Everything Is Just Dandy!

Is There a Language of Terrorists? A Comparative Manifesto Analysis

Phenomenology and Existentialism
Julia Ebner

Assessing the language and narratives in the written communications of terrorists, including manifestos associated with violent attacks, may help to establish a foundation for improving detection and prevention practices in the field of counter-terrorism. Over the past decades, security and intelligence services, as well as social media companies, have tried to gain a better understanding of the communication patterns demonstrated amongst violent extremist and terrorists.1 Despite the high level of interest and stakes involved with establishing linguistic cues that can serve as early detection mechanisms to help prevent violence, there continues to be only a limited amount of published research in this area. This systematic manifesto analysis seeks to help address this gap in the research literature. Although we do not claim that our framework will have the predictive power to reliably single out every would-be terrorist, it could provide a useful tool within linguistic diagnostics, complemented with other forms of intelligence to narrow down potential targets of investigations. Given that security forces have limited capacity to conduct investigations and there is often an excess of textual data available, producing effective linguistic analysis tools is crucial to helping counter terrorism units to act more efficiently. Most importantly, the tool we have developed is theory driven, based on extensive previous research on the psychological drivers of violent extremism.

The likelihood that someone puts their life on the line for a group has been linked to identity fusion – an extreme form of group alignment whereby an individual’s personal identity merges entirely with the group identity.2 Numerous previous studies have shown that high levels of identity fusion predict willingness to fight and die for the group when it is imperiled.3 Both non-state armed insurgents and conventional military groups in theaters of conflict in the Middle East have repeatedly demonstrated that fusion is a reliable predictor of actual violent self-sacrifice, as opposed to merely the endorsement of it.4 The same phenomenon has been observed with groups as diverse as Islamist fundamentalists in Indonesia5, farmers in Cameroon6, college fraternities in the U.S.7 and Football hooligans in Britain and Brazil.8 Despite the growing body of evidence linking fusion to violent self-sacrifice, its potential applications in counter-extremism efforts remain underexplored.

Our study is based on the fusion-plus-threat model9 which integrates previous empirical findings on group alignment and identity, psychological kinship, and parochial altruism to form a theory that makes specific predictions about the role of in-group identity fusion and out-group threat perception as drivers of violent extremism. According to this theory, extreme ideological commitment or ‘radicalization’, in the absence of fusion plus threat, would be much weaker predictors of violent extremism.10 Predictive power is expected to be increased further when measures of relevant moderators of the relationship between fusion-plus-threat and violent extremism are available, such as violence-condoning norms.11

The model implies that reliable means of measuring identity fusion, perceptions of existential threat, and violence-condoning norms could help identify individuals at risk of carrying out terrorist attacks. Previous studies have used a wide variety of psychometric measures to assess these variables, but all rely on the willing participation of study participants. It is obviously impractical in most situations to persuade members of illegal groups or extremist networks, including those that condone or utilize violence, to provide honest answers to questions using measures of this kind. Consequently, in this paper we explore whether linguistic markers found in actual terrorist manifestos can serve as reliable measures of fusion, threat, and willingness to condone the use of violence and if the predicted relationships are observed between these variables.

With the aim of identifying text-based predictors of terrorist actions, we conducted a qualitative ethnographic assessment of fifteen manifestos produced by individuals who engaged in violent self-sacrifice as well as ideologically extreme and moderate manifestos that were not followed by violent self-sacrifice by their authors. Our study included the manifestos of the terrorists behind the 2011 Norway attacks, the 2014 Isla Vista killings (U.S.), the 2015 Charleston shooting (U.S.), the 2019 Christchurch mosque attacks (N.Z.), the 2019 attacks in Halle (Germany), Poway (U.S.) and El Paso (U.S.) as well as publications by NSDAP leader Adolf Hitler and the prominent jihadist revolutionary Sayyid Qutb. We also included manifestos of ideologically extreme but nonviolent authors such as the writings of Norwegian far-right blogger Fjordman, the Islamist ideologue Yusuf al-Quaradawi and Marx and Engel’s Communist Party Manifesto, as well as ideologically moderate manifestos by civil rights movement leader Martin Luther King Jr, feminist thinker Simone de Beauvoir and climate activist Greta Thunberg.

