Making sense of terrorism: a narrative approach to the study of violent events

ABSTRACT How does violence become understood as terrorism? In this article, we show how a narrative approach to the study of violent events offers a conceptually productive way to understand the process of “seeing” an event as a terrorist act, one that explicitly integrates the phenomenology of violence. While the collective practice of defining terrorism in academia and the policy arena has struggled to produce a universal definition, we identify a set of “common sense” characteristics. We argue that if the framing of violent events prominently features these characteristics as discursive anchors, this primes processes of sensemaking toward interpreting violence as terrorism. While terrorism markers are often articulated as being pragmatic and apolitical indicators of terrorist acts, we show that they are indeed at the core of political contests over historical and physical facts about violent events. The narrative approach we develop in this article underscores that intuitive leanings toward interpreting violence as terrorism are a sign of political agency precisely because they are produced through the stories political agents tell.


Introduction
Terrorism has been the most prominent security issue since the start of the new millennium. 1 Since 2014, series of deadly attacks in France, Belgium, Germany, the United States, and Great Britain have fueled new anxieties about terrorist incidents, particularly in Western countries. That the terrorism label has remained ambiguous and contentious, with hundreds of definitions on offer in the ivory tower and the policy arena, has not limited its potency to attract public attention and shape state policy (see e.g. Barrinha 2011). While we can perhaps identify cases of unprincipled, illicit "bad" rhetorical deployments (Finlay 2009, 751-3), there is no universally agreed upon apolitical "right" usage of the terrorism label. Rather, there is an implicit belief that what Supreme Court justice Potter Stewart heard in 1964 from his clerk about pornography also holds for terrorism -"I know it when I see it" (US Supreme Court 1964, 197).
But what precisely enables us to see terrorism? The narrative approach we develop here illustrates that interpretations of violence as terrorism, while linked to the phenomenology of the event, are the result of a process of meaning-making in which language plays a crucial role. Whether violence becomes widely understood as a terrorist attack depends to a significant degree on the rhetorical choices made by political agents. While it is now well-established in the existing literature that the stories political agents craft to represent what is happening in a given situation, and why, are politically consequential (Barnett 1999, 8-9, 15; see also Kettel 2013;Homolar 2011;Nabers 2009;Ben-Porath 2007), we argue that concentrating on the side of the speaker does not provide sufficient insight into how and why stories work. The article therefore moves beyond works that fall within the broad remit of securitization theory that focus on the centrality of speech acts in the construction of security threats by political leaders (Buzan, Waever, and De Wilde 1998;McDonald 2008;Hansen 2012), as well as research on the instrumentalisation and projection of strategic narratives, primarily concerned with how these forge shared meanings among policymakers and world leaders (cf. Miskimmon, O'Loughlin, and Roselle 2018, viii). While being in direct conversation with such scholarship, in contrast to these existing works in the study of international politics we put our emphasis on better understanding how narratives serve to structure information and experiences as a core element of the cognitive processes of sensemaking (e.g . Bruner 1991;Weick 1995;Weick, Sutcliffe, and Obstfeld 1999;cf. Brown, Stacey, andNandhakumar 2008, 1037). As we explain further below, this approach to studying violent events opens up a route to (re)politicizing what is often perceived as a "common sense" understanding of terrorism.
The article proceeds as follows. The first section introduces a narrative approach to the study of violence, and demonstrates the importance of identifying and understanding the discursive anchors in sensemaking processes that encourage us to cognitively attach a terrorism-centric meaning to events. Section two maps the core markers of terrorist attacks that are found across the study of terrorism despite the absence of a universally agreed-upon definition. Because of their high degree of visibility and repetition in everyday debates over terrorist violence, we conceptualize these markers as discursive anchors that foster the normalization of an event as an act of terrorism and the crowding out of alternative interpretations. The final section critically examines three sets of violent events that share common characteristics in terms of time, space, perpetrators, and victims, and demonstrates that phenomenologically similar incidents can be narrated very differently. We argue that it is only when terrorism markers are explicitly acknowledged and employed in the rhetorical claims made by political agents as part of a terrorism narrative that an event becomes broadly understood as an act of terror. In short, the labeling of violence as terrorism is a political act.

A narrative approach to terrorism
In the study of international politics, an analytical focus on the intersection between narratives and security dynamics is generally associated with discursive approaches which emphasize the productive power of language. Over the past few years, such scholarship has seen the onset of a "narrative turn", a gradual shift away from a broad understanding of discourse toward the stories that political agents tell to advance (or constrain) policy agendas (see e.g. Galai 2017;Subotić 2016;Homolar and Scholz 2019). Recent work has explored, for example, how "strategic narratives" provide a tool for political agents to shape behavior in the international arena and the "discursive environment in which they operate" Roselle 2013, 2, 2018;Krebs 2015;cf. Shepherd 2015). These works share the assumption that narratives have a deeper function in society and politics than simply being one of many communicative ways for "someone telling someone else that something happened" (Herrnstein Smith 1981, 228).
