Like Dust before the Wind, or, the Winds of Change? The Influence of International Criminal Tribunals on Narratives and Media Frames

When created, international criminal tribunals (ICTs) were not only expected to do justice but also to provide stabilization to postconflict regions, contribute to reconciliation and curb the potential denial of atrocities. Based on media content analysis, this article examines whether controversial ICT decisions triggered changes in narratives or frames about the conflicts which formed the background of the respective ICT decisions. There is no evidence for dramatic changes in the preexisting narratives about these conflicts, but we found some cases in which tribunal decisions caused changes in media frames and in elements of such frames, mostly by emphasizing outgroup victimhood and individual responsibility of ingroup perpetrators, as well as triggering effects of collective guilt externalization. Although frame changes were often observed in both democratic and nondemocratic countries, only in democratic countries with pluralist and competitive media systems could they be attributed to tribunal decisions. K E Y W O R D S : international criminal tribunals, media, frames, narratives, reconciliation * Professor, SWPS University of Social Sciences and Humanities, Warsaw, Poland. Email: kbachmann@ swps.edu.pl † Professor of Law, Stellenbosch University, South Africa. ** Researcher, Institute of Social Sciences, Belgrade, Serbia. †† Research Fellow, Institute of Culture and Memory Studies, Research Centre of the Slovenian Academy of Sciences and Arts (ZRC SAZU), Ljubljana, Slovenia. *** EURIAS Postdoctoral Fellow, Polish Institute for Advanced Studies, Warsaw, Poland. ††† Attorney of Law, Kraków, Poland. **** Academic and Researcher, Protestant Institute of Arts and Social Sciences, Butare, Rwanda. †††† Mohammed Ali Mohammed Ahmet worked as a professional Arabic–English translator and interpreter for foreign media, international relief organizations and the UN in Sudan. He passed away in 2016. ***** Sociologist, Faculty of Philosophy, University of Prishtina, Kosovo. 1 Research for this article was sponsored by the National Research Center (Narodowe Centrum Nauki) of Poland, NCN grant no. UMO-2012/06/A/HS5/00249, under the research project ‘International Criminal Tribunals as Actors of Domestic Change, 2013–2018.’ Gerhard Kemp and Klaus Bachmann are grateful to the Stellenbosch Institute for Advanced Study, Wallenberg Research Centre at Stellenbosch University, South Africa, for hosting them between March and May 2018. VC The Author(s) (2019). Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 1 International Journal of Transitional Justice, 2019, 0, 1–19 doi: 10.1093/ijtj/ijz005 Article D ow naded rom http/academ ic.p.com /ijtj/advance-articledoi/10.1093/ijtj/ijz005/5372410 by gest on 24 April 2019 When the International Criminal Tribunal for the former Yugoslavia (ICTY) and the International Criminal Tribunal for Rwanda (ICTR) were created, they were first and foremost expected to do justice and prosecute perpetrators of the most heinous crimes. However, in the respective UN General Assembly (UNGA) resolutions and the resolutions of the UN Security Council (UNSC), we also find the expectation that these tribunals might contribute to peacekeeping, stabilize the countries under their jurisdiction and even – in the case of Rwanda – contribute to reconciliation. Later on, judges at both tribunals referred to these expectations, invoking reconciliation in order to justify some of their judgments and decisions, pointing to the (alleged) will of victims or the (also alleged) need to reintegrate perpetrators. One of the more specific expectations in this context was the wish that the tribunals might ‘shrink the space for denial.’ The wish to curb denial was the main motive behind the creation of an outreach programme by the ad hoc international criminal tribunals (ICTs). At the International Criminal Court (ICC), outreach and extrajudicial tasks had less importance due to the limited scope of judicial intervention which the ICC can undertake. The UNSC can refer a situation to the ICC prosecutor, but without such a referral, the ICC’s possible interference with signatories to the statute is limited to cases where a state is either unwilling or unable to prosecute international crimes under the Rome Statute. As long as a state party can show its willingness and capacity to hold perpetrators of such crimes accountable, there is neither a need nor a way for the ICC to step in. Additionally, the gravity threshold of the Rome Statute in principle also restricts the ICC’s possible interference to cases of mass atrocities and crimes committed by high-ranking perpetrators. Unlike the ICTY and the ICTR, the ICC does not work under a primacy principle which would give it the right to claim any suspect and take over any case it deems appropriate. It functions as a kind of ‘court of last resort’ which can only take over cases that are not being investigated by the respective country. The determination of whether a country is willing and able to prosecute a crime usually takes quite a lot of time, therefore the ICC often investigates crimes which took place many years ago. All this makes it much more difficult for the ICC to ‘shrink the space for denial,’ or, in general, to fulfil extrajudicial tasks like reconciliation and postconflict stabilization. The literature about ICTs and international criminal justice is dominated by lawyers, who focus on ICT tasks, the material and institutional law applied by ICTs and discussions on the functioning of the tribunals and the coherence of their jurisprudence. A part of this literature meanders between law and sociological institutionalism and tracks decisions, trying to reveal how and why certain decisions at ICTs 1 Klaus Bachmann, Thomas Sparrow-Botero and Peter Lambertz, When Justice Meets Politics: Independence and Autonomy of Ad Hoc International Criminal Tribunals (Frankfurt: Peter Lang, 2013). 2 Diane F. Orentlicher, Shrinking the Space for Denial: The Impact of the ICTY in Serbia (New York: Open Society Institute, 2008). 3 Klaus Bachmann telephonic interview with Gabrielle Kirk McDonald, the second ICTY president, April 2012; personal interview, Refik Hod zi c, ICTY press spokesperson under Kirk McDonald, New York, April 2012; personal interview, Innocent Kamenzi, outreach officer at the ICTR, Kigali, Rwanda, July 2012. 2 K. Bachmann et al. D ow naded rom http/academ ic.p.com /ijtj/advance-articledoi/10.1093/ijtj/ijz005/5372410 by gest on 24 April 2019 were taken and how they were influenced by external and internal factors. This body of research dealing with ‘process control’ is supplemented by a growing literature on the relations between intra-court decision making and the wider public and often concentrates either on issues of judicial behaviour (how judges and courts are influenced by external factors) or on the question of how courts and judges influence the public’s perceptions and media coverage. This is at the core of this article’s interest. Its purpose is to find whether – and if so, how – tribunal decisions impact public opinion, or, more precisely, media frames about the conflicts which formed the background of the respective tribunals’ jurisprudence. Our aim is to contribute to a scholarly debate which has lasted for several centuries (starting with the antislavery tribunals) and whose core concerns the argument in Kathryn Sikkink’s ‘justice cascade.’ As Sikkink, Thomas Risse and others have shown, legal change may be fostered by nongovernmental organizations (NGOs) and networks of so-called norm entrepreneurs that initiate transnational change through lobbying and their access to media. The process starts with small groups of activists in countries hostile to legal innovation (like new human rights legislation) campaigning and connecting with external and international networks of norm entrepreneurs. They in turn use their influence on their norm-friendly governments to extract concessions from normhostile governments in negotiations about trade or political issues which are important for the norm-hostile government. Lobbying takes place in the form of traditional campaigns and through the use of traditional and new media (e.g. internet, social media, transnational television). Once a government that is hostile to new human rights legislation concedes to include some innovations into its legal system, local activists take up the ball and launch strategic litigation campaigns. In this way, they infuse new human rights norms which the government did not want to submit to parliament, into the courts. Domestic judges act as occasional legislators, adding new interpretations of existing legislation in accordance with the norms promoted by the litigators. At the end of this process, new human rights norms are applied despite the government’s reluctance or (in autocratic systems) even hostility to endorse them. This model has mostly been applied to explain the introduction of new human rights legislation in countries with weak human rights records and autocratic political systems. Furthermore, it explains why nondemocratic governments hostile to human 4 Nancy A. Combs, Fact-Finding without Facts: The Uncertain Evidentiary Foundations of International Criminal Convictions (Cambridge: Cambridge University Press, 2007); Gideon Boas, The Milo sevi c Trial: Lessons for the Conduct of Complex International Criminal Proceedings (Cambridge: Cambridge University Press, 2010); Klaus Bachmann and Aleksandar Fati c, The UN International Criminal Tribunals: Transition without Justice? (London: Routledge, 2015). 5 Barbora Holá, Alette Smeulers and Catrien Bijleveld, ‘Is ICTY Sentencing Predictable? An Empirical Analysis of ICTY Sentencing Practice,’ Leiden Journal of International Law 22 (2009): 79–97. From the same authors: ‘International Sentencing Facts and Figures: Sentencing Practice at the ICTY and ICTR,’ Journal of International Criminal Justice 9(2) (2011): 411–439; and about the ICTR: ‘Punishment for Genocide: Exploratory Analysis of ICTR Sentencing,’ International Criminal Law Review 11(4) (2011): 745–773. 6 Kathryn Sikkink, The Justice Cascade: How Human Rights Prosecutions Are Changing World Politics (New York: W.W. Norton, 2010). 7 Thomas Risse, Stephen C. Ropp and Kathryn Sikkink, eds., The Power of Human Rights: International Norms and Domestic Change (Cambridge: Cambridge University Press, 1999). International Criminal Tribunals to ICTs 3 D ow naded rom http/academ ic.p.com /ijtj/advance-articledoi/10.1093/ijtj/ijz005/5372410 by gest on 24 April 2019 rights end up with modern human rights laws, which – in the long run – may even undermine their rule. Sikkink’s model has been praised as well as criticized. In this article we test a very narrow hypothesis, derived from the ‘cascade’ model, and ask whether ICTs function as ‘norm entrepreneurs’ and initiate changes in countries whose public and governments are hostile to some or all of the norms promoted by the respective ICTs. We do so in a very small field, where impact is relatively easy to detect and can be described in detail – at the junction between ‘the tribunal’ and ‘the media.’ Sikkink’s model hypothesizes direct influence to take place only after a very long time, when the water from the ‘justice cascade’ has trickled down from the international level, through the media and public opinion of norm-friendly countries to the media and public sphere of norm-hostile countries, and then finally, after years or even decades, it might or might not have changed the human rights law applied in countries under ICT jurisdiction. There is a methodological problem with such an approach: if the change which the model anticipates is supposed to occur after such a long time (the length of which the model does not predict), it is impossible to measure and disentangle it from other factors. Thus, even if we observe an authoritarian country showing increasing respect for human rights, how can we be sure after such a long time that it was the result of the ‘justice cascade’ rather than something else? Thus our interest in detecting short-time changes caused by ICT decisions on media frames about the underlying conflict. We are not interested in changes in media frames about the respective tribunals – such research has been done exhaustively. Usually, ingroup members and media praise a tribunal when it indicts or sentences a protagonist of the outgroup, and discard its findings when they shed a bad light on ingroup members. This article focuses only on those media frames which deal with the underlying conflict, the elements of which were investigated by the relevant ICT and formed the background to the ICT trial. Our aim was to explore how certain decisions which include specific interpretations of past events affected the way in which the media described these events after the respective ICT decisions. This impact was measured by applying qualitative frame analysis: after an indictment was published or a judgment issued, we compared the frames about the underlying events to the frames 8 Tricia Olsen, Leigh Payne and Andrew Reiter, Transitional Justice in Balance: Comparing Processes, Weighing Efficacy (Washington, DC: US Institute of Peace, 2010); Padraig McAuliffe, ‘The Roots of Transitional Accountability: Interrogating the “Justice Cascade”,’ International Journal of Law in Context 9(1) (2013):

When the International Criminal Tribunal for the former Yugoslavia (ICTY) and the International Criminal Tribunal for Rwanda (ICTR) were created, they were first and foremost expected to do justice and prosecute perpetrators of the most heinous crimes. However, in the respective UN General Assembly (UNGA) resolutions and the resolutions of the UN Security Council (UNSC), we also find the expectation that these tribunals might contribute to peacekeeping, stabilize the countries under their jurisdiction and even -in the case of Rwanda -contribute to reconciliation. Later on, judges at both tribunals referred to these expectations, invoking reconciliation in order to justify some of their judgments and decisions, pointing to the (alleged) will of victims or the (also alleged) need to reintegrate perpetrators. 1 One of the more specific expectations in this context was the wish that the tribunals might 'shrink the space for denial.' 2 The wish to curb denial was the main motive behind the creation of an outreach programme by the ad hoc international criminal tribunals (ICTs). 3 At the International Criminal Court (ICC), outreach and extrajudicial tasks had less importance due to the limited scope of judicial intervention which the ICC can undertake. The UNSC can refer a situation to the ICC prosecutor, but without such a referral, the ICC's possible interference with signatories to the statute is limited to cases where a state is either unwilling or unable to prosecute international crimes under the Rome Statute. As long as a state party can show its willingness and capacity to hold perpetrators of such crimes accountable, there is neither a need nor a way for the ICC to step in. Additionally, the gravity threshold of the Rome Statute in principle also restricts the ICC's possible interference to cases of mass atrocities and crimes committed by high-ranking perpetrators. Unlike the ICTY and the ICTR, the ICC does not work under a primacy principle which would give it the right to claim any suspect and take over any case it deems appropriate. It functions as a kind of 'court of last resort' which can only take over cases that are not being investigated by the respective country. The determination of whether a country is willing and able to prosecute a crime usually takes quite a lot of time, therefore the ICC often investigates crimes which took place many years ago. All this makes it much more difficult for the ICC to 'shrink the space for denial,' or, in general, to fulfil extrajudicial tasks like reconciliation and postconflict stabilization.
