Genocide has been called the “crime of crimes” and an “odious scourge.”Footnote 1 With millions of victims in the last century alone, it is one of the great moral and political challenges of our age. Its significance has generated extensive research over the past fifteen years as the violence in Rwanda and Bosnia drove scholars to expand their focus beyond the Holocaust, which had long been the primary case study. Today, researchers across disciplines are conducting comparative genocide research, exploring its necessary conditions and patterns. This is evident in the development of several journals, textbooks, readers, encyclopedias, conferences, and professional organizations devoted to its study. Its policy importance is also apparent in the expansion of government agencies, non-governmental organizations, and a special UN office devoted to its prevention.Footnote 2
This new generation of scholarship has crystallized into the interdisciplinary field of “genocide studies,” a community of scholars and practitioners dedicated to researching and preventing genocide. However, genocide studies has emerged as its own research field, developing in parallel rather than in conversation with work on other areas of political violence. Aside from a few important exceptions, mainstream political scientists rarely engage with the most recent work on comparative genocide. Some of the newest genocide research appears in topic-specific conferences and journals like Genocide Studies and Prevention and the Journal of Genocide Research, but not in political science venues.Footnote 3 The reasons for this separation are complex, but partly stem from the field's roots in the humanities (especially history) and reliance on methodological approaches that have had little resonance in mainstream political science, as well as the field's explicit commitment to humanitarian activism and praxis. Earlier generations of political scientists and sociologists who studied genocide often found little interest for their work among dominant political science journals and book publishers; they instead opted to establish their own journals and professional organizations. Although the field has grown enormously over the past decade and a half, genocide scholarship still rarely appears in mainstream disciplinary journals.Footnote 4 Five years after Scott Straus's assessment of the field, there remain a number of underexplored areas where political scientists can make contributions.Footnote 5
This article outlines a research agenda for advancing theoretical knowledge about genocide. The agenda focuses on analyzing the conditions under which genocide is likely to occur, the multilevel processes of violent escalation and de-escalation, and the ways in which these processes are shaped by, connect to, reinforce, accelerate and impede one another. The agenda seeks to explain, in other words, variability in genocidal outcomes. The article is centered on a set of underdeveloped questions that political scientists are especially well equipped to address given our discipline's substantive and methodological advances on related topics (civil war, interstate war, state repression, etc.). I argue that scholars should 1) model elite and follower radicalization processes by disaggregating genocidal “intent” over time and space and exploring how it emerges, rather than taking it as pregiven. Doing so will permit researchers to 2) situate genocide research within a broader context of political violence to understand how they are related temporally (in terms of sequencing) and spatially, as well as decenter analytical domains beyond the standard country-level and single victim group to gain insight in the dynamics of genocide, including how perpetrator policies vary by group; 3) draw on recent advances in microanalyses of civil war to theorize about subnational patterns of violence diffusion; 4) move beyond reductive contrasts between ideology and rationality to analyze how ideologies frame the strategic choices “available” to genocidal elites.
While genocide scholarship has furthered the understanding of particular historical contingencies of genocide, much of it still lacks theoretical frameworks for analyzing dynamic changes in violence over time and across space.Footnote 6 Genocide research is only now gaining more interest in the political science discipline. Perhaps the most prominent approaches have been elite rationality theories. Benjamin Valentino, for instance, highlights the strategic rationality of radical elites. Rather than focusing on broad patterns of prejudice or deep ethnic cleavages, Valentino argues that genocide occurs when extremist leaders who consistently face challenges to their plans adopt increasingly radical policies, culminating in genocide. Genocide is thus an instrumentally rational tool of policy attainment. Manus Midlarsky also underscores the importance of strategic rationality, but drawing from prospect theory, he argues that genocide occurs when leaders are faced with significant losses, especially territorial losses, and thus turn to genocide for “loss compensation.”Footnote 7
I agree that elite strategic behavior is crucial for understanding genocide, but I argue that it is important to move beyond elite theories to systematically explore variation in genocidal violence within cases, something that is not addressed in macrolevel studies. Furthermore, I contend that subnational analyses can inform—and revise—elite theories by showing how micro- and meso-processes may be only loosely linked to elite policies. Finally, I question the sharp distinction between “rationality” and “ideology” often found in strategic rationality theories, and argue that the two concepts are not mutually exclusive but can be used profitably together in theory formation.
In these respects, the research agenda that I outline is close to Scott Straus's work in mapping subnational variation in violence in Rwanda.Footnote 8 At the same time, I argue for a robustly comparative approach to questions of variation. Such an approach can advance research in political science in several ways. First, by treating genocide as a process rather than an outcome, we can better model the contexts and scenarios in which genocide occurs, and thus understand what conditions and interactions may make genocide more likely. Second, placing genocide firmly in the domain of political violence significantly widens the cases available for comparative study and permits researchers to assess which causal processes are unique to genocide and which are not, advancing theoretical knowledge on various trajectories of the onset, diffusion and abeyance of mass violence. Third, this agenda's focus on subnational and microanalytical perspectives furthers our understanding of basic concepts in the study of large-scale political behavior, including “intentionality,” “ideology,” “rationality,” and “identity,” which are often reified in country-level research on violence with problematic consequences for causal theorizing.