The approach to the study of violent events that we introduce here speaks to such scholarship. Yet, rather than investigating the motivations and goals of political agents as producers of narrative meaning, the article zooms in on how, when, and why the stories they tell about violent events can foster the interpretation of an incident as a terrorist attack (or not). This focus falls within the broad remit of the critical branches of terrorism scholarship that treat terrorist acts as a socially constructed phenomena in which discursive conditions and processes play a key role and which therefore reject approaching terrorism as both an extra-discursive, objective entity and a source of insecurity with clearly identifiable causes and solutions (Hulsse and Spencer 2008, 575;Jarvis 2009, 14, 18;Dixit and Stump 2011, 502, 509;Patrick 2014). While viewing terrorism as a narrative phenomenon is not new (see for example, Cobb 2013; Pemberton and Aarten 2018;Graef, Da Silva, and Lemay-Hebert 2018), to better understand how rhetorical choices shape political behavior at the intersection between society and the individual, as well as between agency and structure, we put an explicit emphasis on cognitive processes of sensemaking.
Narratives are discursive systems of meaning-making that are linked to the material world surrounding us. They are an omnipresent feature of social life, as Barthes (1975, 237) suggested more than four decades ago. People rely upon narratives to make sense of the world, to reduce its complexity, and to comprehend new information. Narratives render matters "real" and give meaning to what would otherwise be incomprehensible (Selbin 2010;Mayer 2014, 66, 71). In short, they play a central role in constructing and symbolizing experienced reality at the level of both sensegiving and sensemaking. 2 As vehicles for structuring information, they are a "mode of verbal representation" (White 1987, 26). While they are susceptible to change through both negotiation and retrospective recasting (Tilly 2002, 9), narratives enable us to both capture the many complex relationships and events that are integral to our everyday lives and to make sense of seemingly unconnected phenomena (Fludernik 2009, 1;Miskimmon, O'Loughlin, and Roselle 2013, 5).
Applying the logic of narrative meaning-making to episodes of violence suggests that without narratives, such events remain unintelligible. Whether and how specific incidents are constructed as acts of terrorism depends on the meaning discursively assigned to what is happening, to whom, and why, rather than being simply determined by the raw material of the event. Terrorism, then, is not a quality that merely resides within a violent event, and there is no metaphysical essence that makes its appearance in a terrorist act. Rather, it is both an interpretation that we subjectively project onto violent events through narratives and a process in which we mould our understanding of the occurrence to fit pre-existing discursive maps. Violent events acquire a terrorist meaning through creating discursive links between what would otherwise remain isolated and largely unintelligible episodes (Graef, Da Silva, and Lemay-Hebert 2018, 2).
But if there is nothing intrinsic about terrorism, what is it that enables us to "see" events as terrorist acts? We argue that a significant part of the answer is found in the way in which narratives foster making sense of unfamiliar material and events by presenting what is novel within a familiar code (Spradley 1980, 100;Altman 2008, 18;Scovel 1978, 129). Even if we do not accept any characterization of a terrorist act as either universal or dictated by the nature of the event per se, we suggest that the many debates within the academic and policy communities over what distinct features should serve to objectively categorize events as terrorism and a unique type of political violence (cf. Boyle 2012, 529) impact upon processes of narrative meaning-making because they are integral to this code. In particular, the comparatively salient elements of convergence within terrorism definitions discussed below function as discursive anchors that transmit "conceptual residue" and link past experiences of violence with a new situation (Weick 1995, 50;Fillmore 1982, 111;Gamson and Lasch 1983;Rein and Schön 1993, 151).
In contrast to existing research that underscores how terrorism definitions are subjective and prone to selection bias (Horgan and Boyle 2008), we therefore suggest that they deserve our attention not because they are right or wrong but because they provide the discursive setting in which the interpretation of an event as terrorist act takes place and biases our understanding of what happens, to whom, and why in the direction of salient "terrorist" reference points that we have previously been exposed to. If employed in the message text, they act as cognitive cues that prompt memories of terrorist violence, fostering unreflective knowledge toward interpreting a violent event as an act of terrorism.
A narrative approach to understanding violent incidents thus serves to reveal that people do notand, indeed, cannotmake judgments about the nature of an event in a discursive vacuum. Whether we are likely to "know terrorism when we see it" when we encounter stories about violent incidents depends on whether the narrative interpretation of this event contains discursive anchors that act as implicit cognitive benchmarks. Such discursive anchors prime us to routinely evaluate new information within familiar frames and to link together contemporary violent events with earlier episodes, even if this involves disregarding or downplaying the substantive differences between them.
In a narrative analysis of violence, therefore, the debates over how terrorism should be defined matter because they enduringly and regularly expose audiences to claims about constitutive characteristics. That they have failed to produce a universally accepted definition does not have to be resolved to understand their narrative grip (cf. Stampnitzky 2017Stampnitzky , 2013. Indeed, even if definitions remained marred by contractions such as suggesting "that states can or cannot commit terrorism, or that terrorism is only, or not only, violence against civilians" (Stampnitzky 2017; original emphasis), they nevertheless expose us to the very same discursive anchors. As Lakoff (2004) points out, even if a frame is negated, it is still activated; and it is also amplified through repetition in definitional debates. The reiteration and routinization of the core elements contained in the plethora of terrorism definitionsas well as the broader public resonanceserve as cognitive devices to anchor a new instance of violence as an act of terrorism and provide the symbolic connection to past terrorist attacks.