Patterns in the narratives and language used in the manifestos were identified, distilled, and categorized by the lead author to allow for a comparative analysis. 100 sample phrases identified from the extracted manifesto content which had been identified as representative of relevant coding categories were subsequently collected. These phrases were then coded independently to investigate the coherence of the coding categories and the reliability of independent assessments. More specifically, an Intercoder Reliability (ICR) analysis was conducted – first by two subject area experts and second, by twenty-four non-expert coders.12

Preventing acts of terrorism requires a reliable method of identifying would-be terrorists before they carry out attacks. Even if extreme ideology is frequently associated with terrorist violence it is not a necessary condition and its causal role has not been well established. Moreover, in recent years, a number of researchers have warned that the role of ideology in radicalization has been exaggerated in counter-extremism approaches.13

As the effort to link terrorist and extremist profiles has been inconclusive14, researchers studying radicalization have increasingly turned their attention to potential psycho-social drivers. Wikotorowicz called the matching of individual factors and group factors ‘frame alignment’ – i.e. when an individual’s ‘cognitive opening’ coincides with the narrative offered by an extremist group.15 Sageman divided radicalization into a four-stages process: “a sense of moral outrage, interpreted in a specific way, which resonates with one’s personal experiences, and is channelled through group dynamics, both face-to-face and online”.16 Likewise, Ranstorp described radicalization as a “multifaceted combination of push-pull factors involving a combination of sociopsychological factors, political grievance, religious motivation and discourse, identity politics and triggering mechanisms”.17

We use the term “self-sacrificial violence” to describe violent pro-group behaviors that entail risk to life and limb for the perpetrators, regardless of whether they subscribe to ideological extreme beliefs or not. Self-sacrificial violence may therefore include acts of terrorism as a form of homicidal self-sacrifice and other expressions of violent outgroup hostility.18 Our study applies the term “ideological extremism” based on the Institute for Strategic Dialogue’s extremism definition:

Extremism is the advocacy of a system of belief that claims the superiority and dominance of one identity-based ‘in-group’ over all ‘out-groups.’ It advances a dehumanising ‘othering’ mind-set incompatible with pluralism and universal human rights.19

Using Language to Detect Would-Be Terrorists

Quantitative and systematic linguistic analyses of violent and nonviolent manifestos are scarce. There have been several publications that analyzed the themes, narratives and language of individual terrorist manifestos but these previous works did not perform a comparative analysis.20 The most extensive linguistic analysis of terrorist manifestos was conducted by researchers at Uppsala University in 2016.21 Their study focuses on “leakage warning behavior” (or simply “leakage”), which describes the intentional or unintentional signaling of planned violence in public or nonpublic communications. The purpose of the leakage that was observed in the study with the perpetrators ranges from attention-seeking, coping with anxiety related to the impending act, or the desire to memorialize one’s deed beyond one’s death.

The Uppsala group used the text analysis tool Linguistic Inquiry and Word Count (LIWC) to assess the patterns in written documents authored by ten different lone terror offenders prior to their engagement in violence. Words were coded in psychologically meaningful categories, yielding eight indicators of the drives and emotions that preceded violent attacks: the use of big words, i.e. words longer than six letters have been found to be indicative of psychological detachment (e.g. degeneration, resistance, overconsumption), the use of personal pronouns, in particular third personal plurals (e.g. they, their, them), expressions of emotions, including the ratio of positive emotions (e.g. happy, good, pretty) to negative emotions (e.g. hate, enemy, worthless) and anger words (e.g. hate, kill, annoyed), words related to friends (e.g. buddy, neighbor, brother), certainty (always, never) and power (e.g. superior, bully, weak). The identified communication features were also compared with those found in natural conversations.22

