Nationalism and Ethnicity in International Relations

Lars-Erik Cederman. Handbook of International Relations. Editor: Walter Carlsnaes, Thomas Risse, Beth A Simmons. Sage Publication. 2002.

Today, at the dawn of the twenty-first century, most international relations (IR) specialists accept that ethnicity and nationalism are highly relevant to their field. Yet, only a decade ago, IR studies on these topics were few and far between. It was only with considerable delay that the IR literature vindicated Donald Horowitz’s (1985) comment in the mid-1980s about ethnicity’s having ‘fought and bled and burned its way into public and scholarly consciousness.’ In fact, it took an extraordinary amount of nationalist activity in the early 1990s for scholars to begin to grapple with the challenge posed by these topics. The decade that followed after the collapse of communism has indeed been marked by remarkable outbursts of nationalist violence, most notably, but far from exclusively, in the former Yugoslavia and Rwanda. Despite the Gulf War, it is becoming increasingly clear that the importance of inter-state war is declining while that of nationalist conflict is increasing (Wallensteen and Sollenberg, 1999).

This chapter explores primarily the response to these recent events for the simple reason that not much else exists to review in the IR literature. A quick glance at recently published textbooks and handbooks speaks a clear language. Indeed, one of the few political-science experts in the area summed up the situation in the mid-1980s by referring to the study of politics and culture as ‘moribund’ (Laitin, 1986: 171).

Granted the sustained relevance of ethnic and nationalist conflict since the French Revolution, this silence calls for an explanation. With the exception of the decolonization process which ended by the late 1970s, the Zeitgeist of the postSecond World War period triggered a scholarly rejection of nationalist themes: ‘Nationalism was blamed for the onset of war in 1939; as statesmen paid no attention to national self-determination when dividing Germany and Korea, so scholars in their turn ignored nationalism’ (Hall, 1998b: 1). Furthermore, the strong impact of economics on the social sciences has tended to obscure topics such as culture and collective identity formation. Finally, as a result of its state-centric focus, IR theory has become strangely ‘depopulated’ as a result of ignoring the role of individuals and non-state groups (though see Wolfers, 1962). Fortunately, this is beginning to change.

Yet it would be a mistake to limit the scope to the first decade following the fall of the Berlin Wall. Because the scholarly output appeared as late as it did, and mainly focused quite narrowly on the post-Cold War situation, it is necessary to cast a wider net. First, the IR treatments of these topics constitute a very small fraction of a broader and more mature, interdisciplinary literature on nationalism. At any rate, border-transgressing phenomena such as nationalism make a mockery of any attempt to demarcate IR as a distinct field separate from comparative politics. But even this widening of the disciplinary scope remains insufficient, for most of our knowledge of nationalism and ethnicity stems from academic writings located entirely outside political science. This burgeoning literature brings together a motley collection of disciplinary perspectives, including history, historical sociology, cultural anthropology, social psychology and social theory.

Far from being historical constants, ethnic and nationalist processes antedate the current era by a wide margin. To argue that the post-Cold War cases of conflict along nationalist lines represent a profoundly novel development in world politics would thus be tantamount to a glaring anachronism (Ayres, 2000; Fearon and Laitin, 1996; Gurr, 1993). Such a post hoc rationalization squanders a valuable opportunity to engage in reconstructive theorizing of the discipline’s conceptual foundations. As I will argue, such a need for conceptual reconstruction relates to widespread terminological confusion that has even turned the name of the sub-field into a misnomer, for the actual practice of ‘international relations’ would be better characterized as ‘interstate relations’ (Connor, 1972).

Rather than merely summarizing the recent work in IR, then, I adopt a wider, interdisciplinary approach as a way to put the IR literature in its proper theoretical and macro-historical perspective. This opening will facilitate the assessment of its strengths and weaknesses and help us identify future avenues to a deeper understanding of the complex processes underpinning the phenomena under scrutiny.

To my knowledge, there have been few, if any, attempts comprehensively to take stock of the structure and insights of the IR literature on nationalism and ethnicity. Because of the relative scarcity of studies in IR, previous reviews of these topics that have appeared in political science publications have tended to focus on extra-disciplinary work (Haas, 1986; Waldron, 1985). In addition to these useful, but now somewhat dated, reviews, Brubaker and Laitin (1998) provide an excellent survey of the literature in sociology and political science (see also Calhoun, 1993, 1997; Hall, 1995; Smith, 1991, 1995). The current chapter complements their insightful, stock-taking exercise by focusing primarily on IR applications. Yet, as opposed to Brubaker and Laitin’s methodological classification, this chapter follows a distinctly theoretical outline that classifies the literature according to the ontological status of the key concepts.

While I have made an effort to encompass as much of the recent IR literature as possible, the current wealth of writings has forced me to limit the scope to:

  • Explanatory and analytical research, at the expense of normative studies and practical prescriptions (e.g. Beiner, 1999; Brown, 1997: Part II; Lake and Rothchild, 1998a: Part 4);
  • Predominantly theoretical and analytical studies, rather than purely empirical ones, with the exception of such texts that elucidate general analytical points;
  • Mostly conflictual cases due to their impact on the international system;
  • Political science writings, whereas other fields receive attention to the extent they elucidate the IR problems under scrutiny; and
  • Primarily, but not exclusively, US scholarship, because of the relative lack of publications in the non-American IR literature.

This chapter is organized as follows. The next section attempts to bring some clarity to the thorny issue of conceptualization. The following section introduces an ontological 3 × 3 table, which will serve as the main taxonomic device used in this chapter, followed by four sections, each one covering a ‘field of construction.’ A concluding section evaluates the main trends characterizing IR work on nationalism and ethnicity, and comments on the general direction of future research.

Defining the Master Concepts: The State, the Nation and Nationalism

It has become a cliché to characterize the key concepts relating to ethnicity and nationalism as hopelessly elusive. Nevertheless, despite some signs of improvement, observations to this effect remain pertinent. What is more, political science seems curiously resistant to repeated attempts at clarification of the fundamental ontology.7 I will therefore start by discussing the two most serious misconceptions (cf. Miller, 1995: ch. 2).

The most acute one pertains to two central concepts, namely the state and the nation. To disentangle these terms, it is helpful to start with Max Weber’s (1946) classical definition of the state as a territorial organization exercising legitimate control over its own bounded territory, unchallenged by internal power competition or external intervention. Against this conceptualization, Weber pits the nation, which he defines as ‘a community of sentiment which would adequately manifest itself in a state of its own’ and hence ‘tends to produce a state of its own’ (176; see also Alter, 1989; Cederman, 1997: 16–19; Haas, 1986: 726; 1997: ch. 2).

This preliminary definition of the nation calls for several remarks. First, although the nation sometimes coincides with the state, such a coincidence should be treated as a historical contingency rather than as a case of conceptual unity (Connor, 1972, 1978). Where the state and the nation do coincide, it is legitimate to refer to the nation-state. Second, Weber’s definition requires the presence of an intersubjective understanding of belonging. This is precisely what Benedict Anderson (1991) aptly labels an ‘imagined community.’ In addition, the nation must be a bounded community defining citizenship for the masses. Third, Weber’s nation concept depends directly on the state. By definition, there can be no nations independently of the state system (though see Smith, 1991: 14, for a cultural definition). Since some nations do not fight for full sovereignty but instead claim wide-ranging powers, it makes sense to relax Weber’s demanding definition by including self-defined communities that pursue autonomy within a state framework (e.g. Brubaker, 1998: 276; Snyder, 2000: 19).