The anatomy of terrorism
Terrorism remains a fundamental taboo in world politics that is placed outside the boundary of legitimate political contestation and dissent (cf. Loadenthal 2013, 93). Indeed, there is a broad international normative consensus that any "criminal acts intended or calculated to provoke a state of terror in the general public, a group of persons or particular persons for political purposes" are unjustifiable (United Nations 1994). Yet apart from a negative definition of terrorism as violence that is not insurgency, guerrilla action, ethnic riots, or hate crimes, and that non-state actors should be considered as the primary culprits (Laqueur 1986), its exact meaning remains a matter of dispute in the international community. While Brannan, Esler, and Anders Strindberg (2001) pointedly observed a "perverse situation" where a phenomenonterrorismis extensively studied although its essence has not been pinned down, traditional terrorism scholarship adds some depth to this minimalist definitional compromise, providing us with the narrative building blocks to give sense to violent events as terrorist incidents.
In 1988, Schmid and Jongman undertook a prominent survey of key terrorism scholars, which continues to serve as a guide to understanding what constitutes a terrorist act. Uncovering 109 different definitions of terrorism, they were also able to identify a number of recurring elements. Specifically, Schmid and Jongman 1988, 5) located references to violence and/or force in over eighty percent of all definitions, while the idea that terrorism has a political motivation and is related to fear and terror came in second and third places, respectively. More recently, Schmid (2011, 86-87) offered an update on the core elements of a terrorist act, which again emphasizes lethal and lifethreatening physical violence as a key component but also places emphasis on a general lack of legal or moral restraints as important terrorism signpost. The political character of terrorist acts, either through the motivation of the perpetrators or the social repercussions of the event, features centrallyas does the element of intentionally creating fear beyond the immediate victims. This updated "consensus" contained within the surplus of definitions understands terrorist acts furthermore as "threat-based communication processes" that target "civilians, non-combatants or other innocent and defenceless persons", often as part of a "campaign of violence" rather than as merely an isolated incident, to achieve the desired political outcome (ibid.).
These broad scholarly parameters for classifying violence as terrorism are echoed in the way data about violent incidents is collected. Despite some criticism (e.g. Pape et al. 2014), the Global Terrorism Database (GTD), which provides information on "terrorist" violence from 1970 onward and portrays itself as the "most comprehensive unclassified database on terrorist events in the world" (GTD 2015), is a popular resource for terrorism scholars, policymakers, and the media. Overlapping with core criteria provided by the scholarly community, the GTD codebook defines a terrorist attack broadly as "the threatened or actual use of illegal force and violence by a nonstate actor to attain a political, economic, religious, or social goal through fear, coercion, or intimidation" (GTD 2014, 8). To be counted as a terrorist incident, a violent act must also fulfill two out of three criteria that are related to the goal, purpose, and legality of the violence employed: (1) the act "must be aimed at attaining a political, economic, religious, or social goal", with the economic aim being profound or seeking to achieve systematic economic change; (2) an "intention to coerce, intimidate, or convey some other message to a larger audience (or audiences) than the immediate victims" must be evident; and (3) the act of violence "must be outside the context of legitimate warfare activities. That is, the act must be outside the parameters permitted by international humanitarian law (particularly the prohibition against deliberately targeting civilians or non-combatants)" (ibid.).
Critical terrorism scholars have long taken issue with such a limited focus on definitions and criteria that are ignorant of global and local historical contexts (Marie 2007;Breen Smyth et al. 2008;Gunning 2007), and they also point to the prevalence of a selection bias with respect to the agents of political violence that discourages the study of state terrorism (Horgan and Boyle 2008, 52-53;Jackson 2009;Jarvis 2009). Indeed, to understand more holistically how terrorism emerges as a problem or threat in the political arena, scholarship within this "critical turn" in the study of terrorism has been committed to deconstructing mainstream understandings of terrorism and unveiling the significance of subjugated knowledge (Jarvis 2009;Jackson 2012). Yet while such works showcase alternative ontological and epistemological commitments, they nevertheless frequently engage in definitional games, including with their orthodox counterparts. Indeed, although they offer competing conceptions of what should be understood a terrorist event, the discussion still integrates the very same key markers. Notably, Jackson's (2011, 123;cf. Stump 2013) redefinition features a target audience beyond direct victims and political objective, whereas he critiques the inclusion of "civilian" and the "random" markers as part of the forest of misconceptions in scholarly definitions (ibid, 119-21). Finn and Momani's (2017, 838-4) "minimal foundationalist" conceptualization of terrorism, in turn, explicitly includes civilian victims as well as the "indiscriminate" and "random" markers in their critical discussion. Other examples of such reiteration practices within the critical branches of terrorism studies include Gunning's (2007) "political use of terror"; Burke's (2008) normative preference for a definition of terrorism as "political violence directed against civilians"; and Smyth's (2007, 262) emphasis that a critical approach sees "political terror [as] antithetical to human security and wellbeing". Works associated with the critical terrorism studies field thus echo the key markers contained within their orthodox counterpart's definitions, even if it is in disagreement. From the perspective of discursive anchoring, then, such critical scholarship contributes torather than untwinestheir cognitive association with the notion of terrorism.
The above discussion suggests that despite the ongoing definitional contests, existing scholarship has produced a number of terrorism descriptors that appear, phenomenologically, as "common sense" characteristics. The essential "material" ingredient for labelling an event as a terrorist act is physical violence described as arbitrary, indiscriminate, and extraordinaryviolence that is constructed as illegal and morally unjustifiable, and as targeting civilians and non-combatants for a (under-defined) political purpose. A terrorist attack is also associated with a high degree of premeditation, in terms of planning and coordination, and with the clandestine and serial character of violence that is aimed at addressing a secondary audience rather than the primary victims.