The literature about ICTs and international criminal justice is dominated by lawyers, who focus on ICT tasks, the material and institutional law applied by ICTs and discussions on the functioning of the tribunals and the coherence of their jurisprudence. A part of this literature meanders between law and sociological institutionalism and tracks decisions, trying to reveal how and why certain decisions at ICTs were taken and how they were influenced by external and internal factors. 4 This body of research dealing with 'process control' is supplemented by a growing literature on the relations between intra-court decision making and the wider public and often concentrates either on issues of judicial behaviour (how judges and courts are influenced by external factors) 5 or on the question of how courts and judges influence the public's perceptions and media coverage. This is at the core of this article's interest. Its purpose is to find whether -and if so, how -tribunal decisions impact public opinion, or, more precisely, media frames about the conflicts which formed the background of the respective tribunals' jurisprudence. Our aim is to contribute to a scholarly debate which has lasted for several centuries (starting with the antislavery tribunals) and whose core concerns the argument in Kathryn Sikkink's 'justice cascade.' 6 As Sikkink, Thomas Risse and others have shown, legal change may be fostered by nongovernmental organizations (NGOs) and networks of so-called norm entrepreneurs that initiate transnational change through lobbying and their access to media. 7 The process starts with small groups of activists in countries hostile to legal innovation (like new human rights legislation) campaigning and connecting with external and international networks of norm entrepreneurs. They in turn use their influence on their norm-friendly governments to extract concessions from normhostile governments in negotiations about trade or political issues which are important for the norm-hostile government. Lobbying takes place in the form of traditional campaigns and through the use of traditional and new media (e.g. internet, social media, transnational television). Once a government that is hostile to new human rights legislation concedes to include some innovations into its legal system, local activists take up the ball and launch strategic litigation campaigns. In this way, they infuse new human rights norms which the government did not want to submit to parliament, into the courts. Domestic judges act as occasional legislators, adding new interpretations of existing legislation in accordance with the norms promoted by the litigators. At the end of this process, new human rights norms are applied despite the government's reluctance or (in autocratic systems) even hostility to endorse them. This model has mostly been applied to explain the introduction of new human rights legislation in countries with weak human rights records and autocratic political systems. Furthermore, it explains why nondemocratic governments hostile to human rights end up with modern human rights laws, which -in the long run -may even undermine their rule. Sikkink's model has been praised as well as criticized. 8 In this article we test a very narrow hypothesis, derived from the 'cascade' model, and ask whether ICTs function as 'norm entrepreneurs' and initiate changes in countries whose public and governments are hostile to some or all of the norms promoted by the respective ICTs. We do so in a very small field, where impact is relatively easy to detect and can be described in detail -at the junction between 'the tribunal' and 'the media.' Sikkink's model hypothesizes direct influence to take place only after a very long time, when the water from the 'justice cascade' has trickled down from the international level, through the media and public opinion of norm-friendly countries to the media and public sphere of norm-hostile countries, and then finally, after years or even decades, it might or might not have changed the human rights law applied in countries under ICT jurisdiction. There is a methodological problem with such an approach: if the change which the model anticipates is supposed to occur after such a long time (the length of which the model does not predict), it is impossible to measure and disentangle it from other factors. Thus, even if we observe an authoritarian country showing increasing respect for human rights, how can we be sure after such a long time that it was the result of the 'justice cascade' rather than something else? Thus our interest in detecting short-time changes caused by ICT decisions on media frames about the underlying conflict. We are not interested in changes in media frames about the respective tribunals -such research has been done exhaustively. 9 Usually, ingroup members and media praise a tribunal when it indicts or sentences a protagonist of the outgroup, and discard its findings when they shed a bad light on ingroup members.
