- 1 Kalyvas 2006.
- 2 I would like to thank Agustin Cosovschi, Gilles de Rapper, Ivana Spasić, Fabrice Virgili and Polyme (...)
1In 2006, Cambridge University Press published The Logic of Violence in Civil War by the Greek political scientist Stathis Kalyvas.1 Does it make sense to take a fresh look at this book, which is almost twenty years old, at a time when political scientists’ analyses of civil wars and violence have developed considerably, and Kalyvas himself has published numerous articles to build on and sometimes adjust his approach? I think so, for two reasons. The first, objective, reason is that Kalyvas’s book is a seminal one, systematically cited in political science scholarship, and has inspired a series of other analyses of civil wars and violence. The second, more subjective, is that my own reading of this book has evolved over time. At first I read it with enthusiasm, but each time I have gone back to it, new questions arose. This paper is therefore an attempt to formulate my doubts in a clear and, I hope, coherent way, about a book that I still consider to be a major work and which I recommend to anyone interested in political violence. I shall therefore attempt to summarise Kalyvas’s analyses of selective violence in the context of civil war, before presenting my own reservations about his use of sources, his recourse to the theory of rational choice, and his portrayal of peasant societies in wartime. On this basis, I will present other possible ways of analysing such societies, also focusing only on the Greek case.2
2As Stathis Kalyvas makes clear, his book is not about civil wars and their causes, but about the use of violence in a civil war context. He defines this as an armed conflict taking place within the borders of a sovereign political entity and leading to a challenge to the state’s monopoly of legitimate violence. In such a context, a division between the territories controlled by the incumbents and those controlled by the insurgents appears. Kalyvas excludes genocidal violence from his scope of analysis; he focuses instead on civil wars in which at least one protagonist seeks to control the non-combatant population rather than to expel or exterminate it. More specifically, he is interested in the selective violence that each belligerent uses to reinforce this control. Indiscriminate violence is excluded from this analysis because it is counterproductive and tends to be abandoned as the conflict moves forward.
3This analysis of selective violence is based on a twofold rejection. On the one hand, it rejects macro-historical analyses centred on the structural causes of conflict, reasoning on the basis of pre‑war social, political and cultural conditions and bringing into play collective entities such as ethno-national communities, social classes or political groupings. On the other hand, it is opposed to representations of civil war inspired by Thomas Hobbes, whereby the collapse of state structures and social norms leads to a war of all against all, and to those derived from Carl Schmitt, according to whom the ideological polarisation of the belligerents gives rise to an ever-increasing level of violence. Kalyvas, for his part, focuses on the endogenous causes of wartime violence, using a microhistory of selective violence to analyse variations in space and time. He also looks at non-combatants’ allegiance to one side or the other. These individuals are motivated by their own survival and other personal interests, but their collaboration is essential to enable armed actors to identify and eliminate opponents on their territory.
4On this basis, Kalyvas demonstrates that the greatest levels of violence are not to be found, as might be expected, in the most hotly disputed regions, but rather in those where one of the protagonists exercises dominant but not hegemonic control. To reach this conclusion, he constructs a model based on rational choice theory, distinguishing five different zones, ranging from the hegemonic control of the incumbents (zone 1) to that of the insurgents (zone 5), via two zones of non-hegemonic domination (zones 2 and 4) and a zone of contested control (zone 3). It turns out that, in zones 1 and 5, non-combatant individuals are willing to denounce opponents, who are by definition few in number, but the armed actors do not need this information. In zone 3, on the other hand, the armed actors are looking for information but the non-combatants are unwilling to provide it for fear of reprisals or counter-denunciation. It is therefore in zones 2 and 4 that the behaviour of armed actors (searching for information) and non-combatants (willing to denounce) converge to produce a high level of selective violence.