Yet the "things" used to identify a terrorist event do not merely act as "empty receptacles of discourses nor do they have 'essential' characteristics …but [rather] emerge in relation with material-discursive practices" Aradau (2010). They are neither immutable nor passive or fixed (Barad 2007, 151). And while neither discursive practices nor material phenomena should be understood as ontologically or epistemologically prior (Barad 2003, 822), as Hardy and Thomas 2015, 686) point out, "material objects and discourses are intertwined, with the former acquiring its identity through the discourses in which it is situated". In this intersubjective process of meaning-making, narratives function as the key device to hold meaning and matter together. Although the phenomenology of violence matters in identifying "senseless" violence as terrorism, its discursive construction thus plays the lead role in giving sense to it. In what follows, we illustrate that the material basis of violence does not automatically serve to cognitively anchor an event as a terrorist attack in processes of sensemakingeven if it significantly overlaps with the predominant terrorism markers contained within the surplus of definitions. Physical properties alone do not make us "see" particular forms of violence as terrorism. Rather, it is the invocation of a terrorism repertoire in stories about violent events that subconsciously pushes us in this direction, and a focus on narratives opens up a pathway to look beyond what "happens" toward the context of meaning-making that establishes an act of violence as terrorism.

The politics of labelling violent events
Political agents influence how stories about violent events are told. This includes whether terrorist markers are built in as discursive anchors to tilt the sensemaking scale toward terrorism interpretations of violence. While rhetorical choices may not necessarily be the result of deliberate planning or extensive strategic calculation, they nonetheless signal political agency. As Heath-Kelly (2018) underscores, political discourse makes a threat object hyper-significant while simultaneously remaining ambivalent and replaceable. The narrative meaning assigned to violent events that share the same physical properties that are commonly used to categorize violence as terrorism can thus differ substantially, influencing both the category and the intensity of the state's response.

Violence against civilians
Stories about terrorist events evoke images of massacre, chaos, destruction, agony, disorder, and carnage. They are filled with moral evaluations that make reference to the innocence of their victims and the evil, dark, barbaric, uncivilized behavior of the perpetrators. Uniting two core discursive markers of a terrorist attack, such terrorism narratives are usually spun around violence against civilians in which the latter are the main target rather than merely collateral damage. Like other fundamental "taboos" in world politics (Price 1995;Tannenwald 1999), the deliberate targeting of civilians is now a violation of norms of civilized international society and lies at the core of the international community's critique of terrorist events. Yet whether victims of a violent event fall into the category of civilians is not always clear-cut, and often becomes a matter of interpretation. This ambiguity extends to the reach and scope of international agreements that govern the use of force.
The international legal framework prohibiting violence against civilians has been a cornerstone of the inter-state system for nearly seventy years. Article 3 of the Fourth Geneva Convention on the Protection of Civilian Persons in Time Of War (ICRC 1949, 169-170) notes that its provisions apply to armed conflicts only if they occur in the territory of the contracting partners, even if they are not of an international character. The Convention defines civilians as those persons who are "taking no active part in the hostilities, including members of armed forces who have laid down their arms" (ibid.). Prohibited acts include violence to life and person, the taking of hostages, and violations of personal dignity. The Rome Statute of the International Criminal Court, which was adopted in 1998 and entered into force in 2002, reasserts that "intentionally directing attacks against the civilian population as such or against individual civilians not taking direct part in hostilities" constitutes a war crime in international armed conflicts (United Nations 1998, 5).
The distinction between civilians and combatants in the use of force is also an integral element of jus in bello and one of the basic rules of customary international humanitarian law. Yet there is no provision for the civilian category in intrastate conflicts that are not broadly recognized as armed conflicts or civil wars. The International Committee of the Red Cross (ICRC 2016, Rule 3, 5, and 6) underscores that customary international human rights law is unclear on "whether members of armed opposition groups are civilians"and as such subject to the loss of protection from attack only in cases of direct participationor whether they are generally "liable to attack" even if they do not partake in hostilities.
In cases of non-international conflict, then, who should be classified as civilian is not as clear-cut as contemporary political discourse on terrorism suggests. If there is no armed conflict, the combatant status vanishesbut so does the category of the civilian. In "peacetime", both the perpetrators and victims of the use of force can be construed as civilians. The violence-against-civilians discursive marker of terrorism thus creates two critical openings for narrative contestation over the meaning of an event: (1) how the status of the victims is represented; and (2) how the context of violence is defined.