In what follows I consider the four methodological issues listed above and then present ways to advance our theoretical knowledge. Given space constraints, I do not examine the growing literature on prevention models or preventive policy. Better methodology on the onset and diffusion of genocide can aid prevention by giving us clearer understandings of early indicators and accelerators of mass violence. But prevention also involves a host of additional factors that cannot be addressed here, including political will and third-party interests.Footnote 9
Comparative Genocide Research Problems: Definitions and Case Selection
The jurist Raphael Lemkin invented the term genocide in the 1940s. His definition was inclusive, and focused on “a coordinated plan of different actions aiming at the destruction of essential foundations of the life of national groups, with the aim of annihilating the groups themselves.”Footnote 10 Although influenced by Lemkin, the United Nations adopted in 1948 a more restricted notion of genocide to mean:
any of the following acts committed with intent to destroy, in whole or in part, a national, ethnical [sic], racial or religious group, as such:
(a) Killing members of the group;
(b) Causing serious bodily or mental harm to members of the group;
(c) Deliberately inflicting on the group conditions of life calculated to bring about its physical destruction in whole or in part;
(d) Imposing measures intended to prevent births within the group;
(e) Forcibly transferring children of the group to another group.Footnote 11
As the guiding legal definition, the UN conception of genocide ought to provide a point of common definitional agreement and it indeed serves as the definition in international criminal law. However, from a theoretical perspective, it suffers from a variety of conceptual problems. First, it designates a rather arbitrary distinction between victim categories, leaving out other historically targeted groups such as political and economic victims. Second, it fails to explain what constitutes the threshold of “in part” destruction; third, it leaves unclear the degree of preplanning and coordination necessary to satisfy the “intent” component, which has typically been defined explicitly. Finally, it includes a variety of actions that can qualify as genocidal (direct killing, serious mental or physical harm, and forcibly transferring children) but fails to address the commonalities that make them genocidal per se.Footnote 12
Scholars have responded to these problems in a variety of ways, with the goal of giving greater specificity and coherence to the concept of genocide. Several of these efforts have been especially sophisticated. Legal scholar David Scheffer places genocide along with crimes against humanity, war crimes and “ethnic cleansing” under the rubric of “atrocity crimes,” which constitute fundamental human rights violations of a large magnitude (such as mass killings) committed by a ruling or otherwise powerful elite and which are recognized in current international human rights law. For Scheffer, the priority of law is to develop systematic norms and doctrines to facilitate the prosecution of atrocity crimes.Footnote 13 Martin Shaw's sociological approach contends that genocide is not merely the physical destruction of a group, but also includes the destruction of group identity and represents a form of “degenerate warfare” where targeted civilians are treated as enemy combatants.Footnote 14 Mark Osiel and Larry May have developed complex philosophical accounts of what constitutes a “group,” the elements of destruction “in part,” and the place of individual culpability in large-scale, coordinated killing.Footnote 15 Nevertheless, there remains no consensus on the term genocide in the current literature.
Some scholars have adopted a restricted definition of genocide, focusing only on cases where extermination was driven by an explicit ideology of purificationFootnote 16 or targeted a specific type of group, such as an ethnic community.Footnote 17 Other scholars focus less on ideology as a bounding concept and explain large-scale extermination by focusing on the destruction of groups, regardless of victim identity or perpetrator motivation.Footnote 18 Others still have preferred to create complex categorizations of violence that include urbicide, politicide, classicide, democide, ecocide, femicide, gendercide, fratricide, cultural genocide, linguicide, omnicide, “ethnic cleansing,” murderous cleansing, and auto-genocide to analyze various violent phenomena that seem to have, in Wittgensteinian terms, family resemblances.Footnote 19 A non-exhaustive survey of the field counts over twenty scholarly definitions of genocide as well as many more cognates.Footnote 20 These definitional battles have preoccupied genocide scholarship for over twenty years.
Variation in genocide conceptualization means that scholars choose different cases based on their definitions, and thus their theories are difficult to compare. The “core” accepted modern cases of genocide include Armenia, the Holocaust and Rwanda; Cambodia, Bosnia and Darfur remain somewhat more contested, and many more cases fall on the boundaries (for instance, Bangladesh 1971, Biafra 1968–1970, Indonesia 1965–66, Tibet 1959, etc.). Variation is evident in some of the most sophisticated studies of genocide: Mark Levene provides a rich study of Vendée massacres, European colonial extermination, the Armenian and Jewish genocides, Stalin's attack on “kulaks,” and Cambodia and Rwanda, among others.Footnote 21 Ben Kiernan traces similarities across cases ranging from Sparta to Darfur, using the UN definition to connect all of these conceptually.Footnote 22 Midlarsky, however, uses a restricted definition of genocide and thus develops a theory that focuses on three contemporary cases: Armenia, the Holocaust and Rwanda, while excluding Cambodia and other cases of “politicide.”Footnote 23 Michael Mann provides an inclusive theory of “murderous ethnic cleansing” and genocide that includes genocidal democracies in the New World, the Ottoman Empire, the Holocaust, a variety of communist regimes, Yugoslavia and Rwanda.Footnote 24 Eric Weitz focuses on genocide and ethnic cleansing, and includes the Holocaust, Soviet kulaks, Cambodia and Bosnia, but not Rwanda, while Benjamin Valentino combines “mass killings” and genocide in his analysis of not only the Holocaust and Rwanda, but also Guatemala, Afghanistan, the Chinese communist revolution and other cases.Footnote 25 Given this variation in definitions, objects of study, and consequent outcomes, it becomes difficult to compare various causal mechanisms, processes and explanatory theories.
Definitional debates have been going on for several decades and are unlikely to stop. Nevertheless, I suggest it is possible to retain the analytical concept of genocide and move forward with comparative research. The element of “intent to destroy,” for instance, is common to nearly all scholarly definitions, but the presence or non-presence of intentionality has problematically served to delimit which cases are open to genocide research. I suggest below that by disaggregating intentionality we can bring genocide into the comparative study of political violence, to the mutual benefit of genocide scholarship and comparative politics.