The last remark highlights a second pernicious terminological blurring characterizing much of the political science literature, namely the failure to separate nationalism from ethnicity. Too often the two are treated as if they were synonymous. To enhance conceptual precision, it is therefore preferable to define an ethnic group or an ethnie as a cultural community based on a common belief in real or putative descent (Weber, 1978; cf. Smith, 1991: 21). Although nations sometimes correspond closely to ethnic groups, in which case they are usually referred to as nations’ ‘ethnic cores,’ other nations contain many ethnic groups and may in some cases be devoid of any sense of ethnic belonging (cf. Barrington, 1997: 731). Ethnic groups, in turn, differ from ethnic categories, which are based on cultural markers imposed by outside observers without the members necessarily attaching any importance to the characteristics (Brass, 1976: 226). Sometimes kinship is also differentiated from ethnicity, but this will not concern me in this chapter (cf. Calhoun, 1997: 39–40).

Figure 21.1 Three historical types of nationalism

Having defined the basic units, I now turn to the notion of nationalism. Again there is a plethora of definitions to choose from (see Calhoun, 1997: 4–5; A.D. Smith, 1991: 72), but I will limit the use of the term to a specific ideology with European origins stating that each nation should possess its own state or at least some degree of territorial self-determination. This definition is somewhat wider than Gellner’s (1983: 1) terse formula stating that ‘nationalism is primarily a political doctrine, which holds that the political and the national unit should be congruent’ (cf. Haas, 1997). Viewed as a corollary, this conceptualization implies (1) that the world is divided into nations, (2) that the nation is the source of all political power, and (3) that national loyalty overrides all other allegiances as a ‘trump card’ (cf. Calhoun, 1993, 1997; Smith, 1991: 72).

To put the pieces together, I propose a simple, historically inspired taxonomy that will help to identify the main types of nationalism. Following up the distinction between the state and the nation given above, Figure 21.1 introduces a two-dimensional scheme depicting stylized historical situations. Logically speaking, such situations can be characterized by either the presence or absence of a common state and/or nation.

Starting with the anarchical configuration in the upper left quadrant, where neither entity dominates the entire area, I sketch three developmental paths that can, but do not have to, lead to the formation of a nation-state in the lower right corner. This dynamic categorization draws on Theodor Schieder’s (1991) three ideal types of nationalism (cf. Alter, 1989; Cederman, 1997: ch. 6; Gellner, 1997: ch. 6). Depending on the initial geo-cultural configuration, either a common state emerges before the nation or nation-building precedes state-formation. The former ‘Western European’ possibility leads to state-framed nationalism (Brubaker, 1998: 300). In Germany and Italy, the order was arguably reversed, such that a cultural nation preceded the state thus generating unification nationalism. Finally, when the East European empires collapsed, separatist nationalism appeared in their wake. While originally associated with European cases, these three trajectories are quite general, and could be applied to both decolonization and post-communist nationalism.

Since nationalism requires national and territorial units to coincide, pressures will be exerted on existing state frameworks where this is not the case, that is, in the ‘mixed situations’ corresponding to the upper right and the lower left cells. In a very simplified way, the diagram thus illustrates that nationalist activity can both produce political integration and territorial fragmentation.

Contextualizing the IR Literature: Fields of Construction

The remainder of this chapter considers to what extent the territorial and popular dimensions of world politics are problematized. A few words are in order to justify the use of ‘fields of construction’ as an ontological map since it may be less familiar to an IR audience used to the traditional levels-of-analysis scheme.

While Waltz’s (1959) three images have been used to explain ethnic conflict (e.g. Kaufman, 1996), such applications are problematic. For one, it is unclear how to fit ethnic groups and nations into the traditional state-centric scheme. Moreover, the levels-of-analysis model usually entails a one-shot and level-against-level notion of causation that would fail to do full justice to the complex processes under scrutiny. Third, given the importance of identities, a classification of theories of nationalism should be able to represent different ‘levels of construction’ (Wendt, 1999: 244).

Here three levels of analytical problematization will be considered. Logically speaking, theoretical frameworks can view social actors as either being absent, reified, or constructed. Absence means that the theory in question ignores the unit or presupposes it implicitly. Reification corresponds to the practice of treating actors as if they were objects, that is, given entities that are held constant throughout the analysis (Berger and Luckmann, 1966). Not only assumptions, but also independent variables and parameters, belong to this category. Finally, I consider a collective actor to be constructed to the extent that its identity plays the role of a dependent variable or at least varies with time. In short, it has to be a ‘moving part’ of the explanatory machinery even if it is not fully explained.

The two-dimensional exposition thus allows me to discern different meanings of constructivism in IR. While the term is often reserved for theories that endogenize states’ identities, constructivist theories can also endogenize both national and ethnic identities. Applied to such identities, the construction–reification dichotomy has the advantage of circumventing the potentially pejorative term primordialism. Scholars often use this label as a description of nationalists’ ideological efforts to convince others (and themselves) that their own nations are naturally given entities. But they also employ it as a characterization of social science research. Yet, while many analysts engage in ‘methodological reification’ for pragmatic rather than ideological reasons, it would be wrong to assume that they accept the nationalists’ argumentation (although in some cases this cannot be excluded).

Neither the reified nor the constructivist perspective on national identities follows a single logic. Indeed, there is considerable variance on both sides of the ontological divide. The assumption of nations’ invariance in reifying approaches can be justified by references to biology (e.g. van den Berghe, 1982), cultural stability over long time periods (e.g. Smith, 1986), or analytical convenience for a shorter period not exhibiting identity change (e.g. Kaufmann, 1996).

Constructivist approaches differ even more widely. There are many ways to explain how national identities form, change, dissipate, or remain stable. The last explanatory task may seem to be the turf of reifying theories, but these assume identities to be constants rather than explaining why they fail to vary (Wendt, 1999: 238). Furthermore, two distinct logics drive constructivist explanations. Whereas some analysts view identity-formation in exclusively instrumentalist terms, thus following March and Olsen’s (1998) ‘logic of consequences,’ others emphasize rule-following and identity-reproducing institutions, that is, ‘the logic of appropriateness.’ Although this analytical distinction does not always coincide with the question of identity malleability, instrumentalists typically assume identities to be fluid and subject to relatively free individual choice. Sociological institutionalists, by contrast, are more prone to regard identity as being ascribed rather than chosen, and if an element of choice is involved, it is heavily embedded in an interpretative frame of constitutive rules locking individuals into a path-dependent collective identification process (Calhoun, 1997: 30–3). Thus, it would be a mistake to treat rationalist and constructivist approaches to identity-formation as inherently competing perspectives (Brubaker, 1998: 291).

Together the distinctions relating to actor type and levels of construction open up a two-dimensional landscape of theoretical possibilities consisting of nine cells (see Figure 21.2), ranging from theories in which both states and national (or ethnic) identities are absent (type A) to those that attempt to problematize them both (type 4). Without confusing ethnicity and nationalism, cells D and E in the first column refer to ethnic groups rather than nations because the latter cannot by definition exist in the absence of the state.

Rather than defining self-contained ‘paradigms,’ the figure depicts analytical combinations adopted by various approaches, some of which cover more than one cell. Since their utility depends on the substantive goal of the research, none can be said to be better than the other without specifying the substantive task at hand. Nevertheless, the highlighted four cells in the lower right corner correspond closely to the topic of this chapter (types 1–4). In the following, I limit the scope to studies that render both state and national identities explicit. Since states figure in all these combinations, the main focus will be on nationalism rather than on ethnicity in its own right.