This article focuses only on those media frames which deal with the underlying conflict, the elements of which were investigated by the relevant ICT and formed the background to the ICT trial. Our aim was to explore how certain decisions which include specific interpretations of past events affected the way in which the media described these events after the respective ICT decisions. This impact was measured by applying qualitative frame analysis: after an indictment was published or a judgment issued, we compared the frames about the underlying events to the frames 106-123. 9 In most cases, media coverage about ICT decisions was very predictable. If an ICT took a decision favourable to an accused, it was praised by the accused's supporters but those who sided with the actual or alleged victims were outraged. If the decision was unfavourable to the accused, victim communities praised it and the accused's supporters were outraged. In the case of the ICTR, the issue was even simpler: the public, the media and victim groups in Rwanda were always in favour of the harshest possible punishment and were always outraged when a sentence was lenient. This was because the ICTR only judged perpetra- used by the same media before the respective decisions in order to establish whether the frame had changed. In this article, an 'ICT decision' refers not only to an indictment or verdict, but to any decision made by an ICT organ which resonated in the public sphere. For example, the ICC never issued any verdict against a Sudanese suspect nor against any of the politicians and journalists involved in the Kenya situation. 10 The ICTY, for its part, tried to solve the problems arising from lack of state cooperation by issuing secret indictments. In these cases, indictments of which the public and the media were unaware could hardly be expected to impact on media frames. Instead, other decisions -public indictments and verdicts at the ICTY and the ICTR as well as arrest warrants by the ICC -were more likely to cause shifts in media frames. But there were also exceptions to this rule. The ICC decision to end proceedings in the Kenyan cases had as much impact on media coverage as the decision to issue arrest warrants had done before. This is why we chose to examine the impact of different types of decisions on media frames rather than sticking to the narrow and legalistic notions of indictment and judgment.

GRAND NARRATIVES, FRAMES AND FRAME ELEMENTS ABOUT THE
UNDERLYING CONFLICTS Consumers of media outlets tend to perceive narratives and opinions in newspapers and electronic media as true and consistent with established facts and common knowledge. However, rather than assuming that reporting is necessarily derived from some prior, objective and univariate reality, media analysts ask why certain media describe things in one way while others do so in a different way. The basic assumption of this approach is that every media outlet and every journalist can choose from a multitude of options when deciding to present a specific issue. By writing about an identified issue, media draw upon and refer to cognitive schemes, called 'frames,' which provide categories, order and chronology that allow the reader (and author) to organize fragmented information and attribute meaning to it. By emphasizing certain aspects of an event and embedding their description in headlines, pictures, the structure of a page, a film or a radio programme, media instruct us both what to think and how to think about it. 11 Such frames act as patterns of interpretation which affect problem definition, moral evaluation and causal interpretation, and may even imply particular solutions to the identified problem.
In order to render frames comprehensible to media consumers, an overarching narrative must exist in which to embed the respective frame; the event to be framed has to possess certain features that allow journalists to ascribe meaning to it and to interpret it normatively. When events lack these conditions, journalists tend to neglect them.
Three levels of analysis must be distinguished here, all drawn from media content analysis. In every public sphere a grand narrative channels the flow of scattered and unrelated information about the past into a comprehensive story, which is shared by all or most citizens and to which media coverage about current events relates, often by taking the general knowledge about this story for granted. Thanks to such a grand narrative, reference in reports and op eds can be limited to a few hints, which are often incomprehensible to outsiders. The story describing 1994 in Rwanda as a genocide, during which Hutu killed Tutsi while the international community failed to intervene, is one such grand narrative, currently shared within Rwanda as well as by most of the media coverage and popular science accounts outside Rwanda. The Homeland War in Croatia is another such grand narrative, claiming that in 1995 Croatian forces rightly drove out Serb forces, thus liberating Croatian territory from aggressors.
Underneath such overarching accounts, which attribute meaning and moral judgement to a whole conflict or an entire historical period, there are 'media frames' which build on these narratives. Frames help journalists and readers to update the grand narrative and to adapt it to new events and information. Both narratives and frames contain moral judgements; identify causes and consequences, culprits and victims, guilt and innocence; and often lead to conclusions about what needs to be done. However, frames provide interpretation to only a small fragment of the grand narrative -they assess a shorter period of time, often only one incident, such as a battle within a larger war, a crime committed in the framework of a military campaign or a political decision within the context of a historical development. For example, journalists' assessment of the shelling of the city of Knin in Croatia is such a frame, resting on readers' knowledge of the general narrative about the Homeland War, as a result of which journalists did not need to justify the shelling as the grand narrative already included the assumption that the war was a 'just and legitimate' one. Finally, there may be cases where a frame is not replaced by another one, but rather changes with respect to one or several elements, for example, when the legitimacy of a war is not questioned but the media outlet suddenly starts to mention victims, perpetrators or features of the fighting which it did not mention before.
Some authors treat frames as independent variables which may lead to specific consequences (framing effects); others regard frames as dependent variables which are the outcome of decisions and events beyond the reach of the framing journalist. In this article, we use frame analysis to determine changes to the dependent variable, specifically the impact of ICT decisions on grand narratives, media frames and small changes within media frames.

CASE SELECTION CRITERIA
In the framework of the case studies, the results of which are presented in this article, the main criterion for choosing an ICT decision was its salience for the public in the respective country or entity. In order to be selected, a case had to be sufficiently controversial and the suspect of high enough rank in the hierarchy of the country or entity to ensure that it triggered enough media coverage (and therefore frames 12 ) about the underlying conflict for our analysis (Table 1). We assumed that frame changes mainly take place after (public) indictments, trial judgments and appeals judgments and therefore only included cases that would provide us with a maximum number of such decisions, that is, cases which were already closed by a final ICT verdict. Because our project's aim was to find out whether ICT decisions can trigger changes in frames and narratives and to identify the circumstances in which such changes may occur, we concentrated on cases with a high first-sight probability of such changes. We hypothesized that such changes would be likely when the media covered cases ending in guilty verdicts against suspects from their ingroup and in acquittals of suspects from their outgroup. There were a number of exceptions to this rule. We included the Plav si c case 13 (which ended with a plea agreement) because it was the first case of a high-ranking Bosnian Serb politician admitting guilt at the ICTY and was therefore likely to trigger guilt externalization and change some of the preexisting and deeply embedded frames in Bosnian Serb media about the war in Bosnia. By way of comparison, we also included the Luki c and Luki c case (also called the 'Vi segrad case') 14 in order to determine whether low-ranking perpetrators triggered similar effects among the same (Bosnian Serb) media as had been the case in Plav si c. In Kenya, there were no cases concluded which we could analyze. We therefore examined the media coverage of the 'Kenya I' and 'Kenya II' cases, 15 both of which ended without trial judgments.