The results of this computerized textual analysis of manifestos offers an illustration of the use in exploring the language and narrative patterns in terrorist manifestos. One significant limitation of the Uppsala method, however, is that the approach was not well-grounded theoretically and did not test predictions based on previous evidence-based research. For instance, although the authors highlight the category “friends” as an important one that emerges from their analysis, and even quote “brother” and “buddy” as example terms, they do not reference any previous work on psychological kinship or fusion theory. Likewise, the study cites linguistic markers of cognitive processes – in particular “the simplification of complex matters” – without linking its findings to established measures of cognitive complexity such as Integrative Complexity (IC).23

Another limitation of this study stems from the timing of the publication. As the analysis was completed in 2015, it does not include the manifestos of more recent high-profile lone actors, such as those that were posted in the run-up to the attacks in Christchurch, Poway, Halle, and El Paso. The authors similarly acknowledge that future work looking at linguistic cues should seek to include a range of manifestos for analysis, as well as comparing differences individually with their categories as a potential foundation.

Another important and more recent contribution to the analysis of the content of terrorist manifestos was made by Jacob Ware of the International Center for Counter-Terrorism who carried out a qualitative analysis of the manifestos of extreme-right terrorists.24 According to Ware, recurring themes in the manifestos were an alleged “clash of civilisations” between races or religions, the importance of the preservation of European culture, the political climate, and the portrayal of terrorism as both self-defense and a last resort. While this analysis focused on thematic patterns, it did not assess technical linguistic characteristics.25

Thus, in contrast to these previous comparative manifesto-related studies which were not clearly theoretically grounded, the following analysis presented in this paper seeks to ground our qualitative and linguistic analytical approach using the recently developed fusion-plus-threat model.26 In so doing we seek to provide a potentially more generalizable approach to identifying linguistic markers in narratives that are capable of predicting violent as opposed to nonviolent outcomes. Our hope is that we can build on the valuable existing research that we outline above and trigger wider discussion concerning the content and linguistic cues in violent versus nonviolent manifestos, rooting our analysis in an evidence-based framework.

A plausible diagnostic marker for high levels of fusion in written or verbal form is the presence of language emphasizing relations of kinship or familial ties (e.g. siblingship) among group members. Research on identity fusion from the outset has emphasized the link with family ties and feelings of brotherhood and sisterhood in fused groups.27 Studies comparing pathways to fusion among large samples of identical and fraternal twins indicate that perceptions of shared biology as well as shared experience contribute to identity fusion.28 Studies among survivors of atrocities also show that feelings of psychological kinship mediate the relationship between fusion and self-sacrificial behaviors.29 The pervasive use of metaphors of kinship when talking about the in-group and its members can thus indicate heightened levels of identity fusion, and associated willingness to engage in extreme pro-group behavior, including violence and terrorism.30 There is therefore a compelling theoretical rationale for treating expressions of kinship relatedness among members of groups as indicators of high levels of fusion.

Fusion is thought to be more effective at motivating extreme pro-group behavior than group identification.31 A long tradition of research in the field of social identity examining how people bond with groups, indicates that group identification is a depersonalizing process.32 Thus, when a person highly identifies with a given group identity and that identity is made salient, their personal identity becomes less salient implying a hydraulic relationship between the personal and group identities.33 However, more recent research showed that identification is primarily based on sharing identity markers with other members of a group such as beliefs and practices, while identity fusion derives from sharing traits or experiences that define the personal self, such as life-defining experiences as in the case of persecuted or embattled groups.34

Consequently, a strong conflict between self-interest and group needs can generate a dilemma that may more readily be resolved in a self-preserving manner when identification rather than fusion is the dominant form of group alignment and depersonalization is not evident. Fused individuals, by contrast, due to the synergistic nature of the self and group bonds appear more willing to take extreme actions to protect the group as if it is equivalent to their personal self.35 While strong group identification often occurs in tandem with high levels of identity fusion, our theoretical framework based on previous research36 leads to that prediction that in the presence of threat fusion rather than identification is more often a potent driver of violent extremism.