The status of victims
Non-state perpetrators tend to depict their victims either as combatants or as members of the military apparatus and government officials "who are part of the presumed oppressive apparatus" and therefore constitute legitimate targets (Walzer 1977, 202). Governments and political leaders, in turn, emphasize the civilian status of victims in the same event to represent violence as outside the boundary of legitimate conduct. Consider the case of British Army soldier Fusilier Lee Rigby, who was hit by a car and then killed with knives and a meat cleaver by Michael Adebolajo and Michael Adebowale in Southeast London on 22 May 2013. Political debate initially centered on whether this violent incident constituted an act of terrorism. The primary opening for narrative contestation rested on the victim's status as a soldier of a state that was engaged in the global "War on Terror": was he therefore a "legitimate" target? One of the killers, Adebolajo, stressed that the victim was a combatant of a country "chasing us" and "killing us" (quoted in The Telegraph 2013). In the immediate aftermath of the attack on Rigby, he justified his actions by stating: "This British soldier is an eye for an eye, a tooth for a tooth" (ibid.). The most prevalent counter-narrative emphasized instead that at the time of the assault on Rigby's life, he had the status of a noncombatant as he was neither in service nor wearing a uniform (see Greenwald 2013). While the latter interpretation became dominant, the initial narrative contestation was revived when the Guardian journalist Seumas Milne argued that because Rigby had taken part in combat operations in Afghanistan, the attack on him "wasn't terrorism in the normal sense of an indiscriminate attack on civilians" (Milne 2013).
Whether political violence is labeled as terrorism or not can matter greatly for how such episodes subsequently feature in policymaking processes, perceptions of risk, or as justification for new state security initiatives. Following Rigby's murder, the Intelligence and Security Committee (ISC) of the British Parliament published a report on counter-terrorism shortfalls in November 2014 (ISC 2014). The ISC used the incident as a cornerstone for arguing that existing UK counter-terrorism programs were either "not working", or, in case of the PREVENT initiative to counter Islamist extremism, had not been given "sufficient priority" (ISC 2014, 2; on the UK's PREVENT strategy, see Heath-Kelly 2013). The British government declared Fusilier Rigby's killing "a sickening act of terrorism on our streets", and responded by announcing efforts to have communication service providers report "imminent threats" to law enforcement agencies, a more efficient handling of "intrusive surveillance", and the management of "low level subjects of interest" (Cabinet Office 2015, 3). Narrating Rigby's murder as an act of terrorism came with a price tag of an "extra £130 million available to strengthen our ability to combat terrorism" and found its way into the 2015 Counter-Terrorism and Security Act (ibid.).
While Rigby's murder was highly politicized with significant policy reverberations, an event in Wales that shared many of the violent features of the attack on the Fusilier saw little narrative contestation. In June 2015, Zack Davies attacked Dr. Sarandev Bhambra with a machete and a claw hammer and caused life-changing injuries. Davis stated that the aim of assaulting Bhambrawhom Judge Rhys Rowlands described as an "innocent man"was to avenge the death of Lee Rigby, and that he was inspired by both right-wing ideologies and the infamous Islamic State extremist "Jihadi John" (Morris 2015). Although the victim had no connection to Rigby's murderers and was not a soldier but a dentist, the incident was not interpreted as an act of terrorism, but instead classed by the police as "an attempted murder, racially motivated" (Harley 2015) and did not receive any specific policy response from the UK government. The victim's community criticized the hesitance of the Police, Crown Prosecution Service, Politicians, and the media to narrate his death within the terrorism framework, suggesting it had the characteristics of a terrorist attack, and demanding the allocation of resources to protect the Sikh community (Asian Image 2015; Sikh Channel 2015). The victim's brother put particular emphasis on the opposing racial disposition in the two related incidences; had it been reversed, he argued, the case would have been interpreted as "an act of terror" (BBC 2015).

The context of violence
The second critical opening for narrative contests over the meaning of an event relating to the violence-against-civilians marker is the context in which violence is understood to occur. Of primary importance is whether an act of violence committed by a non-state actor is considered to be part of a "proper" armed conflict. In cases of violence within state boundaries, provisions of international law apply only to conflicts that take place between governmental forces and non-governmental armed groups or between different armed groups that operate outside the state. Yet, the Additional Protocol II to the Geneva Conventions from 8 June 1977 implies that, to be defined as a "noninternational conflict" and thus to fall within the remits of international law, "dissident armed forces or other organized armed groups" involved must be "under responsible command, exercise such control over a part of its territory as to enable them to carry out sustained and concerted military operations" (UN 1979, 661, emphasis added). The Protocol asserts that its regulations do not apply to situations of "internal disturbances and tensions, such as riots, isolated and sporadic acts of violence and other acts of a similar nature, as not being armed conflicts" (ibid., emphasis added). The classification of intrastate violence as an armed conflict is thus dependent on the conditions of protracted hostilities and that non-state armed groups must be organized. International law, however, does not provide a precise "threshold at which violence amounts to an armed conflict' in the context of intrastate violence" (Sassoli 2006, 963). This includes situations that at the international level are widely perceived as civil wars where all warring parties fight in uniform with light infantry-type weaponry and adopt commando-style tactics. A current example of narrative contests over the classification of an intrastate conflict is the ongoing hostilities in Syria, for which president Bassar Al Assad has systematically rejected the label of "civil war" between two domestic contenders (Mroue 2012), and instead drawn a parallel to the terrorist attacks in Paris in November 2015 that took place in the absence of intra-state conflict (Bacchi 2015).
What happens within state boundaries in "peacetime" largely remains outside the scope of international legal frameworks, despite attempts to enshrine a responsibility to protect and the principle of conditional sovereignty within international law. This is partly owing to the tendency of states to want to limit the ability of the international community to intrude in their domestic affairs and avoid the international spotlight from violent intrastate conflicts, because it could tie their hands (Scharf 2001). If the situation in which an event takes place is interpreted predominantly as an armed conflict, this poses a significant challenge for political agents to frame a violent attack as an act of terrorism. There is an incentive for governments to avoid representations of violence as non-international armed conflicts, and instead to represent violent incidents as sporadic and isolated.