With respect to national identities, analytical approaches can be either reifying or constructivist. The second question pertains to the analytical status of state identities. Whereas the nation-reifying perspectives freeze state identities as well (type 1), or endogenize these (type 2), those that adopt a constructivist approach to national identities either limit their constructivist outlook to nations while treating states as if they were fixed objects (type 3) or embrace ‘double constructivism’ with respect to both states and nations (type 4).

The remaining combinations correspond to ‘contiguous’ fields that constitute the intellectual context of this chapter’s main area of focus (types A–E). The type A category is silent about both the state and the nation. Despite attempts to ‘bring the state back in’ (Evans et al., 1985), students of domestic politics have frequently resorted to this analytical starting point, as for example in pluralist studies of interest groups and other societal actors (see references in Nettl, 1968). Type-1 research also includes analysis of historical periods during which neither the state nor the nation was present.

Figure 21.2 Classifying analytical perspectives: nine fields of construction. In cells D and E, the vertical axis refers to ethnic groups rather than nations.

Type-B research combines an explicit but reified treatment of the state with no, or minimal, references to ethnic or nationalist actors in their own right. Most mainstream IR writings fit comfortably in this state-essentialist cell since both neorealism and neoliberalism tend to represent states as given and unchanging entities while ignoring nationalism and ethnicity. Relaxing this position, type-C studies problematize state identities while still bracketing nationalism and ethnicity. While social constructivism has secured a safe position on the state-centric IR agenda, they continue to pay surprisingly little attention to other actor types, such as ethnic or national communities. Ironically enough, while arguably better placed to break new ground in analyzing nationalism than any other meta-theoretical paradigm within IR, ‘conventional’ constructivism has had little to say about this important phenomenon. The strong focus on the purely intra-scientific goal of criticizing and surpassing neorealism has detracted from the more important theoretical objective of analyzing major substantive developments in world politics (e.g. Wendt, 1999).

Reifying ethnic groups while bracketing states, type-D research corresponds to traditional, ‘pre-Barthian’ approaches to anthropology, especially those that analyze primitive, pre-modern communities as objectified collections of cultural traits (Neumann, 1999: 4). While often misclassified as a primordialist theory of nationalism, Clifford Geertz’s (1973) approach to ethnicity does reify ethnic groups (Calhoun, 1997: 31–2). Reacting to such reified conceptions of ethnicity, Fredrik Barth (1969) shifted the attention from cultural ‘essences’ to ethnogenesis through boundary formation, thus pioneering the type-E research. According to this type of constructivist anthropology, groups do not consist of objective cultural traits but need to be viewed through the self-categorization of its members. Although recent anthropological studies also stress ‘cultural contents,’ contemporary anthropology is deeply indebted to Barth’s formal boundary approach.

Reifying Both States and Nations: ‘Double Essentialism’ (Type 1)

Since so little ‘native’ theorizing of nationalism existed in IR at the beginning of the 1990s, it is hardly surprising that most analysts started by extending their type-B frameworks. While retaining a reified notion of the state, studies of this type incorporate nations and/or ethnic groups as equally objectified entities, either encapsulated within states or plugging cultural units into the slot previously held by states. Recently, some studies have tried to disaggregate ethnic groups. Each possibility will be explored in turn.

Explaining Inter-state Behavior

In the years that followed immediately upon the collapse of communism, most IR scholars retained their state-centric and state-reifying frameworks. Viewed through this rather narrow lens, nationalism appears to be primarily a popularity-enhancing device reinforcing states’ internal cohesion and external power projection. For instance, Barry Posen (1993a) investigates the spread of nationalism in the nineteenth century as a means to enhance states’ extractive and mobilizing power. Drawing on Waltz’s (1979) arguments about competition and socialization inducing unit likeness, Posen explains the post-Napoleonic spread of mass armies.

But not all behavior fits into the matrix of raison d’état. Committed to rationalistic assumptions, some realist scholars resort to the term ‘hypernationalism’ for episodes of what they consider to be nationalist over-reactions. Defined as an artificially generated ‘belief that other nations or nation-states are both inferior and threatening,’ hypernationalism is thought to undermine carefully calibrated balance of power calculations (Mearsheimer, 1992: 221; see also Van Evera, 1990/1). Yet, it is tempting to use hypernationalism in order to explain away whatever does not fit neorealist rationalism (Lapid and Kratochwil, 1996b: 112). Indeed, the definitional section above has shown that nationalism does not always strengthen the state or its policies.

In a dramatic and highly publicized departure from the neorealist tradition, Samuel Huntington (1993a, 1996) claims that state behavior in the post-Cold War period will no longer reflect primarily power calculations or ideology but, rather, civilizational affiliations. It is along the ‘fault lines’ between the world religions that conflict will be most prone to erupt. In particular, the ‘kin-group syndrome’ prompts intervention by distant cultural relatives, as illustrated by Russian and Greek nationalists aligning themselves with Serbia. Needless to say, Huntington’s thesis, with its overtly normative overtones, has come in for fervent criticism because of its attempt to reify civilizations as large-scale ethnic categories (see Huntington’s, 1993b rebuttal to the critiques in the same issue).

Explaining the Behavior of Given Ethnic/Nationalist Groups

In the post-Cold War period, a number of scholars have abandoned the exclusive focus on the state system and applied the unitary rational actor assumption to ethnic groups. There are four clusters of analyses: first, some studies explain inter-group conflict as a direct consequence of long-standing antipathies among ethnic groups. Second, rejecting this view as too culturally deterministic, others draw on the ‘security dilemma.’ Third, rational-choice theorists have joined these efforts. Finally, statistics on ethnic minority groups have been collected and analyzed.

We start by considering conflict-reifying studies. Reacting to the painful images broadcast from the former Yugoslavia and Rwanda, politicians and journalists have tended to attribute the appalling acts of violence to longstanding, ‘century-old’ hatreds between the ethnic groups. This argument assumes that conflict erupts as soon as the state’s power wanes: once the lid is lifted the ‘seething cauldron’ boils over (see critique in Brubaker, 1998: 281–5).

Although most academic commentators disagree with this version of events, a minority of scholars have followed similar lines. These studies rely heavily on culture as the main factor perpetuating the animosities. In an often-cited interpretation of the Yugoslav conflict, Robert Kaplan (1993) attributes the violence to specific traits characterizing the entire region’s allegedly belligerent culture. Similarly, Huntington’s (1993a) civilizational thesis not only applies to inter-state exchanges, but can also be seen as a theory of culturally defined groups. In particular, Islamic culture allegedly provokes more conflicts than other civilizations.

The tendency to reify conflict, together with the cultural units perpetrating it, is also strongly present in an article by Chaim Kaufmann (1996), who recommends partition as a solution to ethnic civil wars (see also Mearsheimer and Van Evera, 1995). Contending that such conflicts differ fundamentally from ideological contests, Kaufmann asserts that ethnic divisions render conflict resolution next to impossible. Once started, violent exchanges quickly harden ethnic identities to such an extent that reification presumably becomes a fully justifiable analytical assumptions. In fact, Kaufmann even thinks that ethnic boundaries are inert (1996: 269) and that ‘atrocity histories cannot be reconstructed’ (1996: 283).