1. Intent
The UN definition stipulates that genocide is an intentional act, not merely the byproduct of other policies or actions. The centrality of intent stems from the Holocaust, which was originally seen as the outcome of Hitler's preplanned obsession with eradicating Jews.Footnote 26 Later critics framed the Holocaust as the contingent outcome of numerous competing and increasingly radical bureaucratic strategies and policies that had only limited detailed and direct orders from the Nazi leadership.Footnote 27 These “intentionalist v. functionalist” debates were largely exhausted in the 1990s, and more recent research shows that the Holocaust was the result of a process of increasing radicalization, where the intention to exterminate Jews and others emerged to replace earlier policies of discrimination, expulsion, concentration and smaller scale murders that were perceived as “unsatisfactory.”Footnote 28 Unfortunately, this more complex notion of intentionality has been largely underplayed in general understandings of genocide. Genocide is still broadly understood to require rather explicit preplanning to satisfy the requirement of intentionality, reflecting an outdated understanding with roots in earlier Holocaust historiography and reinforced in human rights jurisprudence, where the standard of proof for intentionality remains high.Footnote 29 This latter influence is particularly unfortunate since the purpose of a legal definition is to identify and prosecute crimes rather than to provide a social scientific representation of complex violent phenomena.
Intentionality remains at the center of most genocide research.Footnote 30 Nevertheless, the assumption of explicit prior intentionality as the benchmark for genocide is problematic not only in the Holocaust but in other cases as well: sophisticated case studies on ArmeniaFootnote 31 and RwandaFootnote 32 have shown how elite interpretations of unfolding events, including war, radicalize state behavior in complex and dynamic ways. Preexisting explicit intent is often difficult to reconstruct and to connect to specific behavior; intent has proven so vexing analytically that some scholars have chosen to jettison it altogether and search instead for “relations of genocide.”Footnote 33
The analytical problem of inferring intentionality is less pronounced in other areas of political violence research (civil war, interstate war, state repression, etc.), where causal theories focus primarily on actor capacity and behavior, and only secondarily on intentions and motives. Methodologically, this makes analysis easier; the main observables are actor capacity and behavior, not intent. Indeed, the focus on intentionality has kept genocide studies unconnected to other research of mass violence. Given that intent often emerges over time and is hard to detect in “real time” (and often afterwards, as well), using it as the primary determinant for which cases to study and which to leave out inhibits comparative work and rather arbitrarily places genocide outside of the political violence literature.
What should be done? Scholars should not jettison intentionality, but instead employ a more phenomenologically complex conception in place of the reductive interpretations that are a holdover from early studies of the Holocaust and human rights law, and have served to isolate genocide research from work on political violence. In some cases genocide preplanning is evident, and in others there are only general pronouncements by leaders calling for violence against certain groups but little indication that a plan for intentional extermination, as such, exists. There is significant variation across cases and comparative work should take this into account. One way to address this is to focus on emergent intentionality, as I call it here, in order to address more complex conceptions of intent by emphasizing the contingent ways in which genocide develops, showing how leaders and their subordinates perceive and respond to supposed threats and changing conditions.
Analytically, emergent intentionality requires focusing both on perpetrators' capacity to inflict violence as well as their actual behavior. Behavior includes at least three dimensions: 1. level of lethality (to what extent violence is destructive rather than repressive of the group); 2. degree of coordination (how systematic, coordinated and sustained violence appears to be, such as with the use of similar destructive tactics in a wide area); and, 3. scope (the extent to which coordinated lethal violence is applied against all or a substantial part of a victim group). Note that such analysis addresses not only levels of coordinated lethality in general, but also group-directed violence—that is, violence aimed at destroying the group. Indeed, genocide is “group selective” (and not only combatant selective or indiscriminate) and “group destructive” (rather than only harmful, for instance).Footnote 34 To the extent that there are high and sustained levels of group-targeted lethality, intentionality is evident. Barring clear orders or statements calling for extermination, we can infer an intentional plan to destroy a group to the extent that violence becomes more lethal, appears coordinated and sustained over time, and targets an increasingly wider proportion of the victim group. In most cases, radical measures emerge over time; elites do not exterminate as a first choice. Rather, genocide develops as other strategies and policies are considered inadequate for addressing whatever “threat” leaders perceive. As other less violent strategies are discarded, the choices become increasingly more extremist. Elites may first seek to respond to a perceived ethnic or other “threat” through political, legal and social exclusion, and then perhaps expulsion, forced assimilation or selective massacres and other forms of repression. In the context of increased violence and general instability such as war, leaders may use more extensive and violent repressive measures, including larger massacres of targeted civilian populations and forced “deportations,” all the while reducing their reliance on lesser forms of repression. As these new efforts “fail,” leaders opt for more radical measures, moving from episodic massacres to widespread killings. This is still genocide. But the plan of extermination emerges over time and in the face of changing circumstances, as leaders and their followers interpret prior efforts as failures. The most recent Holocaust research shows the dynamics of emergent intentionality. The policy to physically exterminate Europe's Jews coalesced during the period between fall 1939, with the German occupation of Poland, and fall 1941, with the launch of the war against the Soviet Union. Prior to this period, the Nazi leadership pursued a variety of policies against German-, Austrian- and other Jews who fell under their control, including legal, economic and political segregation, significant harassment and incarceration, and forced emigration. Germany's invasion of Poland triggered increasingly violent “resettlement” and isolation policies of unwanted populations, including Polish Jews, but this proved to be a failure. With the invasion of the Soviet Union, Germany mobilized special killing units (Einsatzgruppen) to target Jews and partisans. These units were highly lethal, coordinated, and focused on killing Jews as a group. Indeed, by fall of 1941, the intention to exterminate Europe's Jews was established, and the remaining challenges were logistical (hence the introduction of the extermination camps). Prior to the invasion of Poland, there does not appear to have been an intentional plan to physically exterminate Jews.Footnote 35
Emergent intentionality requires analytical specification and sensitivity to context. Radicalization is shaped by leaders' specific goals and motives, which can vary. For instance, leaders seeking to radically transform society in conformity with their ideals and beliefs about desirable national identity are likely to view targets as existential enemies, whereas in other cases genocide is pursued as a strategic way of maintaining political control and eliminating a military “threat” framed in ethnic terms (typical of counterinsurgency programs that degenerate into coordinated and widespread massacres of minority or peripheral civilian populations). Motives may overlap and reinforce one another over time.