In the contest to own the narrative, violent opposition groups tend to both advance an interpretation of violence as part of wider struggle, and seek recognition as an army or political organization engaged in non-international armed conflict to emphasize the legitimacy, even legality, of their actions. "When you kill in war, it is an act that is allowed", noted Yigal Amir, who assassinated Israeli Prime Minister Yitzhak Rabin (quoted in CNN 1995). Dissident groups often adopt names to suggest their status as non-governmental forces engaged in "proper" armed conflict, ranging from the Revolutionary Armed Forces of Colombia (FARC) and the Irish Republican Army (IRA) to the German Red Army Faction (RAF), a tendency which increased significantly with the increasing stigma attached to the terrorist label after the Second World War (Rapoport 2004). Likewise, the Basque Fatherland and Liberty (ETA) separatist group carries out its attacks such as bombing shopping malls and politicians' cars as "commandos" and explicitly under the headline of "armed conflict" (Veres 2006).
Despite their efforts to represent themselves as a legitimate warring party, intra-state dissident groups are often labelled as terrorist organisations. In line with his conceptualization of the Syrian war as a terrorism problem, President Assad has put in place both a "Counterterrorism Law" and "Counterterrorism Court" to handle dissidence. This move has invited widespread accusations of Assad using legal platforms against human rights defenders and other peaceful activists in the name of counterterrorism (Al Arabiya 2012; HRW 2013). Likewise, the inclusion of dissident groups such as the ETA, IRA, and FARC in the US list of Foreign Terrorist Organizations has significant consequences for suspected members, including restrictions on their financial assets or isolating them internationally (US Department of State 2017). This is in stark contrast to the recent peace efforts between the Colombian government and the FARC. The initial 2012 peace agreement opened up a less confrontational route of engagement between the two parties by explicitly refraining from employing terrorism terminology and instead prioritizing terms such as "conflict", "bilateral hostilities", and "guerrilla organizations" (Gobierno de Colombia 2012). In 2016, the Colombian government continued to describe the FARC as a "guerrilla" and "armed movement" rather than labeling it a terrorist organization, and encouraged the group to "do politics without weapons" (Gobierno de Colombia 2016).

Arbitrary violence
Delineating terrorism through the character of an act of violence is not limited to the question of whether such violence claims innocent civilian lives. Analytically intertwined with this is the degree of target selectivity and forewarning that is involved. The more the use of force is narrated as arbitrary, random, and unexpected, the more it fosters the post-hoc socio-political construction of a violent event as an act of terrorism. One of the primary reasons for this is that striking targets in an indiscriminate fashion and without warning lies outside the international normative framework for the use of force. International law explicitly prohibits methods and means that are notor cannotdiscriminate between military targets and either civilians or civilian objects, and this holds for intrastate conflict (Crawford and Pert 2015, 198;United Nations 1998, 5).
The technology used to carry out violent attacks and the types of spaces targeted are two core features to discursively establish whether the use of force is intentionally indiscriminate, in particular in domestic contexts. In general, public venues like transport hubs or food markets are associated with the presence of civilians. Spaces that are presumed to shelter combatants or which, while being public, are occupied by soldiers at the moment of the attack, do not as readily enable their representation as sharing a comparable civilian character, thus making the construction of an episode of violence as a terrorist attack less straightforward. Yet while some advanced technologies commonly associated with acts of terrorism may appear more indiscriminate by design than others, such as explosives and weapons of mass destruction, rudimentary devices can also be used in an arbitrary fashion, as is the case in knife-wielding and vehicle-mediated attacks. Nevertheless, the use of explosives in a metropolitan train station crowded with travelers tends to be understood more readily as an indiscriminate act of terrorist violence, whereas the use of boxing gloves to physically hurt passengers waiting on a remote railway platform lends less discursive weight to the representation of violence as a terrorist plot.
These discursive markers have to be actively integrated into stories about the violent event in order to help fix its meaning as a terrorist attack. Take two incidents in China in 2013 as an example, neither of which was predominantly narrated as an act of terrorism despite sharing many of the characteristics often harnessed to instigate a terrorism narrative. In July 2013, the detonation of a homemade explosive device, without causing fatalities, at Beijing airport by a man who protested his ill-treatment by the police was not constructed as an act of terrorism by the Chinese government (Chen 2013;Zheng and Baijie 2013). Neither did the incident lead to specific policy responses despite the fact that it triggered a "massive public discussion" about violence resulting from social conflicts (Chang 2013). As in the case of the airport bombing, the official Chinese story did not designate as terrorism the explosion of nine bombs outside the headquarters of the Communist Party of China in Taiyuan in the Shanxi province in November 2013, which killed one person and left eight others injured. The perpetrator, an ex-convict and local resident Feng Zhijun, had instigated the violence after his petitions failed to stop a series of demolition orders for his home (Meng 2014). Again, rather than narrating the incident as an act of terrorism, the Chinese authorities played down the Taiyuan attack (Xinhua 2013). The muted government reactions to these "non-terrorism" violent incidents are in stark contrast to the strong official response to similar events in the Muslim province of Xinjiang, where the Chinese state narrates violence as terrorism, almost without exception (see PRC 2002).