It is theoretically and methodologically important to be able to distinguish between the two forms of group alignment when attempting to explore signals of willingness to conduct personally life-threatening acts of violent outgroup hostility, rather than merely endorsing such acts philosophically or issuing hollow threats. An important caveat is that, according to our framework, fusion alone is not sufficient to motivate outgroup violence. We also do not claim that the fusion-plus-threat model is the only way of explaining violent self-sacrifice, since it is possible that other motivators may be sufficient, ranging from violence endorsing group norms to psychosis and suicidalism37. Nevertheless, to the extent that the fusion-plus-threat model explains much of the variation in this domain, it is vital to develop a set of diagnostic tools for early prevention.

This manifesto analysis applies the predictions of fusion theory by exploring whether the language found in different types of manifestos could be used as diagnostic tool in prediction of violent extremism. In the process, we take care to disambiguate linguistic markers of fusion and identification, and to clarify which clustering of other relevant variables demonstrate mediating connections between group bonding indicators and violent outcomes, and which do not. Using the fusion-plus-threat framework for predicting violent self-sacrifice,38 the hypothesis is that a higher degree of identity fusion language, coupled with potential moderator variables (detailed in Figure 2), increases the likelihood that individuals will carry out violent pro-group action. Importantly, the fusion plus threat framework as the name indicates, argues that fusion alone is unlikely to predict violent self-sacrifice without threat or relevant group values. This means that manifestos that led to acts of self-sacrificial violence would be predicted to include more fusion language than those in the categories “ideologically extreme non-violent” and “moderate non-violent”.

To capture potential variables of interest, including proposed moderators of the pathway from fusion to violence, we coded for linguistic indicators of i) Group Alignment, ii) Out-Group Entitativity, iii) Out-Group Threat, iv) Violence Condoning Norms, and v) Threats of Violence. See for full descriptions and Figures 1 and 2 for visual representations of the proposed relationships between relevant variables. Of particular note due to being frequently cited factors that might contribute to an escalation toward violence include:

Is There a Language of Terrorists? A Comparative Manifesto Analysis

Published online:

09 August 2022

Figure 1. Proposed relationships between group alignment and behavioral outcome.

Figure 1. Proposed relationships between group alignment and behavioral outcome.

Is There a Language of Terrorists? A Comparative Manifesto Analysis

Published online:

09 August 2022

Figure 2. Proposed relationship between fusion and violence.

Figure 2. Proposed relationship between fusion and violence.

Is There a Language of Terrorists? A Comparative Manifesto Analysis

Published online:

09 August 2022

Table 2. Narrative classifications and definitions.

  1. Perceived out-group entitativity,39 which will be traced via the use of “us versus them” narratives40, in particular the use of language that insults, demonizes or dehumanizes an entire out-group,41

  2. perceived out-group threat,42 which may manifest itself in narratives of an existential threat posed to the in-group,43 the belief in a conspiracy of the outgroup,44 or the belief in an inevitable war between the in- and out-group,45 and

  3. violence condoning norms,46 which may include the justification of violence, the glorification of violence via martyrdom narratives or the so-called “warrior mentality”,47 the identification with a violent role model,48 and perceived hopelessness of alternative solutions.49

We treated calls to violence as distinct from violence condoning norms, since adhering to norms condoning violence does not necessarily mean that one considers violence the best method of securing the group’s present interests. Moreover, it is important to recognize that merely threatening violence is no guarantee that one will actually engage in violence. While many governments and intelligence agencies traditionally focus on detecting explicit threats of violence and expressed support for terrorism in their prevention and intervention efforts,50 we argue that this variable alone may not represent a reliable predictor of actual violence. In online communities many calls to violence remain no more than empty threats or efforts at sarcasm, while an increasing number of today’s violent extremists have become more cautious in their use of threatening language to circumvent detection mechanisms and removal policies.51 Our fusion-plus-threat based framework thus aims to serve as an alternative or complementary prediction model that can potentially identify higher risk combinations of language. Tracking the direct calls to violence in terrorist manifestos, however, remains important as it allows us to examine how this variable interacts with other factors and clarify when this variable correlates with actual violence.

