The Abyss of Ethnic Division: Two Decades of Discussing War in the Parliament of Bosnia and Herzegovina

Abstract

The article analyses over two decades of parliamentary debates in Bosnia–Herzegovina in order to understand the role of war past in the political reconciliation of Bosnian elites. We show that the discourse of war identified in the Parliament of Bosnia–Herzegovina structurally differs from the mainstream notion of Bosnian politics. The patterns detected in the parliamentary debates indicate that the central conflict exists primarily alongside Bosniak–Serb grievances, with Croat MPs being far less engaged. We argue that the three-sided conflict, often portrayed by literature as the major obstacle to reconciliation in Bosnia–Herzegovina, needs to be re-evaluated as a 2 + 1 model in which Croat MPs play a balancing role in maintaining the Post-Dayton status quo.

Introduction

War changes people, communities, and states. Its devastating effects can last for decades and become embedded in the collective memory for generations (Kijewski, 2020). Politicians like to use the memory of war instrumentally either as part of broader discursive strategies or as situational rhetoric that supposedly gives them a higher ground and moral superiority (Gelpi & Feaver, 2002; Smith & Barkhof, 2018). It often leads to perpetual commemorative practices making the past sacrifices sacred. The process is common across different models of divided societies with dire consequences for the development and social cohesion. If there is one country embodying this profile on multiple fronts, it is Bosnia–Herzegovina. The siege of Sarajevo, Srebrenica massacre, mass killings, and atrocities shocked Europe in a way not seen since the end of WWII. The horrors of war left the country devastated and deeply divided. As a result, post-Dayton Bosnia–Herzegovina is a witness to the omnipresent memory of the Bosnian war that is remembered by its three constituting nations in their own way (Lovrenović, 2016; Pehar, 2019).

Sadly, the majority of Bosnians continue to embrace homogeneous ethnonational narratives of the war (Kostić, 2012). The literature agrees that rather than talking about the prevailing discourse of war, there are three separate discourses, each maintained by one of the three constituting entities – Bosniaks, Croats, and Serbs. All three groups uphold their own historical narrative combining recent history with centuries-long grievances (Moll, 2013; Sokol, 2014). However, the question remains how these patterns, often discussed in the context of extraordinary events and media stunts, manifest themselves under a system that embedded the relative strength of fighting parties into the institutional setting of post-Dayton settlement. What are the patterns of parliamentary war-related discourse in Bosnia–Herzegovina? Have these patterns changed over time? What are the implications and lessons learned? Answering these questions should help us, first, to get a better picture of the war-related discourse of powerholders in Bosnia–Herzegovina from the late 1990s until today; second, to assess the legacy of Dayton settlement in the highest levels of Bosnian politics represented by all elected MPs; and third, to further theorise the long-term legacies of war in the context of political competition in divided societies.

The article analyses over twenty years of parliamentary debates (1998-2018) collected from the official website of the Parliament of Bosnia–Herzegovina for both the House of Representatives (Predstavnički dom/Zastupnički dom) and the House of Peoples (Dom naroda). We use an innovative research design combining advanced natural language processing (NLP) tools for modelling patterns in political discourse with qualitative content analysis in order to explore how war is discussed by politicians when the message is intended as a position-taking signal (Proksch & Slapin, 2015). The article shows that the discourse of war identified in the Parliament of Bosnia–Herzegovina structurally differs from the mainstream notion of Bosnian politics (both domestic and international), which presupposes an equal role of all ethnonationalist political elites in promulgating ethnic tensions and conflict-prone war memory (both in an academic and non-academic sense; see Mijatović, 2017; Seizovic, 2019; Selimbegović, 2017; USAID, 2022; Vuković, 2015). The patterns detected in the parliamentary debates indicate that the central conflict exists primarily alongside Bosniak-Serb grievances, with Croat MPs being far less engaged. We theorise that this effect of ‘holding back’ results from strategic decisions Croat elites in Bosnia–Herzegovina have made over the years to accommodate their political aspirations and strategic positioning. As a result, Croat MPs are far more cautious when it comes to deliberations about the war and war legacies making the main conflict line between Bosniaks and Serbs far more dominant than it might appear from existing literature.

These findings generally hold through time, giving us an additional layer to the discussion about how the war past resonates in the political arenas of Bosnia–Herzegovina and how it has (not) evolved since the late 1990s. On the one hand, it is a story of failed reconciliation at the top levels of Bosnian politics (Bosniak–Serb grievances); on the other, it shows how strategic concealing might work under the setting of situational alliances, external pressure, and the right kind of potential (future) subsidies (Croat MPs). Although these findings are not automatically generalisable to every aspect of the Bosnian political landscape, the analysis of the war discourse uncovers a less known dimension of power relations in Bosnia–Herzegovina, which might spill over to other policy areas and issues regularly.

Discussing War

Analysis of discourse of war has a long tradition in political science ranging from a critical examination of only a handful of political speeches (Bratberg, 2011; Kennedy-Pipe & Vickers, 2007; Pujante & Morales-López, 2008) to a systematic review of complex corpora of political debates counting thousands of documents (Mochtak et al., 2020; Morley & Bayley, 2009). Scholars have focused on topics such as military intervention (Boucher, 2009), media frames (Hackett & Zhao, 1994), war veterans (Taylor et al., 2019), military spending (Lewis & Hunt, 2011), commemoration events (Pavlaković & Pauković, 2019), past grievances (Fraser, 2012), gender based violence (Abramowitz & Moran, 2012), war crimes (Carpenter, 2000), or refugees (Baker & McEnery, 2005). Big part of this research is driven by a notion of symbolic and identitary aspects of politics, how power is legitimised, how it is executed, and what implications it has (Lebow et al., 2006).

Although direct violent conflicts have avoided most Western societies since the end of WWII, part of the research on war discourse is almost exclusive to them. While oral histories, critical discourse analyses, and interpretative studies can be found across different contexts and fields, quantitative discourse analysis is heavily skewed towards the developed world or transnational institutions (e.g. US Congress or British Parliament on the one hand and European Union and the United Nations on the other). At least two reasons for that need to be mentioned here. First, many post-conflict societies do not offer resources on par with their Western counterparts. It translates to a lack of high-quality databases, which must be created with the support of ruling elites committed to transparency and general openness. Second, as wars are not events exclusively occurring in English speaking world, English-focused out-of-the-box analytical tools have little to no use for low-resourced languages such as Bosnian-Croatian-Serbian (BCS). As a result, it creates an artificial barrier to doing such research in countries actually affected by violent conflicts and their legacies. The missing resources and the limited analytical tools then support the existing gap between what we know about war discourse on different levels of aggregation. These insights are, however, important for a more complex understanding of post-conflict societies and the challenges they face years after the wars are over. While small-N studies can tell us a lot about individual perspectives on war and war past, large-N studies provide a more general view on trends and patterns in political discourse. For many post-conflict societies, this macro perspective is entirely missing, highlighting the gap between what we know about particular actors and their positions on the one hand and more general structures they are part of on the other.

