Political opening and nationalist media On the path to civil conflict

Transcription

1 National Centre of Competence in Research (NCCR) Challenges to Democracy in the 21 st Century Working Paper No. 67 Political opening and nationalist media On the path to civil conflict Nino Abzianidze NCCR Democracy University of Zurich Affolternstr. 56 CH-8050 Zurich January 2014

2 Abstract Building up on the debate about the conflict-inducing nature of the democratization process this paper delves deeper into the mechanism linking these two variables. Previous research has been mostly salient in acknowledging exclusionary nationalism in media as one of the main channels transmitting risks of conflict in democratizing environment. However, the extent and the nature of these nationalistic appeals in media content have not been analyzed systematically so far. To make sure that this crucial entity of the hypothesized mechanism is present empirically the present single case study focuses on identifying (1) if the instances of increased political participation (periods of national elections and street protests) lead to the intensified nationalistic rhetoric in media content and (2) what is the actor constellation around these nationalist appeals present in media. For this purpose a quantitative, strongly actor-centered, content analysis of the Georgian print media is conducted. The case of Georgia is selected as the most-likely case of democratization and civil conflict. Exclusionary nationalist appeals are operationalized and measured by the original codebook. The analysis comprises period of democratization in Georgia from 1990 up to the re-occurrence of the violent conflict in Key words: Nationalism; Content Analysis, Democratization; Civil Conflict; Causal Mechanisms. 1

3 Table of Contents Research question puzzle, relevance and aim... 3 Point of departure... 5 Theoretical framework... 6 Disaggregation - New approach to democratization-conflict link... 6 Exclusionary and Hostile nationalism... 8 Nationalism and conflict Hypotheses Elections and nationalism in the media content Elections, protest and nationalism in the media content Diverse actors diverse rhetoric Methodology Georgia the case of democratization and conflict Content analysis of exclusionary and hostile nationalism Sampling strategies Operationalization of exclusionary and hostile nationalism Coding strategy Appendix Appendix Appendix List of References

4 The present working paper summarizes the research design of my doctoral dissertation. Several preliminary hypotheses are proposed and the strategies for the data collection to test those hypotheses are presented in details. Research question puzzle, relevance and aim Did Mansfield and Snyder make the right conclusion in the mid-1990s when they wrote the following: Promoting this kind of [professional] journalistic infrastructure is probably the most highly leveraged investment that the West can make in a peaceful democratic transition (Mansfield and Snyder 1995)? This conclusion followed their findings about the risks of conflicts which democratization, defined as the process of transition from an authoritarian regime to a more democratic political system, bears in itself. The centerpiece of American foreign policy in the 1990s was directed towards the worldwide promotion of democracy and it triggered the intense debate over the hypothesis claiming that democratization can lead to a violent conflict. A number of scholars has found statistical evidence supporting this proposition (Mansfield and Snyder 1995, 2005; Gleditsch 2002; Cederman, Hug, and Krebs 2010) 1. The casual mechanism behind the link between the democratization process and inter-state as well as intra-state conflicts strongly relies on the argument that the opening of the political arena can lead to elites pursuing an intensified exclusionary nationalist rhetoric on the marketplace of ideas, particularly in the media. This in turn increases the risks of conflict among contestant groups. This theoretical argument is widely recognized to be applicable to violent inter-state conflicts (Mansfield and Snyder 1995; Snyder and Ballentine 1996; Snyder 2000; Mansfield and Snyder 2002) as well as to violent civil/ethnic conflicts (Cederman, Hug, and Krebs 2010; Gleditsch 2002; Mann 2005) in the democratization context. While exclusionary nationalism in the media is acknowledged as one of the main channels transmitting conflict risks during the democratization period, it is the least studied feature of this theoretical framework so far. Most of the debate around the theory is shaped by the critique of insufficient measurement strategies for the concept of democratization. Scholars refer to arbitrary measures for identifying regime changes, not being precise enough in distinguishing the effects of democratization from those of regime change generally and not considering the difference among the types of democratization (Thompson and Tucker 1997a, 1997b; Enterline 1996; Wolf 1996; Ward and Gleditsch 1998; Hegre et al. 2001; 1 For a comprehensive review see (Cederman, Hug, and Wenger 2008) 3