In sum, rather than take intent as the given starting point for comparative study, which has previously resulted in “silo-ing” genocide analysis off of other research areas, intent should be disaggregated and scrutinized analytically. Political scientists can model the conditions and decision points where intentionality emerges and employ counterfactuals to explain limitations on or absence of policy radicalization; by doing so, they can situate genocide within the comparative political violence research. Placing genocide in this broader context gives our theories more analytical leverage by highlighting what conditions lead to and sustain different forms of mass violence and allows for comparing similarly violent phenomena that are otherwise kept separated by adherence to the problematic UN definition.Footnote 36 We should not abandon the analytical concept of genocide, but some of the most problematic consequences of two decades of definitional battles can be mitigated by looking at violent phenomena comparatively and inquiring what explains their similarities and differences. The current research agenda develops this question of analytical contextualization in the sections below.
2. Genocide and Political Violence
The problems outlined above are related to a deeper issue of how scholars situate genocide within a wider constellation of political violence. Because many comparative genocide scholars focus on the particular outcome of genocide (that is, they tend to select on the dependent variable), insufficient attention is given to how genocide is connected to other forms of political violence and whether the proposed causal mechanisms and processes actually explain genocide or potentially other forms of violence as well. Naturally, this is not a problem for all research.The selection bias may be less troubling in pragmatic contexts of advocacy where human rights activists and researchers are focused on examining and publicizing a particular case. But for comparative research, selecting only genocides for study will likely lead to choosing those cases that confirm our theoretical assumptions, and thus raise questions of whether stipulated causal mechanisms are indeed the primary causal factors. For example, selecting only highly ideological genocides for research and then positing that radical ideologies are the primary cause of genocide fails to address how ideologies may be present in less violent outcomes or not present in other cases that have many of the empirical features of genocide but do not qualify as such under the UN definition.
There are several ways forward. First, we must broaden the types of violence that are studied and thus situate genocide more thoroughly within the political violence literature. This requires exploring advances in other research areas on political violence, such as civil war and non-genocidal state repression. Related to this, we can draw on the burgeoning micro-analytical work on violence (such as civil wars) as a way of better understanding the dissemination, intensity and patterns of violence within cases. Second, we must explore more systematically the ways in which genocidal elites and their followers perceive and construct their “enemies,” to include not only victims of genocide but also other targeted groups. In essence, this is a call to connect the comparative study of genocide to the broader domain of conflict studies.
2A. Contexts of Political Violence
A more systematic and inclusive approach would investigate the broader host of perpetrator repressive and destructive policies and use a dynamic framework that takes into account all relevant actors. This is important because it may shed light on the general processes of perpetrator policy radicalization and show at different steps in the continuum of repression how perpetrators choose and implement increasingly violent measures. Additionally, we need to understand why genocide does not occur in certain cases, and why violence may remain constant rather than increase to genocidal dimensions in yet other cases (such as in Kenya in 2008). The current comparative genocide literature does not explain very well the lack of genocide or instances of violent constancy.Footnote 37 If we do not connect comparative genocide research with a wider domain of research on political violence, we risk producing causal theories that may either misrepresent the sources of genocide or overdetermine outcomes.Footnote 38 We miss, in other words, the contributions made by scholars of other forms of violence that shed light on why political violence escalates, remains constant, or declines.
The state repression literature is helpful on this score. Scholars from a wide array of disciplines have sought to understand under what conditions states increase and decrease repressive behavior.Footnote 39 Why and when do states move from institutional coercive practices, such as legalized civil and political discrimination or language restrictions, to wider forms of collective suppression, such as total language and religious prohibitions? Under what conditions do state-tolerated episodic riots turn into sustained organized attacks, and targeted forced displacements into large-scale, violent “deportations”? Why do some states settle on forced conversion or removal of target populations, while others choose extermination? The state repression literature itself is highly heterogeneous and includes a variety of methodological approaches, but it generally shares a commitment to exploring the emergence, maintenance, decline and variation of repression and violence in a multiplicity of forms and over time. Furthermore, much of the best repression literature is dynamic; it seeks to understand how interactions between the state and non-state actors (armed and unarmed) alter the political landscape, in some instances reinforcing violent state practice and in others mitigating the likelihood of increased conflict.Footnote 40 By repositioning the study of genocide within the broader domain of political violence, analysts gain theoretical purchase on change and variability in violent outcomes, something that is largely absent in genocide research.