5To test the validity of this theoretical construct, Kalyvas focuses on the 61 villages of the Argolida, a region in the north-eastern Peloponnese, at the end of the Second World War (September 1943-October 1944). He meticulously reconstructs the variations in selective violence in time and space, drawing on secondary literature, court archives and numerous interviews conducted in the villages of the region. Kalyvas divides his 61 villages into the five previously-defined zones of control, and also distinguishes between four different periods, thus identifying a total of 20 distinct situations, ranging from zone 1-time 1 to zone 5-time 4. This analysis shows that his theory of selective violence is very much confirmed by the case of the Argolida. Kalyvas then takes a briefer look at the region of Almopia in Greek Macedonia, presents the results of several research studies conducted in rural Greece after 1945, and then analyses a wealth of historical, political and anthropological literature dealing with multiple civil wars. Each time, his theoretical model stands up to the test of facts.
6Then, Stathis Kalyvas focuses on the motivations that drive non-combatant individuals to denounce opponents (when the military situation is conducive), and on how their personal interests interact with the more political interests of the armed actors. A recurring theme in the literature on civil wars is the intimate nature of the violence perpetrated against neighbours, friends and even members of one’s own family. Here again, Kalyvas opposes the Hobbesian and Schmittian readings, favouring an approach in terms of rational interests, i.e., the resolution of interpersonal conflicts that predate war. It turns out that the intimate nature of violence can be explained paradoxically by the density of social relations in rural areas: as Stathis Kalyvas writes, intimate violence is the ‘dark side of social capital’. This first paradox is immediately compounded by two others. On the one hand, non-combatants will try to manipulate armed actors for their own private interests: during civil wars, we therefore witness a ‘privatisation of politics’ that is just as marked as the opposite process of the ‘politicisation of private life’ sought by armed actors. On the other hand – and this is the supreme paradox – the practice of denunciation is not explained by any particular propensity to violence by the non-combatants, but on the contrary, by their reluctance to exert direct violence against their personal enemies, a task which they are therefore entrusting to the armed actors.
7According to Kalyvas, this is the source of the ‘alliance’ and ‘joint violence’ that bind armed actors to non-combatants: the former provide the ‘muscle’ that the latter need to settle their interpersonal conflicts, but in return, they demand the denunciation of their opponents. As a result, non-combatants tend to inscribe their interpersonal conflicts in the political categories put forward by the armed actors and give the misleading impression of a politicisation of individuals who in fact remain focused inward on their local world and their private interests. This model therefore makes it possible to link the centre and the periphery, national and local dynamics, the agency of armed actors and the agency of non-combatant individuals. It also makes it possible to avoid interpreting civil wars in terms of collective actors based on a ‘complete interchangeability of individuals, either as participants and perpetrators or as targets’,3 and favours a reading of selective violence in the context of civil war as a complex interaction between rational actors.
8Stathis Kalyvas’s approach to selective violence in civil war has many strengths: it draws on numerous primary and secondary sources, the demonstration is clear and almost irresistible, violence is no longer synonymous with chaos or hatred, individuals’ agency is substituted for collective entities that are too often essentialised, and civil wars are placed in the context of the rural world that they primarily affect. Yet on all these points, a number of questions arise.
- 4 These are the following monographs: Aschenbrenner 1987, p. 105‑125; Damianakos 1996; Du Boulay 1974 (...)
9Such is the case with Kalyvas’s use of sources. First of all, he resorts to a particularly rich secondary literature, and finds facts and analyses that support his theory of selective violence. Yet this raises a question: does this secondary literature have nothing else to say about violence and civil wars? The example of the English- and French-language monographs on various Greek villages cited by Stathis Kalyvas suggests that this is not the case.4 Admittedly, these monographs speak of private settling of scores, betrayals and denunciations, a right‑left divide that overlaps with conflicts between families or between village factions – all phenomena that confirm Kalyvas’s theory. However, they also emphasise family structures, local political and religious leaders, clientelist relations with state representatives, temporary or permanent migration and opening up to the wider world – all aspects of village society that are largely absent from Kalyvas’s analyses, as we shall see later.
- 5 Handman 1983, p. 40.