With its three constituting nations and still unresolved grievances, Bosnia–Herzegovina fits the presented pattern well. In the past two decades, most of the research on war discourse in Bosnia–Herzegovina has been dominated by qualitative research designs often combining oral histories with critical discourse analysis. Taking it as a relatively grounded reference point, the existing literature on the war past in Bosnia–Herzegovina highlights predominantly three patterns of how war and war past are discussed and remembered (Moll, 2013; Sokol, 2014). First and foremost, there is the tradition of glorifying one's own battles and heroes combined with a strong accent on own victimhood when it comes to the justification of violent means (Ramet, 2007b, 2007a). The second pattern operates with a strong reference to WWII and is focused on the redefinition of the role of partisan movement and the anti-fascist struggle and their relevance in the process of disintegration of Yugoslavia and the subsequent wars (Kuljić, 2010, pp. 85–115). The third pattern is based on the homogenous historical continuity through which the ethnic groups construct their selves. In this context, history is perceived as a process of endless repetition with the path-dependency of past events to present-day societies (Bijelić, 2005, pp. 81–83).

When it comes to the discursive space of Bosniaks in Bosnia–Herzegovina, the literature agrees that it is dominated by the concepts of ‘aggression’ and ‘genocide’ (Jacobs, 2017). The war of 1992–1995 is predominantly remembered as an act of aggression committed by Serbs and Croats against the Muslim population and as an attempt to destroy the newly independent state of Bosnia–Herzegovina (Duijzings, 2007; Moll, 2013). On the other hand, Bosnian Serbs see the 1992–1995 conflict as a defensive war for their fatherland (otadžbina) led against the Muslim and Croat threat. They perceive the ‘secession’ efforts of Bosniaks and Croats as a betrayal of Yugoslavia and the legacy of WWII (Kostovicova, 2004). The war memory of the smallest constituting ethnic group in Bosnia–Herzegovina, Croats, is very similar to native Croatia (Moll, 2013). The central component of the narrative is built around the myth of ‘Homeland war’ and its political ramifications. The significance of the homeland reference (domovina) is embedded in its cognitive connotation of protecting the home, evoking defense and liberation rather than aggression and hostility (Sokolić, 2019). It is important to emphasise that the concepts of Serbian fatherland and Croatian homeland are not only two different references embedded in the notion of natural living space for Serbs and Croats in Bosnia–Herzegovina but also two radically opposed ideological and historical constructs (Milosavljević, 2002, 2006; Sokol, 2014).

Taking the general idea of fragmented memoryspace as a broadly accepted argument, literature tends to portray political strife in Bosnia–Herzegovina as a three-dimensional conflict that leads to political stalemates and endless obstructions (Keil & Perry, 2015) with troubling policy ramifications (Dragovic-Soso, 2016; Muehlmann, 2008; Sebastian, 2009). The way how politicians confront each other in public space (typically through media) further creates an impression that all three ethnic groups invest equal time and effort in defending their political positions (Mijatović, 2017; Mujagić, 2010; Vuković, 2015). This is also supported by academic research, which selectively pays attention to the most vocal representatives of political parties while trying to balance their presence and the attention they get (Hasić, 2020; Majstorović et al., 2010; Majstorović & Turjačanin, 2013; Moll, 2013). Although highly insightful and relevant, this approach portrays only a narrow sample of the most active politicians, with a significant ‘grey area’ remaining overlooked. Recalling the theoretical discussion above, the current state-of-the-art is almost exclusively formed by qualitative studies with limitations translating to only a partially complete picture of the war discourse in Bosnia–Herzegovina.

We address this gap by broadening the scope with original data from the Parliament of Bosnia–Herzegovina and focusing on average war narratives defined as combinations of all speech acts presented by all representatives of the same ethnic group. Although Bosnian politics is dominated by strong leaders making bold claims (Barton Hronešová, 2022; Majstorović et al., 2010), we approach war narratives from a different angle focusing on a more holistic perspective of political discourse. Using transcripts of political debates, we create a map of over two decades of war discourse in the Parliament of Bosnia–Herzegovina as formed by the three constituting ethnopolitical groups rather than their most vocal representatives. To our best knowledge, it is the first attempt to do such analysis in a post-conflict society using quantitative discourse analysis on over two decades of parliamentary debates. The goal is to understand what patterns of parliamentary war-related discourse prevail in the Parliament of Bosnian-Herzegovina over time and their implications for everyday political reality.

Data and Methods

The article analyses an original corpus of parliamentary debates from Bosnia–Herzegovina which contains parliamentary speeches from both lower and upper houses of the federal Parliament – the House of Representatives (Predstavnički dom / Zastupnički dom) and the House of Peoples (Dom naroda). Transcripts of parliamentary debates were collected from the official website of the Parliamentary Assembly of Bosnia and Herzegovina (Parlamentarna skupština BiH, 2020a) and cover the period from 1998 to 2018.¹ Records were originally stored as machine-readable PDF files with a loose structure and fluid form over different terms. Each document was parsed and text-mined using regular expressions (RegEx) in order to construct a proto dataset with a simple structure having just two entries: a speaker (most often first and last name) and a speech (a string of text capturing transcribed spoken word in Bosnian-Croatian-Serbian). It was then further populated with meta-information assigned to its parent file – House of Parliament, date, and session number. Finally, the names of MPs were linked with their party affiliation as another level for grouping the presented speeches. The final corpus counts 127,713 speech entries (roughly 9.7 million words) with a unique overview of political discourse that captures not only war and its legacies but the whole policy space of the country in its modern history.²

Using this corpus of relatively modest size, we employ two sets of analytical tools. The first of them utilises the latest advancements in NLP, especially word embeddings, in order to model the prevailing patterns in language used by parliamentarians. The approach is based on training vector models, which can approximate the semantic as well as the contextual meaning of words (Mikolov et al., 2013; Rodman, 2020). The intuition behind the modelling comes from an influential position in lexical semantics which argues that semantic representations for words in the form of numeric vectors can be derived from the analysis of patterns of lexical co-occurrence in large language corpora (Sadeghi et al., 2015). John R. Firth (1957) famously coined this position with an argument that ‘you shall know a word by the company it keeps’. The architecture of these models is language agnostic, meaning it performs well across different language families. We train several vector models using text2vec package with GloVe algorithm (Selivanov & Wang, 2019). Rather than using raw textual data, we pre-process the collected speeches in Bosnian-Croatian-Serbian (we do not translate any part of the corpus), so the context of words as well as their form is standardised for further analysis: the speeches are split into sentences and lemmatised, cleaned off of numbers, punctuation, and whitespaces, and turned to lower case. Only words that occur at least ten times in the corpus are kept, effectively disregarding infrequent words as well as misspellings.