Certainly, the genocide literature is replete with case studies that provide historically rich descriptions of incrementally repressive practices, but these historical reconstructions of radicalization are often somewhat teleological: genocide appears as the final, almost unavoidable, stage of elite policy, socio-psychological processes or mass political mobilization.Footnote 41 These and other works fail to draw general theoretical insights about radicalization from case-specific historical contingencies. Given that so little of comparative genocide research places its analytical framework within broader explanatory models of political violence, it is perhaps unsurprising that genocide seems overdetermined. The frameworks provide few theoretical resources for understanding violence escalation and de-escalation. There are important exceptions, certainly. In sociology and political science, Mann and Midlarsky have advanced our theoretical knowledge by explaining and testing the conditions across cases under which leaders move from targeted killings and repression into large-scale, systematic extermination.Footnote 42 Their contributions mark an important step in understanding elite adoption of increasingly lethal policies, but they represent a minority approach in genocide research. Anchoring our causal theories within the broader political violence literature would provide us with the conceptual tools to theorize more broadly about the dynamics of escalation and de-escalation.
2B. Decentering the Analysis: Domains and Victim Groups
Without a doubt, states carry out most genocides. They have the military and logistical capacity to do so and often face minimal organized resistance from the targeted population.Footnote 43 Consequently, many comparative genocide works focus on individual country cases. Nevertheless, this focus on the country-level is potentially problematic. It may arbitrarily separate instances of genocide and mass violence according to national boundaries when in fact the violence spills over borders. In the process, we risk truncating our analysis and losing the ability to understand how violence disseminates across multiple territories.
Specifically, scholars should resist predetermining their scope of analysis to a particular country. The proper analytical domain for explanatory theories is the space over which genocide and related forms of violence occur; frequently, this goes beyond national borders.Footnote 44 With the exception of Holocaust research,Footnote 45 case studies often remain wedded to country-level analyses that misrepresent the causes, nature and dynamics of violence. For example, what is often referred to as the Darfuri genocide in fact includes violent engagements within Darfur and across the Sudanese border into Chad (and even into Libya). Chadian civilians have also been displaced into Darfur as a result of N'Djamena's persecution of eastern ethnic groups, and both countries support insurgents in the other's territories.Footnote 46 No account of the violence in Darfur should ignore these cross border dynamics and how they interact with the targeting of Darfuri civilians. The Rwandan genocide also occurred within a broader zone of violence that included Burundi and parts of eastern Zaire, and the aftermath of the genocide is still felt in the eastern DR Congo provinces.Footnote 47 Likewise, Cambodia is only properly understood in the context of the revolutionary struggles in South East Asia, and the Khmer Rouge's violent policies occasionally extended into neighboring Vietnam.Footnote 48
The country-specific nature of much comparative genocide research is replicated in the focus on only one primary victim group, while excluding or only superficially exploring the wider set of targeted groups. For instance, comparative works that include the destruction of Ottoman Armenians rarely analyze the broader set of policies directed against other Ottoman targets.Footnote 49 Analyses of Rwanda have mostly focused on Tutsi victims of the genocide but provide little exploration of intra-Hutu violence occurring at the same time, and how these latter dynamics influenced and were influenced by violence against Tutsi.Footnote 50 Until recently, comparative works that included the Holocaust paid relatively little attention to victim groups other than Jews. More generally, comparative work tends to compare only genocidal violence across cases, while excluding systematic intra-case comparisons that could shed light on the internal variation of types and intensity of violence, perpetrator elites' intentions (purpose) and motivations (justification), their general perceptions and selections of opponents, and the increasingly violent policies employed against various target groups.
Current historiography on World War II has made important advances in this area, showing the complex vectors of violence and how Nazi Germany and its allies targeted a diverse group of victims using a variety of violent strategies.Footnote 51 Many of these strategies were related: the creation of the Einsatzgruppen mobile killing units in occupied Eastern Europe targeted not only Jews but also communist party members and partisans. Early gassing efforts were carried out on Soviet POWs before being used extensively on Jews, Roma and Sinti and others. Similarly, the leaders of the Ottoman Empire employed forced deportations, massacres, and other forms of violent repression against a host of domestic groups aside from the Armenians, including Pontic and other Greeks and “Assyrians.”Footnote 52 A more inclusive approach to the actors involved—especially the variety of targeted groups—can inform our understandings of elite policy escalation, since elites are rarely responding to only one group “threat” but instead are reacting to a variety of “enemies.”
3. Microanalytical Complementarity
Microanalyses can contribute to explaining the dissemination and intensity of violence within cases. Some country-level analyses argue that general preexisting ethnic fear provides elites with useful discourses for the mass mobilization of political violence. Mobilization may be framed as an appeal to traditional myths of victimhood that are emotionally resonantFootnote 53 or individual rational choice.Footnote 54 Here, major crises including war and political instability raise levels of insecurity, sharpen ethnic differences, and lower moral prohibitions on violence against outgroups. Against this backdrop, ethnic animosity takes a violent cast: Hutus are more likely to kill Tutsis, Turks to kill Armenians, and Serbs to kill Bosnian Muslims.Footnote 55
The country-level study of causes of political violence has been reproduced in comparative genocide research. Vahakn Dadrian, Leo Kuper and Richard Hovannisian focus on the role of general preexisting cleavages in society; Irving Horowitz, Rudolph Rummel and Hannah Arendt draw attention to regime type to explain political violence; Robert Melson and Barbara Harff expand this by analyzing the destabilizing effects of national political crises; Valentino focuses on the interests and goals of national elites.Footnote 56 Mann traces the process of elite and follower radicalization against the background of salvation ideologies and the rise of modern mass democracy.Footnote 57 Widely used quantitative data sets on political violence also generally focus on country level data, using “country-year” as the unit of analysis.Footnote 58 Nevertheless, outside of idiographic anthropological and historical studies,Footnote 59 there is still relatively little work, especially in political science, that systematically explains internal differences across space and time. Genocide is understood as an aggregate outcome of country level factors, ignoring variation within states and regions experiencing violence. This can lead to over-representing the role of national elites and macro-level state failure, and employing static and reified conceptions of “masses” and “ethnicity” at the expense of understanding sub-national and local dynamics and patterns of violence (and, crucially, non-violence).
