-symbol complex. "5
SymboHt Poltth to) / fhnk War
/ thiur war, finally, is a war in which the key issues at stake—that is, the -ss reason political power is being contested—involve either ethnic markers such as language or religion or the status of ethnic groups them-. I \ cs. A war is organized armed combat between at least two belligerent •tides in which at least one thousand people are killed.
Theories of Ethnic Group Origins and Violence
\n.mpts to explain the origins of ethnic groups and the reasons they fight l.i 11 into two basic but diverse categories: rational choice explanations and psychological arguments.
rational choice approaches Ethnic groups as instrumental According to the "instrumentalist" approach, ethnic groups are merely coalitions formed in a rational attempt to compete for scarce goods in the 11 mtext of social changes brought about by modernization.6 As evidence, in-i mmentalists point to ethnic conflicts in Africa, which are often essentially l .inflicts between ethnically defined patron-client networks over economic goods distributed by the state. This is even true, they argue, when the rhetoric of leaders emphasizes such noneconomic issues as cultural autonomy.7 In this view, ethnic nations formed in the last two centuries when a "national" language was chosen from among a variety of different dialects, written down, and made the basis for mass literacy in specific states.8 This was done because industrial society requires a standard means of communication, thus a single standard is set for an entire territory.9 In empires with many unrelated languages, certain economic classes of subordinate groups promoted literacy and ethnic nationalism in their linguistic groups, thereby creating new ethnic groups.
In sum, the key to this explanation is the self-interested basis of ethnic group formation: elites form people into ethnic nations because of a combination of linguistic and class interests. Furthermore, while all group members share these interests, the elites have an additional interest in organizing tlie group to pursue them—the elites gain power by leading that pursuit.
Rationalist theories of ethnic war Rational choice theorists starting from this instrumental understanding of group origins have produced several different theories of ethnic war. I will focus here on three: the "economic rivalry" approach, the "hard rationalist" school, and the "soft rationalist" account.
[16]
[17]
lie approach. Arguments that ethnic wars result primarily from economic concerns come in several different versions. One version has it that the issue is distribution: poorer areas might want to secede because they consider the central government to be discriminating against them; and richer areas might want to secede because they do not want to be burdened by union with the poorer ones.
Such arguments have a poor track record empirically. They cannot explain ethnic mobilization in general: analyses find that ethnic groups mobilize under virtually all economic circumstances.10 Statistical studies come to a similar conclusion: economic discrimination has no significant effect on ethnic group mobilization or on demands for group autonomy.11 These economic arguments also fail to account for the pattern of violence in the dying Soviet Union, as Table 2.1 shows. Rather, all four of the non-Russian republics that suffered ethnic war—Armenia, Azerbaijan, Georgia, and Moldova—were middle-income republics. There is no pattern among secessionist subregions either: some, such as the Transnistria region of Moldova, were more industrialized and prosperous than the unit they were seceding from; while others, such as South Ossetia, were less so.
Another prominent approach focuses on relative deprivation,12 and specifically on declines in standard of living. This logic would seem to apply to Yugoslavia and the Soviet Union, both of which experienced serious declines in standard of living in the years before their collapse. It does not, however, explain the pattern of violence, which should show that those areas which suffered more were more prone to ethnic violence.
As Table 2.1 shows, Tajikistan fits the relative deprivation logic—it was second-worst off in both economic performance and national income, and it did experience a civil war—but its conflict was not an ethnic war. On the other hand, Armenia and Georgia, which did experience ethnic wars, actually saw their economies grow relative to Russia's in the years before their wars, while the declines in Azerbaijan and Moldova were minimal. Since the Soviet economy overall grew at a reasonable pace in the 1960s and 1970s, and was stagnant from 1980-88 (the rapid decline began only in 1989),13 even the mild declines relative to Russia suffered by Moldova and Azerbaijan may not imply absolute decline. The data seem clear: ethnic violence does not correlate closely with economic hardship or economic decline.
In addition to this trouble with the evidence, economic explanations of ethnic war face the additional problem of logical inconsistency: for most people, fighting an ethnic civil war is not economically rational. The reason is simple common sense: fighting a civil war over the benefits of a fragile national economy is obviously likely to destroy that economy. Peaceful ethnic mobilization may make economic sense, but running a serious risk of ethnic war does not because the economic risks vastly outweigh the potential gains, except for would-be profiteers. The Yugoslav conflict of the
Belarusian SSR Lithuanian SSR Georgian SSR Armenian SSR Russian SSR Kazakh SSR Azerbaijani SSR Moldovan SSR Turkmen SSR Latvian SSR KyrgyzSSR Ukranian SbK Estonian SSR Tajik SSR Uzbek SSR
63.6 84.8 69.2 68.8 100.0 61.2 67.1
73.1
61.2 112.8
55.6 88.9 114.8 50.8 55.8
1005 95.0 77.4 75.6 100.0 61.0
65.6 71.0 56.9 107.3
48.7 80.0
105.3 39.3 43.0
+37.2 +10.2 +8.2 +6.8 0.0 -0.2 -1.5 -2.1 -fc3 -5.5 -6.9 -8.9 -9.5 -11.5 -12.8
3 5
7
8
4 11 10
9
12.0
1
13 6
2 15
14
Uzbek bbK
Source O. G. Dmitrieva, Regional'naia ekonomicheskaia diagnostická (St. Petersburg: Limbus Press, 1992), p. 79.
early 1990s illustrates the point. That conflict was to a considerable degree triggered by disputes over the distribution of economic goods. However, it very quickly became apparent that almost everyone would lose economi-1 .illy once war began disrupting trade, destroying factories and infrastructure, and turning workers into fighters, refugees, or casualties. In fact, virtually everyone in the former Yugoslavia and the South Caucasus (except for a few profiteers and looters) lost economically from war by every measure. Given the economic costs of postwar hostility, even most of the winners ended up worse off absolutely, worse off than they would have been had the status quo continued, and even worse off than they would have been had they acceded to the rival group's demands. Hard rationalist approaches. Another rationalist explanation for ethnic war, promoted by Russell Hardin and others, takes a step away from the instrumentalist understanding of ethnicity, arguing that ethnic war is explicable as a function of individuals' rational pursuit not of material benefits but of personal security.14 The argument suggests that in cases of "emergent anarchy," when the state will not or cannot guarantee people's safety from violence, it is rational for groups to start mobilizing in preemptive self-defense. The result is a security dilemma, in which each group's acts of self-defense threaten other groups, leading to escalating preparations for violence and ultimately to the outbreak of fighting. "Risk-aversion is enough," in this account, "to motivate murderous violence."15
[18]
'rl 11,III
Tlit* process of mobilization, in this view, is a "tipping process" driven by peer pressure: the mow people join in ethnic mobilization, the more they can pressure others into joining, and the more credibly they can argue that it pays to join because their growing movement is likely to succeed. Thus the mobilization process snowballs. And if a group norm in favor of violence somehow emerges, it becomes rational for individual members of the group to engage in murder.16
A third argument Hardin makes relies on selective incentives. In this account, people may engage in violent conflict because they are promised rewards for killing or threatened with punishment if they do not kill. To the extent that the participants in Yugoslavia's conflict, for example, were conscripted, this argument explains the conflict as the result of the coercive power of the state, especially Serbia. Criminals may also be useful for aggressive governments, as they can be promised loot or locations for mafia-style operations in exchange for leading their gangs (armed by the state) into battle.17
David Lake and Donald Rothchild suggest two additional reasons to think that "emergent anarchy" by itself might be enough to motivate violence.18 One is information failures: each side has reason to conceal its true desires and strength, but doing so makes it harder to reach a negotiated agreement. The second issue is the problem of credible commitment: even if the sides make promises in good faith, there may be no way to guarantee the promises will be fulfilled later. Added to the security dilemma, these problems make negotiated agreements hard to reach and war more likely.
However, Lake's and Rothchild's understanding of the security dilemma assumes that both sides want to avoid war, and that they are willing to compromise to avoid it. War results, in this logic, only because the sides do not trust each other.19 The problem with this argument is that I know of no case in which ethnic war resulted from such a process. In every case I am familiar with, security dilemmas were the result of the sides' openly stated pursuit of dominance, not the result of overzealous self-defense under "emergent anarchy." Each side's goals, and expectations about the other, came from hostile interpretations of history encoded in each group's "myth-symbol complex." In such cases, information failures and problems of credible commitment are irrelevant: the sides would rather fight than compromise for reasons that better information and stronger commitments cannot change.