The second set of analytical tools builds on an exploratory analysis of the pre-trained vector model. We use the prevailing patterns observed in the corpus across the whole studied period to navigate us back to the raw text and explore actual narratives qualitatively. This allows us to understand both the macrostructure of war legacies in the Bosnian Parliament and follow much more subtle nuances present in the speeches of individual parliamentarians. In other words, language modelling (quantitative part) tells us where to look and what to look for, while close reading (qualitative part) takes us back to actual speeches. Using this approach, we perform a qualitative discursive analysis focused on specific discourse concerning war modelled as prevailing patterns in over 20 years of political debates.

War Discourse on the Level of the Bosnian Parliament

The following section investigates the relations of most prominent war-related concepts and maps them into a network-like structure. The goal is to summarise over two decades of parliamentary debates in Bosnia–Herzegovina. The word embeddings model is trained on the whole corpus (both houses of the Parliament; period 1998-2018) in order to capture semantic relations of words and explore the prevailing patterns they create. After several rounds of testing using grid-search, we settled on two main hyperparameters for the model – the number of dimensions [D] is 100, and the context window is eight words. All other hyper-parameters remain default (Selivanov & Wang, 2019). This setting is both customary and appropriate for a corpus of this size (Rodriguez & Spirling, 2022). The 100D model we explore is stabilised using bootstrapping with 100 iterations (Rodman, 2020). To capture a discursive network of war, we assume that the positions words occupy in a high dimensional vector space have semantic meanings (i.e. words with similar meaning or context have similar vector representation, hence are close to each other). By exploring the neighbourhood of keywords we are interested in and connecting them into a more complex network-like structure, we can reveal macro relations among wartime vocabulary. To do that, we manually explore the first 15,000 most frequent words in the Bosnia–Herzegovina corpus. Following Zipf's law, this relatively small number actually covers 97.6% of all words present in the corpus (over 9.6M words), which substantially covers any empirically relevant concept ever discussed by the MPs. This process helps us identify 269 words that strongly link to Bosnia-Herzegovina's war past.

To explore the overall landscape of war discourse in Bosnia–Herzegovina, we use the list of war-related words and their neighbours extracted from the stabilised 100D GloVe model and connect them through direct links or shared neighbours into a discursive network (Figure 2). Rather than focusing on plain n-closest words, we include all stable words which are among the 100 closest words (using cosine similarity) in at least 50 iterations of bootstrapping. The visualisation is done using Gephi 0.92 and ForceAtlas2 algorithm (Bastian et al., 2009).

Figures 1 and 2 present an overview of war discourse in the parliamentary debates of Bosnia–Herzegovina, focusing on war vocabulary and its respective within-issue relations. The network structurally identifies four distinct clusters of war-related agenda (see encircled areas in Figure 1 and their zoomed overview in Figure 2): (1) war crimes and their prosecution [Cluster A], (2) locations and their role in the war [Cluster B], (3) peacekeeping and refugees [Cluster C], and (4) political and military leaders during the war [Cluster D] (see Figure 2). All four clusters have words that are both frequent and prominent in the corpus (colour hue/size of nodes). The identified topics generally mirror findings of qualitatively oriented research that focuses on either peacekeeping aspects of the post-conflict reconstruction [Cluster C] or a process of reconciliation, including prosecution of war crimes and their perpetrators [Cluster A]. When it comes to names, only a handful of them are predominantly associated with the war, while the majority is separated from the central clusters, hence dominated by different framing [Cluster D]. Karadžić and Mladić are an exception here. They are placed close to words like aggression [agresija] and criminal [zločinac], referring to their responsibility for the atrocities during the war in Bosnia. Srebrenica appears in this context to be the most defining word associated with the generic word war [rat]. At the same time, both are positioned between the cluster of war atrocities and their prosecution [Cluster A] and the most important locations referring to war [Cluster B] (see the network in high-resolution in Appendix [Figure A1]).

Figure 1. Visualisation of war discourse in the Parliament of Bosnia-Herzegovina Note: Size of nodes and colour hue capture the number of sentences in which a word occurs (prominence in the corpus).

Figure 2. Clusters of war discourse in the Parliamentary debates of Bosnia-Herzegovina

This overview perfectly resonates with the existing research on Bosnian politics, war memories, and the past (Majstorović et al., 2010; Majstorović & Turjačanin, 2013; Moll, 2013; Sokol, 2014). As a robustness check, it further supports the internal validity of constructed vector model and its meaningful representation of semantic relations of the war-related corpus. This intermediary step is important for a more substantial analysis presented in the following section where we split the original corpus into three sub-corpora mapping war discourse for the three constituting ethnic groups of Bosnia–Herzegovina. It replicates both the institutional divisions created by the post-Dayton settlement and the everyday political cleavages existing alongside the ethnic lines.

War Discourse among Ethnic Groups in the Bosnian Parliament

To understand how war is represented in the parliamentary discourse of Bosniak, Croat, and Serb MPs, we employ the same approach from the previous section and map prevailing patterns in words referring to war past. We split the corpus into three subsets based on party/ethnic affiliations³ of MPs and ran the same kind of analysis we did for the whole Parliament. War vocabulary is then extracted together with its network relations and visualised using Gephi (Figure 3; high-resolution graphs are also placed in the Appendix). The graphs are explored for prevailing patterns (e.g. topological proximity of words) which are then traced in the raw corpus using a logic of keywords-in-context (KWIC) analysis for extracting substantially relevant speeches (i.e. speeches containing various combinations of the followed keywords). These are then analyzed for the general message, representing the relational patterns of words in the graphs and the general narrative of extracted speeches through time. The data-driven approach allows us to select narratives supported by meta structures modelled on the level of the studied corpus (i.e. co-occurring words having high cosine similarity).

Figure 3. War discourse of Bosniak, Croat, and Serb MPs

Note: Red colour visualises words associated with the wars in the 1990s, while blue capture more general war terminology (manual colouring).

Using the presented approach, we identify three seemingly irreconcilable interpretations of the character of the Bosnian war we organise our analysis around: international aggression against Bosnia–Herzegovina championed by Bosniaks, civil war framing pushed forward by Serbs, and Homeland war references made by Croats. These dominant framings of the war past seem to be consistent throughout all six terms (1998-2018), showing their prominence and continuity in the discourse of war and their internal consistency with the previous research. To demonstrate the temporal dimension, the referenced speeches are selected and dated with the purpose of ‘reading history forward’ (Møller, 2021). Although we use them primarily as an illustration for the patterns modelled on the level of ethnic groups (general pattern → example of actual narrative), their dating constructs a virtual timeline, which helps us keep track of potential changes over time.