Microlevel analyses can contribute to correcting these tendencies. Microanalytic approaches may supplement macroanalyses without necessarily replacing them. Political scientists should refocus the scope of analysis to the area over which genocide and mass atrocities occur, whether subnational, national or regional, and employ microanalytical approaches to explore the causal mechanisms and processes that operate across this broader analytical field. The civil war literature, for example, has begun disaggregating country-level factors and exploring microlevel processes, showing that relying on national indicators of ethnic fragmentation and state capacity cannot satisfactorily explain civil conflict and in fact miss the importance of local cleavages and geographic differences in fostering and sustaining violence.Footnote 60 For instance, Elisabeth Wood has collected extensive ethnographic data to explain the intensity and variation of civil war violence, and Sidney Tarrow has modeled subnational contentious politics in individual country cases.Footnote 61 In a different context, Ashutosh Varshney has investigated communal violence by looking at types of inter- and intra elite relations in different Indian cities to explain the likelihood, intensity and targeting of “ethnic” rioting.Footnote 62
For comparative genocide research, this requires analyzing spatial and temporal variations within case studies (countries or regions, such as the African Great Lakes region), which would provide better understandings of why genocidal violence occurs in some places prior to others, and what the micro or meso causal mechanisms that determine internal variation might be. For example, how is identity understood and acted upon in particular circumstances? The master ideological narratives articulated by elites—whether ethnic, political, regional, religious, or other—often have complex and rather indirect relations to violence in situ. Certainly, the primary agents of violence—military and paramilitary units, extermination teams and other organized killers—may show broad fealty to master narratives and consistency in their selection and targeting of victims. However, even though perpetrators devote significant energy to defining enemies, the ways in which individuals are perceived and “othered” often differs within a particular genocide. Skin color, height, or facial features may serve as phenotypical markers of distinction in official propaganda, but in practice they may be less clearly defined. Unsurprisingly, few European Jews fit the stereotypes espoused by Nazis (as in Rwanda). Accent, the use of particular words or slang, and linguistic competence may operate as a relevant signifier in other instances, but perpetrators often interpret identity in myriad ways. Indeed, more tangential markers, such as dress, occupation, and neighborhood of residence may work as proxies for primary identity. And when victim groups are displaced and fleeing, violator interpretations are even looser: often, targeting becomes more inclusive to ensure no possible “enemies” are missed.Footnote 63
How does geography affect the onset and diffusion of violence? We know that geography played an important role in explaining successful Armenian resistance in Musa Dagh and elsewhere, in Tutsis' and Hutus' abilities to escape killings, and in the Khmer Rouge's uneven control across various national Zones, but we have no sophisticated understanding of the role of geography in genocide as such.Footnote 64 Or take state power (or capacity), often put forth as crucial to explaining genocide. Rummel puts it succinctly: “Power kills; absolute power kills absolutely.”Footnote 65 However, measured at the national level as done by Rummel, state power tells us little about when genocide is likely to occur or how it is likely to spread.Footnote 66 The state may project power unevenly across the country given its coercive and intelligence gathering capacity, which explains why violence may be highly targeted in some areas and wild and all-encompassing in others. Greater sensitivity to state coercive capacity across space and time, as well as the use of local allies and proxies, can provide us with more nuanced understandings of repression and mass killings. Lee Ann Fujii's ethnography on the Rwandan genocide explains variation in killings as a result of local state power and two social mechanisms: the density of local ties among potential perpetrators and thus practical opportunities for recruitment (affected by geography), and social-psychological group dynamics that shaped individual behavior through enormous peer pressure to conform to group expectations, including collaboration in killing.Footnote 67 Fujii shows how radicalized views of ethnicity were as much an outcome as a cause of the killings, and that perpetrators often framed identity in local rather than elite ideological terms. Her findings contest dominant theories focusing on macro variables like widespread “ethnic hatred” or authoritarian Rwandan culture, and instead explain how genocidal dynamics were shaped by the state's local power, the density of community ties, and group pressure. The payoff is a better theoretical understanding of why massacres occur in some places and not others.
Some political scientists are already moving in this direction with sophisticated idiographic studies.Footnote 68 Nevertheless, these microanalytical methodological perspectives have yet to be integrated into broader comparative genocide research. Work on the Holocaust is perhaps furthest along in this area given the attention it has received for sixty years, but we need more of this type of analysis to inform our cross-case comparative theories (especially needed, for example, in studying Armenia). Particularly in the context of war, when genocide normally occurs, violence is multidirectional and includes a host of armed and unarmed actors, with variations in level, organization and types of violence across space and time.Footnote 69
To be clear, microlevel and country level studies can ask and answer different types of questions, and certainly country level studies should not be abandoned; some questions can be studied at both levels and others at only one level, and thus these analyses can complement one another. However, the general dearth of microlevel comparative analysis unfortunately leads us to ignore some of its specific benefits. First, as Straus notes, it expands the number of observations available for study, providing richer accounts of violent dynamics than those found in works focusing on national level events and actors.Footnote 70 Second, it problematizes our general descriptions of violent phenomena. Rather than take for granted that a conflict is primarily ethnic or racial, greater sensitivity to local variation forces scholars to rethink how master narratives may (or may not) connect to local violence.Footnote 71 Third, it questions normally accepted beginning and endpoints of violent phenomena like genocide, showing continuity and discontinuity of violent patterns before and after event-defining dates and thus avoiding their procrustean truncation.