In fact, by focusing on "emergent anarchy," these theorists ignore evidence that they have the causal chain backwards: in ethnic conflict, the security dilemma causes anarchy to emerge, not vice versa. In the Soviet case, for example, there was little evidence of "emergent anarchy" before the
//;, Symbati Politict o) i thnic War
j ii il>.«gh crisis exploded into violence in February 14HK. It was, rather, the uimil.ition of ethnic and nationalist disputes across the Soviet Union Hi ii eroded the effectiveness of the Soviet state, and finally caused it to lue.ik up more than three years later. In most cases—in ten of fifteen Soviet I * 1 Mil '11. . and twenty of twenty-one ethnic regions in Russia, all of them elliiiu ally mixed—no ethnic war occurred in spite of the "emergent anar-
• h\ mi the Soviet breakup. When violence did erupt, the issue was usually Rl 'i fear about what rival groups might do, but an open struggle for domi-n.in, e This insight explains why nationalism is, in Jack Snyder's words,
• in' "default option" for identification when empires break up:20 empires In. ik up because the forces of nationalism are already in control.
another weakness of these "hard rationalist" arguments is that al-
ii.....gh they begin as arguments about individual self-interest, they end
u|< depending on the emergence of group norms. If groups formed instru-Bientally merely to seek security, it would be rational for people on the n eaker side to switch groups, and for their opponents to let them. It might Mdinetimes be rational, however, for one's own (selfish) co-ethnics to be-
• i.iv one, rendering the whole idea of group self-defense problematic.21 1'hese problems can only be avoided by assuming that effective group norms prevent such conversions. The switch to group norms weakens the theory, but is supported by the evidence: studies suggest that the values thai motivate people to engage in political violence are collective interests, not individual self-interest.22 The trouble is that these theories do not ex-I'l.iin why this happens.
Additionally, this argument requires that group norms justify extreme \ uilence, which then spreads due to a "tipping process." But it would seem more rational for people threatened by violence to reinforce norms against 11 Norms can define self-defense in a narrow and immediate sense that diK's not encourage preemptive mobilization or violence. Rationalists such as Hardin do not explain why violent norms emerge instead. Appealing to .1 history of conflict does not help: most neighboring groups have some history of past violence that can be used and exaggerated. Alternatively, neighbors can be conflated with mythical enemies, as when Serbs associate Bosnian Muslims with the depredations of the Ottoman Turks.
Soft rationalist approaches. The soft rationalist approach, pioneered by Alvin Rabushka and Kenneth Shepsle, concedes that extremist group values are the core causes of ethnic violence and builds on that assumption.23 "Rationality" in these models means the rationally calculated pursuit of any consistently defined goals, including those defined by a nationalist ideology. In this version, it is easier to explain why people are willing to join in ethnic movements, since one can appeal to nationalist values to explain
[21]
people's !vh.i\ 101 R.ihushk.i and Shepsli .-.       m-iu is in.it lr people have extreme preferences—for example, they want to try to dominate the other i group even if trying is very risky—then extremist politicians are likely to | outpoll moderates. These extremists, once in power, can then use the ma-chinery of government to organize their willing followers for war.
When preferences are extreme, a "softer" version of Hardin's tipping argument can also be used to explain ethnic wars that start from the bottom up, without government help. Alexander Motyl points out that some people are more strongly nationalistic than others, and they can be divided into "martyrs," "fanatics," "true believers," and "believers of convenience."24 I Nationalist martyrs and fanatics can be counted on to be politically active ! even in the face of certain repression, so there is always likely to be someone promoting nationalist policies. If political repression decreases, true believers will join in and the nationalist movement will grow. If the incumbent government then shows signs of responding, more moderate nationalists will join as well. The more who join, the greater the incentives for others to join, as a larger movement is both safer to join and more likely to succeed. Martyrs or fanatics may also play an important role by starting the cycle of violence, provoking retaliation which shocks moderates into the belief that extremist policies are necessary.
The inadequacy of rationalist approaches Available rationalist explanations of ethnic war are inadequate. The economic rivalry argument is logically inconsistent and fails to explain the pattern of ethnic violence. Hardin-style "hard rationalist" arguments fail to explain why ethnic groups hold together. They are also wrong empirically: "emergent anarchy" does not cause ethnic security dilemmas; rather, mutual hostility causes insecurity, ethnic mobilization, state breakdown, and therefore anarchy. Deprived of this argument, these hard rationalist explanations for violence collapse. While the soft rationalist approach works, it does so only by assuming group norms and hostile attitudes, which is most of what the theory is meant to explain.
Rationalist approaches do offer important insights, however. First, they point out that elites, who powerfully shape the course of ethnic conflict, may use ethnicity instrumentally in pursuit of their own personal interests. Second, the mass-led "tipping" process and selective incentives provided by elites are convincing explanations for how mutually hostile groups might mobilize. Finally, information failures and problems of credible commitment can exacerbate interethnic security dilemmas, though rationalists misunderstand the origins of those security dilemmas.
To explain why ethnic groups hold together and hostile attitudes emerge to create security dilemmas, we must turn to psychological theories.
Polltiao) i /'"in Wo
psychological theories
Psychological roots of ethnicity
Jhitltordialism. A powerful statement of the primordialist position is 11 M. Isaacs's Idols of the Tribe, which states, "basic group identity con-1*1 .1 the ready-made set of endowments and identifications that every M"lr- idual shares with others from the moment of birth by the chance of Ihr family into which he is born."25 The markers of group identity, Isaacs
I......i. out, are basic personal characteristics that fundamentally shape
how individuals view the world and how the world views them. Group i I' nliiv is indeed for most people ascriptive—that is, assigned at birth— H ii I is often marked on the body, either naturally as racial characteristics ■" i arved on by circumcision, tattoo, or other artificial process. Native lan-Mge provides the words, and therefore the conceptual lens, through fcrhii 11 one understands the world. And religion shapes values, identity, mil therefore how people define their wants and needs. Walker Connor, who takes a position similar to primordialism, argues that the sum of these markers defines a group which acts like, and indeed claims to be, a kin group.26 According to this logic, ethnic loyalty taps some fundamental I >m (logical drives, such as defense of kin and territoriality. It is these mutually reinforcing bonds which give ethnicity its power in the primordialist view.
The trouble with this argument is that the history and kinship ties are usually fictitious, while most national identities are new. Thus in France, the epitome of a European nation-state, many peasants felt only local rather than national loyalties down to the end of the nineteenth century.27 N.itional identity can hardly be "primordial" if it is new; and it cannot be genetic if its members are not related.
Constructivism. The constructivist position begins from the insight that the meaning of an ethnic identity—who is included in the group, what its values are, and so on—is a set of ideas. Those ideas, constructivists point out, are generally either newly invented or newly interpreted by ethnic or nationalist intellectuals. It is therefore these intellectuals who "construct" ethnic identity, sometimes by inventing group history from whole cloth.28 In some parts of Africa and the Soviet Union, even ethnic labels and literary languages were first created by outsiders such as missionaries or anthropologists, and the resulting identity came to be accepted by the groups only after governments began applying the label to them.
This fundamental point is actually conceded by both primordialists and instrumentalists. Thus, the above-mentioned instrumentalist account of the creation of nineteenth-century European nationalisms is also a construc-
|22]
Modern Hatreds
tivist account, dependent on the role of ethnic elites and intellectuals in "constructing" nations. And Clifford Geertz, considered the founder of the primordial school, agrees that ethnic markers are culturally rather than biologically determined.29 The primordialist point is that there is a limit to the plasticity of ethnic identity. Although intellectuals can feel free to invent, reinvent, or obscure their group's past to a large extent, they must build on certain preexisting foundations, such as language, religion, culture, and territory. The trick for the intellectual "cultural entrepreneur" is to link a name and cluster of cultural elements to a history and mythology that both creates ethnic symbols and answers contemporary group needs. In short, while ethnic groups are not permanent and fixed, they are not infinitely malleable either.30 Weakly supported national identities such as the Yugoslav and Soviet do not survive.
In some cases, ethnic identities may be used to justify conflicts not intended by their creators. For example, myths of "ancient" Sinhalese-Tamil conflict in Sri Lanka were recreated by Sinhalese activist Anagarika Dharmapala in the late nineteenth century to help promote a cultural and Buddhist religious revival among the Sinhalese people, and secondarily to build political opposition to the British colonialists. Decades later, S.W.R.D. Bandaranaike used those myths to mobilize Sinhalese around an anti-Tamil program in the 1956 election campaign.31 Thus the politicians who exploit ethnic myths may not be the thinkers who created them.