5 e.g.narang and Nelson 2009) 2. Unlike this strand of research, Snyder and Ballentine put more emphasis on the issues linked with the marketplace of ideas and the media in particular (Snyder 2000; Snyder and Ballentine 1996). However, their analysis is mainly concerned with structural factors such as media ownership issues and journalistic professionalism and leaves out the systematic analysis of media content itself. In his detailed case studies Snyder relies on the qualitative analysis conducted by Thompson (1994) for the case of Yugoslavia and provides some scattered citations in some of the other cases. This information, however, does not show systematic evidence whether the content of the media yields the patterns of nationalism hypothesized by the theoretical argument. This, instead, leaves the following questions without a proper answer: is exclusionary nationalism present in the actual media content? To what extent is it hostile and thus conflict-inducing? Who are the actors pursuing this kind of nationalistic rhetoric? Or is there any variance in kind or in degree of nationalistic rhetoric across different actors? While the structural factors are indeed important, if they have any influence, it is exactly the content of media through which they do. Therefore, it is instrumental to conduct a systematic analysis of media content with regard to the mechanism linking democratization to the conflict onset in order to either strengthen or maybe question the propositions of this mechanism. This is exactly the aim of the present interdisciplinary single case study, which combines premises of the fields of political and communication sciences. If the instances of intensified exclusionary and hostile (thus conflict-inducing) nationalism are identified in the media content, then the next question is what the conditions are leading to this intensification. The casual mechanism described above begins with the opening of the political arena for more participation. Moreover, due to the immense complexity of the concept, the most recent research proposed to disaggregate democratization and to focus the analysis on its particular factor, namely the competitive elections, which can be causally instrumental with regard to the risk of conflict (Cederman, Gleditsch, and Hug 2012). This approach is effective also because, by focusing on elections, which themselves are a form of political participation, it directly addresses the most important casual factor of the mechanism leading to conflict. In line with this approach, the present study concentrates its analysis on the instances of increased political participation and considers the latter as the most favorable condition for the intensification of exclusionary and hostile nationalistic 2 Since these debates are concerned particularly with inter-state wars and my proposal addresses intra-state wars, I leave the extensive review of this literature out. For more details see the indicated articles. 4

6 rhetoric in the media. Political participation can be institutionalized and exercised as elections but it can also be non-institutionalized and manifested in street protests and demonstrations. Both types of participation are considered in the present study; it looks at whether and in what constellation these instances lead to intensified nationalism in media content. Nationalism is analyzed with regard to two conflict-inducing aspects exclusion and hostility. Based on the argumentation presented above the present study will address the following research questions: RQ1: Do specific instances of the democratization process, such as increased political participation, lead to intensified exclusionary and hostile nationalism in media content? RQ2: What constellation of institutionalized and non-institutionalized political participation leads to intensified exclusionary and hostile nationalism in media content? Point of departure Following the Willsonian proposition about making the world safe for democracy, the end of Cold War was marked by the promotion of democratization by the United States of America in the former communist countries, with the hope that it would serve as an antidote to international war and civil strife (Snyder 2000; Cederman, Hug, and Wenger 2008). Perhaps this hope was picked up in Fukuyama s prediction of the end of history in terms of universalization of Western liberal democracy as the final form of human government (Fukuyama 1989). Surprising or not, from today s perspective, this optimism seems a bit overstated since the path of transition appeared to be longer and rockier than expected. The third wave of democratization coincided with chronic nationalist conflicts (Snyder, 2000: 15). Only the number of conflicts that occurred in the post-communist space can be sufficient to cast doubt about the road to democracy being mined (Przeworski, 1991: 51). The likelihood of violent conflict triggered by the process of democratization was explained by the causal mechanism resembling Huntington s (1968) proposition concerning the gap between high level of political participation and weak state institutions. Mansfield and Snyder (1995, 2000, 2005) hypothesized the mechanism which rests on diversionary elites manipulating the media to stimulate nationalist feelings among the public. More precisely, the authors argue that after the breakdown of authoritarian regime, political 5