Manifesto Selection and Coding

Fifteen manifestos were selected for the analysis on a scale from violent self-sacrificial to nonviolent to allow for comparative analysis across different types of writings. The selected ideologically extreme (violent and nonviolent) manifestos differ in ideological leanings and cover anti-Muslim, anti-Semitic, misogynist and jihadist documents while the moderate (nonviolent) manifestos cover prominent feminist, environmental and anti-racist texts. All assessed text-based documents are comparable in three aspects: their declaration of political or ideological views, the expressed urgency of their political or ideological message, and their targeting of public audiences.1 While all of the authors called for radical change, the tactics they pursued to reach them varied widely following the publication of their statements. We limited the total number of analyzed manifestos to fifteen due to the limited availability of terrorist manifestos, access restrictions to such documents, as well as researcher time considerations. A total of over 4000 pages were manually reviewed.

A range of terrorist manifestos were selected for inclusion to capture relevant high-profile violent attacks that occurred after the 2011 Breivik attack in Norway, where the perpetrators have left behind manifestos or extensive letters about their motivations. As Jacob Ware observed in a recent report, the use of manifestos by terrorists has increased in recent years.52 However, it is worth noting that manifestos are more commonly used by extreme-right terrorists than by violence perpetrators of other ideological leanings. Our observation was that while ISIS frequently claimed responsibility for attacks ex-post via their Amaq News Agency or its propagandists, terrorist manifestos similar to the ones examined in this study are predominately a far-right phenomenon.2

One explanation of the discrepancy in the use of manifestos by perpetrators of different ideological backgrounds may lie in the frequency of copy-cat terrorism, where a perpetrator helps to inspire the next through their manifestos and attacks, which is more prevalent among the extreme-right. When the Norwegian terrorist Anders Behring Breivik published his 1500-pages manifesto before launching his attack on 22 July 2011 he initiated a phenomenon of posting documentation prior to far-right attacks that cannot be observed with jihadist perpetrators (Hoffman 2020). A New York Times investigation found that the majority of white terrorists since 2011 referred to previous right-wing attackers and their manifestos in their written communications (Cai & Landon 2019). Indeed, during the drafting of this manuscript there was another mass shooting in Buffalo, New York, that resulted in 10 fatalities. The accused shooter, Payton S. Gendron, composed a 180-page manifesto which was reported to have plagiarized about 57 percent of the ideological sections from previous sources, especially the Christchurch shooter manifesto.53

While jihadist attackers have often been found to be inspired by political and theological manifestos written by Islamist ideologues, they do not tend to leave behind manifestos themselves (Ayad, Amarasingam & Alexander 2021). Violent self-sacrifice is not limited to terrorists however. Although not classifiable as terrorist manifestos, Hitler’s Mein Kampf and Sayyid Qutb’s Milestones are writings by authors who subsequently participated in the planning of acts of extreme violence.

All the manifestos used in this study were coded on a spectrum from violent to nonviolent texts which reflects the different pathways based on the fusion-plus-threat model, our analysis then distinguishes between three types of manifestos:

  • Violent self-sacrificial manifestos: texts written by authors who committed acts of self-sacrificial violence.

  • Ideologically extreme (nonviolent) manifestos: political texts written by authors who did not commit acts of self-sacrificial violence but adhered to extremist group ideologies.

  • Moderate (nonviolent) manifestos: political texts written by authors who did not commit acts of self-sacrificial violence and did not adhere to extremist group ideologies ().