Nevertheless, outside the security literature, most political scientists appear to reject these perspectives as over-simplified (cf. Laitin, 1998: ch. 12; Snyder, 2000: ch. 1). Unfortunately, some of the rebuttals are quite overstated. From the rejection of conflict-reifying determinism it does not follow that all arguments involving culture are tantamount to primordialism. Indeed, it cannot be excluded that specific cultural types of discourse are violence-inducing (Brubaker, 1998: 283), nor does it seem implausible that violence often hardens group boundaries (Simmel, 1955). Nevertheless, recent statistical evidence appears to contradict the effectiveness of partition as a policy, thus undermining the credibility of essentialist theorizing in general (Sambanis, 2000).

The fact that the most violent conflicts usually cannot be traced back to ‘century-old’ conflict seriously undermines ‘ancient hatred’ accounts. To illustrate, most Yugoslavs lived peacefully side by side until shortly before the violence erupted (Gagnon 1994/5; Ignatieff, 1994; Sekulic et al., 1994; Woodward, 1995). In order to resolve this anomaly, some IR theorists have attempted to account for conflictual outcomes by replacing the strong assumptions of long-standing hatred by weaker postulates.

For example, Barry Posen (1993b) suggests that the ethnic conflict in post-communist Europe should be seen as a situation of ‘emerging anarchy.’ Using the same logic as in a traditional interstate setting, Posen contends that belligerent ethnic groups, such as Croats and Serbs in the early 1990s, are subject to a security dilemma (Jervis, 1978), because without stable stateled enforcement, they cannot trust each other and commit to liberal policies. On this view, ethnic conflict erupts due to offensive strategies and high degrees of uncertainty.

Especially when used as an alternative to conflict-reifying interpretations of ethnic conflict, security-dilemma analysis usefully highlights action–reaction effects locking the participants into an escalating process of increasing radicalism and violence. These insights come at a high price, however, since questionable assumptions are carried over from type-B theorizing. For example, Posen’s interpretation of the Yugoslav case reifies the ethnic groups while ‘neglecting both the role of the state in constructing these identities and the cynical rewriting of history that is taking place to fit present political purposes’ (Lapid and Kratochwil, 1996b: 115). Moreover, the substitution of ethnic groups for states confuses the fundamental nature of the units and exaggerates the cohesion of ethnic groups (cf. Crawford, 1998; Gagnon, 1994/5; Laitin, 1998: ch. 12). Statistical work casts further doubt on the viability of ‘security dilemma’ applications (Sambanis, 2000).

Formal modelers have not been lagging far behind their qualitative colleagues in reacting to the changes following the Cold War. Most of the initial modeling effort formalized and elaborated on the security-dilemma logic at the level of ethnic groups. Viewed more broadly, this rationalistic literature highlights information failures and problems of credible commitments (Lake and Rothchild, 1998b).

James Fearon’s (1994) paper represents one of the first systematic efforts to use game-theoretic tools to model ethnic conflict in IR. Inspired by Posen’s theory of the ethnic security dilemma, Fearon provides an alternative to the ‘ancient hatred’ explanation of the ethnic conflict between Serbs and Croats. Subsequently published in revised form, Fearon’s (1998) model features a government and an ethnically distinct minority group. Once the geopolitical situation shifts from hierarchy to anarchy, inter-ethnic contracts can no longer be expected to hold. Faced with this structural transformation, both actors now face a serious commitment problem (see also Bates et al., 1998 and Weingast, 1998).

In a more philosophical essay, Russell Hardin (1995) proposes a qualitative rational-choice framework as an antidote to ‘primordialist’ explanations of ethnic conflict. Following rationalistic principles, Hardin comes to the conclusion that ethnicity can solve collective-action problems within groups while creating sub-optimal outcomes between them driven by pre-emption and a lack of centralized control: ‘Even people of good will can be panicked into escalating moves by the fear of failing to respond to an aggressive adversary, especially when pre-emptive responses might be vastly more beneficial than later responses’ (1995: 160).

As Posen’s interpretation, these rational-choice models usefully highlight the strategic aspects of ethnic politics. Due to the strong isomorphism with the security-dilemma setup, however, the formal approach suffers from the same structural weaknesses as the qualitative version. First, the reification of ethnic groups as unitary rational actors is problematic, because, in the absence of centralized enforcement, it is unclear how such loosely aggregated communities could exist in the first place despite free-riding at the individual level. This problem does not disappear by moving the strategic analysis up to the group level (e.g. Hardin, 1995: 144). Second, because of the commitment to non-cooperative game theory, ethnifi-cation of politics and the breakdown of trust remain outside the scope of rational-choice analysis.11Third, while formal approaches of this type go beyond qualitative theorizing in offering causal micro-mechanisms, it does not mean that they are empirically accurate or even particularly plausible (Elster, 2000).

Compared to the small number of statistical studies exploring the impact of ethnicity on interstate war, the quantitative literature on ethnic minorities within states has advanced much further. These advances are due to Ted Gurr’s (1993) massive data collection effort entitled Minorities at Risk. As its title suggests, the project contains data on the relations between states and their culturally distinct minorities. Rather than trying to summarize the rather complicated causal model that Gurr proposes, however, I will here turn directly to Fearon and Laitin (1999), who improve and reinterpret Gurr’s original data set. Applying a simple multivariate model to the modified data, Fearon and Laitin conclude that violence tends to increase robustly with economic indicators, such as GDP per capita and growth, settlement patterns including a rural base, and the presence of rough terrain. In addition, they report less robust results on minority size and the presence of nearby co-ethnic states. More importantly though, they fail to find a relationship between ethnic violence on the one hand, and the size of the cultural differences, the degree of the state’s discrimination, and democracy on the other hand (though see Vanhanen, 1999).

While providing useful information about large-scale conflict patterns, quantification also poses problems. Apart from obvious difficulties of measurement and observability (cf. Gurr and Harff, 1994: 93), there are tricky conceptual issues associated with the operationalization of culture and identity. For example, Yee (1996: 102) suggests that the behaviorists’ ‘commitment to empirical analyses of observable behavior that can be tested and falsified renders them reluctant and ill-equipped to analyze the intersubjective meanings and symbolic discourses that give ideas their causal effects.’ More specifically, hypotheses linking ‘cultural difference’ to political action, such as nationalist warfare, implicitly accept an essentialist notion of culture and thus cannot problematize ‘politically relevant identities’ (Crawford, 1998: 18; see also Barth, 1969). Thus, it is necessary to complement these studies with qualitative process tracing. Selection bias is another serious problem haunting projects relying on the data set, for it is far from clear what ‘at risk’ really means (Fearon and Laitin, 1999). Other tricky problems pertain to the absence of ‘majority’ groups, such as Russia within the former Soviet Union, as well as the failure to consider the endogeneity due to historical path-dependence and action ‘spilling over’ state boundaries (see types 2 and 4 below).

Explaining the Behavior of Individuals in Inter-ethnic Relations

Not all type-2 studies treat ethnic groups as unitary rational actors. A technically sophisticated solution is to let the individual group members act strategically. While keeping the macro-identities reified, Fearon and Laitin (1996, 1999) propose models that disaggregate micro-behavior. In these cases, individuals belonging to either of two groups do the acting. Attacking the puzzle of inter-ethnic cooperation, Fearon and Laitin (1996) model group interaction as a series of pairwise encounters in a ‘matching game.’ They show that such a setup yields two equilibria, one leading to conflictual spirals and the other supporting decentralized, in-group policing. In addition, Fearon and Laitin (1999) attempt to explain the empirical regularities measured at the group-level by pitting a unitary state against an n-member minority.