7 participation increases although state institutions are too weak to regulate this increased participation. As a result, old elites feel threatened by the emerging new elites. Both old and new elites start to mobilize all resources at their hand to defend their threatened positions. Under the situation of imperfect institutions, they exploit their power to control the political agenda and turn to nationalist appeals in order to mobilize mass allies. According to the authors, these appeals can be based on almost any cleavages but almost always elites use exclusionary nationalism (Snyder 2000) to shape the content of information media and thus fuel the conflict with external forces. More recent studies argue that the same theoretical argument applies equally well to civil conflicts (Cederman, 2010: 378). Moreover, external actors come into play only after the processes described above lead to a civil conflict, which in turn diffuses further into inter-state clashes (Gleditsch 2002). Mann (2005, : 22) looks at the cases of ethnic cleansing and also finds democratizing countries being most vulnerable to ethno-nationalism. However, the mechanism he puts forward proposes alternative motivations of elites in forging nationalist conflicts. Namely, according to his argument these incentives are shaped with the will to entwine the demos with dominant ethnos generating organic conceptions of the nation and the state (2005: 3). While Mann understands elites incentives in a different manner, a special importance of political entrepreneurs in pushing (ethno-) nationalism and thus fueling the conflict in the democratization period is obvious in his argument as well. Theoretical framework Disaggregation - New approach to democratization-conflict link Disaggregating a casual mechanism. To lay some ontological foundation in the present study, it should be clarified that a causal mechanism is defined as a dynamic complex system, which produces an outcome through the interaction of series of interlocking parts, where each part is transmitting causal force (Glennan 1996, 2002; Beach and Pedersen 2013; Bunge 1997). For a hypothesized mechanism to work it is of a crucial importance that all its entities and properties are present (Hedström and Ylikoski 2010). Otherwise, in the absence of one part, the existence of the whole machine can be questioned (Beach and Pedersen 2013). Therefore, the essential element in any mechanism-based explanation is to confirm each entity of the mechanism with empirical evidence because it is exactly the rigorous empirical check of assumed entities that distinguishes a proper mechanism-based study from the mere one (Hedström and Ylikoski 2010). As Hedström and Ylikoski put it the empirical evidence 6

8 turns the possible mechanism in a plausible mechanism and may eventually lead to the identification of the actual mechanism (2010, :52-53). The theoretical argument behind the relationship between democratization and civil conflict yields a very complex causal mechanism, which is based on the elite's manipulation of the media for nationalist appeals under the conditions of increased political participation and weak state institutions. It combines in itself what Hedstörm et al. (1996, : 295) label as situational mechanism (how beliefs and preferences are formed), individual action mechanism (how beliefs and preferences form certain type of actions on individual level) and transformational mechanism (how individual actions can be transformed in a collective action). This complexity points to different research fields (such as psychology and sociology for instance) and makes it enormously difficult to confirm the existence of each of its entities with the systematic empirical evidence in a single study. Therefore, my approach to studying the link between democratization and conflict rests on examining a particular, yet crucial, entity of the hypothesized mechanism between these two variables. In other words, the mechanism in this study is disaggregated to its particular property, namely, the increased exclusionary nationalism in the media content under certain instances of democratization, which in turn is supposed to lead to a conflict. Two main factors make the analysis of media content justified at this point of debate. First, the mechanism states that the nationalist appeals pursued by elites in the democratization period are transmitted to the public through the marketplace of ideas (Snyder and Ballentine 1996), in which mass media plays a pivotal role since it is the source of information which reaches the broadest public in the shortest period of time. While structural factors such as partial monopoly of the media outlets and journalistic professionalism are important instances to consider as factors reinforcing elites ability of media manipulation, they do not necessarily account for the existence of certain patterns and extents of nationalist appeals in the media content. Therefore, identifying the latter empirically is crucial to improve the understanding of the hypothesized causal mechanism. The second reason is related to the first one: media content has never been systematically analyzed with regard to this theory. Disaggregating Democratization. As stated above, the present study will pursue the approach of disaggregating the concept of democratization to its specific casual factor. There are two main reasons for this decision. First, since the goal of the study is to test systematically a specific chain of a causal mechanism underlying a theory, it makes sense to concentrate the analysis on the most important causal factor. Second, the theory which has to 7

9 be tested in this study clearly posits increased political participation at the beginning of democratization process as the most important causal factor which drives the whole mechanism all the way to conflict (Mansfield and Snyder, 1995; 2000; 2005, 2007). 3 Therefore, instead of examining the effect of overwhelmingly broad concept of democratization on the extent of exclusionary nationalism in the media content, I will focus my analysis on its particular instance, namely political participation. While the latter is not the only feature of democratization, surely it is one of the most important ones, stressed by nearly all the definitions of democracy (e.g.dahl 1971; O Donnell and Schmitter 1986; Diamond, Linz, and Lipset 1990; Huntington 1991; Collier and Levitsky 1997; Levitsky and Way 2010; Schumpeter 1942; Alvarez et al. 1996). In addition, the Democracy Barometer has identified political participation as one of the main pillars of democracy (Bühlmann et al. 2011). Political participation can be institutionalized and occur in the form of universal suffrage, but it can also be non-institutionalized, reflected in anti-government political protests and demonstrations. My analysis will take into account both types of participation elections 4 and anti-government protests 5. Political participation will be considered as increased when election periods are accompanied by protests and demonstrations. The extent of nationalism in the media's content will be tested first, only with regard to the election periods and second, with regard to the periods of increased political participation, i.e. elections and protests coinciding. Exclusionary and Hostile nationalism Nationalism belongs to the terms characterized by some as definitional anarchy (Abdelal et al. 2009) in literature. Given that measuring the extent of nationalism in the media content will be the primary empirical contribution of this study, clarifying my approach to this concept is one of the most important tasks. The starting point of the conceptualization of nationalism in this study is the definition by Mansfield and Snyder, according to whom nationalism refers to a doctrine that people who see themselves as distinct in their culture, history, institutions, or principles should rule themselves in a political system that expresses and protects those distinctive characteristics (Mansfield and Snyder, 1996: 9-10; Snyder, 2000: 23). 3 Note the titles of Snyder s book From Voting to Violence and Mansfield s Electing to Fight 4 Only elections on national level will be considered. More precisely the analysis will be focused on presidential and parliamentary elections. 5 Anti-government protests are relevant since by definition civil conflict implies that government is one of the conflicting sides. 8