Certainly, microlevel analyses face their own challenges. Without a broader comparative framework it may be difficult to assess whether conflict dynamics are peculiar to a case or reflect general patterns. However, a focus on microlevel factors is not meant to replace higher-level analysis. Clearly, genocide requires elite commitment, complex logistical coordination, and trained killers. The claim here is not that genocidal violence erupts spontaneously and randomly in communities, outside of a broader social and political context. Rather, I am arguing that micro approaches can complement macroanalysis by giving us finer grained and more sensitive understandings of the onset and diffusion of genocide, which in turn forces us to rethink our broader analytical frameworks. Without a better understanding of local dynamics and circumstances, as well as their interactions with larger scale events and processes, we are likely to continue to produce generalized causal theorizes that misrepresent the cases we purport to explain, or otherwise minimize real and important local differences.Footnote 72
These calls to reinterpret our analytical frames and problematize our levels of analysis may seem to push in two incompatible directions: one micro (subnational) and the other regional (or beyond the country level). Are these calls compatible? Can we seek to investigate both micro-dynamics and dynamics that occur beyond the country? I believe so, provided we are clear on how they are related. The domain of analysis for a relevant theory on the causes of genocide is the spatial range of genocidal and similar events: that is, the territory or territories over which genocide occurs. Our causal theories should focus on explaining genocide across this space, as well as comparable spaces where it did not occur. The analytical domain should not be theoretically predetermined; we should not assume that genocides occur within neatly defined national boundaries, marking off case studies according to national borders. The relevant range might be regional and encapsulate several entire countries or only parts of many. Given how porous national boundaries are in violence-prone countries (all the more so under conditions of war and mass population displacement), we need to ensure that our theoretical frameworks capture this and our analysis is not limited by relying on cartographical divisions that have little empirical resonance. It is within this reconceptualized space for theoretical analysis (that is, the relevant domain), that we may find that different types of microdynamics are reproduced.
4. Genocide: Rational, Ideological, or Something Else?
Since the beginning of genocide research, ideology has played a central explanatory role. Some of the most prominent recent scholarship still emphasizes ideology.Footnote 73 Many scholars, however, have criticized ideologically-based theories as providing too imprecise an account of the relation between belief and behavior.Footnote 74 Crucially, they contend that ideology does not explain timing: if elites and their followers are ideologically driven to exterminate, why don't they do so as soon as they seize power? Why is it, in other words, that leaders often choose genocide after pursuing a host of less violent strategies? Furthermore, ideological theories are criticized for being tautological: leaders carry out genocide for ideological reasons, and the proof that leaders are ideological is their genocidal behavior. In response, some scholars, particularly in political science, have adopted rationalist theories that focus on the role of the strategic behavior of elites. Valentino, for example, argues that elites pursue genocide as a means to achieve some specific radical policy aim. In doing so, they employ genocide against a particular group in order to force victims to acquiesce; genocide is employed when earlier, less radical policies have failed or are no longer practical.Footnote 75 For Valentino, ideology is of only secondary importance. The fact that extremist political ideologies are more common than genocide and that ideology theories cannot explain radicalization highlights the need to look at other ways of modeling causes, such as through strategic rationality. Employing prospect theory, Midlarsky has similarly argued that elites pursue genocide when the nation has experienced significant loss and prior state strategies have “failed” to address the sources of perceived “threat.”Footnote 76
Valentino and Midlarsky provide sophisticated criticisms of ideology theories. Nevertheless, framing this debate as an either/or choice (ideology or some form of rationality) underplays the possibility of combining the two approaches in a useful manner. A rationalist model tells us little about why elites frame perceived threats in certain ways and the types of options they believe are available to them in order to respond to those threats. Behavior is only rational within a particular context of meaning. And while ideological frameworks may not be able to predict subsequent behavior in any specific way, political ideology can be seen as constituting a general framework of related normative beliefs that structure our understanding of the political world and our place in it. Within this framework we interpret and make sense of other actors. Of course, political ideologies are normally bounded to and exclude someone or something, such as another religion, “race,” ethnicity, political creed, and so forth.Footnote 77 However, exclusivist ideologies are significantly more extreme: an exclusivist ideology is a totalizing system of meaning based on pronounced in-group and out-group distinctions permitting no shared forms of identification between groups and premised on a radical devaluation of the out-group.
Discourse organized around extremist in-group/out-group identifications has two consequences: (a) it reifies artificial differences between groups, ascribing elements of uniqueness to each that make them ontologically and thus ineradicably different, and (b) it results in the devaluation of one group by another, so that the differences are normatively laden. Exclusivist ideology is dangerous not only because it constructs insurmountable walls between different groups, but also because it explicitly targets and devalues those on the other side of the wall. It achieves its greatest resonance when it exaggerates and perverts already recognized differences between groups.