Though representing an advance over the group-reifying models, these papers still fail to problematize the groups themselves, since their boundaries are imposed exogenously by the initial labeling of the individual actors. Furthermore, like other reasonably complex game-theoretic models, they assume high levels of rationality without proving an intuitively plausible mechanism tracing how the equilibria can actually be reached. In that sense, neither the creation of trust nor its breakdown is explained.

Problematizing States While Reifying Nations: ‘Systemic Nation-Essentialism’ (Type 2)

One way of relaxing the double reification assumed by type-1 approaches is to let state identities vary while national identities remain fixed. In many ways, this category of work follows the example of IR constructivism (type C). Due to the state-centric assumptions, however, these scholars have endogenized state identities rather than national or ethnic ones. For example, Wendt (1994: 387) states that ‘nationalism may be in part “primordial” and thus inherent to societies’ self-conception as distinct groups’ (see Pasic, 1996 for a critique).

To find nation-reifying studies that endogenize states’ boundaries, it is necessary to turn to scholarship that has normally not been associated with IR constructivism. While exogenizing national identities, these perspectives focus on disintegration and irredentism.

Disintegration

Analyses of this type explain the presence or lack of states’ fragmentation in reified cultural terms. Simply put, territorial rule is assumed to require ethnic cohesion, implying that in its absence state collapse becomes likely (Collins, 1986: ch. 8). Disintegration comprises a spectrum of possibilities ranging from weak types of decentralization, such as separatist claims for cultural autonomy, to full-fledged secession (Horowitz, 1985: 231). Empirical evidence shows that while many secessionist attempts were made during the Cold War period, before the last wave of imperial dissolution, secession has rarely been successful (e.g. Hechter, 1992), though this has changed in this post-Cold War period.

The main debate concerns the role of cultural identities as a driving force in secessionist projects. While some theorists emphasize the non-instrumental nature of secession (e.g. Connor, 1994: ch. 6), others are more prone to stress material factors (e.g. Gourevitch, 1979; Hechter, 1992). Despite the elegance of parsimonious explanations, however, mixed ones probably provide a better account of the underlying processes than more one-sided explanations. Offering an interpretation that draws on both cultural and rationalist reasoning, Donald Horowitz (1985) suggests that while advanced groups engage in interest-based calculations, their backward counterparts are more likely to secede regardless of the costs.

Irredentism

The concept of irredentism stems from the Italian state’s claim to ‘redeem’ its ethnic brethren in the Habsburg provinces Venice and Trento at the end of the nineteenth century. It can be defined as the attempt to break loose and integrate a territory populated by a ‘kin’ population into the state of its ethnic ‘homeland’ (e.g. Chazan, 1991b). Either the homeland state already exists (as in the case of newly unified Italy) or it has yet to be created (as in the Kurdish case). Thus, irredentism can be seen as a combination of nationalist secession and integration: ‘Irredentism involves subtracting from one state and adding to another state, new or already existing; secession involves subtracting alone’ (Horowitz, 1991: 10).

The differences between irredentism and secession do not imply that the two phenomena are unconnected (Horowitz, 1985, 1991). Indeed, a successful bid for irredentism must include secession. The question, however, is whether the process ends there or whether nationalist integration ensues. The decision to secede and unify hinges on a chain of conditions, including power-related considerations. For example, it could be that the diaspora minority’s leaders are unwilling to give up power to the homelands’ politicians or that the latter are reluctant to upset the domestic balance (Chazan, 1991b).

The factors governing the spread of irredentism are typically both regional and systemic. Regional explanations attribute irredentist pressures to a discrepancy between the cultural and political maps in a particular area. The more ‘stranded diasporas’ there are, the more likely secessionist campaigns become (Chazan, 1991b; Van Evera, 1994). Systemic accounts trace the spread of secessionism from one state to the other. Whereas some analysts warn against the risk of diffusion (e.g. Halperin, 1998; Heraclides, 1990; Kuran, 1998), others are less alarmed (e.g. Fearon, 1998; Saideman, 1998).

Diffusion of irredentist politics can be attributed to several mechanisms. First, ethnic conflict may proliferate due to physical externalities through the spread of violence and refugees from one state to the other. While the particularist logic of ethnic nationalism is to some extent ‘self limiting’ (Fearon, 1998), the possibility of imagining large pan-national units should not be entirely discounted. Second, behavioral demonstration effects may also drive diffusion. Both minority élites and homeland politicians could get inspiration from successful irredentism elsewhere (cf. Kuran, 1998).

Despite this list of explanations, culturally reified approaches overlook one of the most important sources of diffusion, namely demonstration effects pertaining to the very principle of nationalism. Once the direct link between cultural ‘raw material’ and political identities is broken, it becomes possible to imagine a mobilization effect spreading the ‘modular’ idea of nationalism (Anderson, 1991) to unmobilized areas where it had not previously been present. While further discussion of this topic will have to be postponed to the survey of type-4 studies, the approaches described in the next section take an important step in this direction.

Problematizing Nations While Reifying States: ‘Comparative Nation-Constructivism’ (Type 3)

Given their close interest in societal dynamics, it is unsurprising that comparativists have got farther than IR specialists on the ‘long road from primordialism to a more constructed view of nations’ (Laitin, 1998: 334). Due to the emphasis on country comparisons, however, these contributions usually reify the states as cases:

The boundaries of states are territorially defined, and despite border wars, remain fixed over time. Classic theories of international relations assume fixed boundaries. But the boundaries of nations are defined by the cultural stocks of people, and these boundaries are forever ambiguous. (Laitin, 1998: 340)

As opposed to reified perspectives that resort to ‘groupism’ (Brubaker, 1998: 292), this ‘sociational’ focus on the very existence of national collective identities forces the analyst to consider the puzzle of how nationalism ties together large numbers of people, spans over long time periods and vast territories (Simmel, [1908] 1992).14 The best theoretical solution to this puzzle remains Ernest Gellner’s (1964) classic analysis in Thought and Change. Gellner asserts that, in contrast to pre-modern society which was based on direct interpersonal relationships, the large scale of the nation requires abstract categorization: ‘In modern societies, culture does not so much underline structure: rather it replaces it’ (Gellner, 1964: 155; see also Anderson, 1991; Calhoun, 1991; Gellner, 1983). As opposed to illiterate peasants, modern citizens need to possess a modular communicative capacity that can only be acquired through formal education in a high language.

Thus, modern national identities should not be confused with pre-modern, ethnic cores: ‘Nationalism is not the awakening of nations to self-consciousness: it invents nations where they do not exist–but it does need some pre-existing differentiating marks to work on, even if … these are purely negative’ (Gellner, 1964: 168). This fundamental constructivist point shows that there is no simple one-to-one correspondence between ancient, cultural groups and modern political identities (Calhoun, 1997: 48; cf. Barth, 1969: 14; Cederman, 2001a).

Following the seminal conceptualization of Gellner’s compatriot Mroslav Hroch (1985, 1993), nationalization projects can be divided into three consecutive phases: cultural manipulation, nationalist politicization and nationalist mass mobilization, each of which will be discussed in turn.