10 From the first inspection of this definition it becomes clear that the notion of nationalism is tightly linked with political system. Some authors equate this political system with the state (Weber 1946), while others argue that nations do not necessarily coincide with state boundaries but instead can demand wide-raging powers such as autonomy (Brubacker 1998; Cederman 2002). What is more important, this emphasis on the political is what helps to overcome the terminological confusion characteristic of political science literature in separating nation from ethnicity. While nations are envisioned as intrinsically political communities as source of sovereignty (Calhoun 1993, : 229), ethnies are seen as a cultural community defined by a shared belief in real or putative descent (Weber 1978; Smith 1991, :21; Cederman 2002, : 211). Markers such as common language, adherence to the same religion and common history are usually used to indicate such shared culture and ancestry (Wimmer, Cederman, and Min 2009). Some nations might comprise one ethnie while others can be multiethnic (Cederman, 2002). Nationalism per se might not be conflict inducing. For example, civic nationalism is supposed to dampen rather than increase the possibility of conflict since it is inclusive within the territory of state and accommodates all citizens under a certain institutional framework (Snyder, 2000: 80). What makes nationalism likely to trigger conflict is the extent of its exclusionary and hostile nature. I build my argument around this assumption and focus my analysis on these two dimensions of the nationalism concept. Exclusionary nationalism. Two main mechanisms linking democratization with conflict onset considered in this study stress two types of incentives elites can have for pursuing (ethno) nationalist appeals in media the need to hold on power (Mansfield and Snyder, 1995; Snyder, 2000) and the will to make the nation coincide with a particular ethnic group (Mann, 2005). Both of these incentives inevitably imply that a certain group is excluded from political power. Although the difference between the two is that in the latter case exclusion always occurs on ethnic lines while in the former case the line of exclusion is not exclusively defined in ethnic terms. More precisely, it resembles the diversionary theory of war according to which elites can reinforce mass support by raising tensions with external actors. In this case, exclusionary rhetoric is directed towards groups whether an anti-government opposition or a contestant ethnic group being blamed for supporting the external force against the nation. I conceptualize exclusionary nationalism along the lines of these incentives. Namely, the concept of exclusionary nationalism will refer to the type of rhetoric which stresses the distinctiveness of one group in terms of either ethnic (language, religion 9

11 and ancestry), historical or institutional terms, attaches certain interests and goals to this category and calls for excluding those groups from certain political or civic rights which either do not belong to it or act against those interests and goals. Hostile nationalism. If nationalist rhetoric is supposed to lead to conflict another premise which should be apparent in the rhetoric along with its exclusionary nature, is the extent of hostility directed towards the to-be-excluded group. By hostile nationalism I refer to the type of rhetoric yielding negative attitudes towards a contestant group, recalling previous events of enmity, eliminating opportunities for peaceful co-existence in the future and the allegation of war and violence as the only way of solving a problem. Exclusionary and hostile nationalism can exist in media content independently from each other, or might be present simultaneously with regard to the same group. However, it is equally important to determine the extent to which each of them dominates across the time of analysis since both of these dimensions are assumed to bear conflict-inducing risks. Nationalism and conflict While it is not the goal of this study to test empirically the effect of nationalism on conflict onset, it is important to discuss what are the factors driving this effect. In other words what patterns of nationalism should be identified in the media content to have implications for further conflict. It is argued that the impact of nationalism on war increases through factors such as a permanent remembering of past crimes by victims, responsibility attribution to the groups still on the scene, lack of contrition and repentance, allegation of hegemonic goals especially towards minority rights and divergent believes in mutual history of groups (Evera 1994). Varshney (2003) argues that any ethnic or national mobilization implies an ex ante possibility of violence since it relies not only on favoring passionately one group for domination but also confronting the other group. Based on a natural experiment Dumitru and Johnson (2011) show how states can trigger the sense of in-group affinity through inclusive policies while otherwise activate violent actions. Mansfield and Snyder argue that the nationalist rhetoric makes war likely through the construction of a nation s view with regard to the feasibility of success in cooperation and underestimating the costs of violence. Further, the risk of conflict resulting from increased nationalism is identified as a function of weak political institutions of accountability such as for instance impartial electoral commissions, well-organized political parties, corrupted bureaucracy, competent legislatures, etc. Due to their infancy, these institutions do not have enough capacity to regulate the increased political 10