With the technological and spatial characteristics of terrorism prone to conflicting interpretations in the stories constructed about violent events, a third discursive marker is often applied to underscore the arbitrary character of the violence: the element of surprise. What singles out a violent event as an act of terrorism is the launching of violence without confrontation or detection prior to the attack (Collins 2008, 400). Up to the moment of the attack, the "terrorist" is unnoticed; how and why the violent event takes place remains hidden until it erupts. The lack of anticipation is primarily owing to the clandestine makeup and modus operandi of the perpetrators, who are presumed to hold no territorial base, wear no uniform, and conceal their weapons, all of which lies outside the boundaries of international law (Merari 1993).
The key here is the unexpected and random disruption of everyday life, which challenges the international normative consensus on the use of force by violating conditions of preemption and self-defense. As Sassoli (2006, 9) puts it, "it is difficult to conceive how an act could be labeled as terrorist if … the author identifies himself as an attacker while he is visible to the combatants he attacks during and prior to the launching of the attack". Ethnic riots or violent clashes between street protesters and the authorities lend themselves less readily to being interpreted as terrorist incidents because they are usually preceded by a confrontational situation that escalates (e.g. Collins 2008). Within the context of such "highly patterned" events, violence is neither random nor unpredictable and surprising (Horowitz 2001, 1). In contrast, a high degree of surprise in discursive representations of a violent event, particularly in cases of narrative contestation, tends to tilt the balance in favor of terrorism interpretations.

Political violence
The construction of episodes of indiscriminate, unexpected violence against civilians as terrorist incidents usually depends on whether they are interpreted as inspired by an actively political agenda or reduced to a personal matter. The "political" terrorism marker draws attention to the agenda that underlies an act of violence in order to designate it as a terrorist event. Yet it is difficult to set boundaries between political motives and those which are not (see e.g. Cassese 2006). As Wieviorka 2007, 96) suggests, "the political dimensions of terrorism are permanently fuelled or invaded by other logics where meaning is lost or overloaded by new elements". The two deadliest shooting incidents in modern Californian history are a case in point.
On 18 July 1984, James Oliver Huberty committed a mass shooting in San Diego at the San Ysidro branch of the McDonald's fast food chain. This attack killed twenty-one people and injured nineteen others before a SWAT team member fatally shot the perpetrator. Justice Work described the incident, which took the victims completely by surprise, as a "maniacal, mass murderous assault" and an "hour of terror" (Court of Appeals of California 1987). The Justice pointed out that immediately after entering the restaurant, Huberty "began indiscriminately slaughtering patrons and employees within the glass-enclosed structure" (ibid.). While the precise motive for the attack remains unclear, the perpetrator had long been disgruntled with what he saw as the deterioration of America because of the US government's interference with business practices and a "cabal of international bankers" (Brown 2012, 194). Despite explicitly referring to characteristics such as terror, surprise, and arbitrariness as well as dissatisfaction with the existing socio-political order in the US, the perpetrator was not portrayed as a terrorist. Instead, the incident was predominantly narrated as a shooting rampage committed by a "violence-prone loner" who had been dismissed from his job a week before the shooting and who had problems with his wife (The Gazette 1984). Perhaps unsurprisingly, the San Ysidro shooting did not trigger any high-profile political response. It did, however, serve to rationalize the creation of a special unit in the San Diego police department that is aimed at developing more efficient weapons and tactics for the police officers to face future shooting rampages (Kavanagh 2009). This is in stark contrast to what happened two decades later.
California's second-deadliest mass shooting took place at the Inland Regional Center in San Bernadino on 2 December 2015 where the attackers killed fourteen people and seriously injured twenty-two others in less than four minutes. The perpetrators were the married couple Syed Rizwan Farook, an American-born US citizen, and Tashfeen Malik, a Pakistani-born lawful permanent resident. Both were fatally shot by police during an exchange of fire several hours after the couple had fled the scene in a rental car. A day after the attack, the Federal Bureau of Investigation opened a counter-terrorism investigation. While enquiries into their private communications revealed a "joint commitment to jihad and to martyrdom", the young couple did not have a history of promoting such ideas on social media platformsonly posting a hastily composed pledge of allegiance to the Islamic State shortly before they were killed (Serrano 2015). Despite the absence of a public dedication to a radical Islamic philosophy, narratives quickly zoomed in on the perpetrators' inspiration by foreign terrorist organizations and self-radicalization.
In the following weeks, the readily available descriptors for the attack itselfcivilian victims, indiscriminate violence, and clandestine perpetrators, as well as elements of surprise and premeditationplayed little role in characterizing the violent incident. Instead, the religion of the couple took center stage in fixing the meaning of the event as a terrorist attack. US President Barack Obama (2015) emphasized that the perpetrators "had gone down the dark path of radicalization". Although Farook, just like Huberty, had a troubled childhood that included his parents' divorce and potential workplace grievances could not be ruled out, the dominant narrative centered on his potential past plotting of violence (Finnegan 2016). Likewise, media reports focused on Malik's religious devotion and orthodoxy rather than on either her experiences of social alienation growing up in Saudi Arabia's guest worker community or retaining her outsider status later on in life as the "Saudi girl" (Bengali and Linthicum 2015).