- 6 Aschenbrenner 1987, p. 106.
10Stathis Kalyvas relies on local court archives as another source, but he does not tell us much about how he used them, or about possible biases in trials held at a time when (right-wing) incumbents were waging a merciless fight against (left-wing) insurgents. Lastly, he also makes use of interviews he conducted in the late 1990s in the 61 villages of the Argolida; this is perhaps his most extensive source. If we divide the time he spent in the field (ten months) by the number of villages studied, we come to the conclusion that Stathis Kalyvas spent an average of five days in each of them. Similarly, if we divide the number of interviewees (116 in the Argolida) by the number of villages surveyed, we get an average of two interviewees per village. In both cases, this may be sufficient for political science research, but it is relatively sparse for an oral microhistory or anthropological approach. Indeed, among the authors of the village monographs mentioned above, Marie-Elisabeth Handman writes: ‘It took me more than two visits to the village for the dramas of the interwar period, like those of the Occupation and the Civil War, to begin to emerge from the depths of consciousness where they were kept shut away’,5 and Stanley Aschenbrenner speaks of 70 months of fieldwork between 1969 and 1983 to unravel the mysteries of the war in the village of Karpofora.6 Admittedly, Kalyvas’s interviews were conducted at a particularly favourable time, when some of the eyewitnesses to the events of the 1940s were still alive, and when more than two decades of democracy and peaceful changes of majority in the Parliament had eased tensions and fears. But at no point does Kalyvas seem to question the way in which memories are transformed over time, and in particular how the civil war (1945-1949) impacted the memory of the Second World War. One can therefore wonder how he manages to reconstruct an exhaustive list of the executions committed in the 61 villages between September 1943 and October 1944, and to draw a clear line between indiscriminate and selective violence.
- 7 Kalyvas 2006, p. 420‑421.
11In fact, the issue does not lie with the data collected; after all, no one can fault Stathis Kalyvas for having had the good fortune or the talent to unearth very well-preserved court archives and particularly talkative interviewees. In these sources, however, the German occupation and the violence perpetrated by German troops are strikingly absent. Besides, the exceptional quality of the data collected in the Argolida makes it difficult to reproduce this experience elsewhere. Above all, however, the issue lies in the interpretation: Kalyvas categorises all the executions without exception as either indiscriminate or selective violence, even as he emphasises the messy and shifting nature of violence in civil wars. In some cases, he expresses doubts about the nature of the violence, but eventually classifies it as indiscriminate or selective.7 Yet don’t some cases defy classification? Is there no violence that, while planned as selective violence, suddenly becomes indiscriminate? When an armed actor massacres an entire village, is this indiscriminate violence, or is it selective violence of a different kind, insofar as each village is perceived as an indivisible whole? Does this violence stem from a lack of information, or from an economy of violence that requires local notables to control the behaviour of their community? And shouldn’t selective violence and indiscriminate violence have been integrated into a single theoretical model?
- 8 See Taylor 2006; Douglas, Ney 1998.
12Of course, Stathis Kalyvas answers that the building of a theory necessarily implies a simplification of reality; this is undeniable. But the individual at the heart of Kalyvas’s model is a being of purely instrumental rationality, preoccupied only by his or her own survival (and not, for example, that of his family), free of any social commitments, devoid of emotion and moral standards, as is generally the case in the theory of rational choice.8 As such, this individual is characterised by an even more radical interchangeability than the individual in analyses of violence in terms of collective actors. Furthermore, the individual described by Kalyvas has no spatial mobility: he cannot, for example, have been driven out of zone 1 or 2 and cut off from his immediate environment, or denounce his neighbour in zone 3 before taking refuge in zone 4 or 5, even though civil wars generally create very high levels of mobility, individual or collective, voluntary or forced (provided this last distinction is always clear‑cut).