Cultural Manipulation

Since high culture seldom comes in pre-fabricated bundles ready to be politically employed, the cultural ‘raw material’ first needs to be ‘engineered.’ This is what Hroch (1985) calls the period of ‘scholarly interest.’ The typical situation features cultural fragmentation that is overcome through centralized assimilation or ‘horizontal’ standardization (Miller, 1995: 33). Either one culture is chosen and then promoted through more or less coercive means, or an entirely new conglomerate of preexisting bits and pieces are fitted together (as exemplified by the Norwegian linguist Å sen’s invention of the new high language Nynorsk, an amalgam of various dialects). In the case of cultural differentiation, new cultural barriers are erected, such as seemingly arbitrary linguistic rules emphasizing the distinction between the Serb and Croat communities (for more examples, see Smith, 1991).

Both integrative and disintegrative projects require a fair amount of intellectual work. Although diffuse political purposes may figure in the background, this phase is usually the responsibility of historians, linguists and other humanists (Hroch, 1985). Mythological historiography is of particular importance since cultural standardization calls for not just remembering but also selective forgetting (Connor, 1972; Renan, [1882] 1996).

Nationalist Politicization

Culture can be politicized in two ways: either it develops as a state-framed project or its role is state-seeking. The latter case includes both unification nationalism and national mobilization in opposition to a pre-existing, multinational state. Whether the state plays an opposing or supporting role, type-3 theories view it as an exogenous factor influencing nationalist activities.

In most scholarly texts, nationalist mobilization starts with a small élite of politicians who turn the cultural agenda into a political one (though see Kaufman, 1996 for a bottom-up account). Ideal-typically, the key actors are either intellectuals extending their cultural projects or political opportunists who use cultural identities for their own instrumental purposes. Most nationalist movements contain a mix of both nationalist activists and political entrepreneurs (Lake and Rothchild, 1998a: 19; see also Snyder, 2000). Exemplified by Garibaldi and Cavour respectively, romantic activism or cynical élite manipulation complement each other (Brubaker, 1998; Bunce, 1999).

In addition to conscious attempts to politicize culture, the explicitly political phase of nationalist mobilization is sometimes triggered inadvertently by a central state’s mobilizing for other purposes than national identity formation. State leaders may opt for assimilation policies in the name of political participation, or economic and administrative efficiency (Deutsch, 1961) while overlooking the identity repercussions of such projects (Cederman, 1997: ch. 7). Wherever such mobilization processes outpace cultural assimilation, counter-state nationalism is likely to follow (Deutsch, 1953a, 1953b). The classic case involves the introduction of the dominant nation’s language as a standardized lingua franca in a multi-ethnic setting, as illustrated by Vienna’s attempt to replace Latin with German as the standard administrative language within the Habsburg Empire (Breuilly, 1982). Other examples include the inadvertently Russifying effects of Gorbachev’s political reforms (e.g. Bunce, 1999; Gitelman, 1992) and the colonial powers’ self-defeating exploitation of the Third World territories (Breuilly, 1982; Mayall, 1990).

According to Gellner’s (1964, 1983) original theory, a state’s centralized mobilization will trigger centrifugal nationalism in the cases where the minority élite’s access to the dominant high culture is blocked (see also Deutsch, 1953b). While some cultural traits are harder to change than others, ultimately what matters is not so much the cultural differences themselves, as the political selection of the traits to be included in a national identity. Applying this Gellnerian logic to nationalism in the postSoviet republics, Laitin (1998) argues that, to a large degree, the form of nationalist activity in these areas reflected the Soviet Union’s nationalist policies.

Nationalist Mass Mobilization

The need to conceptualize national identities as abstract categories becomes particularly acute once the question of mass mobilization is addressed. Rather than being based exclusively on a tight network of interpersonal communication among élites, the nation acquires its cohesion thanks to institutional mechanisms that bring together large groups of people most of whom have never met each other and never will (Anderson, 1991).

As in the process of politicization, much hinges on whether the nationalist movement controls state institutions or not. If the nationalist élite controls territorially based organizations, the task of reaching out to large numbers of people obviously becomes much simpler than without such resources. These instruments include state policies in a broad set of areas. As stressed by Gellner (1964, 1983), education is the most obvious of these. The nation-state presupposes a mechanism ‘generating citizens,’ and no other state institution contributes as intensively to political and cultural socialization as universal schooling (Boli, 1989).

As we have seen, such a project requires a medium. Thus, few nationalizing states can do without a language policy (Brass, 1974). From a historical standpoint, the commercial printing press dominated the earlier phases of nationalism (Anderson, 1991). Today, modern mass media, especially radio and TV stations, contribute to national identity formation (Schlesinger, 1991). In addition, party systems both reflect and channel cultural affiliations within civil society (Rokkan, 1999). Less obviously identity-conferring projects, such as road building, legal unification and bureaucratic and monetary standardization, may also enhance mobility considerably and thus accelerate national mass mobilization (Deutsch, 1953a, 1953b; Weber, 1979).

In addition to the internal mechanisms covered so far, there is a set of external ones (Simmel, 1955). Historical evidence shows that many nation-states unified internally mainly through inter-state war (Colley, 1992; Mann, 1992, 1993). Beyond violence, other types of interaction may encourage nationalism, including commerce and immigration (Brubaker, 1992). It should be noted that internal and external institutional mechanisms depend on each other. For example, creating a national enemy may require negative stereotyping in school curricula.

Counter-state nationalists have to operate under much tougher conditions than those movements that enjoy the support of their ‘own’ state. Therefore, state-opposing mobilization tends to rely primarily on ethnic rather than civic mobilization, such as oral tradition, extended kinship networks, and autonomous churches. Yet, as noted above, multinational state élites sometimes unwittingly generate separatist nationalism by improving the infrastructural conditions for nationalist activity (Brass, 1980: 47; Olzak, 1983: 359). Even more ironically, stateled repression itself sometimes has a long-time disintegrating effect by creating a tougher selection environment in which only more inclusive, and thus more powerful, peripheral nationalist platforms survive (Cederman, 1997: ch. 8; Hannan, 1979). Concessions to nationalists equipping them with ‘ethno-federal’ institutions that attempt to buy short-term ethnic peace often compromise long-term integration, as illustrated by the Soviet and Yugoslav break-ups (Brubaker, 1996; Bunce, 1999; Motyl, 1990: ch. 6; Roeder, 1991).

The three-step logic of national mobilization begs the question how the nationalization process shifts from one step to the other. It would be a mistake to view Hroch’s scheme as a deterministic sequence that all national movements have to traverse. To be sure, some nationalist campaigns never make it to the last, mass-based stage. Others skip over the cultural construction phase thanks to preexisting cultural identities.

In spite of these differences, however, it is possible to identify general mechanisms that drive the transition from politicization to mass mobilization. As we have seen above, Russell Hardin (1995) suggests that nationalist coordination may help overcome collective-action dilemmas. Unlike pre-modern rebellions that can be more easily dealt with by the state’s repressive institutions through a strategy of ‘divide and conquer,’ national mobilization coordinates the resistance thus setting the ‘snow ball’ of collective action in motion. Once the critical threshold is reached, the process ‘tips over’ so that the expectation of more protest begets even more activity, and so forth (Granovetter, 1978; Schelling, 1978; see also Kuran, 1998; Laitin, 1998).