Similar to the San Ysidro incident, the San Bernardino attack was followed by debate on the use of military equipment, particularly in those states where this had previously been limited by the US government (Kates 2015). However, the explicit construction of the San Bernardino shooting as an "act of terrorism" (Obama 2015) caused wider waves in US security policy. First, it served to rationalize and legitimize the administration's fight against the Islamic State in Syria and Iraq. For example, Obama rhetorically linked the violent event in California to the military coalition fighting IS, and promised further airstrikes against Daesh as a response (Harris and Shear 2015). Four months after the mass shooting, the Obama administration also approved an increase in the US Special Operations forces in Syria, citing the unity in "the defense of our way of life" in reference to the San Bernardino incident (Jaffe, Ryan, and DeYoung 2016). Second, the shooting prompted a war of words among Republican candidates during the 2015-16 US presidential election campaign on how to regulate Muslim immigration in order to prevent terrorist attacks (Oliphant and Whitesides 2015). Once in office, President Donald Trump, signed an executive order that sought to ban refugees and citizens from seven Muslimmajority countries from entry into the US. Although neither of the San Bernardino perpetrators would have been affected by the ban, Trump administration officials explicitly used the event to justify the controversial order (Shear and Cooper 2017). Finally, echoing security measures adopted in the UK in response to Rigby's murder, the San Bernardino violence triggered a debate in the US on the implications of encrypted communication technology on counter-terrorism strategies (Nakashima 2015). For example, in direct response to the incident, California Democratic Senator Dianne Feinstein, presented a bill requiring any company "engaged in providing an electronic communication service or a remote computing service to the public" to report any knowledge of terrorist activities to authorities (Cockerham 2015).
This exploration into the politics of violent events highlights that the use of cognitive cues that foster terrorist understandings of violent events is far from an objective, automatic, or impartial process. Political agents may opt to integrate elements that are considered to be symptomatic of terrorism to represent violent episodes as terrorist attackssuch as a wider "campaign of violence" as well as a high degree of planning and organization. Yet, depending on specific settings and policy agendas, they may instead choose to fall back on the very same elements to fix the meaning of a violent event to the contrary. As our examples showed, although the discursive construction of an event as a terrorist attack generally becomes more challenging when the violence targets victims other than those readily described as "civilians", the story may very well zoom in on different characteristics, such as the ethnicity and religion of the individuals involved. Despite the high level of audience recognition of the intentional targeting of civilians as a core marker of terrorist attacks, a violent event still has to be narrated and interpreted in this sense to become widely understood as an act of terrorism. Likewise, whether a violent act is interpreted as politically motivated is highly ambiguous. Indeed, it appears that when the aim of a violent event is perceived as external in terms of geography, identity or culture, then ideological and religious agendas are more readily assumed to constitute a political aim. There is thus an "othering" labeling bias in the narrative construction of political violence, in particular when the event makes it possible to establish associations between Islam, violence, and terrorism.

Conclusion
Processes of narration are necessarily selective; there is always more than one story to tell. This implies that political agents may offer competing interpretations of violence in their attempts to "own" the narrative. Stories always stand in relation to other storiesincluding their counter-narrativeswith political agents' struggling over having their frame accepted (Boin, Hart, and McConnell 2009, 82; see also Holland 2009). Terrorism is no exception, and in the "war of words" naming particular entities as terrorists has political, economic and military implications (Turk 2004, 271-272).
But how do some acts of violence become widely understood as terrorist events, while others do not? In this article, we have explored the narrative construction of terrorist events and the role that discursive anchoring plays in this political process. We argued that while the collective practice of defining terrorism has done little to conceptually pin down what terrorism is, the core characteristics the definitional exercise has produced over time serve as discursive anchors that cognitively bias us toward "seeing" an event as a terrorist act. When these markers are used so regularly to designate a terrorist attack that they appear as intuitive, they form an integral part in the social mechanism that primes processes of sensemaking in this direction. This matters because they also implicitly set forth what type of violence is morally acceptable and at what point the use of force enters the red zone of deplorable behavior, thereby setting the boundaries of political possibility.
"Common sense" terrorism markers are at the core of language games about the meaning of violent events precisely because they can be articulated as being pragmatic, rational, and apolitical indicators. While the phenomenology of violence plays an important role in terrorism stories, the meanings attached to events are subject to contestation and political agents may offer different interpretations in their attempts to "own" the narrative. The contest over historical and physical factsthe attempt by different parties to establish the "true" story that underlies and explains the eventreveals the importance of understanding terrorist events through a narrative analytical lens. As our analysis has shown, what are often perceived as "intuitive" characteristics of a violent event do not inevitably trigger an interpretation of violence as an act of terrorthis depends instead on the types of rhetorical claims made by political agents. How audiences comprehend violence, while linked to the materiality of the event, depends on how it is framed and whether the narrative contains discursive anchors that push our sense-making in this way. Notes 1. This work was supported by funding from the Economic and Social Research Council under grant number ES/K008684/1. We would like to thank the three anonymous reviewers as well as André Broome, Christopher Browning, and Matthias Kranke for their valuable suggestions. 2. Sensegiving can be defined as the framing efforts of political agents to steer meaning-making toward a preferred (re)definition of experienced reality (Gioia and Chittipeddi 1991, 442). Sensemaking, in turn, is understood as the cognitive processes of attributing meaning to experienced reality (Weick 1995); it is, as Brown, Stacey, andNandhakumar (2008, 1038) put it, "a kind of creative authoring on the part of those who construct meaning".

Disclosure statement
No potential conflict of interest was reported by the authors.