In countries and entities allowing for media pluralism, we included at least one outlet from each ideological current (left, right, liberal, close to or critical of the government and the respective ICT). This proved impossible in Rwanda because there are no media outlets that are critical of the government. It was possible in Sudan, though our access was often limited by external constraints, for example the unavailability of archives, high access fees or the destruction of media archives during the conflict.
Initially, we adopted a homogeneous methodological framework for the whole project, urging all case study researchers to include at least 10 items 16 from each examined medium before and after each respective ICT decision. We hoped thereby to ensure that changes in media frames were more or less stable and not the result of one author's sudden mood change. In the course of the project, it became clear that such a rigorous framework would be impossible to apply coherently across all cases. In many cases, the 10 items after an indictment overlapped at least partly with the 10 items preceding a verdict, and sometimes the items containing frames about a firstinstance judgment overlapped with those preceding a second-instance judgment. 17 In other cases, the media did not produce enough frame-containing items to reach the 10-item margin. In almost all countries, TV and radio archives were inaccessible to our researchers for various reasons: archives were not maintained by some media in Bosnia and Herzegovina (BiH) and Rwanda, and by any in Sudan and South Sudan; the archives were technically not searchable (Rwanda); or the archives could 13 ICTY, IT-00-39 and 40/1. 14 ICTY, IT-98-32/1, 'Vi segrad.' 15 ICC-01/09 and ICC-01/09-02/11. 16 We understand an 'item' to be a text published in a newspaper or a TV or radio programme containing at least one frame about the underlying conflict which gave rise to the court case, and in whose framework an ICT decision was taken. 17 The ICTY, ICTR and ICC conduct their trials on two levels, trial chamber and appeals chamber, but in some cases (e.g. the ICTY Vukovar trial and the ICTY Haradinaj trials) the appeals chamber orders a trial whose verdict can then be challenged on appeal, ultimately leading to three verdicts.
only be accessed after paying outrageously high fees, for which we had no financial cover in the project budget (Croatia and BiH).
DEFINING AND MEASURING IMPACT For this study, we assumed the influence of ICT decisions on media frames when such frames (or elements thereof) changed after an ICT decision had been issued and the change was stable over time, that is, it occurred consistently in several or all of the 10 items examined. At the same time, the new media frames detected after an ICT decision had to overlap content-wise with frame elements invoked in the ICT decision. For example, the interpretation of the conflict and the change in media frame in Sudan, from one describing the Darfur conflict as tribal to one based on underdevelopment, can only be considered a result of an ICC decision if the latter also contained this 'underdevelopment frame' or at least elements thereof. 18 Additionally, in every case study we required our field researchers to check for other factors which could have triggered the frame change, such as a change of ownership of the respective media outlet, a change of government censorship policy or a decision of another national or international body. 19 Only when a frame shift was closer in time to the ICT decision than to a third factor, and when the new frame's content overlapped at least partly with the content of the ICT decision, did we assume the shift to be triggered by the ICT decision.

IMPACT OF ICT DECISIONS ON GRAND NARRATIVES
ICTY decisions and judgments did not change the Croatian narrative about the Homeland War as a just and legal endeavour to liberate the country from Serb aggression, nor did it incline Serbs to regard the wars in Bosnia or Kosovo as a justified liberation campaign against Serb oppression. ICT decisions did not shatter the grand narrative about the 1994 genocide in Rwanda, and they did not change what Kenyan journalists wrote about the postelection violence in 2008. In Sudan, censorship limited Arab-language newspapers' possibilities to cover and interpret the Darfur conflict to such an extent that no grand narrative even came into being. Unlike other countries and entities, where governments promoted their own narrative about a past conflict (e.g. in Kosovo, Croatia and Rwanda), the Sudanese government was 18 Sudanese media tended to present the conflict in Darfur as being a result of a policy of keeping Darfur underdeveloped. This policy, in their opinion, had been conducted from colonial times and was being continued by the Omar al-Bashir government. According to this frame, rebels in Darfur had risen in order to protest against this policy and achieve improvements in infrastructure and service delivery, which, if undertaken, would end the conflict. A competing frame claimed the conflict in Darfur to be caused by tribal animosities, in which the central government had no important role other than being a potential mediator. 19 This is why the ICJ appears in our research. Initially, it was not part of the research design as it is not an ICT, but its ruling in Bosnia vs. Serbia appeared as the main factor (see Table 1) explaining why changes in Serb media frames, which appeared after the Scorpions video about Serbia's alleged guilt for the Srebrenica genocide, were later partly reversed by some of these media. not interested in imposing its own interpretation of the Darfur conflict on the Sudanese media and instead did its utmost to silence any information about the events in Darfur. Media coverage was scarce, fragmentary and lacked clear patterns of interpretation. Even after the constitutional reform in 2005, only a few media frames about Darfur could be detected.