Both quantitative modelling and close reading of the parliamentary speeches show that Bosniak MPs are committed to the strong belief that the Bosnian war was an international aggression against a newly formed state led by its neighbours, Serbia and Croatia, with pretensions of territorial expansion (Duijzings, 2007; Jacobs, 2017). While acknowledging that elements of civil war and ethnic and religious conflict played an important part, Bosniaks’ war memory is principally structured around the idea that the Bosniaks fought an uphill battle against their neighbours who strived to carve out the country and split it between themselves. Speeches made by Šefik Džaferović and Asim Sarajlić from the Party of Democratic Action (Stranka demokratske akcije, SDA) document that in a very narrow sense:

… [T]here was no civil war in B&H. There was an aggression against B&H […]. There are at least three verdicts of the Hague Tribunal confirming that. If you want, I can show you several UN Security Council resolutions from which you will clearly conclude that aggression was carried out against B&H. Moreover, watch the recordings of various TV channels from 1992 to 1995, and you will see where the tanks did come from to B&H, columns of tanks. (Šefik Džaferović, SDA, 14/6/2001)

To discuss the State of the Security in B&H like this is just ignoring the facts; it is an attempt to detach it from reality. We do not live in a Disneyland. There was a war in this country, there was an aggression, and there was a genocide. (Asim Sarajlić, SDA, 29/8/2913)

The view of the Bosnian war almost unanimously shared by Serb MPs is fundamentally opposed to this international aggression framing. For Serb MPs, the tragic war in Bosnia–Herzegovina is undeniably a civil war that happened as a consequence of one ethnic group (Bosniaks) forcing their political ambitions (independence of Bosnia–Herzegovina from Yugoslavia) onto others (Barton Hronešová, 2022; Kostovicova, 2004). In the early terms of the Parliament, this rhetoric was championed by the Serb Democratic Party (Srpska demokratska stranka, SDS), the leading Serb party at the time. Gradually the SDS attempted to soften its attitude under the leadership of Dragan Čavić primarily due to pressure from the Office of the High Representative (OHR) and to reform its image as torchbearers of radical right-wing Serbian nationalism (Majstorović et al., 2010). As a result, the vacated position was soon seised by the Alliance of Independent Social Democrats (Savez nezavisnih socijaldemokrata, SNSD), which took over the role of the true defender of Serbs and Republika Srpska. Unburdened by the actual war responsibility, the rhetoric of MPs from the SNSD became increasingly radicalised after coming into power, effectively preserving the legacy of SDS in the highest level of Serbian political representation. We demonstrate the change with the excerpts from the speeches made by the members of SDS and SNSD from the late 1990s to the mid-2010s:

… [F]or the sake of building trust among peoples of B&H, I sincerely ask you not to mention an aggression. Whose aggression and against whom? Who can say that the RS is a fascist creation which originated from the aggression of the SRY against B&H? (Mirko Banjac, SDS, 5/10/1999)

Do not use the vocabulary from 1990 in 2002. The wounds are still fresh and painful. Get rid of the term aggression. (Momir Tošić, SDS, 29/5/2002)

Which court has ruled that there was an aggression against B&H? As soon as some court does that, I will be the first to acknowledge it, same as I did for Srebrenica. There was a genocide in Srebrenica, International Criminal Tribunal for the former Yugoslavia (ICTY) confirmed that, and none of us dispute it. I accept the court's rulings, but which court has ever ruled that there was an aggression? Because there was not. (Slavko Slavuj Jovičić, SNSD, 22/10/2008)

What happened in Bosnia–Herzegovina was a civil war and a conflict, and I am saying that not just for the sake of it but because I believe in it. I stand by my words. What happened here was a conflict among three groups of people. What you say and think is your own business, but you cannot ask others outside of this Parliament to interpret our history […] the way you perceive it. I see it this way and Serbs in Bosnia–Herzegovina see the same thing. It was a civil war and not an aggression or any other kind of conflict. (Milica Marković, SNSD, 11/5/2016)

The main frame Croat MPs use when talking about war is similar to the dominant narrative presented in mainland Croatia (Mochtak, 2020). Their notion of war is characterised through a framework of a ‘Homeland war’ as a defensive and protective struggle against external threats. The significance of the reference usually highlights the noble cause of protecting home/homeland, while the reference itself signifies the sacred legacy of the fight. This can be supported by the speeches presented by members of the two main Croatian parties in Bosnia–Herzegovina, the Croatian Democratic Union of Bosnia and Herzegovina (Hrvatska demokratska zajednica Bosne i Hercegovine, HDZ BiH) and the Croatian Democratic Union 1990 (Hrvatska demokratska zajednica 1990, HDZ 1990):

Why did the Government of Federation of B&H allow its Ministry for Issues of the Veterans and Disabled Veterans of the Defensive-Liberation War to omit the category of demobilized defenders and remove the term ‘Homeland War’ during the drafting of The Law on Fundamental Rights of the Disabled Veterans and Families of the Killed Soldiers? It has caused public outrage. (Mirko Grabovac-Titan, HDZ BiH, 20/11/2001)

… [A]s a member of the HDZ 1990 I am a pure-blood Croat, both from my father's and my mother's side […] and in comparison to many of you sitting here and calling yourselves Croats, I am a volunteer and a veteran of the Homeland War. (Diana Zelenika, HDZ 1990, 27/8/2015)

The reconciliatory tone predominantly used when talking about the war in neutral terms gives away to a more defensive attitude once Croat MPs are faced with the actual historical legacy of Herzeg-Bosnia and the HVO (Croatian Defence Council [Hrvatsko vijeće obrane]), the main military organisation of Croats during the Croat–Bosniak conflict:

… [W]e often celebrate the Day of the Croat Republic of Herzeg-Bosnia, which was with all of its political and legal heritage recognized by the Washington Agreement and is one of the building blocks of this country. Unfortunately, we do not celebrate this day enough, and hopefully, we will do better in the future. (Predrag Kožul, HDZ BiH, 30/8/2016)

The unsuccessful attempt to establish an exclusively Croat state within the territory of Bosnia–Herzegovina is not brought up often and it is far less prominent in the Croatian discourse than might be expected. Even when defending the idea that Croats were the ‘good guys’ of the Bosnian war, Croat MPs rarely shift blame to the other sides the same way that is common for Bosniak and Serb representatives:

Mr. Sokolović, an argument that genocide happened in Mostar during the war does not deserve a reaction. Mostar is a truly multi-ethnic city and during the fiercest clashes between HVO forces and the Army of the Republic of Bosnia and Herzegovina, more than 10.000 Bosniaks and other ethnicities lived there. (Božo Ljubić, HDZ, 1990, 12/6/2014)