12 participation and allow elites to act in the name of people without being fully accountable to them (Snyder 2000, : 55; Mansfield and Snyder 2002, : 303). Under these conditions, war can erupt as an indirect by-product of nationalist politics in democratizing regimes (Mansfield and Snyder 2002, : 11). While it is not the aim of this study to measure weaknesses or strengths of all the political institutions mentioned above, it will be possible to draw the implications of nationalist rhetoric in media content for further conflict if the empirical analysis takes all these discursive factors, discussed above, into consideration. Even more importantly, these implications can be drawn if the analysis is strongly actor-centered. Namely, analyzing statements of certain actors directed towards others and measuring the extent of exclusionary and hostile nationalism in them will make it possible not only to identify the extent of nationalism in media content but also to trace the dynamics of this rhetoric among actors namely to trace the process by which actors rhetoric becomes more exclusionary and hostile, who are the targets of this hostility, when do certain actors become more hostile, and, most importantly, whether the most hostile actors and most frequent targets of nationalist appeals resemble those groups involved in armed conflicts that followed. Hypotheses Elections and nationalism in the media content One of the main implications of Mansfield and Snyder s (1995, 2005, 2002; 2000) argument is the questioning of election timing in the countries that took over the road of transition from authoritarian regimes to democratic system. In the concluding remark of their most influential article, they state that holding elections in a country which lacks the political capacity of surviving it may institutionalize a highly divisive pattern of ethnic, sectarian, economic populist, or militarist politics that may not only risk war but hinder subsequent attempts to consolidate liberal democracy (1995, : 32). This was one of the main issues driving the consequent debate about the timing of elections. Although Carothers (2007) opposed the sequencing approach to democratization, i.e. creating certain institutional preconditions before holding open elections, later studies confirmed that early competitive elections 6 in post-authoritarian countries might indeed bear in themselves a threat to stability 6 By competitive election most of the studies refer to the formulation by (Hyde and Marinov 2010). According to them elections are competitive if opposition is allowed, if more than one party is legal and if there is a choice of candidates on the ballot. 11

13 (Brancati and Snyder 2010; Brancari and Snyder 2011; Cheibub and Hays 2010; Gustafson 2010; Cederman, Gleditsch, and Hug 2012; Strand 2005) 7. Especially when the stakes for elections in democratizing countries are higher than in mature democracies namely, there are greater benefits for winners and higher costs for losers (Höglund, Jastanrd, and Kovacs 2009). In addition, the outcome of the elections might become another factor triggering conflict. Not having recognized the official results of elections, sore-losers might initiate violence (Collier 2009; Przeworski 1991). If elections in dangerous places are not subject to certain rules of conduct contestants are driven to extremes (Collier 2009, : 15). Whatever incentives elites have for being extreme, whether need to hold on power (Mansfield and Snyder, 1995; Snyder, 2000) or to make a nation coincide with a particular ethnic group (Mann, 2005), they do need to gain mass allies in order to implement their goals. For this purpose, in both cases mass media become the most important platform for them to reach their target publics and deliver their messages. As both mechanisms, described in the previous section, yield these messages are mostly shaped by (ethno) nationalist appeals. Given that election periods (including pre-election campaign and post-election period) increase the competitiveness between contestant elites, I expect that the extent of nationalism in the media content will be high in the election period. Consequently the first hypothesis of the study reads the following: H1: in democratizing countries the extent of exclusionary and hostile nationalist appeals in media content intensifies during election periods as compared to those of no polling Elections, protest and nationalism in the media content While elections are the sine qua none of democracy (Huntington, 1991), it is not the only form of political participation. Given that democratization implies opening a political arena, it gives way not only to the institutionalized type of participation but also to the noninstitutionalized one (Kaase 1999). Namely, the opening of the political arena amongst others comprises allowing for freedom of expression and gathering. The latter frequently is reflected in anti-governmental protests in the environment where elections are not the only rule of the game to replace one government with another. As Przeworski (1991, : 59) points out when there are no institutions where newly emerged groups can present their views and negotiate their interests the only place left, where these groups can struggle for their values, is the street. If these two types of participation, elections and anti-government protest, occur at the 7 For alternative argument about the stabilizing effect of elections in democratization period see (Birnir 2007; Lindberg 2009) 12