In some cases, we identified changes in the grand narrative invoked by a media outlet, but they were all attributed to third factors. After the 2000 trial judgment in the Bla ski c case at the ICTY, 20 which outlined the crimes committed and sentenced the accused to 45 years of imprisonment, some Croatian and Bosnian Croat newspapers started to blame the former Croatian leadership under Franjo Tudman for its interference in BiH and the crimes committed by Croatian units there. 21 The newspapers had not done so before, instead presenting the fights of Croatian units in Bosnia as a kind of self-defence against an attack by Bosniak troops. Hrvatska rije c, a paper very close to the Bosnian Croat leadership, even argued that the Bosniak crimes were worse than those committed by Croats. 22 The blaming of Tudman is a clear example of the externalization of guilt. 23 However, despite the impression created, it was not triggered by the ICTY's Bla ski c judgment, but by government change in Croatia. Tudman died in 1999, and in 2000 Stjepan Mesi c, a dissident from Tudman's party, was elected president. The new government, led by Ivica Ra can, which emerged from the parliamentary elections, started to cooperate with the ICTY and delivered files from the state archives. They helped to exonerate Tihomir Bla ski c from some of the crimes for which he had been sentenced, thus shifting the blame to members of the previous government. This helped Mesi c to discredit Tudman's party, led to a more lenient appeals judgment for Bla ski c and put a new narrative in place, which admitted Croatia's intervention in the war in Bosnia and the commission of crimes, but blamed the former government, not Croatia or the Croatian nation. This new narrative led to frame changes in some Bosnian Croat media but they cannot be attributed to the ICTY, whose trial judgment was issued a month after Mesi c had taken office, by which time the frames about the Croatian intervention in Bosnia had already changed. A similar change of frame elements took place after the appeals judgment in the Naser Ori c case at the ICTY, 24 when the Bosnian newspaper Nezavisne novine started to dedicate more space to Bosnian Serb victims' commemorations and printed statements of Serb victims of atrocities committed by Bosniak fighters. However, a closer look at the situation of the newspaper revealed that this change was triggered by a shift in the editor's policy. In 2007 the owner of the newspaper came under the influence of the government of Republika Srpska and started to push his paper in a pro-Serb direction, also giving more space for denial of the genocide committed in Srebrenica.
In parts of the Sudanese Arab-language media (Al Ayam, Ray Al-Shaab and Ajrasselhurriya), entire frames about the Darfur conflict were changed. These could not, however, be the result of an ICC decision because the respective ICC arrest warrants did not contain any elements of the new frames. The frame changes happened long after the warrants had been issued. In 2005, the UNSC referred the situation in Darfur to the ICC, allowing the ICC prosecutor to investigate the crimes committed there. 25 Two years later, the ICC issued the first Sudan-related arrest warrant, after which the first frame blaming the government for the conflict in Darfur started to appear in the daily Al Ayam, which so far had only superficially and in a very limited way dealt with the conflict, claiming that the government in Khartoum had lost control over Darfur because of a lack of appropriate information. Al Ayam's new approach was due to the enactment of a new constitution in 2005 as a result of the peace negotiations between the government in Khartoum and the leaders of the South Sudan opposition. The result provided for the inclusion of South Sudanese politicians in the national government. These changes increased media pluralism and allowed Arab-language media to comment on Darfur relatively freely. But these new frames produced by the media did not overlap with the way the UNSC referral and the first ICC arrest warrant had described the conflict.
In some countries, neither changes of narratives nor frame changes took place. In Kosovo, Montenegro and the Serb part of Bosnia, the Republika Srpska, ICTY decisions reinforced existing frames which saw the ICTY trials in a context of nation building. Convictions of accused war heroes were interpreted as an attack on the entire nation and convictions of outgroup members were seen as a confirmation of the outgroup's moral inferiority. Something similar happened in English-language media in Sudan decision the ICC issued, it was interpreted as proof of the veracity of the frames which these media had created before and which were usually hostile to the government of Omar al-Bashir. The conflict in Darfur was denounced as a genocide committed by the Al-Bashir regime against either 'Darfurians,' 'ethnic African tribes' or 'unarmed civilians.' 27 In Kosovo, Kenya and Rwanda, there was yet another reason why frames did not change. Social control proved stronger and more successful in preventing changes of frames than even censorship did. The ICTR did not cause any changes in the genocide frames of the Rwandan media because the Rwandan government tended to impose its own compulsory frames on the media and to monitor whether they actually used them. 28 In Kosovo and Kenya, the governments do not have any direct grip on media coverage and media tend to be pluralistic with regard to many issues. But in Kosovo they were homogeneous in their approach to the war with Serbia at the end of the 1990s. No ICTY decision changed any of the media frames, which presented the Kosovo Liberation Army's struggle as a righteous liberation fight against Serb oppressors. In Kenya, social control went in the opposite direction, resulting in any reference to ethnicity vanishing from media coverage about the 2008 postelection violence in the country which the ICC investigated. Anxious not to stir up ethnic hatred, journalists avoided ethnic labels in their reports and refused to identify specific political actors with ethnic groups, regardless of whether they appeared as victims or perpetrators in the text. 29 This tendency was strong enough to make some reports incomprehensible to outsiders and to readers who were not familiar with the ethnic fabric of the situation portrayed by the outlet. 30  Conflict and Communication 15(1) (2016): 1-13. It would be easy to explain the absence of ethnic frames in the coverage of the genocide commemorations in Rwanda by pointing to censorship, but at least a part of this reluctance seemed to be due to a similar anxiety as in Kenya about stirring up ethnic hatred by using ethnic labels in media coverage. This explains the paradox that Rwanda has a strongly regulated media environment whereas Kenya does not, but in both countries media do their utmost to avoid ethnic labels in their reporting. 30 One exception to this rule was The East African, which has many readers outside Kenya (in Tanzania, Rwanda and Uganda). Its reporters often referred to ethnic labels.
which had befallen Kenya, without mentioning who had been perpetrator and who had been victim. 31 This changed with the ICC trial hearings in 2014, at least with regard to the most popular media: The Standard, The Daily Nation and KTN TV. The East African, which has a considerable audience in neighbouring countries, frequently used ethnic labels in its coverage of the postelection violence, as well as throughout and after the ICC hearings. With the start of these hearings, the Kenyan media could no longer refrain from associating accused and witnesses with ethnic groups -not doing so would have rendered their trial coverage totally incomprehensible. The ICC trial hearings, broadcast to Kenya online and in TV coverage, thus triggered quite considerable frame changes, turning the 2008 postelection violence from a political conflict into an ethnic one. This change cannot, however, be attributed to an indictment or verdict, but only to the beginning of the trial hearings. Nevertheless, if one regards the start of the hearings as an ICC decision, then there was a clear ICC influence on media coverage in Kenya. 32 In the cases of some Croat and Serb media, elements of existing and wellestablished frames changed slightly. In these cases, the media did not question the general interpretation of collective responsibility -who was considered collective victims or collective perpetrators during the conflict. However, the trial judgments in While stories about outgroup victims had been neglected and silenced before the respective ICT decisions, victims were mentioned (sometimes even with empathy) in media reports after these ICTY decisions.