Is There a Language of Terrorists? A Comparative Manifesto Analysis

Published online:

09 August 2022

Table 1. Manifesto coding.

Narrative Coding

The manifestos were analyzed with a view to detecting fusion, threat, as well as other variables of interest that might impact or complement the fusion-violence link (see Figure 1). Drawing on this theoretically-grounded assessment, we used the following five meta category narratives and thirteen sub-category narratives (see ) for our systematic coding of manifesto content.3

Based on the above identified meta-category narratives and sub-category narratives, each manifesto was coded to indicate whether the identified narrative categories were present or absent in each manifesto. We did not include specific counts of occurrences at this stage, as the quantitative assessment of all manifestos will be performed in a subsequent NLP analysis. To allow for an Intercoder Reliability Analysis in the next step, we collected at least one evidence sentence from each coded narrative occurrence in each manifesto.

Intercoder Reliability Analysis

To test the coherence of our coding framework and assess the reliability of all narrative categories, we conducted an Intercoder Reliability (ICR) analysis. Using our set of collected evidence sentences representative of each coded narrative (overall 265 evidence sentences), we extracted 100 randomly selected coded phrases from all fifteen manifestos. All narrative categories were represented in our set of coded sample phrases. We randomized the sequence in which the sentences appeared and removed any mentions of authors and pages.

During the first stage of the ICR analysis, the lead author as well as two independent expert coders received the 100 anonymized, randomized sample phrases, which they needed to code based on a coding manual that outlined the coding procedure and explained the pre-selected narrative classifications.

Coder 1 was Graham Macklin, Assistant Professor at the Center for Research on Extremism (C-REX) at the University of Oslo. Coder 2 was Jacob Davey, Research Manager specialized in far-right extremism at the Institute for Strategic Dialogue. The expert coders received the anonymized Excel sheets and the coding manual and returned their coding of the 100 random sample phrases. For each sample phrase the coders could select up to two narrative classifications from a dropdown list or select the answers “others” or “none”, in case they felt that none of the pre-selected narratives fit the sample phrase. In the end, the percent agreement of the independent expert coding with the lead author’s original coding was calculated.

This first expert intercoder review yielded an intercoder reliability of over 80 percent for most narrative sub-categories, and over 90 percent for most meta-categories. The results for identity fusion, calls to violence, out-group dehumanization, out-group demonization, existential threat, martyrdom narrative and violent role model were particularly promising. A debrief with the expert coders was used to address areas of disagreement or misunderstanding related to the coding.

To further assess the reliability and coherence of our categories, the same intercoder review process was then repeated with twenty-four independent non-expert coders. In addition to providing an independent test this second analysis allowed us to compare coding consistency levels with non-expert coders possessing limited expertise and experience with the narratives and language used by violent extremists. Based on feedback provided by the expert coders, the category “generalization” was replaced with the new narrative category “out-group slurs”. The final ICR results based on the calculated percent agreement of the 24 non-expert coders showed a reliability rate of over 90 percent for almost all narrative meta categories and sub categories ().

Is There a Language of Terrorists? A Comparative Manifesto Analysis

Published online:

09 August 2022

Table 3. Intercoder reliability (ICR) analysis results: experts and non-experts.


Our coding of the manifestos exposed a range of overarching trends in the narratives and language that run through violent self-sacrificial manifestos as opposed to those found in ideologically extreme or moderate nonviolent manifestos.