14 same time the level of overall participation is higher. Consequently, elites who seek to obtain mass allies are more pressured to use all the means at hand to win a political contest. Therefore, their incentives to mobilize supporters are stronger during increased political participation. Furthermore, this is exactly the situation which, according to the theoretical argument behind the democratization-conflict link, leads to a conflict particularly when political institutions are week. I expect that the conflict-inducing rhetoric should be most intense in this instance. This discussion coupled with the argument that (ethno) nationalism is the universal element in mobilizing supporters in democratizing countries brings me to the following hypotheses, which are related to the first one: H2a: In democratizing countries the extent of exclusionary nationalist appeals in media content is the strongest when election periods are accompanied by anti-government protests compared to the election periods which are not so or to the periods of no polling. H2b: Exclusionary nationalist appeals in media content become more hostile when election periods are accompanied by anti-government protests compared to the election periods which are not so or to the periods of no polling. Diverse actors diverse rhetoric Both concepts of exclusionary and hostile nationalism, by definition, imply a certain type of relationship between or among different actors. Neither exclusion nor hostility can occur independently from actors. In order to exclude someone there should always be an actor which excludes and an actor which is excluded. Similarly, hostility only occurs when there is an actor which expresses hostility and an actor towards whom hostility is expressed. Given this nature of the concepts I deal with in this study, my analysis is strongly actor centered. Literature on the democratization-conflict link is almost consentient in attaching particular importance to actors. Usually, authors refer to actors in a cumulative way by talking about political, military, cultural, sometimes ethnic elites. However, they never question if there is any kind of variation, either in degree or in kind, in the rhetoric of these different types of actors. Because actors differ in their interests and motivations, their rhetoric and thus, the extent of exclusionary and hostile nationalism in it, might vary. For example, factors such as being elected or representing a governmental institution might affect the extent of exclusion and hostility in actors rhetoric. In addition, the latter might also depend on whether an actor represents military forces or cultural elite. 13

15 By analyzing the content of print media it is possible to fill in this gap and to determine which types of actors use exclusion and hostility as a mean of their rhetoric more frequently, which do so less, which actors are frequent targets, etc. In addition, it will be possible to observe if periods of elections or street protests lead to an increase of exclusion and hostility in the rhetoric of certain types of actors, while decreasing in that of others. With regard to this argument, actors will be compared based on the following factors: Elected vs. Non-elected; Governmental vs. Non-governmental institutions; Political vs. Military elites; Political vs. Cultural elites; Majority vs. Minority ethnic groups. This argument can be summarized as the following broad hypothesis: H3: The extent of exclusionary and hostile nationalism in media content will vary in the rhetoric of different types of actors. Methodology In this section I will present my methodological approach for the study. Namely, how the study will be conducted in practical terms. As I already described, the major goal of this study is to contribute to better understanding of the mechanism behind the democratizationconflict link. Given the complexity of this mechanism discussed in the previous chapter, I focus the analysis on this particular entity the media content - and examine whether a specific instance of democratization, namely, increased political participation leads to intensified exclusionary and hostile nationalism in the media content. Given that existence of each entity is crucial for every mechanism to work (Hedström and Ylikoski 2010), if the hypothesized pattern of nationalism is not present empirically then the plausibility of the whole theoretical mechanism can be questioned at least with regard to this particular link. It is exactly the goal of this research project to test whether this link of the mechanism is at work or not. For this purpose the study pursues what Neuendorf et al.(2010) call the integrative model of content analysis, which calls for the collation of message-centric data with other available empirical information regarding source, receiver, channel or other contextual state, whenever possible (p. 209). The study will be pursued in a single case study manner. There are four main reasons to this approach: first and the foremost, the goal of the study is to test the mechanism of a theory which has already been tested in large N studies, which instead calls for an in-depth analysis of a specific case. The investigation of a single case may allow for testing the casual 14