The symbolist synthesis. The different schools of thought about ethnicity are more compatible than normally understood. The barrier to theoretical reconciliation has been the misperception that primordialists see ethnic ties as immutable, genetic links. Understanding that most primordialists see ethnicity as culturally rather than genetically determined opens the way to combining all three schools of thought. The outlines of a synthesis were sketched out by Crawford Young in the mid-1970s.32
The synthesis begins with the argument that group loyalty is likely to be evolutionarily favored: those who can in a crisis count on fellow group-members' loyalty—including such non-kin as in-laws—were presumably advantaged over strict egoists, who would have died when their groups fractured under stress. "Nationalism gets its force," in this view, "by drawing on [this] primordial sociality." This argument is sometimes caricatured as a simplistic "killer ape" theory of human nature, but in fact its more sophisticated version is consistent with constructivist logic. The argument is not that ethnic violence is somehow encoded in human genes, but that cultural tendencies toward collective group self-defense are evolutionarily favored.33
Ethnic or national leaders then create nationalist identities, using ethnic symbols to mimic the cues that originally invoked a genuine kinship/
[24]
I lu- Symbolic Politics of Ethnic War
C.ioiip defense response—hence the "motherland" and "fatherland" sym-!•! ilni ommonly used by nationalists to combine the notions of home territory >ii 1.1 family.3* The core of the ethnic identity is the "myth-symbol complex"— (In i ombination of myths, memories, values, and symbols that defines not . nily who is a member of the group but what it means to be a member.35 The Wlstence, status, and security of the group thus come to be seen to depend on the status of group symbols, which is why people are willing to fight and In idi them—and why they are willing to follow leaders who manipulate those symbols for dubious or selfish purposes.
Nationalism is further strengthened by the religious nature of ethnic or nationalist ideologies. In this view, what nationalist ideologues create is in He sense a religion. For the nationalist, the nation is a god—a jealous ■od—to whom one pays homage, venerating its temples (monuments), 1. lies (battle flags), and theology (including a mythical history); and receiving in return a sort of immortality as a participant in what is conceived as in eternal nation.36
This synthesis explains why ethnicity is so powerful and ubiquitous: it 111 aws its power from many sources. If cleverly cast, an ethnic or nationalist appeal can claim that the ethnic warrior is fighting simultaneously for self-respect (identity), self-interest (material goods), clan survival, clan territory, the propagation of the faith, and country; and if the fight is successful, the warrior will have achieved immortality (through martyrdom and the ■ li'lense of progeny) even in death. Ethnic group or nation is, therefore, a god so powerful that it is irresistible to invent him wherever he does not r\ist. Attempts to create ethnic loyalties are only successful, however, when the symbolic claims seem credible and relevant.
This account also explains why ethnicity is increasingly important today, though ethnic groups have existed for millennia.37 The key changes are mass literacy and mass media, which made it possible to align the appeal of state and (redefined) tribe, and to mobilize the entire group around the combined ethno-nationalist theme.38 At the same time, the processes of modernization throw together people from different places, making the markers of ethnicity more salient in daily life, while the increasing role of the modern state makes it more important to mobilize politically—hence the incentive to mobilize on ethnic lines. Finally, international acceptance of nationalism as the main prin-«1 pie legitimizing independent statehood encourages emulation, as groups are induced to say, "we too want our own nation-state." All of these factors contribute to the rising importance of ethnicity.
Psychological explanations of ethnic war The case for emotional motivations in ethnic war is most effectively argued by Crawford Young and Donald Horowitz.39 Horowitz starts with Henri Tajfel's finding that, when offered the choice between maximizing benefits
[25]
Modern Hiiti
lor llit'ir own group or maximizing the difference between their group and another, people tend to choose to maximize the difference. In other words, people gave up some (potential) benefits for their own group to ensure that the other group gained even less—even when the group is a randomly created one. Horowitz adds that when ethnic conflict turns into such competition for group advantage, the result is frequently a contest for dominance of the state as a way for groups to try to show their superior group worth. The language of the contest is legitimacy, as each group tries to prove its moral and historical claims give it the legitimate right to political dominance in their own homeland.40 The issue in ethnic conflict, then, is not so much specific economic, linguistic, or other specific benefits, but relative status—superiority over other groups.
Prejudice plays a role as well. In many cases, a group's myth-symbol complex includes prejudice against the other group—that is, stereotypes about the rival group, enhanced by negative feelings about that group. Again, the emotional dimension of prejudice is important: research suggests that negative feelings about the other group are more important than stereotypes in explaining attitudes toward outgroups.41
The thinking involved in such conflicts is illustrated by the anecdote of the Russian peasant who finds a genie in a bottle. Offered a single wish, the peasant muses: "Well, my neighbor has a cow and I have none. So—Kill my neighbor's cow." This story illustrates the logic of ethnic war—hostility trumps acquisitiveness. The peasant could have improved both his absolute and his relative position by asking for two cows, but he considered the opportunity to gloat over his neighbor's loss to be worth more than two cows (or a bag of gold) for himself.
Horowitz adds that in cases where conflict leads to ethnic war, there is an additional motivation beyond the contest for dominance: anxiety-laden (that is, exaggerated) fears of group extinction. Such fears tend also to be based on demographic fears and a history of domination by the rival group.42 Horowitz also agrees with the primordialist view that ethnicity has a "kinship with kinship": ethnic extinction matters, even if one's personal safety is not imperiled, because the future of the kinship group is imperiled. The fear of group extinction, Horowitz argues in sum, leads to feelings of hostility, and then to group violence. Young adds that the atmosphere of hostility and threat is likely to increase group solidarity, encourage the groups to perceive events in ethnic terms, and promote mis-perceptions across group boundaries.43
The insight that the motivation for ethnic war is in part emotional— interethnic hostility resulting from fear of ethnic extinction—helps fill in some of the logical gaps left by rationalist theory. If the issues at stake are defined not merely by economic or linguistic interests, but by a contest for status defined as domination and by anxiety-laden fears of extinction,
[26]
The Symbolic I'olitw- «11 llinu War
ílu n il is ť.isv to sec why attitudes would be hostile—that is, why groups prefer to weaken or harm other groups even at some cost to their own material welfare.
I Ix-se insights do not, however, form a coherent theory. Building them |Rto one requires the addition of a theory of choice that incorporates the ■motional nature of ethnic bonds. Murray Edelman's theory of symbolic pniilies offers a way to do so.
A Symbolic Politics Theory of Ethnic War
symbolic choice vs. rational choice
K.itional choice theory is based on several fundamental assumptions. Two ol them are that people have stable, ordered preferences; and that in choosing they try rationally to maximize their utility as defined by those prefer-tnces. Those assumptions are useful for some purposes, such as explaining most kinds of economic behavior, ordinary diplomacy, or interest-group politics. However, as symbolic politics theory points out, those rationalist rtssumptions are often false.
Concerning the first assumption, psychologists Irving Janis and Leon M.mn argue that people are "reluctant decision maker[s]" because deciding is stressful: people find it hard to handle complexity—especially the . i unplexity of trading off incommensurable values—and they fear that their decision might be wrong.44 Therefore, people frequently do not make the tradeoff decisions. Because different criteria lead to different prefer-ences, the result is that people's opinions on complex issues are often un-Itable. For example, polls in 1966 found majorities of Americans in favor both of President Johnson's policy in Vietnam, which involved escalating 1 he war, and of the idea of de-escalating the war.45 In this case, what determined people's views was how the issue was framed.46 If the question was one of support for the president, people were inclined to go along. If the question was about what to do in Vietnam, people were inclined to favor de-escalation. The reason for the uncertainty was partly the difficulty of making a tradeoff in incommensurable values: am I sure enough that I dislike this policy to disagree with the president on it?
Regarding the second assumption, psychologists argue that when people choose—and especially when they choose to act—they often do so emotionally rather than rationally. As Susan Fiske and Shelley Taylor put it, "emotion commits one to action more than does the cost-benefit calculation of intellective cognition."47 Thus people are more likely to participate in a protest rally or write to a politician on an issue that angers them or otherwise stirs them emotionally—an issue that stirs a "hot cognition"—than on an issue that only involves their material interests.48 This finding is important because of a
[27]
Mtxlern lltttml-.
second effect of emotional decision-making: "emotions...divert people from pursuing one goal and point them toward pursuing another goal that has meanwhile increased in importance."4'' Donald Kinder has found, for example, that Americans' feelings of anger against Saddam Hussein during the Persian Gulf War correlated with fewer qualms about the civilian casualties caused by U.S. action, and with greater support for having U.S. troops march to Baghdad to oust Saddam.50 Emotions change preferences: in this case, anger promoted support for hawkish policies.