13Consequently, situations where the level of violence does not correspond to the predictions of the theoretical model are not, as Stathis Kalyvas writes, ‘the best proof of the value of a well-specified theory’.9 He does not tell us how many villages diverge from this model overall, but the three particular situations he analyses – namely, zone 3-time 1, zone 1-time 3 and zone 2-time 2 – are presumably those with the highest proportion of divergent cases (9 villages out of 21 in the first case, 6 villages out of 30 in the second, and 12 out of 21 in the third). Kalyvas therefore focuses on these villages to work out a qualitative study that is more similar to microhistory and anthropology. Yet these detailed descriptions are full of family solidarity, which generally resists violence and is even bolstered by the desire for revenge. More incidentally, they also mention various local notables (such as village mayors, military commanders, a priest, a doctor, a teacher) and individuals who had moved to the city before the war or fled their village because of it. Lastly, the divergent situations are partly explained by various individual or collective emotions (fear, anger, etc.) and moral standards (solidarity, reciprocity, etc.), tending to create ‘virtuous cycles of non-violence’.
14Although his theory of selective violence is based on rational and instrumental action, Stathis Kalyvas acknowledges that ‘two types of non-instrumental mechanisms that are explicitly excluded from the theory may play a residual but nevertheless important role, namely norms and emotions’.10 This ‘residual but nevertheless important’ characteristic remains open to discussion, but that’s not the main point. Kalyvas acknowledges the role of emotions and norms in the behaviour of village inhabitants who diverge from his theoretical model. This immediately sparks a question: What about those who live in villages that conform to this model? Are they devoid of all emotion and all norms, which seems unlikely, or do they have other emotions and other norms that explain their inability to escape the logic of selective violence? In this case, Kalyvas’s qualitative analysis is based on a conception of the individual that is radically different from that used in his quantitative study, and the use of rational choice theory does more than simplify reality, it distorts it.
- 11 Kalyvas 2006, p. 14.
- 12 Kalyvas 2006, p. 85.
- 13 Kalyvas 2006, p. 114.
15In his local qualitative studies and in his use of secondary literature, Stathis Kalyvas is much more cautious and nuanced than in the development of his theoretical model. He writes: ‘it is extremely difficult to uncover with an acceptable level of accuracy the individual motives behind violent acts’;11 ‘violence is a complex phenomenon, and it clearly encompasses multiple processes and mechanisms’;12 and ‘loyalty can be acquired in a variety of ways, including the provision of social benefits and ideological appeals’.13 Yet his description of individual, local and national dynamics in a rural context remains influenced by his theoretical model of selective violence and by the biases that inevitably follow from rational choice theory.
- 14 Kalyvas 2006, p. 364.
- 15 Kalyvas 2006, p. 365.
- 16 Kalyvas 2006, p. 44.
16A first persistent problem in Stathis Kalyvas’s analysis is the fact that the individual and local dimensions are not always clearly distinguished. He states that ‘individuals and local communities tend to take advantage of the prevailing situation to settle private and local conflicts’14 and refers to ‘the local or individual quest for personal and local advantage’.15 This blurred perception is significant because it makes it possible to exclude inequalities of status within village communities from the field of vision, to erase the social, cultural and political mediations specific to the local scale, and to continue to reason on the basis of a rational individual isolated from his surroundings. Admittedly, Kalyvas writes elsewhere that ‘the local is not the provincial or the parochial but rather the social’,16 but the main quality he attributes to the local is that it is a non-political universe. Here again, this choice makes it possible to ignore local notables, their ability or inability to regulate the village community, and their clientelist or partisan ties with national players. This view of village society as radically and eternally apolitical is the sine qua non of the idea that non-combatants resort to denunciation only in order to settle interpersonal conflicts, and can therefore change their behaviour in the context of war, but never their motivation.