As it extends to the masses, nationalist mobilization depends crucially on categorical, abstract identification connecting the nationalist and activist entrepreneurs with the entire population. Thus this macro process cannot be reduced to interpersonal ‘games’ featuring conscious ‘choices’ of identity. In general, exclusive reliance on such a voluntarist approach underestimates the structural framing effects embedding identity-formation (Horowitz, 1975: 121). More nuanced, partly extra-rational explanations are thus needed to show how participants are ‘reprogrammed’ through media, educational systems, and other identity-conferring mechanisms and how the ‘selection environment’ favors certain identities over others.

Problematizing Both States and Nations: ‘Systemic Nation-Constructivism’ (Type 4)

Our ontological exploration has finally reached the most general ontological position that calls for a doubly constructivist perspective, endogenizing not only nations but also states. Very few political scientists have ventured this far in the quest for theoretical flexibility. As shown by the last section, most constructivist theories of nationalism analyze the emergence of national consciousness and its political repercussions within a given state. Gellner’s theory is no exception from this tendency: ‘Its explanatory capacities are implicitly limited to nationalist conflicts within states; it has little to say about nationalist conflicts between states’ (O’Leary, 1998: 61).

Nevertheless, in situations characterized by drastic boundary transformations, it makes little sense to talk about ‘domestic’ and ‘international’ realms. As the former Yugoslavia broke up, what was literally ‘inside’ became ‘outside’ and vice versa. Theoretical schemes and disciplinary categorizations that ‘hard-wire’ the internal–external distinction into units of analysis and causal explanations, rule out change of territorial and national boundaries by assumption.

Endogenization is not an end in itself for it carries with it conceptual costs due to the analytical framework’s many moving parts. Nevertheless, a sound research strategy needs to confront substantive puzzles with as few preconceived notions about the ontological status of the basic units as possible. Type-4 constructivism becomes indispensable when studying periods characterized by tumultuous and far-reaching change during which both states and national communities co-evolve (Cederman, 2000; Waever, 1993). It is also the most sensible theoretical choice for general macro-historical IR theories featuring changes of the basic actor types (Gilpin, 1981).

Explaining Turbulent Periods Marked by Co-evolving States and Nations

While transformation periods can be marked both by integration and disintegration, space constraints prevent me from covering the latter. The discussion of systemic nation-essentialism (type 2) illustrated that nation-reifying theories suggest candidate explanations of separatism and secession. But the leap from cultural diversity to polity-destroying action is far from automatic. Some of the nation-constructivist frameworks (type 3) discussed above go beyond intra-state mobilization by explicitly explaining secession. For example, rather than viewing the national components of the communist empires as pre-given units, Bunce (1999) suggests that the central regime itself contributed to creating these counter-state identities, and that together with historically contingent opportunity structures, the communist parties and the state’s culturally decentralizing and identity-reifying institutions paved the way for secession (see also Brubaker, 1996).

While enlightening, this constructivist core–periphery logic does not go far enough in cases where the disintegration process affects an entire system of states. In a pioneering article introducing the notion of the ‘Macedonian syndrome,’ Myron Weiner (1971) showed that nationalization often spills over state borders. This realization highlights a triad of actor types including nationalizing states, national minorities and national homelands (see also Brubaker, 1996: ch. 3). Inspired by this Balkan analogy, Weiner sketches a stylized scenario in which an irredentist claim by homeland or minority leaders unleashes a dynamic that radicalizes politics along national lines in the entire region thus either creating or reinforcing the three community types. Because of the emotional polarization, democracy and territorial moderation usually fall victim to processes of this kind. In the end, violence and, ultimately, secession typically follows.

In some particularly conflict-prone situations, there is not just one irredentist state but two, as illustrated by the recent case of Serbia and Croatia in the early 1990s. Such cases of ‘competing nationalization’ are hard to square with security-dilemma accounts, not merely because of these perspectives’ dyadic actor typology, but also because of the conflict-seeking preferences on both sides. In the Yugoslav case, both Serb and Croatian élites preferred conflict as a way to reconfigure the ethno-political map.

Most dramatically, ‘ethnocidal jerrymandering’ features authoritarian power-holders threatened by sudden democratization, and unsurprisingly has represented some of the worst human rights abuses since the Second World War. In Rwanda, the Kigali government’s meticulously planned genocide served to reshape the nation by simply murdering the Tutsi ethnic group and the entire political opposition whether Tutsi or Hutu (Prunier, 1995). Indeed, even in the most extreme cases, there is a perversely ‘constructive’ aspect of violence that can be exploited by ruthless leaders: ‘Genocide, after all, is an exercise in community building … Killing Tutsis was a political tradition in postcolonial Rwanda. It brought people together’ (Gourevitch, 1998: 95–6).

In the Yugoslav case, Milosevic’s notorious decision in 1990–91 to combine nationalization of his own federation with its territorial ‘amputation’ in order to vindicate the principle of ethnic nationalism illustrates the same point (Gagnon, 1998). Despite the ensuing warfare, these priorities suited the late Croatian president Tudjman as well. Thus, explanations that reify state and national boundaries cannot provide an accurate description of these strategies. More ominously, treating groups’ corporate identities as if they were unproblematic even risks playing into the hands of the perpetrators of genocide by retroactively legitimizing their preferred group boundaries (Gagnon, 1998; Lapid and Kratochwil, 1996b).

Macro-historical IR Theory

Having so far considered theories of ‘the middle range,’ it is appropriate to end this chapter on a more ambitious note. Rather than attempting to capture ‘covering laws’ supposedly governing interactions between nicely compartmentalized, ahistorical units of analysis, a focus on nationalism invites us to view IR in broadly macro-historical terms: ‘The study of international relations, like much of social science, is a branch of history’ (March and Olsen, 1998: 969).

Expressed schematically, the challenges involve capturing three macro-historical transitions in state–nation space (see Figure 21.3). As opposed to Figure 21.2, the quadrants here describe world-historical configurations rather than regional cases of nationalism. This holistic schema allows us to sketch the evolution of world politics since the Middle Ages in terms of actor types. Thanks to steadily improving communication technology and administrative and organizational innovations, the infrastructural reach of power centers gradually improved during this period (Mann, 1986).

Figure 21.3 Three macro-historical transitions in state–nation space

This gigantic process began with a medieval situation in which neither the state nor the nation existed in their modern forms (Ferguson and Mansbach, 1996). The first transition saw the birth of the modern, territorial state in Western Europe led by absolutist monarchs and finally justified by the doctrine of territorial sovereignty as enshrined in the Westphalian Peace (e.g. Spruyt, 1994; Strayer, 1970; Tilly, 1990; though see Krasner, 1999). The next major transition featured the move from territorial to popular sovereignty (e.g. Hinsley, 1986; Holsti, 1991). The American and French Revolutions irrevocably reversed the descending logic of dynastic legitimacy into an ascending notion vesting sovereign power in the nation (Calhoun, 1997: 4–7) and created the doctrine of national self-determination (Mayall, 1990; Neuberger, 1995). Partly implementing these historical novelties, the state system since the Napoleonic wars has developed in acute tension between territorial and national sovereignty (Barkin and Cronin, 1994).

By explicitly endogenizing both states and nations, perhaps Otto Hintze ([1902] 1975: 161) provides the most concise summary of this co-evolutionary two-stage process:

States are created by war, colonization, conquest and peaceful settlement, through amalgamation of different parts and through their separating from each other; and all this is bound up with an alternating process of intermingling and separation of races and civilizations, and languages. The European peoples have only gradually developed their nationalities; they are not a simple product of nature but are themselves a product of the creation of states.