This peculiar behaviour is probably most apparent when discussing the political and military leadership of the Herzeg-Bosnia proto-state. Being convicted by the International Criminal Tribunal for the former Yugoslavia (ICTY) on account of participating in a joint criminal enterprise (International Criminal Tribunal for the former Yugoslavia, 2013), the whole issue is very uncomfortable for both Croat political leadership in Bosnia–Herzegovina as well as in Croatia. While Serb MPs are more inclined to engage in debates and defend the war legacy of the Army of Republika Srpska with the intention of protecting the legitimacy of Republika Srpska irrespective of the rulings of the ICTY, Croat MPs have little to gain from advancing the discourse on the historical legacies of Herzeg-Bosnia. On rare occasions when they directly address accusations of war crimes committed on the territory of Herzeg-Bosnia by the HVO forces, they do it in a noticeably less confrontational way even when still attempting to dispute these claims and shift the focus away from the actual events. As an example, we provide an excerpt from a speech made by Mario Karamatić from the Croatian Peasant Party (Hrvatska seljačka stranka, HSS) which documents the pattern:

I do not understand how it is possible that members of the Croat Defence Council get systematically qualified as a joint-criminal enterprise. Gentlemen, the territory that was part of the Croat Republic of Herzeg-Bosnia is the only part of the country where no mass graves were found, and this is a fact. (Mario Karamatić, HSS, 17/9/2015)

Contrary to a theoretically driven expectation that representatives of all three ethnic groups might invest equal effort in preserving their war-related memoryscapes, our analysis shows that this is not true. Strategic use of war memory for all kinds of short-term political goals, from delegitimizing opponents to pre-emptive accusations, is a common practice by MPs of all three ethnic groups. However, the main line of discursive confrontation on the subject of war in the Parliament is between Bosniak and Serb MPs (see also Campbell, 1998). The graph of Croat MPs war discourse illustrates this difference (Figure 3; Figure A3 in the Appendix). Besides the fact that Croat MPs’ war discourse is not clearly split into two separate clusters of war-related words, which is something that we found in both Bosniak and Serb discourse (one more closely related to war past and one with more general war terminology), war discourse of Croat MPs is also less loaded with references to war atrocities. This is not to say that Croat MPs are completely exempt from using graphic descriptions of war crimes, but those are far less prominent in their discourse when compared to Bosniaks and Serbs.

When it comes to the key discursive alliances, both ‘Srebrenica’ and ‘genocide’ [genocid] are associated with the word ‘war’ [rat], similar to Bosniaks’ discourse, but expectedly they are not emphasised in the same way since they do not constitute a crucial part of the war narrative for Croats. Adjective ‘civilian’ [građanski], while still being close to the main cluster of war words, is not directly connected or contextualised with the noun ‘war’ [rat] or its adjective [ratni]. Moreover, we did not find any instances of Croat MPs deliberating on the topic of ‘civil war’ [građanski rat], which we identify as a critical intersection point for war debates for both Bosniaks and Serbs.

The same graphs for Bosniak and Serb MPs show that both are heavily invested in the debates on the topic of the character of war. Adjective ‘civilian’ [građanski] and the noun ‘war’ [rat] are positioned close to each other, almost in the centre of Serbs’ war discourse. In the Bosniak discourse, ‘civilian’ [građanski] is used in a similar context as nouns ‘war’ [rat], ‘victim’ [žrtva], and ‘conflict’ [sukob] and the adjective ‘war’ [ratni]. The complete exclusion of Croat MPs from one of the most fierce and persistent debates about the Bosnian war demonstrates that Croat MPs not only did not engage with their Bosniak and Serb counterparts but that their interpretation was rarely directly challenged by either of the two. For Croats, the war in Bosnia–Herzegovina could not be disentangled from the Homeland war led in Croatia against Serbia and their experiences are perceived to be closely linked together. When Serb MPs tried to raise the topic of Croatian paramilitaries’ involvement in fighting during the war and referred to the chain of command that existed between Croatian officials and Herzeg-Bosnia forces, Croat MPs stayed mostly silent and did not participate in these discussions.

We argue that this passive position was maintained mainly because the occasional attacks were not directed towards them in the first place, hence not being politically relevant enough to be vocally opposed. The main purpose of the aforementioned Serb argument was to challenge the international aggression narrative championed by Bosniak MPs and to prove that the military involvement of the Yugoslav People's Army (Jugoslovenska Narodna Armija, JNA) is not enough to confirm the international component of the war since combatants from other former Yugoslav republics played a role as well. Although the ruling of the ICTY in the case of Prlić et al. [IT-04-74] (2017b) ultimately points to the other direction, it does change the main point we find symptomatic when it comes to Croat MPs: even when Croats are the main targets of belligerent rhetoric, the actual conflict still reflects the nexus of Bosniak–Serb power relations.

Interestingly, the pictured dynamics highlight a strategic concealing on the side of Croat MPs, potentially reflecting more on their political aspirations rather than their true preferences. We theorise that Croat MPs have a deliberate interest in staying out of the Bosniak–Serb strife as a result of being caught between a rock and a hard place of everyday political reality. Even if they might agree with the Bosniak interpretation of the war, Bosniaks are their main competitors for resources within the Federation as well as they are the main source of their political frustrations (e.g. being the main political barrier preventing the creation of a third entity). This is why Croat elites might play a non-confrontational game with the Serb MPs hoping they will repay the favour and help them eventually to get the third entity, or at least support a change in the electoral law that would stop Bosniak voters from effectively voting for the Croat member of the Presidency.

However, it also might be the case that Croat political aspirations, although fundamentally opposed to Bosniak unitarist tendencies, do not directly influence the moderate war rhetoric of Croat MPs. Both previous attempts to establish an independent Croat entity within Bosnia–Herzegovina, first in 1993 under Mate Boban but also in 2001 under Ante Jelavić failed. Siding with the ethno-(con)federalist agenda championed by Milorad Dodik and Serb SNSD, while undoubtedly an attractive proposition for the dominant western Herzegovina line within HDZ BiH, would be a tough sell for its leadership. Support for such radical pursuit would hardly come from any of the Western countries or even Zagreb. It would also mean turning back on Croats in Posavina, who would be left in the Republic of Srpska as a de facto minority. This is why the idea of creating a ‘third entity’ has received more support from Serbs than Croats when it comes to public stunts. Thus, while the political aspirations of the HDZ BiH leadership are clearly in conflict with those of Bosniaks and, in theory, closer to Serbs, they are kept under tight control and rarely expressed openly in Parliament.