Interestingly, even rational choice theorists are inclined to agree that emotions help determine decisions. Thus Samuel Popkin, whose book is entitled The Reasoning Voter, notes that "data presented in an emotionally compelling way may be given greater consideration and more weight than data that is statistically more valid, but emotionally neutral."51 This insight accords with psychological research: for example, one study subtitled "Preferences Need No Inferences" shows that emotional judgments are quicker and stronger than cognitive judgments, people are more likely to remember emotionally keyed information than purely cognitive information, and people may have affective responses to something—liking or disliking it—without even recognizing what it is.52 Even more important, studies have shown that attitudes which originally formed emotionally are most responsive to emotional appeals.53 This is the heart of the matter: for some kinds of decisions, people are most likely to base their decisions on emotion.
The core assumption of symbolic choice theory is therefore: people choose by responding to the most emotionally potent symbol evoked. According to Murray Edelman, who originated the approach, symbols get their meaning from emotionally laden myths. Myths, as mentioned above, have the role of giving events and actions a particular meaning—typically by defining enemies and heroes and tying ideas of right and wrong to people's identity. Facts, from this point of view, do not matter—either they are redundant, confirming the myth; or else they contradict it and are rejected. To illustrate the point, Edelman uses the example of American attitudes toward the Vietnam War.54 To those who believed the myth of "America the righteous," the meaning of American military action was a stand against communist aggression, and facts that did not fit that image—American misdeeds of various kinds—were ignored or rejected. To others, the war was symbolized by the napalming of children. They therefore accepted the myth of "Vietnam the victim," the ultimate expression of which was Jane Fonda's visit to North Vietnam, and rejected or downplayed any evidence of North Vietnamese misdeeds.
From this point of view, then, political choice is mostly emotional expression, politics is mostly about manipulating people's emotions, and symbols provide the tool for such manipulation. As a practice, "symbolic politics"
[28]
The SytitboUt Politic* of I thnlc War
i-hts to any sort of political activity focused on arousing emotions rather than addressing interests. As anthropologist Zdzislaw Mach aptly sums it up, in politics, "symbols are... selected and combined so as to achieve a de-iii J state of people's minds; to appeal to values, to refer to ideas, to stir ■ in, itions and to stimulate action."55 Politicians manipulate symbols—wave flflgs, refer to heroes, kiss babies—in order to induce people to make choices bused on the values they are promoting (which are evoked by the chosen jmnbols), or to associate themselves with those values.56 Thus politicians KMpporting a war evoke symbols of the nation's greatness and demonize the iminy, while those opposing it try to evoke sympathy for the casualties and reverse the hawks' identification of heroes and villains.
Symbols are so potent because they have both cognitive and emotional pífects.57 Thus when abortion opponents use the imperiled fetus as the symbol of the issue, they are both framing the issue cognitively as one of me life of the fetus and trying to elicit sympathy for what they consider an 1111 u 'cent and vulnerable baby Ethnicity is a rich resource for politicians en-ttged in symbolic politics because it is so emotionally laden. Ethnic groups by definition have myths of shared history, common heroes, and common kinship, as well as symbols that evoke those myths. Furthermore, a threatened ethnic symbol can be used to tap a number of values and emotions simultaneously—especially fellow-feeling among those in the group, shared feelings of superiority over and threat from the out-group—in addi-Imn to perceptions of conflict of interests. In short, Edelman argues, symbolic appeals create around conflicts of interest a myth of struggle against "hostile, alien, or subhuman forces" as a way to mobilize support.58
the symbolic politics of ethnic war
The central assumption of symbolic politics theory—that people make political choices based on emotion and in response to symbols—fits closely with the psychologically driven understanding of ethnic war suggested by the work of Young and Horowitz. Horowitz argues that emotions such as fear of group extinction are what drive ethnic violence; Young emphasizes (he importance of stereotypes (myths) and symbols in sustaining identity and driving group mobilization. And if ethnicity is an emotional bond evoking kinship feelings, then emotional appeals to that bond should be the basic mechanism by which ethnic mobilization works. Research also shows that people experiencing any negative emotion are more prone to feelings of anger and aggression if ideas justifying anger and aggression are brought to mind.59 Therefore, if emotional appeals to ethnic themes are simultaneously appeals to ideas that lead one to blame another group, those appeals are apt simultaneously to arouse the feelings of anger and aggression most likely to motivate people to want to fight.
[29]
I III VI
I'ul together, these ideas create a convincing picture of the psychology of ethnic war. On the individual level, appeals to emotionally laden ethnic symbols are what motivate people to participate in ethnic movements, and appeals to myths blaming other groups are what make people feel aggressive and motivate them to fight in ethnic wars. Those emotional appeals short-circuit the complicated problem of making tradeoff decisions because they encourage people to put ethnic issues ahead of other concerns. At the same time, the social psychology of group membership reinforces these processes for the group as a whole. These processes lead group members to want to gain status relative to other groups, encouraging them to pursue dominance over other groups. Additionally, feelings that the ethnic group is like a kinship group make people willing to fight to defend their group, especially if they believe that the group is truly threatened with extinction.
According to symbolic politics theory, then, understanding whether people will engage in ethnic violence requires that we examine the myths and prejudices that determine which symbols are likely to move them, and what evokes their greatest collective fears.
necessary conditions for ethnic war
Precisely when, then, does ethnic war occur? The key necessary conditions are:
Myths justifying ethnic hostility According to symbolic politics theory, people respond to ethnic symbols and mobilize for war only if a widely known and accepted ethnic myth-symbol complex justifies hostility to the other group. The myths justify hostility if they identify a territory as the group's homeland which must be defended and dominated politically and define a mythical enemy with which the other group can be identified. Chauvinism—the belief that one's own group is superior—is typically part of the motivation for the goal of dominance. And if the group's identity includes a warrior ethos, as the Chechen mythos does, for example, that group is likely to be more prone to ethnic violence.60 In this book, I look for such hostile myths in the key themes in each group's mainstream history texts written before the conflicts began, as well as in dissident sources.
Myths can be and sometimes are recast by chauvinist elites, of course, but this process takes a very long time unless it builds on a myth-symbol complex already made familiar by previous cultural entrepreneurs. Jack Snyder points out that these myths are often molded and propagated by national governments as a way of gaining nationalist legitimacy.61 In Yugoslavia, for example, the long-standing myth of Serbian martyrdom at the hands of Muslims in the Battle of Kosovo was propagated by the Serbian state in the
[30]
nineteenth and twentieth centuries, then supplemented and transformed by Hi. fear-invoking slogan "Only unity saves the Serbs," which formed the hrtnis for justifying hostility against Bosnian Muslims and Kosovo Albanians in 11 u ■ lyHos and 1990s.
I here does not have to be such a well-developed myth-symbol complex (in both sides. Young begins his book with an example of ethnic violence (a . i inflict between Mbala and Pende in Zaire in 1962) in which one group was nl.iin cl\ well-defined, but the opposing side was essentially a loose coali-lnhi of groups that did not long exist as a separate identity. This defines one 1 il the minimum necessary conditions for ethnic war: the existence of at least ime group with a myth-symbol complex justifying the pursuit of ethnic dominance (and thus hostility to any who oppose it), and the existence of mother group or coalition bound together in opposition to the first group.
Ethnic fears
A fundamental factor causing ethnic conflicts to escalate to war is that first tine side, then eventually both sides, come to fear that the existence of their group is at stake. Such extreme fears justify hostile attitudes toward the other group and extreme measures in self-defense, including demands for political dominance. These fears may be anxiety-laden—that is, exaggerated by emotion and by ingroup-outgroup psychology. For example, I lorowitz quotes Nigerian Hausa, members of the country's largest ethnic gmup, as expressing fears at the time of their ethnic war of being swamped" by minorities from the south.62 These psychological tendencies txplain the power of ethnic fear to motivate ethnic mobilization and murderous violence.
The source of such fear is typically the group's myth-symbol complex, portraying the in-group as peculiarly under threat or peculiarly victimized. In these cases, the more the group's historians emphasize the group's past victimization, the more credible are the emotional charges of genocide that arouse gut-level fears and the more appealing are hate-filled cries for vengeance.63 Indeed, in most cases of ethnic war, at least one group has been historically dominated by the other,64 causing fears of ethnic extinc-iion to appear more plausible to the previously dominated, while giving the previously dominant reason to fear revenge. Sometimes both groups have had both experiences: Azerbaijanis fear a return to the Armenian minority's nineteenth-century dominance, while Armenians in Azerbaijan resent twentieth-century Azerbaijani rule.