17Inevitably, a second issue arises with the way in which Stathis Kalyvas interconnects the local and national levels, partly under the influence of James Scott’s work. He describes an ‘alliance’ between apolitical non-combatants and politicised armed actors, but this is a pact between two radically different kinds of actors that never intermingle. Hence it is hard to grasp what Kalyvas is referring to when he speaks, belatedly and almost incidentally, of ‘mobilisation’. Indeed, any mobilisation, whether political or military (and more often than not both), blurs the boundaries between non-combatants and armed actors, private motivations and political motivations. What better way to exact revenge than to join the ranks of the Greek People’s Liberation Army (ELAS) or the pro-German Security Battalions? What better way to climb the social ladder than to join the collaborationist Greek administration or the political apparatus of the National Liberation Front (EAM)? And what better place for politicisation than the military?
- 17 H.R. Kedward, In Search of the Maquis. Rural Resistance in Southern France, 1942-1944, Oxford: Oxfo (...)
- 18 Kalyvas 2006, p. 346.
18This is the whole point of H.R. Kedward’s phrase, quoted by Stathis Kalyvas, that ‘far more maquisards became communist through maquis experience than were communists by motivation at the outset’.17 So it is a pity that Kalyvas only mentions in passing the fact that ‘civil war empowers young people’18 and does not use his exhaustive data on the Argolida to study the distribution by age, sex or occupation of perpetrators and victims. Here again, however, he cannot take into account the social identities of his actors, their transition from non-combatant to combatant status, or the evolution of their motivations over time without jeopardising his entire theory of selective violence. Kalyvas therefore remains stuck in his representation of the peasant as an immobile individual, in physical, psychological and historical terms.
19Scholarship by political scientist Nikos Marantzidis and anthropologist Riki van Boeschoten on ‘little Moscows’, i.e., the villages characterised by strong support for the Communist Party, contrasts with this static representation of the rural world. In his book Little Moscows: A Political and Electoral Analysis of the Communist Presence in the Greek Rural Space,19 Nikos Marantzidis emphasises certain social factors that bolstered support for the Communist Party, such as the presence of peasant-workers in contact with the urban world, the return of soldiers who had taken part in the First World War or the Asia Minor campaign, and the moral authority of schoolteachers. Village society is thus no longer presented as a homogenous, inward-looking universe, with the emphasis instead on its internal diversity and its contacts with the outside world. Above all, Nikos Marantzidis emphasises the fact that, during the Second World War, the presence of the EAM led to a gradual politicisation of the peasants, through new forms of political participation and the establishment of a parallel state fulfilling various judicial or redistributive functions. This politicisation led to the emergence of a rural communism with strong moral and even religious overtones, the effects of which were felt well after 1945.
- 20 van Boeschoten 1991, p. 140.
- 21 van Boeschoten 1991, p. 157.
- 22 van Boeschoten 1991, p. 171.
- 23 van Boeschoten 1991, p. 190.
20However, the work most directly opposed to Stathis Kalyvas’s theory is by anthropologist Riki van Boeschoten. In her doctoral thesis, she shows that the EAM’s political action had helped ‘to neutralise the disruptive forces of a society regulated by family loyalty alone’20 and had created in the villages a ‘climate of confidence which had never existed before’.21 The new forms of social regulation and political participation introduced by the EAM had also transformed the perception of political power ‘from the small-scale, interpersonal level to a wider, collective and more abstract level’.22 In addition to family and village community, new ‘reference groups’ were added, such as political parties, social classes and – last but not least – the Greek nation.23 This was a process of politicisation that at least partly transcended the local logics and interpersonal conflicts that Stathis Kalyvas considered immutable and which he made one of the central components of his theory of selective violence.
- 24 van Boeschoten 1993, p. 612‑639.
- 25 van Boeschoten 1993, p. 615.
- 26 van Boeschoten 1993.