In the end, the restrained usage of war references is actually more beneficial to Bosniak than Serb positions. While any combustible war memory is potentially convenient for Serbs elites to escalate political crises and proclaim the impossibility of a functional central state, Bosniaks, on the other side, have a pronounced interest in making Bosnia–Herzegovina work. Stable Bosniak–Croat relations in this context are crucial to enduring overall stability of the country. For Croats, Bosniak unitarism is simply tolerated in exchange for a favourable share of institutional control. For years Bosniaks have accommodated Croats who held more than a fair share of offices on both state and the level of the Federation of Bosnia–Herzegovina in exchange for Croats’ silent support of Bosniak slow but steady push towards strengthening powers of the central state. In recent years this Bosniak–Croat dynamic came almost to a grinding halt mainly due to the open subversion of the Daytonian order orchestrated by the Social Democratic Party of Bosnia–Herzegovina (Socijaldemokratska Partija Bosne i Hercegovine – SDP BiH). In a highly controversial move, SDP BiH exploited the numerical weakness of Croats and secured the election of Željko Komšić to the Presidency of Bosnia–Herzegovina. Komšić, while himself ethnically Croat, was elected mainly by the Bosniak votes. This move showed a way forward for those on the left who saw an opportunity to attack the Daytonian political order that favours ethnonationalist parties and animates their voters.

The response from HDZ BiH was a resolute refusal to accept anything less than absolute control over what is deemed to be the political right of Croats as constituent peoples. But the insistence on implementing ‘legitimate representation’ (Krišto & Čolak, 2018), while highly destructive to any notion of citizen focus on Bosnian democracy, should not be conflated with attempts to establish a ‘third entity’. As our data shows, no significant change occurred in Croat MPs’ war rhetoric even after the political monopoly of the HDZ BiH was highly compromised. In our view, any serious possibility of the Serb–Croat alliance intent on dissolving Bosnia–Herzegovina by any means necessary would be preceded by a noticeable change in the way Croats talk about the war.

For as long as Croats see Bosniaks as possible partners in Bosnia–Herzegovina, the system under which the country has been operating since the war ended will remain the dominant platform for expressing ethnopolitical interests. After all, when left to their own devices, SDA and HDZ BiH often managed to find common ground in the past. The latest example is their agreement on the Mostar elections in 2020 (Parlamentarna skupština 2020b). But if Bosniak political forces continue using Bosniak numerical superiority to establish dominance over Croats, this could cause serious disturbance to the implicit Daytonian ethnic power-sharing model and force Croats into changing their rhetoric and maybe even switching sides.

One way or another, these dynamics introduce a little-known perspective on how the political representation of the three entities interact when it comes to high politics and what it means to instrumentalizing the war past and its relevance in Bosnian-Herzegovina. Although it would be a mistake to generalise the dynamics concerning the war discourse automatically to the overall political landscape of Bosnian politics, it certainly invites to a more thorough analysis of the power relations among the three constituting ethnic groups. Rather than repeat the premise of a three-sided stalemate, it is better to think about Bosnian politics in a 2 + 1 setting, under which the Bosniak–Serb strife is accompanied by a balancing position of the Croat representation pursuing its own political agenda.

Bosniak–Serb Strife

When it comes to Bosniak and Serb MPs there is very little restraint when debating the character of the Bosnian war. Discussions are frequently confrontational and emotionally charged. Personal stories and anecdotes of MPs who had first-hand war experiences are used to effectively challenge the narratives that do not correspond to the views of their own ethnic group. For instance, Serb MPs often use references to camp ‘Silos’⁴ as a counter-argument leveraging the allegations of war atrocities committed during the war. Personal experiences are often used as evidence against any attempts of the other side to paint a one-sided picture in which Bosniaks are presented as the main victims:

In a prison camp where I was held for 44 months, and do not dare to accuse me of abusing this, late Alija Izetbegović on 247th session in April 1995 said: ‘Serbs are held in Silos camp as a countermeasure, they are not guilty’. Is anyone going to be held responsible for the fact that 25 of my fellow inmates were killed? Is anyone going to be held responsible for the fact that all men between 14–75 were imprisoned without asking any questions? No charges were ever filed. (Slavko Slavuj Jovičić, SNSD, 30/09/2009)

… [I] was here, in Sarajevo, while my family was held in Silos camp for 44 months. Those who accuse me of not being ready for compromises, well, the leader of their party sent a commission to Silos where my mother was held. (Aleksandra Pandurević, SDS, 22/12/2011)

On the Serbian graph, ‘Silos’ is in close proximity to the words ‘Jasenovac’, ‘Dobrovoljačka’ and ‘kolona’. Words’ kolona’ and ‘Dobrovoljačka’ refer to the Yugoslav People's Army (JNA) column incidents in Tuzla and Sarajevo at the beginning of the war in early 1992. On the other hand, Jasenovac is the infamous concentration camp operated by the Ustasha regime during WWII. This discursive retaliation is used frequently by SNSD but also by smaller Serbian parties such as the Party of Democratic Progress (Partija demokratskog progresa, PDP):

I am asking you, how is it possible that there are no names of perpetrators of crimes in Dobrovoljačka in the Resolution? You know it. We were all witnesses. I lived in the city. There is no person I cannot look straight in the eyes. How is it possible there is no mention of the 18 years old kids at Brčanska Malta? Sure, we do not know the name of the perpetrators but let's at least mention them. (Nikola Špirić, SNSD, 3/1/2003)

It is a well-known fact that on the territory of Yugoslavia, Serbs suffered the most during World War II … If you are talking about Europe, then yes, Jews suffered more, nobody should argue against that, but if you are talking about the territory of Yugoslavia at that time, the Serbs suffered the most. (Slavko Slavuj Jovičić, SNSD, 3/2/2010)

Please do not make me remind you of one of the worst execution sites in Europe, Jasenovac. It is still quite recent when it comes to historical context. It is not acceptable that you emphasize only one tragedy that happened in Srebrenica and forget others. In the context and structure of this country, that is not acceptable and it is not right. (Branko Dokić, PDP, 23/3/2010)

Both quantitative modelling and in-depth qualitative analysis show that names of the places where Serbs were victims are frequently contextualised together. Interestingly, these references often appear in speeches mentioning Radovan Karadžić and Ratko Mladić, Serb civilian and military leaders who played an instrumental role in committing atrocities during the Bosnian war (International Criminal Tribunal for the former Yugoslavia, 2016, 2017a). The symbolic link plays an essential role in many of the ‘attempted deliberations’ in which the recognition of Karadžić and Mladić as war criminals by the Serbs MPs is often conditioned by demands for the recognition of suffering that their own ethnic group endured during different periods in history even when there is no explicit connection between the two:

How to explain the news that the proceedings of The Prosecutor's Office of Bosnia and Herzegovina against 50 individuals, mostly members of the SDS, […] have been stopped two years after everyone was well aware that none of them did it? […] There should be some balance: we have on the one hand a stopped investigation against non-existent helpers of Karadžić and on the other the amnesty for those who ordered and executed the massacre in Dobrovoljačka. Do you think we are blind and that we do not see what this is all about? (Mladen Bosić, SDS, 16/2/2012)