Demographic threats may also motivate ethnic fears, most insidiously in cases involving an "ethnic affinity problem" in which the minority in a country (e.g., Israel's Palestinians) is the majority in the broader region.65 The effect of such a situation is that both groups, by viewing borders differently, can think of themselves as potential minorities in danger of ethnic
[31]
extinction. Mixed settlement patterns also may contribute to ethnic fear,66 but they only cause a security dilemma if both communities are threatened. If the minority (e.g., Bombay's Muslim community) is seriously threatened, but the majority (Bombay's Hindus) are not, the result is pogroms instead of a security dilemma.67
Myths play a key role in the interpretation of these factors, however. Histories of domination can be invented, as the Serbs do when they identify the Albanians of Kosovo, who never dominated Serbia, with the Ottoman Turks, who did. More generally, ethnic histories always contain episodes of peace and of war; which is "traditional" depends on tradition—that is, myth or interpretation—rather than on actual past events. Some groups that are genuinely threatened by demographic trends (e.g., the Latvians) may not resort to violence at all, while other groups (e.g., the Hausa) may distort demographic trends to invent a danger that does not exist. Finally, mixed settlement patterns may contribute to ethnic understanding rather than violence; while ethnic war (as in southern Sudan) may occur where groups are geographically separate. What determines a group's response to its situation is its mythology—which determines its expectations—more than the situation itself.
Once ethnic fears become prevalent among the members of any ethnic group, for whatever reason, they justify and motivate a resort to violence in self-defense. Such fears are a necessary condition for ethnic war because people are much more concerned to avoid loss than to pursue gains, so they are usually mobilizable only when confronted by some threat.68 This is why leaders of nations, even when they launch aggressive wars, always justify their actions by claiming that it is aimed at averting some mortal danger. Even the Holocaust was justified by an ideology that Jews were not only inferior, but evil and dangerous.69
Opportunity to mobilize and fight Another requirement for ethnic war is opportunity: ethnic groups must have enough freedom to mobilize politically without being stopped by state coercion.70 Effective policing can prevent violent episodes from escalating, and political repression can prevent ethnic leaders from articulating their demands and mobilizing their followers for conflict. Therefore, as long as a state maintains an effective apparatus of repression and uses it to suppress ethnic mobilization, large-scale ethnic violence cannot occur. Since the relaxation or weakening of political repression opens up political space for all sorts of political entrepreneurs, such relaxation can make ethnic violence more likely. The way Mikhail Gorbachev's policy of glasnost provided the political space for ethnic entrepreneurs to mobilize in the Soviet Union epitomizes this possibility.
(32)
/ lw SifiiiMic Politic* of l.llmk War
HI course, il it is the leaders ol the slate who want to start ethnic violin lhe\ have the opportunity as long as they are in power: Rwanda's lliiin government in iou.i illustrates the point. If the result is to be ethnic Win, however, both sides must have the opportunity to organize and arm iii. in lelves. Such opportunity requires a territorial base: if one side has no ki-.r either inside the disputed area or across .1 friendly holder - where it . .111 organize its army, it cannot fight. If one side, usually the state, has an <>\ 1 1 whelming military advantage, the result is ethnic cleansing, genocide, nr more limited riots rather than war.
When foreign patrons are the key cause of violence, they act by changing, the opportunity structure, usually by offering material assistance to one side.71 Foreign countries or groups cannot directly create mass hostil-1. or extremist elites, but they can provide money, advice, and propa-. m11 la support to help extremist elites mobilize politically and promote • limn hostility. They can also provide money and arms to enable chauvinist elites to initiate violence using the few fanatics the elites can mobilize at in .i Moldova's Transnistria conflict, as I show in chapter 5, is an example I 1 this potential, with Russia playing the role of the third party. Third par-lies i .111 also, of course, act to discourage violence; those possibilities are iIim ussed below.
11 insnistria and southern Sudan are examples of another point con-l .ruing opportunity: ethnic war does not require the existence of government institutions uniting each conflicting group. While groups in conflict Often take over existing regional governments where they can, they can 1 ie,ite such institutions if necessary (as the Transnistrians did), or they can operate simply as a loose coalition of insurgents (as do the rebels of southern Sudan).
A final related point concerns level of economic development. Jack Snyder has argued that ethnonationalist mobilization is most likely at levels of per capita GDP between $1,000 and $6,000 in 1985 dollars. Below $],ooo in income, he argues, people tend to mobilize as part of patronage networks rather than as ethnic groups; while above $6,000 the growth of an educated middle class damps down tendencies toward ethnonationalist extremism.72 This correlation does seem to exist, but both the trend and the exceptions to it are best explained in symbolic terms. Purveying a strong nationalist mythology generally requires an effective educational system associated with at least moderate levels of economic development. Sometimes, however, as in Rwanda and Burundi, strong ethnonationalist mythologies exist at lower levels of economic development, and these are the exception to Snyder's rule. At higher income levels, prosperous middle-class groups in democratic states are not usually inclined to pursue ethnic violence—but in cases where ethnic myths and
[33]
are strong, as in Northern Ireland and Spain's Basque region, they do resort to violence.
processes leading to ethnic war
If the necessary conditions for ethnic war are myths, fears, and opportunity, the timing of war is explained by an increase in fear, opportunity, or hostility justified by the myths. Hostility and fear rise as a result of symbolic events that activate the myths, such as a violent episode that appeals to ethnic stereotypes (e.g., a murder or beating blamed by one group on another); a leader explicitly manipulating symbols (e.g., waving a flag); a threatening shift in political power (e.g., the election of a minority or extremist candidate); or even the emergence of new information (e.g., publication of a census showing a group's population decline). The alternative possibility is that fear and hostility are already high, but a new political opportunity emerges (e.g., emergence of a reforming leader).
The symbolic politics theory holds that if the three preconditions— hostile myths, ethnic fears, and opportunity—are present, ethnic war results if they lead to rising mass hostility, chauvinist mobilization by leaders making extreme symbolic appeals, and a security dilemma between groups. Different kinds of triggering events work by activating either the hostility or chauvinist mobilization. When myths, fears, and hostility are already strong, a new opportunity and a galvanizing event allow a powerful mass-led ethnic movement to emerge. Such movements spur politicians to seek support by making chauvinist symbolic appeals, goading mobilization even if the government opposes it; if the result is a security dilemma, war follows. Other conflicts are elite-led, in which a few powerful elites, typically government officials, harness ethnic myths and symbols to provoke fear, hostility, and a security dilemma and mobilize their group for violence. In either case, war results from a vicious feedback loop in which hostility, extremist symbolic appeals, and a security dilemma all reinforce each other to spur violence (see Figure 2.1). If any of the three processes is absent, however—if hostility rises but politicians avoid extremist appeals, or if the population resists such appeals, or if the sides' demands do not cause a security dilemma—war can be avoided.
The way an ethnic security dilemma works is most similar to Jack Snyder's concept of the "imperialist's dilemma" or to what Robert Jervis has called a "deep security dilemma."73 In contrast to Jervis's more commonly cited formulation, in which the sides prefer not to fight but feel driven to do so by insecurity, the sides in an ethnic security dilemma are openly hostile and are perfectly willing to fight. Violence is not their first choice—they would rather get what they want peacefully. Yet, what each group does to pursue its own security-defined-as-dominance is so threatening to a rival
[34]
group th.it tin- rival increases its senility demands in ways threatening to the first. The tragic element is that the groups are spurred by their mythologies to define their security in mutually incompatible ways when more modest definitions on both sides would permit mutual security instead of a spiral of insecurity and violence.
A peculiar feature of the process leading to ethnic war is that it is a process of positive feedback: all of the causes reinforce each other in an escalating spiral of violence.74 Hostile myths and attitudes are what make chauvinist politics possible, but symbolic appeals to those myths evoke emotions that make attitudes still more hostile. Feelings of insecurity also encourage hostile attitudes, but hostile attitudes on one side cause a security dilemma (and further hostility) for both. Finally, chauvinist political programs promote armed mobilization that can lead to violence; but the violence feeds back to make the chauvinist political programs more popular. This is a key to understanding ethnic war: because all of the causes reinforce each other in an escalating spiral or positive feedback loop, events need not happen in any particular order. The causes are universal, but the paths to ethnic war are multiple.