16 implications of a theory thus contributing with validating evidence to it (Gerring 2007, : 45). Second, although the proposed mechanism rests heavily on media issues, it has never been tested against the methods from the field of communication science. Therefore, the findings from the single case study might be informative in terms of identifying factors or patterns which have not been considered before. Third, a single case study gives an opportunity for observing variation in the longitudinal manner while holding other factors persistent (Gerring 2007, : ; Yin 1994, : 38). And finally, a single case study opens opportunity for conducting a content analysis for a longer period and thus for observing the dynamics of patterns of nationalism over time. If we translate the integrative model of content analysis as defined by Neuendorf (2010: 209) into a case study research we could land in what Blatter and Haverland (2012, : 45) call the intertemporal comparison - a method of comparing situations shortly before and after the change in the independent variable. In other words, what is observed is the co-variation in a temporal order. In addition, a (single) case study can be based on any mix of quantitative and qualitative evidence (Yin, 1994: 14). Applying these approaches back to the specific focus of this study, I should clarify that the empirical research will be pursued in two parts. A systematic analysis of the media content will be conducted in order to identify the patterns of nationalism in the selected case over selected periods. Next the process of increased political participation will be traced in order to identify whether these two trends show a covariance over time as predicted by hypotheses. Georgia the case of democratization and conflict There are some conditions that the selected case should demonstrate in order to fit this analysis. First, since the focus is on the mechanism, i.e. a path through which independent variable leads to a dependent variable, a case should be positive with regard to the theory. In other words, both democratization and conflict should be present in a case. In addition, in order to be a plausible case, it should demonstrate relevant variation in terms of the independent variable, which in the case of this study means that several types of periods should be available in the selected case. Namely, given the propositions of the hypotheses presented above a selected case should exhibit three types of periods election periods accompanied with street protests, election periods without protest in the street and periods of no polling. In addition, since media content has never been systematically analyzed with regard to this theory before, it is important to select the most-likely case because if a pattern suggested by the hypothesized mechanism is found as a result of a systematic content 15

17 analysis then this study can contribute to the better understanding of the mechanism; while, if the suggested pattern is not found in the media content, the whole mechanism might be questioned since the nationalist media is its central channel linking democratization to the conflict through the increased political participation. I select the case of Georgia because of the extent to which it meets the requirements listed above. First, Georgia represents one of the states which underwent (and is still undergoing) a process of transition from communist authoritarian regime to a more democratic system. Second, it is the case, where this transition process appeared to be particularly rocky. Right after the dissolution of the Soviet Union three nearly parallel conflicts erupted a governmental conflict which ended up with a coup d état (December 1991 January, 1992) and two ethno-territorial wars in Abkhazia and South Ossetia ( ). In addition, after nearly a decade of conflicts being frozen, 2008 was marked with the outbreak of a new wave of violence. Although, it should be mentioned that the latter war was different in kind for several reasons: first, even though Russia s involvement in the wars of 1990 was obvious, the war of 2008 turned explicitly into inter-state armed conflict between Georgia and Russia; second, participants from all sides of the armed conflict in 2008 were military forces and the involvement of civilians was minimal. Nevertheless, the fact is that the conflict started within the state, in or close to the regions where the wars in the beginning of 1990s were fought and what is more important the 2008 August war resulted in the recognition of Abkhazia and South Ossetia as independent states by Russia. According to my hypothesis, there are two types of variation which selected periods should demonstrate: election period versus period of no polling and institutionalized participation (elections) versus coincidence of institutionalized and non-institutionalized participation (street protests). Therefore, after selecting a case it is equally important to select time periods accordingly. For this purpose I reviewed the history of presidential and parliamentary elections in the selected case and listed them against the instances of the most important street protests and armed conflicts 8 which occurred during the democratization process. I count the starting point of democratization in Georgia several years before the declaration of independence. As a part of the former Soviet Union Georgia also underwent important shift with the opening of political space as a result of the reform project of glasnost in As Zürcher (2005, : 261) argues, glasnost created a public sphere were political 8 I made sure that the selected periods for no polling were not coincided with the ongoing armed conflict to eliminate conflict as a factor. 16

18 ambitions could be voiced. First and foremost, these changes strongly affected the media in the Soviet Union and its member countries since restrictions were eased on the topics banned before (Spern 1991, : ). Since the opening of the political space and the increased political participation are the crucial points of a theoretical argument driving the mechanism, including this period into analysis of the media content is important. By important street protests I mean those protests which raised political stakes for the elites on all the sides of the contest. For example, those which were followed by certain type of compromise from the government side either certain actor resigned, or the government was changed or extraordinary elections were appointed due to the unrest. After identifying the years, I selected the months based on the election dates: based on the assumption that elites start preparing for elections several month before it, my approach is to select three month before the election date; further, if elections are to determine post-electoral disagreement or protest, one month should be a sufficient time for nationalist rhetoric to erupt in the media content, therefore the election period is defined as election date, three months before it and one month after it (in total months for each election period). To avoid any bias, months during the periods of no elections will be selected randomly. The number of randomly selected months in the period of no polling will correspond to those in the other two periods. The procedure of period selection is represented in the tables in Appendix 1. I limit the scope of this study to the states formed after the dissolution of the former Soviet Union and Yugoslavia. The reason for this approach is the legacy these states inherited in terms of ethno-federal institutional arrangements. An ethno-federal state, as defined by Hale (2002, : 4), is a federal state in which at least one constituent territorial governance unit is intentionally associated with a specific ethnic category. The ethno-federal organization of these states might have had an implication for further development of nationalist sentiments since ethnic identities in this setting are enhanced, territorialized and politicized, which might create incentives for groups for increased demands for autonomy or independence (Bunce 2004; Brubacker 1996; Snyder 2000; Nordlinger 1979; Zürcher, Baev, and Koehler 2005). This situation might be even more dangerous when ethno-federation is asymmetric and grants some ethnic groups with privileges (such as e.g. right of secession, status of titular nation, right for education and media in their own language, etc.) while deprives others from them (Zürcher et al, 2005). In addition legacies with regard to the media systems in these countries are also similar. While the more explicit goal of this study is to generalize to the theory (Yin, 1994: 36), implicitly it aims at generalizing to the scope defined above. To make this possible 17