As expected, we find that Bosniak war discourse revolves heavily around the words’ genocide’ [genocid], ‘victim’ [žrtva], and ‘Srebrenica’. The word ‘Holocaust’ [holokaust] is also in close proximity to words ‘war’ [ratni], ‘conflict’ [sukob], ‘criminal’ [zločinac] and ‘to prosecute’ [procesuirati]. The main cluster of war-related words is defined by the horrific massacre of Bosniaks in July 1995. In comparison, the Serb war discourse places’ genocide’ in the main war cluster as well, but the word ‘Srebrenica’ is contextualised with other names for towns and places and not directly connected to the main war discourse. Unsurprisingly, the association of the word ‘genocide’ with the word ‘Srebrenica’ is not very common among Serbian MPs. Conceivably, this warrants a strong reaction from Bosniak MPs and accusations of negating the evidence on genocide and ethnic cleansing:

By denial of a genocide, by persistently refusing to adopt the Law on Genocide and Holocaust, and finally by not supporting the laws on The Court and The Prosecutor's Office of Bosnia and Herzegovina, you are obstructing the prosecution of criminals who committed war crimes and genocide in Srebrenica. It concerns the whole B&H, including those seven municipalities for which the Hague Tribunal tries Karadžić and Mladić. And precisely because of the genocide, war crimes, and ethnic cleansing, there are million people missing today in the Republika Srpska and in Bosnia–Herzegovina as a whole. Ladies and gentlemen, genocide against Bosniaks in B&H was committed and that is a legal, human, and God's truth. (Šemsudin Mehmedović, SDA, 16/02/2012)

Bosniaks view themselves as the main victims of the Bosnian war and their role, perceived as morally superior, allows them to use strong rhetoric of blame (Burg & Shoup, 2015, p. 181). Genocide in Srebrenica occupies a central role in Bosniaks’ mythogenesis, aspiring to become the Bosnian Muslim Holocaust (Miller, 2006). For this reason, Bosniak MPs tend to constantly nurture their victim status and re-establish the aggression premise as two essential elements of the Bosniak war memoryscape:

If anyone has the right to speak about what happened in B&H, it is Bosniaks. They suffered the most in this war, during an aggression, which indeed happened in B&H. (Abdurahman Malkić, SDA, 5/10/1999)

Both Bosniaks’ and Serbs’ framings of the Bosnian war are greatly dependent on the narrative concerning the Srebrenica genocide. Recognition of horrific events in Srebrenica as genocide completely negates the Serbian framing of the war and this is why the truth about the number of victims or their status (civilian or military/paramilitary) is often questioned:

I repeat, I feel sorry for every victim of the war, but I expect that one day it will be said that in Srebrenica and the areas around Srebrenica, Bratunac, and other municipalities, out of 3,000 victims, 200 or more were Serbs and that someone killed those Serbs. I feel sorry for every lost Bosniak life, but I expect that you separate those who died innocently from those who were criminals and died while killing Serb civilians. I expect you to do this for yourself and those innocent victims. They do not deserve to have their names next to the names of those who committed crimes. (Aleksandra Pandurević, SDS, 23/5/2013)

And even when acceptance of the Srebrenica genocide is given, it is done in a manner that indirectly implies that Srebrenica was not the only genocide that happened during the Bosnian war and thus undeserving of its special status:

The argument that those who committed war crimes did it in the name of the Serb people is absolutely not true. Nobody did anything in my name, and I represent the Serbian people, neither in Srebrenica nor anywhere else where war crimes or genocide were committed. (Momčilo Novaković, SDS, 29/6/2005)

When debates on ‘Law on the Prohibition of Genocide and War Crimes Denial’ were held in both houses of the Parliament in 2011, tempers flared. Serb MPs identified this law as a final confirmation of the dominance of the Bosniak genocide narrative and vigorously fought to prevent it from being adopted:

… I understood the intention of the author of this law to prevent demystification of the events and to preserve the distorted image which was nurtured for 16 years. Why? First, adopting this law would lead to a situation when accused Serbs would not be able to defend themselves against an allegation they committed a genocide. They would be convicts without a trial. (Dušanka Majkić, SNSD, 8/12/2011)

The pressure of being perceived as the main perpetrators of war crimes and responsible for the Srebrenica genocide, both domestically and internationally, weighs heavily on Serb political representatives. Assuming the role of exclusive protectors of the Serbian people and their interests defines the fundamental inability of Serb MPs to reflect on tragic events in which victims were those from the other side. If that account is taken into consideration, it usually comes together with questioning the truthfulness of counter-arguments while entertaining a dose of conspiratorial rhetoric:

… [D]irect witnesses of what happened during the war are getting increasingly doubtful of the black and white picture of the war past in B&H. Let me remind you of an opinion of the former Special Representative of the Secretary-General for the Balkans Yasushi Akashi who said: ‘It is still not clear who targeted the Markale marketplace’. (Dušanka Majkić, SNSD, 8/12/2011)

In the same speech, Dušanka Majkić describes how Bosniak politicians are intentionally preventing the truth about numerous crimes committed against Serbs in Sarajevo from being investigated while questioning the truth about massacres that happened during the shelling of the ‘Markale’ marketplace in Sarajevo:

… [I]t is well known that most killings of the Serb people happened in Sarajevo from 1992 until the end of October 1993. It all started at Kazani and the chain of Serb suffering continued in Dobrinja, Hrasnica, Novo Sarajevo, Otoka, and other illegal prisons. Authorities of the Federation of Bosnia–Herzegovina still do not want to form a commission for investigating the suffering of Serbs in Sarajevo, although the Ministry of Interior of Republika Srpska provided a long list of evidence about committed war crimes. Demands made by the families of Serb victims were ignored, sending a signal that Bosniaks are not interested in the sufferings of Serbs in Sarajevo. Evidence documenting a large number of cases from Dobrovoljačka, Central prison, ‘Viktor Bubanj’ barracks, Tarčin camp, Hrasnica camp, and other killings happening in Sarajevo's municipalities were passed to the relevant institutions. And I stop here. The full truth about war events has not been revealed yet and this is the main precondition for reconciliation. (Dušanka Majkić, SNSD, 8/12/2011)

The presented overview shows that narratives about the war are crafted, used, and manipulated by the political elites in order to achieve political goals, discredit competitors, shift focus away from own failures, and ‘when most powerful, […] fuse together the present and the imagined historical past while disconnecting the present from an actual personal past’ (Lieberman, 2006, p. 307). In Bosnia–Herzegovina, where ethnopolitics have dominated public discourse over interest-based politics ever since the war ended, the significance of control over war memory can not be overstated (Mujkić & Hulsey, 2010). Our analysis of the political discourse of the Bosnian war shows that more than twenty years after it ended, the war past is still a highly contested topic among MPs. As we show, the main conflict line considering the war past is driven alongside Bosniak-Serb grievances, with Croat MPs being far less engaged. Moreover, when it comes to the Croat memoryscape, we identify very cautious attitudes and careful deliberations about the war, further highlighting the contrast with the other two groups of MPs.