The existence of these elements—hostile myths, fears, opportunity—is a more-or-less rather than a yes-or-no proposition. The more they are present, the more conflict and violence will be centered around ethnicity and the worse that violence will become. If government breakdown leads to violence but ethnic mythologies and identities are weak, then the violence is likely to be organized around nonethnic coalitions. Also, because the process operates by feedback, different initial mixes of these ingredients can produce war: very strong hostile myths may need very little political opportunity to cause war; while weaker myths may require reinforcement by political leaders before they can serve for mobilizing a group to fight.
Mass-led violence scenarios Mass-led paths to ethnic war begin with opportunity—the lifting of some previously existing barrier to ethnic self-expression, usually the coercive force of a state—or else with some galvanizing event like a highly publicized murder. In these cases, the other necessary conditions—especially myths justifying ethnic hostility and ethnic fears—are already significant, and nationalism is therefore already the central value of dissident politics. In other words, there are already relatively large numbers of fanatics and true believers in the population.
In these circumstances, long-standing myths justifying group solidarity and identifying threats to group survival suddenly start being articulated publicly. With hostility already high, mass ethnic nationalist movements can spring into being almost overnight, either de novo or by pressing previously apolitical organizations into action as vehicles for ethno-nationalist mobi-
[36]
.....mi A tipping process works to promote that mobilization as the im-
HU'ving s.ifetv and prospects for success motivate more and more moderate iMllon.ilists to join 111. Intr.iethnu politics then becomes a competition in na-11..11.ilr.1 symbolism, in which elites in each ethnic group find themselves »oiiipiling to establish their ethnic bona fides and gain adherents by promot-1 nuine" nationalist goals—including, if their group fears extinction, Hi. Miliordination of other ethnic groups to their own.75 A security dilemma llurally follows from the pursuit of such goals, creating a feedback loop that increases fear, hostility, and extremist symbolic politics on both sides.
There is, however, no single path which all mass-led conflicts follow to . ilmic war. In some cases, hostile masses reward elites proposing chauvinist platforms, thereby creating a security dilemma (because the platforms threaten other groups) and leading to violence. This is the likely pattern in ■  Hi popular chauvinism, when the group mobilizing first is the majority, ** occurred in Georgia (see chapter 4). In other cases, masses engage in widespread but unorganized violence that first creates a security dilemma .11 lI only then leads to the replacement of existing leaders with extremists intent on implementing chauvinist policies. This pattern is more typical for irpressed minorities—cases of mass insurgency such as the Karabagh con-ihct (see chapter 3). Either way, the first episodes of violence play an important role because they provide vivid evidence supporting previously inchoate fears, and later serve as a symbol of the threat to the group. The • ll notional impact moves opinion by reinforcing existing myths, often turning latent chauvinism into open hostility.
Elite-led violence scenarios Flite-led violence involves yet another set of paths. In these cases, leaders motivated either by ideological zeal or by opportunism mobilize their group for ethnic war in pursuit of their own goals. They use the propaganda resources of modern political organizations and mass media to manipulate ethnic symbols and fan ethnic hostility, identifying outgroups with enemies from group mythology and highlighting the "threats" they pose. Thus minor demographic changes can be redefined as mortal threats to group survival, ancient disasters can be recast as current threats, and violent methods can be promoted as the only alternative to group catastrophe.
Additionally, the power of leaders to define the political agenda and control negotiations makes it possible for them to block any potential compromise simply by being intransigent. Blaming the stalemate on the other group then allows them to discredit opponents who promote moderate programs. Eventually, the extremists can organize militias or armies to launch violent provocations which begin a cycle of violence—radicalizing opinion and creating symbols for future use. If the other side responds in kind, a security dilemma spiral fed by violent propaganda takes off. This is
I [37]
mioaern Hatreds
most likely lo occur in cases of gowrnment jingoism—epitomized by Slobodan Milosevic in Serbia (see chapter 6)—in which top government officials are the culprits.
Also possible is a process of elite conspiracy, in which low-level elites or guerrilla leaders, typically aided from the outside, initiate a similar process. The case of Moldova's Transnistria conflict (see chapter 5) exemplifies this pattern.
explaining atrocities
The above logic can explain why people are willing to fight in ethnic wars: because they are frightened, and because they become convinced that their group's political dominance is essential to group survival. Such thinking can logically justify killing, and even massacre in extreme cases. Atrocities, however, require something more. While I cannot present a complete theory of atrocities here, it is possible to suggest an explanation based on the logic of symbolic politics.
Some argue that atrocities occur because they are in the interests of the people who commit or order them. That explanation, however, is not adequate. First, atrocities are not very useful, even if one's goal is ethnic cleansing. The Serbs in Bosnia, for example, were in many cases able to use simple threats of murder, accompanied by a few exemplary killings, to accomplish this aim. Dismembering victims or torturing them adds little to the terror created by threats of murder. Second, atrocities do attract moral opprobrium from third parties: all groups in conflict want outside allies and sympathizers, and engaging in atrocities is undeniably costly in driving away potential supporters. Atrocities, in short, are not a useful policy tool.
The symbolic politics theory would suggest an explanation based less on logical than on psychological factors. If the point of ethnic symbolism is to engage supporters' emotions, and the point of such symbolism during violent conflict is specifically to encourage aggressive emotions, it stands to reason that some proportion of people will react extremely strongly to the aggressive symbolism and express it in extreme ways. At the same time, even atrocities have to have a normative basis, which should consist of two components: a mythical belief that the opponent tends to engage in atrocities and a normative view that retaliatory atrocities are morally acceptable. The key is the last part: ethnic violence is always defined defensively, by the claim that the other group is trying to take away what is "rightfully ours"; atrocities have to be justified by the claim that committing them is a legitimate way to defend what is "rightfully ours."
To be sure, not everyone who engages in atrocities has to accept such beliefs. But some must. Even if atrocities are ordered by political leaders and
tililotveil both by peer pressure and by government sanction, at least some • 1 1 lie people carrying out the orders must believe in their legitimacy if IIv are to exert peer pressure. This is the key point Daniel Goldhagen mi a . about the Holocaust in Hitler's Willing Executioners: while obedience i<> Authority, peer pressure, and other factors do play a role in making such li hn\ n>r systemic, these motivations are not enough to explain cases in u In. 11 people take the initiative in committing atrocities, or in which they v the opportunity to evade such "duties" but carry them out anyway.76 I'vei \ 1 'lea of "they made me do it" must come attached to a case in which Nonieone willingly applied the coercion.
An alternative explanation for atrocities, and for ethnic wars in general, I* llut they are carried out primarily by common criminals, thugs who like Inn ling people and find in such conflicts an excuse to do so. There is plenty 1 il evidence to support this proposition.77 It is not, however, a complete explanation: behind most warlords employing such thugs (such as Serbia's Ai'k.in) is an ethnic nation and leadership (the Serbs and Milosevic) justifying their behavior in terms of national defense. Furthermore, many atroci-lles are carried out by otherwise ordinary people. Thuggishness explains 11 ime of the atrocities, but the motivating force of ethnic symbolism is nec-■  .ary to fill in the rest of the puzzle.
Policy Alternatives for Conflict Resolution the modest requirements for ethnic peace
One argument of this book is that it is not easy in contemporary times to get people to fight ethnic wars, and it is even harder to get them to commit atrocities. Ethnic war is possible only in the presence of hostile myths, opportunity to mobilize, and fear of group extinction, and it breaks out only if these factors create mass hostility, a within-group politics dominated by extreme nationalist symbolism, and a security dilemma between groups. If any major ingredient is missing, ethnic war cannot occur. If a group's myth-symbol complex encourages cooperation with other groups rather than domination or defensiveness—if, that is, people are reasonable and moderate—mobilization for ethnic war cannot begin because extremist politics is not rewarded. If people are fearful but not extremely so, they are unlikely to resort to violence. And even if people are hostile and fearful, if elites can moderate political demands and restrain violent popular impulses—avoiding extremist political symbolism and preventing the emergence of a security dilemma—violence can still be averted. Ethnic wars only happen when the attitudes of elites and masses are aligned in hostility.
I38J
[39]
conflict resolution tools
Given these three key causes of ethnic war—mass hostility, extremist politics, and a security dilemma—a comprehensive approach to conflict resolution would require attempts to ameliorate or prevent all three, either before or after violence breaks out. One way to restrain extremist politics is through peacemaking—pursuit of intergroup negotiations and cooperation, whether through mediated talks, building consociational institutions, or some other device.78 If the problem can be reduced to a classic security dilemma—that is, a situation in which the parties can agree on a preferred outcome but do not trust each other to implement a compromise—then what is needed is reassurance, perhaps including peacekeeping. If, however, the core of the problem is mutually incompatible security requirements based on hostile myths and fears of extinction, then neither peacemaking nor peacekeeping can lead to a resolution. What is required first in these circumstances is peacebuilding—efforts to bring the groups (not just their leaders) together to change their hostile attitudes so they can revise their understanding of their security needs, thereby making peacemaking possible and peacekeeping less necessary.79 Peacebuilding efforts, I will argue, are unwisely underestimated in conflict resolution practice: while vitally important to make conflict resolution work, they get little attention and few resources. Lack of attention to peacebuilding undenriines conflict resolution efforts.