In some cases, small changes in frame elements could be detected. For example, as a result of the trial judgment in the Vukovar case, prisoners who had been killed by militia members no longer appeared as 'soldiers' but as 'wounded civilians' in the Serbian newspaper Vreme. Remarkably, one of these 'civilians' was described as 'a pregnant woman.' 33 After the trial judgment in Gotovina et al., a discussion started in the Croatian newspaper Jutarnji list about whether and how to hold low-ranking perpetrators accountable for atrocities committed against Serb civilians. 34 Before the trial, the very existence of such victims was denied and the entire military campaign presented as a victorious liberation struggle against Serb aggression -with no outgroup victims at all. A similar but much more significant development took place after the release of the so-called Scorpions video, a prosecution piece of evidence used in Milo sevi c 35 showing the execution of surrendered Bosniak men by Serb paramilitaries. The video was released to the media by an influential Serbian human rights NGO, the Humanitarian Law Center from Belgrade. The video managed to shift frames even in media which so far had rejected any responsibility on the part of their country for war crimes and which had kept a fierce anti-ICTY attitude in their reporting. This frame change, however, was not due to a formal ICTY decision but to the action of the Humanitarian Law Center, which decided the timing and context (the Milo sevi c trial) of the video's release. 36 Its impact was later moderated by the International Court of Justice's (ICJ) ruling in the Bosnia vs. Serbia case, 37 when the ICJ found Serbia not guilty of genocide in Bosnia but instead guilty of not preventing it and of having failed to surrender Ratko Mladi c to the ICTY. Two newspapers in Serbia (Vreme and Danas) had used frames blaming their state for genocide in Bosnia but backtracked from them after the ICJ ruling.
Total replacement of one frame by a different one was hardly ever observed. But after some ICTY decisions, new elements appeared in preexisting frames, for instance, information about outgroup victims and about atrocities committed by the forces which the respective media supported. The most explicit frame change was found in the Croatian daily Jutarnji list's coverage of the Gotovina trial. Ante Gotovina and two other high-ranking military officers of wartime Croatia had been indicted for war crimes and crimes against humanity, committed during and after the Croatian army's campaign to regain control over the Croatian-Bosnian borderland and Eastern Slavonia in 1995. Operation Flash and Operation Storm, during which Croatia regained control of the borderland, were regarded as part of a morally justified liberation struggle against foreign aggressors who had occupied the region after 1991. If Croatian media conceded there had also been victims on the Serb side during the fights, they were regarded as the exception to the rule, according to which the main war victims were those Croatian soldiers who had been killed in action. Even before the indictment was published, Jutarnji list admitted the existence of individual Serb victims. Like other Croatian newspapers, the outlet had declared the Homeland War a shining victory, untainted by any crimes against the enemy. The latter part of the frame did not change after the trial judgment, which found all three accused guilty of war crimes and crimes against humanity and emphasized the civilian victims who had been killed after the military operations. But by then Jutarnji list's coverage no longer omitted civilian Serb victims: 'during and after the operation "Storm" hundreds of Serbian civilians were killed. . .many houses were burned and. . .many properties were plundered.' 38 Zoran Pusi c from the Civic Committee for Human Rights estimated the number of victims murdered after Operation Storm at around 600 and claimed that almost nobody had been tried for those crimes. 39 Before the trial judgment, Jutarnji list had regarded Serb victims as individual, isolated incidents, a kind of inevitable collateral damage of the fighting. After the trial judgment, the newspaper transformed them from accidental victims into targets of a discriminatory policy carried out by the former leadership of Croatia. The paper did not point fingers at the three generals as the perpetrators of those crimes, instead calling for the 'real' criminals to be caught and brought to trial in a court. The newspaper did not entirely endorse the frames from the ICTY indictment and the trial judgment, according to which the main aim of Operation Storm had been the creation of an ethnically homogeneous independent Croatia. But the journalists included those frame elements which regarded the repressions against the Serb population as systematic and politically desired in order to drive all Serbs out of these parts of Croatia.
These changes even survived the appeals judgment, which reversed the trial judgment and exonerated all three accused. The day after Gotovina and Mladen Marka c were acquitted of all charges, Jutarnji list opened with the headline 'The Day When a New, Happy Country Was Created' and quoted Gotovina's statement that 'we won the first, and now the second Storm.' 40 After the controversial judgment, which came as a complete surprise to most observers, Jutarnji list still stuck to its frame about Serb victims targeted by state persecution and demanded that those responsible for crimes committed after Operation Storm be brought to justice because, 'without the fulfilment of justice for all victims of crimes, independently from the fact of who they were, the war in Croatia cannot be considered to be completely over.' 41 Neither the trial judgment nor the appeals judgment shattered the Croatian grand narrative about the Homeland War as a victorious liberation struggle against Serb forces, but the trial judgment prompted Jutarnji list to include at least some information about systemic crimes, Serb victims of state persecution and Croatian (direct) perpetrators who had gone unpunished.
Before the Gotovina et al. trial judgment, the Homeland War had been a just war in which the Croatian side had fought back against Serb aggression, without deliberately causing any civilian victims and without committing any atrocities. After the trial judgment, the Homeland War still appeared as a just and victorious war in the Croatian media, but now journalists tried to explain away victims and atrocities, or openly called for the prosecution of Croatian perpetrators. Some commentators relativized the crimes mentioned in the judgment, whose existence they had denied before and during the trial.

WHEN DO ICT DECISIONS MATTER IN THE MEDIA?