Placing the studied narratives on a timeline and ‘reading history forward’, the analysis shows that very little has changed when it comes to the reconciliation of war narratives over two decades of documented political discourse. The situation might be even worse. This finding is worrying not only because it confirms known arguments about a failed reconciliation on the level of competing ethnic groups (Belloni, 2009; McMahon & Western, 2009; Perry, 2019) but mostly because it shows that political elites and the post-Dayton Parliament have not reconciled over the painful past at all. This does not mean that no form of reconciliation has taken place over the past two decades outside the Parliamentary Assembly. Some of the existing research points out that the hope of fostering reconciliatory attitudes in the broader population is at least a real possibility (Brkić et al., 2021; Perišić et al., 2010). However, our data shows that the Parliament as the highest legislative body and its members have done very little in mitigating the existing conflicts and supporting the bottom-up reconciliatory efforts on a systematic level.

Conclusion

Any post-conflict reconstruction requires a significant effort devoted to overcoming the grievances of past wars. It seems that Bosnian parliamentarians have done very little in this regard, effectively preserving the relative strength of fighting parties and their interests as cemented by the Dayton Agreement back in 1995. Although most politicians in Bosnia–Herzegovina like to talk about reconciliation, the European future, and hope, the actual signals they send when it comes to war are bleak at best. We can make this kind of conclusion based on an analysis of over twenty years of parliamentary debates in Bosnia–Herzegovina which maps the discourse of war of at least two generations of politicians. When it comes to overall patterns, the article shows that parliamentary debates roughly resemble the discourse of war presented in media with one important distinction. The main conflict line exists alongside the Bosniak-Serb grievances, with Croat MPs taking the back seat. Although the recent development in parliamentary politics in Bosnia–Herzegovina indicates that the discourse of Croat political elites might have started to alter (Index Vjesti, 2021), our analysis does not confirm this. On the contrary, we find no substantial evidence that the gradual broadening of the political crisis in the Federation of Bosnia–Herzegovina from 2013 onwards between Bosniak SDA and Croat HDZ BiH led to an increase in the prominence of new war-related narratives among Croat MPs. Interestingly, this seems to be the case for the whole studied period adding an important analytical layer to our understanding of the post-conflict situation in Bosnia–Herzegovina. Moreover, it shows the real face of long-term war legacies that affect actual policies and their implications (Dragovic-Soso, 2016; Muehlmann, 2008; Sebastian, 2009).

Our analysis demonstrates how the past conflicts in the Bosnian Parliament are still alive and vivid as ever. Serb rhetoric has become progressively more radical and defensive of the war legacies and their relevance for Serbs in Bosnia–Herzegovina. Defending the semi-autonomous Bosnian Serb Republic has become a sacred goal of the Serb political elite, often used strategically as a bargaining token in the context of separatist ambitions. Bosniaks’ response has grown to be equally radical. The call for a dissolution of Republika Srpska on the grounds of collective assignment of guilt for committed genocide demonstrates a new vision some of the Bosniak elites have for the country. Amid open confrontation between Bosniak and Serb political representatives, the Croat MPs’ strategy of equidistance seems to be side-tracked at best, putting the official political ambitions of the Croatian entity at odds with the political reality of the Bosnian Parliament.

The way how war discourse shapes the Bosnian Parliament has several theoretical implications. Our analysis shows that reconciliation incentives have never really worked on the level of Bosnian powerholders highlighting the dubious legacy of the Dayton Agreement and the institutional constraints it created (Belloni, 2009; Hoogenboom & Vieille, 2010; Kostić, 2012). Moreover, war has always been a useful reference whenever blame needs to be assigned, when past conflicts might have any relevance in present-day affairs, or when a morally higher ground is needed. The generational change that occurred in the past two decades in the Parliament of Bosnia–Herzegovina has not affected the prominence of war discourse and a new group of ethnic leaders uses the same kind of rhetoric their predecessors had used. Memory politics play a crucial role in this generational stiffness, benefiting those who can capitalise on past grievances and unresolved issues (Božić, 2019; Lazic, 2013; Moll, 2013; Sokol, 2014).

As our analysis shows, the war topic in the Bosnian Parliament functions as a symbol of a failed reconciliation on the level of elites. It does not mean that in other political arenas or policy areas the situation is the same. However, when it comes to the memory politics of the Yugoslav wars, politicians tend to co-coexist but the way they interact is still driven by the unresolved conflicts of the 1990s. While new maps of the country are starting to be drawn again, control over narratives of the Bosnian war will only gain importance, and due to their incandescent nature, they will remain a potent source of ethnic polarisation for the foreseeable future. However, rather than portray the political strife in Bosnia–Herzegovina as a three-sided stalemate, our analysis shows that the conflict line between Bosniaks and Serbs appears to be the defining structure for any kind of future reconciliation in Bosnia–Herzegovina. The main finding concerning the comparatively restrained use of war rhetoric by Croat MPs offers an important contribution to the question of why Bosnia–Herzegovina has continued its often-troubled existence in Daytonian constitutional form despite being remarkably stagnant and inefficient. While the Bosniak-Serb conflict line is constantly on the verge of falling into the abyss of open confrontations, Croat MPs rarely engage with any of the sensitive war topics (e.g. the character of the Bosnian war, Srebrenica genocide, or the prosecution of war crimes) that we identified as key in defining the contested landscape of war memory in Bosnia–Herzegovina. The role of Croats as neither the principal aggressor nor the principal victim of the Bosnian war enables these moderate attitudes and prevents the communication channels from completely collapsing. Consequently, the identified positions alter the paradigm of the three-sided conflict and call for a more accurate 2 + 1 model capturing the central Bosniak-Serb strife accompanied by Croat strategic positioning. Although our findings can not be automatically generalised over other policy areas in Bosnia–Herzegovina and used as a catch-all explanation for the complex web of ethnopolitical divisions, they offer a new way of thinking about how the ethnopolitical elites utilise conflict lines from the Bosnian war in the Parliament of Bosnia–Herzegovina.

Supplemental Data

Supplemental data for this article can be accessed at https://doi.org/10.1080/17449057.2022.2120283

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by an ERC Starting Grant Number 714589 for the project ‘Electoral Legacies of War: Political Competition in Postwar Southeast Europe’ (ELWar). H2020 European Research Council

Notes

1 The first postwar term (1996-1998) is not covered as only session minutes are available. The analyzed corpus covers only finished terms.

2 The corpus is freely available via Zenodo data repository at doi:10.5281/zenodo.6517697 (Mochtak et al., 2022).

3 Rather than assigning known or assumed ethnicity on the level of MPs, we do the modeling using party affiliations and their declared ethnic background. Multiethnic parties are excluded.

4 'Silos’ was a concentration camp operated by the Army of the Republic of Bosnia and Herzegovina (ARBiH) (Balkan Insight, 2009; Oberpfalzerová et al., 2019).