The consequences of out-group devaluation can take many forms, ranging from the limitation of citizenship rights to systematic harassment, incarceration, or even expulsion. In its most extreme form, and with sufficient commitment and resources from the state, it can result in massacres, and ultimately in genocide. The stronger the devaluation, the more probable it becomes that members of the in-group will see violence as a legitimate response. Clearly, exclusivist ideologies cannot predict genocide: they provide only generalized frameworks of meaning. Actors pursue their interests within these broader ideological frameworks, but ideologies—and thus actors' understandings of the context in which they behave—can also be transformed (and radicalized) under certain conditions, such as war. Consider, for example, Donald Bloxham's argument that it is unlikely that the Ottoman Turkish leadership (CUP) had a sophisticated blueprint for the Armenian genocide prior to April 1915.Footnote 78 Rather, the genocide emerged from a series of contingent and radicalizing events in which leaders pursued strategies within the prior ideological frameworks they held. Certainly, there are broad historical background factors that informed these ideologies, including historical prejudice and even socially sanctioned violence against Armenians, a prior pattern of humiliating foreign intervention, and a tendency to view Armenians as agents of foreign powers (specifically Russia). But the factors that contributed to the genocide came later. The Ottoman Empire suffered significant contraction in 1912 and 1913, which led to an influx of terrorized European and Caucasian Muslims into the heartland of Anatolia who tended to be more radical and anti-Christian than their Turkish co-religionists and who spread stories of Christian abuses and betrayal in the Empire's periphery. The 1913 coup by extremist CUP leaders and territorial losses at the beginning of WW1 further radicalized the elite. But even though there were massacres and attacks against Armenians in the fall and winter of 1914, there does not appear to have been a coherent plan at extermination at this point. Rather, there were increasing targeted massacres, some wild massacres, forced deportations and other forms of political repression through this period. However, in Spring 1915, a series of events made genocide more likely: first, there was the “uprising” at Van on April 20th, in which Armenian residents refused to leave their homes and be deported to the southern interior. Second, the Gallipoli attack by British and allied forces drove home to the CUP the threat of military defeat and imperial collapse, and radicalized attitudes toward perceived internal “enemies.” Finally, on May 18th, Russian forces occupied Van and relieved Armenians of the siege laid by Ottoman forces.
At this point, the CUP leadership called for widespread and systematic massacres and deportations to the desert, with the understanding that a significant proportion (if not all) of Armenians would die. These deportations represented a radicalization of earlier traditions of deportation employed by the state against internal opponents.
According to Bloxham, the policy of repression expanded to genocide sometime between April and June 1915. Genocide emerged from war-time radicalization, but was not preplanned. Rather, the CUP's prior ideological understandings of Armenians as internal enemies and pro-Russian fifth columnists framed ever fewer and more radical available choices, culminating in the “rational” policy of genocide. So, the point is not whether ideology drives rationality or vice versa. Rather we need a more dynamic understanding of how political ideologies provide general frameworks in which strategic calculations occur, and we need to understand how perpetrator perceptions of changing conditions (such as loss of territory) can further radicalize the options they believe are available to them. We may refer to this process as cascading radicalization.Footnote 79 Following Bloxham, we should reject the “false dichotomy of ideology versus pragmatism,” which has been carried over from early and less sophisticated studies of the Holocaust, affecting many subsequent studies of other genocides.Footnote 80 Leaders and subordinates can be pragmatic and responsive to changing events, but do so within broader ideological systems that frame events, opportunities, and perceptions of other actors (including who is an ally, bystander and enemy).
Conclusion
Genocide research, in short, needs to be brought back into the fold of political conflict analysis and of comparative politics more generally. For while genocide is an extreme form of violence that understandably generates stark moral, political, and legal responses, it is also a complex historical outcome no less amenable to nuanced analysis than other forms of violence or indeed other forms of general political conflict. A theoretically sophisticated account of genocide would treat it as a dynamic, multilevel process that includes numerous actors (not just “perpetrators” and “victims”), and would seek to identify the primary conditions and patterns of violence escalation, maintenance and deescalation. It would disaggregate the concept of “emerging intentionality,” decenter our analytical focus beyond the country-level and single victim group, look to microanalysis to theorize dynamics and interactions, and move beyond problematic oppositions of ideology and rationality.
Such an analytical deconstruction of genocide promises several benefits for empirical research into large-scale violence. It can afford greater insights into the specific tipping points and interactions that result in genocide, identify causal processes to explain why and when genocide (rather than some other violent outcome) occurs, and advance conceptual clarification of key terms like “intentionality,” “ideology,” “rationality” and “identity” that play a central role in the political violence literature.
Such an approach also has implications for political science research more generally. A focus on genocide as a multi-level process encourages greater specification of the relation between theory and evidence, especially in identifying precisely what constitutes an adequate explanatory mechanism, how mechanisms operate and interact, and what kinds of information are necessary for justifying a given causal argument. These theoretical concerns are sometimes less evident in macrolevel studies that assume uniform behavior and dynamics within a case, or those studies that compare only cases with a similar outcome, since internal variation is not as strong a challenge to explanation. Methodologically, disaggregating complex phenomena over time and space contributes to clearer theorization about the dynamics of social change: why, how and when change occurs, and what explains variance in process outcomes.
Disaggregation through concepts such as emergent intentionality and cascading radicalization, also encourages greater sensitivity to how actors themselves interpret and react to complex circumstances. In the search for greater theoretical parsimony, political science scholarship has moved toward what are often reductive formulations of actor motivation based on abstracted conceptions of rational choice. However, if we analyze social phenomena as temporally and spatially dynamic, assumptions of a linear relation between actor intentionality and outcome appear empirically and theoretically unpersuasive. The call for a disaggregated, multi-level approach treats action as socially constituted—and thus always contextualized—by values, beliefs, interests and the behavior of other actors, placing greater attention on the relation between social systems of meaning (such as ideology) and strategic rationality. By avoiding the problematic reduction of action to abstract individual calculus, we can instead inquire about the social meaning of behavior and ask “why a set of otherwise puzzling behaviors might, from the vantage point of those who perform them, make sense.”Footnote 81
The primary benefits of such an approach are analytic and theoretical. At the same time, genocide is not simply a challenge to scientific explanation and political understanding. It is also a practical challenge to widely shared and deeply held human values. And while the theoretical analysis of genocide is not the same thing as the praxis of genocide prevention or humanitarian intervention, it is obvious that the illumination that such analysis promises is of potentially great practical and moral relevance. Prevention, of course, has its own complex theoretical and practical demands. But without clearer understandings of how genocide originates and spreads, such an ultimate goal will remain all the more difficult to achieve.