Peacemaking
For leaders faced with managing highly emotional ethnic conflicts, skill and sensitivity in building coalitions, negotiating agreements, and calming volatile passions may make the difference between success and failure.80 Some political systems may be more likely than others to produce such leaders: the communist systems studied in this book did so only occasionally. The fact that Gorbachev was the first Soviet leader since Lenin to lack leadership experience outside Russia—and lacked therefore the opportunity to develop skill in ethnic conflict management—may be an important reason for the Soviet Union's breakup.
For conflicts that have already exploded into violence, a growing literature explores different modes of peacemaking and different strategies third parties can follow to mediate or broker settlements for ethnic conflicts. One key insight of this literature is that a conflict is "ripe for resolution" only in certain circumstances: if leaders on both sides want an agreement, can find a mutually acceptable formula, are strong enough to "deliver" their constituencies, and have a mutually acceptable process for negotiation. Typically, a mutually hurting stalemate is required for such conditions to emerge.81
In this context, there is a long list of things third-party mediators can do, most of which are relevant before violence breaks out as well as after.82
[40]
Vht Symfcrf* Politic* of 1 thnii Wtu
-1.. 11.1 lots can create a suitable process by convening negotiations, offering A venue, establishing an agenda, and so on. They can suggest possible com-l finises, propose a formula for resolving the conflict, or set up a concilia-
.....1 « ommission to design one. They can make agreement more attractive
l>\ ottering inducements—resources for reconstruction, rewards for conces-■iioii'. -and by emphasizing the costs of failure to agree. Mediators can also lie 1 to verify and guarantee the settlement. They can even help "deliver" 11..  leaders' constituents by conferring international legitimacy on the
1......       . and the outcome—and on the leaders who reach it. More generally,
l»y gaining the parties' trust and helping them communicate, mediators can In |p the parries reach understandings they might have missed unaided, in 1 hi I < v providing relatively unbiased information. If the parties prove recal-. in .nit, mediators can try to coerce them by threatening to impose economic •.,1111 111 ms, to aid the other side, or to intervene militarily. Finally, by drawing • 11 international experience, mediators can help structure an agreement to in Ip reduce the dangers that it will collapse in implementation.83
There is one problem with this approach, however: it usually does not work for resolving ethnic wars. One study shows that of post-World War II identity civil wars" that have been settled in some way, over 70 percent were settled by one side's military victory, not by negotiations: and in two-Ihlrds of those cases in which a negotiated settlement was reached, the set-iliinents collapsed and war resumed.84 Ethnic wars are hard to resolve through ordinary negotiations.
Reassurance
When two sides' goals are compatible but mutual mistrust is an obstacle to resolution, the sides can try to reduce the mistrust with reassuring moves: 1 onfidence-building measures such as military reductions or withdrawals; Agreement on norms regulating competition; or implementation of strategies such as graduated reciprocation in tension-reduction.85 The most powerful tool available is for a leader to make a highly visible symbolic gesture ol reconciliation, such as Egyptian President Sadat's famous trip to Jerusalem, which started the process that led to the Camp David accords, [he trouble, of course, is that by becoming himself the symbol of peace, the leader becomes vulnerable to retaliation by extremists: the assassinations of Sadat and Israeli Prime Minister Rabin are the measure of the incompleteness of Arab-Israeli reconciliation.
For third parties, the most effective tool of reassurance is peacekeeping, the nonviolent use of third-party armed forces to maintain peace among belligerents.86 In general, peacekeeping only works with the consent and cooperation of the key parties to the conflict: that is, if there is a peace to be kept, and the parties to the conflict want it to be kept, peacekeepers can help by preventing minor incidents between mistrustful belligerents from
[41]
escalating I" renewed war." Interestingly, this rather broad definition—a simplified version of the usual one—applies just as well to domestic policing, and indeed good policing can have this effect. If initial violent incidents play a critical role in provoking fear and motivating further violence, then prompt and fair punishment of violent actors by legitimate authorities can go far in calming fears and undercutting the motives for group violence. The conditions of ethnic warfare presuppose that authorities are unable or unwilling to accomplish this role.
Classical peacekeepers operate differently, by reminding the parties of their agreed obligations, reporting violations, and sometimes by inserting themselves between parties in conflict to prevent violence. In some cases— Macedonia in the mid- and late-i990s is an example discussed below— peacekeepers can be inserted preventively into a peaceful situation to help prevent the outbreak of violence. Peacekeeping has, however, two fundamental limits. First, when peacekeepers are inserted to maintain a ceasefire pending final resolution, the resulting stable ceasefire may remove incentives for the sides to resolve the conflict. The Cyprus conflict, where UN peacekeepers remained for decades in such a situation, is the clearest case. The second limitation is that peacekeeping only works where the parties want it to work: if the parties to a conflict are determined to fight, they can ignore or overrun peacekeepers. Making the parties want peace is the realm of peacebuilding.
Peacebuilding
Many theoreticians and diplomats tend to dismiss peacebuilding as either naive or ineffective, but doing so is a mistake. According to symbolic politics theory, the core causes of ethnic war are ethnic hostility and the myths and fears that promote it. What peacebuilders do is to bring people from opposing sides of a conflict together to replace the myths about the other side with better information, to replace the hostility and fear with understanding, and most of all to build cooperative interethnic relationships to replace stereotyped hostile ones.88 From the point of view of the diplomatic practitioner, such efforts among grassroots leaders can build a political constituency for the diplomatic peace process so leaders can "sell" a compromise settlement. For mid-level officials, it includes reconciliation commissions or "track II" diplomacy that helps officials who have access to top leaders to work creatively with the other group to create mutually acceptable formulas for conflict resolution. And by building cooperative intergroup relationships, it creates resources that can be drawn on in implementing a peace agreement once reached.
Peacebuilding is an extremely difficult undertaking that has rarely been attempted except on a small scale. It has, however, already shown some results: peacebuilding can change attitudes. The typical formula is to bring
[42]
The Si/Nibo/u /Wild •• i>! I Ihnu IVnr
Myopic Irom the groups in conflict together at a neutral site, often to live to-RHlicr tor a period of time and to discuss in detail the issues that divide 11ii 111 What they find time and again is that after heated arguments and I11il1.1l resistance, most participants—for example, Israeli and Palestinian lern.ieris (including stone-throwing intifadah participants)8'' or religious |i   Icrs trom northern and southern Sudan—come to an increased mutual I. 1 standing, and in some cases move on to creative efforts to help re-Dh v real conflicts. One of the proudest accomplishment of peacebuilding I* the 1Q93 Israeli-Palestinian Oslo Accords, which began as a "problem-mlving workshop" involving mid-level officials on both sides. In Mo-Bftnibique, a much larger grassroots effort including church-sponsored dialogues and a UNICEF-funded "Circus of Peace" helped create the environment in which the peacemaking efforts of top leaders could succeed.90 I lie biggest problem peacebuilding faces is the "reentry problem": when parti' ipants return to their polarized societies, they find few people recep-v to hearing their new insights. The best way to overcome this problem 1 1111 eiigh a coordinated set of peacebuilding efforts large enough in scale |D help a networked constituency for peace emerge.
Another set of possibilities involves efforts to recast ethno-nationalist myths into cooperative and tolerant ones, especially by promoting the w 1 it ing and teaching of fair-minded history instead of the ethnocentric and • »|>egoating kind.91 For third parties, this means attention by governments and international organizations to school curricula in multiethnic
I ountries and open criticism of and pressure on countries that teach hostile myths to their schoolchildren. Such myths, however, may not come exclu-ilvely or even primarily from history or social studies classes. On the con-n,try: since ethnic myths are primarily affective, they may be more effectively transmitted by literature than by history texts. Attention must therefore also be paid to the stories children read and the poems they recite.
II they learn to hate in school, it may be more likely to happen in literature 1 l.iss than in history class.