Is world history moving out of the post-Napoleonic constellation? It is not just representative democracy and the welfare state that are arguably being threatened by globalization and supranational integration, but the principle of nationalism itself. These trends could generate a reversal to a ‘neo-medieval’ situation devoid of states and nations (Bull, 1977; Ruggie, 1993). Since nationalist ideology serves as the prime legitimation of the system’s constitutive unit, the nation-state (Meyer et al., 1997), it seems unlikely that systems change will ever be properly understood without a macro-theory of nationalism.

Yet, very few IR scholars have attempted to place nationalism in its world-historical context. Karl Deutsch and Ernst Haas belong to the most important exceptions. Karl Deutsch’s (1953b) essay ‘The Growth of Nations’ remains equally refreshing more than half a century later (see also Deutsch, 1953a, 1969). Unlike much of the contemporary literature, Deutsch not only distinguishes between national communities and states, he also investigates their historical co-evolution. For example, the article traces the slow unfolding of cultural assimilation prior to the age of nationalism, and the subsequent acceleration of mobilization after this threshold. Moreover, Deutsch explicitly relates this development to geopolitics by emphasizing the mobilizational impact of nationalism on power politics without losing sight of the potentially centrifugal consequences of supra-national mobilization projects. This is a useful reminder that nationalism can lead to both integration and disintegration of states, a fact evidenced by the end of the Cold War.

Similarly analyzing competing centralizing and decentralizing processes, the first volume of Haas’s (1997) magisterial study explores nationalism in five countries. As would be expected from Haas’s previous work on nationalism and integration, his recent book follows Weberian lines and adopts an instrumental-constructivist perspective. His main argument is that while different forms of nationalism can lead to political integration, progress is only possible on the basis of liberal nationalism. Citing examples such as India and Switzerland, Haas downplays the role played by cultural cohesion as the key to state-building.

Yet it is hard to avoid the conclusion that IR theory remains curiously ill-equipped to describe, let alone explain, ‘systems change,’ defined by Gilpin (1981: 40) as the most fundamental type of historical transformation affecting the very nature of the actors of the international system. Based on this terminology, the nationalist transformation of the state system following the French Revolution constitutes a macro-historical process that could be labeled nationalist systems change. Due to their rigid assumptions of actor-type invariance, structural approaches such as Waltz’s neorealism are incapable of tracing systems change in general, and nationalist systems change in particular (cf. Cederman, 1997; Ruggie, 1983: ch. 6). Despite its focus on dynamics, even Gilpin’s (1981) book refrains from systematically exploring systems change. By classifying the German unification process as a case of ‘systems change at the level of intra-German politics’ (1981: 40), he loses sight of the long-term geopolitical consequences triggered by the French revolution.

By contrast, a ‘sociational’ rewriting of IR theory that problematizes both states and nations as general actor types and specific historical entities promises to cast new light on events that have so far been viewed as mere inter-state exchanges unrelated to nationalism (Simmel, [1908] 1992). For instance, rather than viewing the world wars as failures of a culturally neutral balance of power mechanism, they could be interpreted as reactions to the entire state system’s nationalization with the disruptive collapse of the Austrian-Hungarian Empire together with the emergence of Germany as a newly unified nation-state at the very center of the system (cf. Holsti, 1991; Schroeder, 1972).

Conclusion

What can be concluded from this inventory of recent IR literature on nationalism and ethnicity? Given these topics’ under-theorized status at the end of the Cold War, the current situation offers more reasons for optimism. After a somewhat hesitant start in the early 1990s, IR scholarship on these topics has begun to show signs of maturity. The general tendency of the intellectual development is pointing away from reified and imprecise terminology and toward a more flexible ontology problematizing both territorial and cultural units whenever theoretically fruitful. This marks a potential break from the last decades’ inter-paradigmatic debates that were driven primarily by intra-scientific assumptions. But there is still plenty of room for improvement.

Conceptual wrangles continue to haunt the discipline. While treating states and nations as synonyms and mixing nationalism with ethnicity may make little difference in many situations, this practice causes serious problems where these phenomena do not coincide. Mistaking the Soviet Union for a nation-state, for example, may have contributed to the failure not just to predict, but more fundamentally toimagine, the end of the Cold War (Kratochwil, 1993). Although it would be both utopian and pedantic to expect IR scholars to replace references to the ‘national interest’ by the ‘state interest’ wherever they mean the latter rather than former, increased awareness of the distinctions would prevent unnecessary confusion.

Compared to rationalist and realist scholars, IR constructivists have been surprisingly slow to address nationalism and ethnicity. This is all the more ironic given the obvious relevance of constructivist approaches to nationalism and ethnicity. The reason for this unfortunate delay appears to be yet another inter-paradigmatic distraction that has diverted much of the IR constructivists’ attention toward combating their ‘theoretical others’ rather than worrying about substantive puzzles. This is obviously not just the constructivists’ fault, but is also due to the efforts to establish constructivism as a legitimate approach in IR, which in itself is a precondition for making progress in studying nationalism. Now that it is finally becoming acceptable to use concepts such as identity in mainstream IR, however, constructivists should be able join forces with constructivist colleagues outside IR thus liberating themselves from arbitrary, self-imposed ontological restrictions.

Rationalist studies of ethnicity and nationalism have much to offer, especially because they serve as an antidote to the cultural determinism that has plagued parts of the qualitative literature. More often than not, however, the rational-choice approach has been framed in narrowly voluntarist and utilitarian terms, thus exerting an unfortunate, constraining influence on theory-building. All the same, some of the most advanced rational-choice work acknowledges the relevance of constructivist thinking and strives for tighter conceptual integration of the two perspectives (e.g. Fearon and Laitin, 2000). What is still missing, however, are models that embed both calculative and norm-driven behaviors in a macro-historical framework with explicit representations of endogenized collective identities.

Apart from elucidating the topics themselves, future IR research on nationalism and ethnicity could serve as an analytical wedge breaking up the ontological closure that still characterizes mainstream IR theory (see type B). Few theorists have realized what a fundamental challenge nationalist and ethnic processes pose to standard conceptualizations of world politics. An explicit focus on nationalism, viewed as a macro-historical family of processes featuring abstractly mediated collective categorizations and boundaries, forces the analyst to consider not only the creation and dissipation of collective identities, but also their continued maintenance. Instead of conceiving of the state system as a passive theoretical backdrop, then, comprehensive theories have to trace the historical emergence and interactions of states and national communities.

Finally, the preceding points have direct consequences for policy-making. Whether the issue is supra-statist democratic governance or prevention of nationalist violence, future policy solutions hinge critically on an understanding of the underlying politico-cultural processes. Those who believe that national identities are both inert and ethnic are reduced to recommending conflict-building methods or even drastic separation of groups. If, however, it is assumed that political identities are at least in principle detachable from the cultural ‘raw material,’ identity change might become more than an academic possibility. Even under the assumption of ‘sticky identities,’ explicit analysis of the processes responsible for the creation and maintenance of national identities could help identify realistic ways to manage multicultural polities and to prevent and overcome nationalist conflict. Without tighter conceptual integration of nationalism and ethnicity as central ingredients of IR theory, the discipline will remain hopelessly incapable of identifying future policy challenges.

As a postscript, it could be noted that there are two ways of conceptualizing nationalism and ethnicity within a handbook of IR: either as yet another ‘issue area’ deserving a chapter or as a part of the field’s theoretical core. I leave to the reader to decide which way to approach the topic. IR scholars, whether they wish it or not, will be grappling with these issues over the coming decades.