21Subsequently, Riki van Boeschoten focused her research on the village of Ziakas in Western Macedonia, a ‘little Moscow’ in which she conducted 80 interviews between 1987 and 1996. In a paper entitled ‘The Peasant and the Party’,24 inspired by the works of E.P. Thompson and Eric Hobsbawn, she argues against the ‘sack of potatoes’ theory, which, based on some of Marx’s writings, sees peasant society as a collection of small landowners isolated from one another. To make her point, she first reconstructs the socioeconomic relationships that characterised the village of Ziakas in Ottoman times and the interwar period. She then focuses on the mediators between the village and the outside world in the form of schoolteachers, conscripts and migrants returning from the city or from America. Finally, she shows that the politicisation of villagers began before the Second World War, as demonstrated by the strength of the Agrarian Party between the wars. With regard to the period of the Second World War, she describes how the EAM challenged the power of certain local potentates, the tsorbatzides, and integrated the peasants into new local institutions. The result was a ‘folk communism’ centred on the notions of self-government and people’s justice, and building on earlier practices of village cooperation and peaceful resolution of interpersonal conflicts. Riki van Boeschoten takes a long-term view of the history of the village of Ziakas, emphasising the changes in its social stratification and internal solidarities, and offers a very different view of peasant societies in war from that of Stathis Kalyvas. As she writes, ‘the backbone of the partisan army were peasants and many of the partisan leaders were of peasant origin’25 and, as a result, ‘obviously something had happened to the sack of potatoes’.26
- 27 van Boeschoten 2005.
- 28 van Boeschoten 2005, p. 53.
- 29 Michael Taylor, Rationality and the Ideology of Disconnection, op. cit.
22In another paper entitled ‘“Little Moscow” and the Greek Civil War’,27 Riki van Boeschoten seems to be closer to Stathis Kalyvas’ description of the rural world in the context of civil war. Based on testimony from the inhabitants of Ziakas, she describes the pre-eminence of family ties over political allegiances and the violence ‘between individuals who knew each other well and were linked by bonds of proximity’.28 But rather than focusing on these cases of interpersonal violence, Riki van Boeschoten demonstrates the existence of a village community with its own strict rules of mutual aid and reciprocity. In this context, war does not lead to a proliferation of denunciations and acts of vengeance, but on the contrary to the neutralisation of many interpersonal conflicts and the protection of certain local political opponents. Thus, Riki van Boeschoten does not share the idea that the intimate violence of civil war is the ‘dark side of social capital’. On the other hand, her description of the avoidance of violence in Ziakas is reminiscent of the ‘virtuous cycles of non-violence’ that Stathis Kalyvas observes in villages that diverge from his theoretical model. The village of Ziakas could therefore be included in his theory of selective violence. However, as Michael Taylor points out, various social psychology experiments have shown that altruistic behaviour is more common than predicted by rational choice theory.29 So it may well be that the way the people of Ziakas behave is the rule rather than the exception. In any case, Kalyvas’s depiction of the rural world, which is necessary to justify his use of rational choice theory, turns out to be incomplete and partly erroneous. Indeed, in contrast to his conception of an immobile peasant, various works on rural Greece present the image of a peasant who, in and through war, became a historical subject in his own right.
- 30 See Bougarel 1996; Bougarel 2020; Bougarel 2023.
- 31 Wolf 1969, p. 301.
23We must therefore take another look at the immobile or mobile nature of the peasant in the context of civil war. This, at least, is the conclusion I have reached following my own research into the conflicts of the 1990s and the Second World War in the Yugoslav space.30 In both cases, the various protagonists encountered obvious difficulties in moving rural fighters attached to their villages and their land. Some protagonists were successful, others were not, and their ability to mobilise the peasantry (in the spatial sense of the term) determined, at least partly, whether they would be victorious. The greater spatial mobility of certain armies can be explained not only by objective conditions such as a high number of refugees, but also by their ability to mobilise the peasantry in the political sense of the term, in the name of the fraternity of the Yugoslav peoples in the Partisan movement during the Second World War, or in the name of solidarity with the Ummah (the community of Muslim believers) in the Bosnian army during the 1990s. In both cases, war coincided with a spatial and political mobilisation of the peasantry and, as anthropologist Eric Wolf writes in another work that has become a classic, once the war is over, ‘society will have changed, and the peasantry with it’.31 The fact that the peasantry always loses out or is deprived of the rewards of its victory is just as important – but that is a matter for another debate.