In the shorter term, attention can also be paid to efforts at countering ncapegoaring propaganda—the twisting of ethnic symbols into symbols of ethnic conflict by chauvinist ethnic elites. Here too the means is primarily 1 ipen criticism, in this case of chauvinist politicians. Chauvinists always try to equate their views with "true" nationalism; the challenge for outside critics is to make clear their disapproval of such chauvinists without appearing to oppose more legitimate sorts of nationalism. Doing so, indeed, cuts against the grain of symbolic politics, as it is easy for the chauvinist to ask: what right do they have to dictate the meaning of our national identity? Perhaps the best that can be done is to find ways to support the chauvinists' opponents in order to illustrate and demonstrate the benefits of choosing a more moderate leadership. This approach only works, however,
l43l
while the chauvinists are out of power; once they conn- to power, the country is on its way toward elite-led violence, and efforts to head it off become much more difficult.
Crisis management options Once conflict begins building toward violence, all parties' options for con- : flict resolution narrow dramatically. If the mobilization is elite-led by government leaders, one option is to offer inducements for peace—a package of political and economic concessions that might help the leaders maintain their power without having to resort to hostile ethnic symbolism. This might be reinforced by deterrent threats to support the other side in case of i violence, or by counter-mobilization of peace advocates. If the mobilization is a case of elite conspiracy by leaders of a subordinate group, the most relevant inducement would be co-optation—offering the opposition leaders limited power in exchange for loyalty. Other options include isolating them from outside aid or offering reassurance in the form of new evidence or incentives for government moderation. The trouble with all of these options is that once mobilization takes off, whether under the control of govern- 1 ment leaders or of well-entrenched insurgents, the leaders are usually too committed to conflict to be swayed.
Mass-led conflicts may be more amenable to crisis management—if government leaders want to avoid violence. Minorities and majorities are I likely to respond to different appeals, however. Mass insurgencies by subordinate minorities, whose fear of group extinction is likely to be relatively well-founded, primarily need reassurance, including from trusted police or peacekeepers, that they will be protected. In Macedonia in the 1990s, for example, United Nations peacekeepers were instrumental in providing such reassurance to the Albanian minority. Dominant majorities, who typically inflate fears of losing control into fears of extinction, are harder to reassure because dominance is the only sort of security their mythology allows. However, peacemaking efforts that grant the majority symbolic dominance while conceding the substantive requirements of security to the minority may be possible; if not, deterrence by the direct presence of superior external force may be the only alternative.
Applying the Symbolic Politics Theory
The symbolic politics theory I have proposed here is a general theory of ethnic war, applicable in any region of the world. Indeed, it is based largely on the ideas of Crawford Young and Donald Horowitz, who write primarily about Africa and Asia. Looking at Africa, for example, the argument that hostile myths and fears of extinction are critical causes of ethnic war holds
[44]
• Symbolic Politic* i't I Ihnu War B well. Sudan's long-running ethnic war is clearly rooted in the mutually antagonistic identity myths of North and South, and where the symbol of l.i'. iry" as the threat to the black southerners is based on a reality of slave 11.11 ling by Arab northerners that continued into the twentieth century.92 Similarly, Nigeria's ethnic war of the 1960s was driven by the Hausa-Fulani tear of group extinction, which was based on the stereotype of "pushy" •Miutherners. In Congo/Zaire in the 1960s, in contrast, most violence was Ideologically or regionally rather than ethnically based because myth-|) rnbol complexes, and therefore identities, were relatively weak and undeveloped. And South Africa simmered but did not explode in the 1990s in large part because Nelson Mandela and the African National Congress suc-1 essfully prevented white fears of group extinction from arising, hence avoiding the emergence of a black-white security dilemma (even as elite-led /itlu-Xhosa violence sparked a security dilemma among blacks in Natal).
The rest of this book is aimed at applying the theory to four cases in post-■ ninmunist Europe. I selected these cases in part to allow a "most-similar terns" comparison: I can control for the type of political institutions and lor political opportunity, which were similar in all cases, and focus on how nl her variables, such as ethnic myths and fears and leaders' behavior, • aused ethnic wars. Brief discussions of peaceful counterexamples illustrate how peace was maintained in similar circumstances when myths and fears were weaker or leaders more restrained. Additionally, because many popular theories of ethnic war were formulated to explain the wars in the former Yugoslavia, I want to challenge those theories on their "home ground," showing how the symbolic politics approach offers a more convincing explanation of that case, while also explaining the mass-led wars of the South Caucasus that rival theories cannot explain.
The purpose of these studies is to illustrate the value of the symbolic theory—to show how the theory generates fairly simple, yet different and convincing, explanations for the causes of violence each case. Strictly speaking, they do not represent a test of the symbolic theory because they mostly "search on the dependent variable"—the discussions of cases in which war was avoided are only brief and suggestive. This procedure is warranted, however, because the aim of this book is not primarily theory testing but theory development. The main case studies demonstrate the usefulness of the theory for explaining those important cases, while the brief studies of non-wars provide some suggestive evidence associating ethnic peace with the absence of key variables in the theory. That, plus the theoretical synthesis, is enough for one book; a full test will have to await future study.
One claim I do test is that some conflicts are mass-led while others are i-hte-led. How can they be distinguished? The critical variable is the attitude of incumbent government officials at the time that ethnic mobilization begins to gather strength. If mass mobilization occurs without government
■ [45]
istance, or even m the face of government efforts to repress it, then the nflict has to be mass-led. If incumbent government leaders are replaced by more nationalist figures as mobilization continues, that would be further evidence of a mass-led process. If, however, there is evidence that government leaders support mass ethnic mobilization from the beginning, then the mobilization is probably elite-led. Since the countries in the study were all communist countries with formal party/government control of the media, the attitude of the media is another indicator: media opposition to ethnic mobilization is evidence of a mass-led process; media support for it is evidence of an elite-led process.
The cases also represent a partial test of the theory in that they look for evidence that the key causes of violence identified in the theory are important in each case. The chapters pay little attention to opportunity, as the opportunities opened up by the death of Tito in Yugoslavia and Mikhail Gorbachev's policy of glasnost in the Soviet Union are already well-documented. They do, however, examine in detail the nationalist ideologies and myths in each case and the evidence of ethnic hostility and fear. The explorations of ideology and myth provide the context for examining each group's assertions of its interests and expressions of fear.
The cases also represent a "process-tracing" test, examining in detail how ethnic mobilization happened in order to determine whether the mechanism suggested by the symbolic politics theory actually took place. One key issue is the theory's contention that ethnic mobilization works by a process in which government leaders or dissident intellectuals manipulate ethnic symbols to attract support. By tracing the mobilization process in detail, I can determine whether people responded primarily to appeals to their tangible, individual interests, as rational choice theories assume; or whether they responded to invocations of emotive symbols while ignoring concessions to their material interests, as symbolic politics theory assumes. I also add brief discussions of atrocities to illustrate the plausibility of the symbolic politics explanation of such behavior.
The process-tracing also looks for evidence that ethnic hostility, extremist symbolic politics, and security dilemmas were all present in the cases, and that they all reinforced each other in a positive feedback process to cause the outbreak of violence. If the symbolic politics theory were incorrect, there would be evidence that security dilemmas were the result of groups' uncertainty about each other's motives rather than open hostility, and preparations for war would have been justified with the argument: "we must be prepared because we do not know what they will do." Finally, if the symbolic theory were incorrect, there would be evidence that leaders and followers preferred to avoid war and were willing to compromise to do so, but that structural factors and mistrust caused war to break out anyway.
The Symbolic PoMcs of Ethnic Mr
111' case studies support the symbolic politics theory. One of the striking findings of tins hook is the degree to which emotive symbolic issues—as-(M'lllons of group or language status, the design of flags, the treatment of "•ai nil'' territory—dominated more tangible issues in motivating partici-I mi   I eaders also refer again and again to the importance of emotion in nmhili/ing followers. People's tangible, individual interests—jobs, hous-education—are frequently irrelevant, and often distorted by ethnic ms lie. even when they are relevant. Security dilemmas arise not from un-i < i unity about the other side's intentions but from strong evidence of the • it I it ■ i side's hostility, usually matched by open hostility on one's own side. I In-, hostility always takes the form of political programs aimed at dominating the other group in the disputed region. Thus war results not from mistrust or miscommunication but from a clash of interests, openly stated
m.l defined as irreconcilable.
Measuring the motivations of the followers of chauvinist movements is, . •! • < >urse, difficult. When opinion polls on relevant issues are not available, (In- best that can be done is to look at what politicians, activists, and the Biedia were saying at the time. Since we know people mobilized, presum-llbly the themes and issues used at the time were the ones that motivated them. Future studies aimed at probing public attitudes and experimenting with their reactions to ethnic symbols will be necessary to prove or dis-I 'i * we definitively the value of the symbolic politics approach.
[46]
l47l