The limited number of cases examined (partly a result of the limited number of countries affected by ICT jurisdiction) does not allow for too much generalization, but some conclusions can be drawn from the evidence. Being a democratic, pluralistic and multiparty country increases the probability of ICT decisions triggering changes in media interpretations of a troubled past, whether this concerns grand narratives, alters media frames or only affects elements of preexisting and deeply embedded frames. The strongest influence with regard to the media was observed in Serbia, Croatia, Bosnia and, to a lesser extent, Kenya, all of which are pluralist democracies. This hardly comes as a surprise. Nevertheless, media in countries under authoritarian rule (Sudan, Rwanda) were not immune to changes in frames, which were sometimes the result of grassroots pressure rather than changes in top-down censorship and media policies. The latter was observed in Arab-language media after 2005, when several different frames about Darfur emerged, some of which were critical of or even hostile to the government in Khartoum. However, none of these changes could be attributed to previous ICC decisions. Instead, they took place as a result of political liberalization and, in Rwanda, emerged after constitutional change and a new top-down policy renaming the 'Rwandan genocide' the 'genocide against the Tutsi' in the aftermath of the constitutional revision of 2008. 42 Furthermore, the media content analysis of democratic, pluralist countries with multiparty competition revealed quite a lot of frame shifts (and changes in frame elements) triggered by ICT decisions in some countries and entities, but none in others. This variation can best be explained by distinguishing between those countries and entities which were engaged in nation building during the timeframe of our analysis and those which were not. There were frame changes and frame element changes in Republika Srpska after the Luki c and Luki c trial judgment and in Croatian and Bosnian Croat media after the Bla ski c trial judgment. But they were all due to factors other than the ICTY. In Kosovo, where nation building was under way, the controversial and strongly mediatized Haradinaj et al. trial did not lead to any frame changes. 43 None of these countries and entities had yet completed their nationbuilding process and their publics tended to discard ICT judgments against members of their ingroup as attacks on that process (and to endorse decisions against outgroup members as support for nation building). The effects of guilt externalization, which ICTs can trigger, were therefore unlikely to occur because the public tended to side with accused persons, especially if they are or were military or political leaders. The situation was different in states which are fully recognized as sovereign countries and that wield complete control over their territory and their populace, like Serbia and Croatia. There, ICT decisions against prominent politicians and military leaders sometimes triggered frame changes, which, in the case of Croatia, remained stable even after an ICT decision was reversed (Table 2).
It was not the core objective of our research to find out which factors other than ICT decisions could cause frame changes in the media. We addressed this only in order to exclude ICT influence as a factor in frame changes. Some of these third factors which can trigger changes in media coverage are consistent with conventional wisdom, like changes in government (Croatia), in government policies concerning collective memory (Rwanda) and in the overall political environment (Sudan). However, we also identified some unexpected frame changers when trying to establish whether a change in media coverage could be attributed to anything other than the ICT decision under scrutiny. These included civil society actors like human rights NGOs and advocacy groups, for example the Humanitarian Law Center in Belgrade, which revealed the Scorpions video. In the Vi segrad case, the daily Nezavisne novine started to cover the commemorations organized by (mostly Bosniak) victims' NGOs like Zena-zrtva rata (Woman-victim of the war) and Vi segrad 92, which brought an emphasis on crimes and crime victims into the outlet's coverage. This could not have been a result of an ICT decision, since the first indictment was issued in 2000 and the Luki c brothers were only arrested and tried after 2008. The commemorations launched by the Bosniak NGOs, however, coincided with the change in Nezavisne novine's frame elements. A similar pattern of NGO-induced changes in media frames was observed when the international 'Save Darfur' campaign was launched against the Sudanese government. It included frames which were then taken over by the Sudanese English-language newspapers, long before the ICC confirmed the genocide charges in its second arrest warrant against members of the Sudanese leadership.
Ultimately, ICTs have difficulties in causing changes in media frames and can hope to do so only in pluralist, democratic countries with a competitive media environment -this is the rule, but there are exceptions. ICTs sometimes managed to embed certain interpretations of past events in media coverage when they promoted a simple and judgmental label carrying heavy moral censure. The ICTY did so when it condemned the atrocities after the fall of Srebrenica in its rule 61 hearings. 44 It also did so in the Krsti c trial and appeals judgments, 45 when it found the massacres after the fall of Srebrenica genocidal, thus attaching the genocide label to Srebrenica for good. At the beginning of its activities in Rwanda, the ICTR refrained from imposing a genocide label. In Karemera, the ICTR finally declared the genocide in Rwanda a 'commonly known fact,' thus relieving the prosecution of the burden of proving that genocide had happened, but that took place 10 years after the ICTR had been established. 46 In the meantime, many other actors, including the UNSC, had thrown the genocide label into the public realm. It became the predominant pattern of describing the massacres of 1994, but not because of the ICTR.
Finally, it should also be acknowledged that even in cases where an ICT managed to trigger a change in frames or frame elements about the underlying conflict, this change was not always in line with the respective ICT's mission. There is a fair amount of controversy in the literature about what such a mission could or should entail. Some authors opt to restrict ICTs' missions to the narrow task of doing justice and judging perpetrators, whereas others see the tasks of ICTs as including the promotion of a culture of human rights and political accountability. 47 Irrespective of which position is taken, it cannot be regarded as consistent with the mission of any ICT if its decision is followed by media frames which morally exonerate a suspect who was convicted by the ICT, call for solidarity with him or declare him a hero. However, this happened in several cases. When Plav si c surrendered to the ICTY and pled guilty, admitting to crimes against humanity, she did not become a repentant criminal in the Serb Bosnian media, but a hero who had sacrificed herself for her nation. The same happened after the first-instance trial against Gotovina in the coverage of Slobodna Dalmacija, where he was depicted as 'a martyr for his nation' after being sentenced for war crimes and crimes against humanity. 48 One newspaper claimed him 'crucified instead of Tudman.' 49 One may doubt, too, whether the ethnification of the postelection violence in Kenya after the start of the ICC trial hearings was in line with the ICC's mission, which aims at individualizing guilt rather than promoting ethnic cleavages and collective stigmatization. In sum, an ICT decision can trigger changes in and of media frames, but only in democratic countries with party competition and media pluralism, and which do not embark on nation building. However, these frame changes do not always take place in a way which would please judges or prosecutors.