19 the final step of the research will be to complete the findings of this study with the evidence from other cases of former USSR and possibly former Yugoslavia. Content analysis of exclusionary and hostile nationalism The content analysis in the case of Georgia will be based on the print media due to the availability of the archive of all newspapers which have been published during and after the Soviet times in the National Library of Georgia and unavailability of the archives of other type of media outlets (such as TV programs, radio and online news). The fact that the archive of newspapers is kept in the library and is not digitalized is the main reason for doing human coding instead of computer assisted coding. In this section, I will present several major questions which should be addressed explicitly in the content analysis. These issues are: sampling strategies, operationalization of core concepts which should be measured and the coding strategy, which brings is the issue of the definition and identification of the unit of analysis in the text. Sampling strategies There are several levels of analysis in my project where sampling issues are of a particular importance. First of all, the relevant newspapers should be selected for the analyses; second, since the periods of analysis count 54 months, selected newspapers should be further sampled into the issues for analysis and finally a sample of articles should be defined. In the following paragraphs, I will present my approach to each of these tasks. The newspaper selection is one of the most complicated tasks to be solved. The main reason for this complexity is the unstable and somewhat chaotic setting of the (print) media landscape over the period of analysis. First and the foremost, right after the dissolution of the Soviet Union, Georgian print media represented mostly a party press, while in the later periods principles of media independence dominated, which makes a huge difference in terms of media systems in different study periods. Second, only few newspapers survived the whole 20 years period of transition, big number of newspapers emerging after the opening of the political arena were stillborn, disappearing after publishing the first issue, others were published irregularly sometimes with the intervals of several months (Bokeria, Targamadze, and Ramishvili 1997).Given these circumstances, selecting newspapers with the same strategy in all of the study periods is not possible. However, this should not create a problem since the present study does not make any comparison between/among newspapers but rather 18

20 aims at identifying a general picture of exclusionary and hostile nationalism in the print media content in the given periods. In addition, this circumstance will not create any bias since each period (defined as (1) Election-street protests, (2) Elections-no street protests, (3) No elections-no street protests) comprises months from 1990s as well as from the later years. To address the complexities listed above, I rely on the following strategies: in the periods when party press dominated (beginning of 1990s), the newspaper selection will be actorcentered, i.e. selected on the basis of actors with whom the newspapers were affiliated; in the periods when party press was abolished and most of the media claimed to be independent from any political influence (since the second part of 1990s), the newspaper selection will be based on their importance in terms of readership and circulation. For feasibility reasons three newspapers per period will be analyzed. For the years in the beginning of the 1990s, I select newspapers based on the following criteria: first, the official governmental newspaper; second, the newspaper of the second strongest party, i.e. party which gained most of the votes after the winner party in the last election prior to the research period, and third, the first nonparty newspaper in Georgia, which paved the way for future independent print media 9. Due to the lack of systematic data on circulation and readership of newspapers in Georgia, particularly in the periods of , the relevant newspapers for these years are identified based on several expert interviews and the available reports by international organizations. The newspaper issue selection is another important task which should be performed effectively so that any kind of bias is avoided. Sampling newspapers based on issues is needed given the big number of months included in the study period. For this purpose a stratified random sampling is applied. More precisely, for weekly newspapers one week per month is selected, for the daily ones - a day per month. In other words, one issue per month will be selected. This approach will guarantee that results are not affected by the difference in the number of newspapers between weeklies (4 per month) and dailies (20-24 per month). The next question is how to determine which week or which day to select? Since newspapers might have certain type of a cyclic regularities (such as for example covering conflict issues every Monday or covering minorities every second week of a month, etc.) it is important to avoid the correlation of the results with this kind of a natural rhythms (Knippendorf, 2004). Therefore, in each period I select different weeks in different months and different days in the 9 Here I mean the media which claimed being independent because given the constant attempts from political powers to control media media independence is a relative concept in this case. 19