Fusion and Threat

In accordance with the fusion-plus-threat model for predicting violent extremism, indicators of fusion were highly prevalent in violent self-sacrificial manifestos. Eight out of nine authors of violent self-sacrificial manifestos used kinship language when referring to other members of the in-group. For example, Breivik asked: “How many of our sisters have and will be raped by Muslims?”.76 He repeatedly referred to fellow white Europeans as “brothers and sisters” who he claimed have been “ravaged robbed, beaten, terrorised”.77 Breivik also warned that race-mixing would lead “the eternal loss of your extended ethnic family”.78 Earnest called the “true anons”, the followers of the far-right online community on the/pol imageboard on 8chnan, his brothers: “To the true anons out there (you know who you are). You are the product of/pol/—the product of unadulterated truth. You are my brothers and the best dudes out there”.79 Ample evidence of kinship language could also be found in the texts of Hitler and Qutb. While Hitler referred to fellow Germans as brothers, Qutb described other Muslim believers as brothers whose bonds become stronger than those of biological kinship. He wrote: “When the relationship of the belief is established, whether there by any relationship of blood or not, the Believers become like brothers.”80Metaphors of shared blood was another fusion indicator that was commonly used in violent self-sacrificial texts. Roof wrote that there is “good white blood” in Uruguay, Argentina, Chile and Brazil that is “worth saving”.81 Earnest directed his manifesto in part to his “brothers in blood”.82

Perhaps the most notable result from our coding was that in-group identity fusion proxies based on our psychological kinship measures did not occur in nonviolent manifestos, whether ideologically extreme or moderate in nature. Instead, fusion proxies were detected in moderate manifestos only when used in reference to the whole of humanity. For instance, Martin Luther King Jr referred to white people as “our brothers” and expressed his hope for black boys and girls “to join hands with white boys and girls as sisters and brothers”.83 He said: “I have a dream that one day on the red hills of Georgia, the sons of former slaves and the sons of former slave owners will be able to sit down together at the table of brotherhood“. Al Qaradawi spoke both of a “brotherhood of Islâm” and a wider “brotherhood in humanity”.84 As noted earlier, fusion itself is not a predictor of violence and only becomes so when linked to outgroup threat, mediated by violence-condoning norms.

In contrast to in-group identity fusion proxies, which were restricted to violent material, forms of in-group identification could be observed across all types of manifestos, including the moderate ones. Simone de Beauvoir’s gender-based identification with women, Martin Luther King Jr’s race-based identification with Black people and Greta Thunberg’s age-based identification with the young generations were akin to Al-Qaradawi’s religious identification with Muslims and Fjordman’s ethno-cultural identification with native Europeans. Figure 3 shows the high degree of fusion-plus-threat (89 percent) detected in the manifestos of violent authors (classified as “self-sacrificial violent”), as well as their absence (0 percent) in the manifestos of nonviolent (“ideologically extreme” and “moderate”) authors. It also illustrates the contrasting results for identification-plus-threat markers, which were common across all types of manifestos. Of the 78 percent of violent self-sacrificial manifestos exhibiting identification-plus-threat 100 percent also exhibited fusion-plus-threat, which is consistent with our interpretation that fusion rather than identification is indeed driving violent behavior.

Is There a Language of Terrorists? A Comparative Manifesto Analysis

Published online:

09 August 2022

Figure 3. Group alignment detected in analyzed manifestos.

Figure 3. Group alignment detected in analyzed manifestos.

The idea of a threat against the in-group was a highly prevalent trait in violent self-sacrificial texts. Eight out of nine violent self-sacrificial manifestos included narratives of an existential threat against the in group. Two in three of the ideologically extreme manifestos and one of the three moderate manifestos (Thunberg) also included an existential threat narrative. However, Thunberg’s speech warned of a threat faced by the whole of humankind rather than an in-group. “I want you to act as if the house is on fire. Because it is“, Thunberg said85 but she also continued: “now we all have a choice. We can create transformational action that will safeguard the future living conditions for humankind. Or we can continue with our business as usual and fail. That is up to you and me“.86

Words such as “exterminated”, “under siege” and replaced” were used to describe an existential threat to the in-group. “Europe is under siege by Islam” 87, Breivik wrote and described an existential threat to Native Europeans from a “genocidal and evil regime committed to wiping out everything European”.88 Earnest warned that” we are running out of time. If this revolution doesn’t happen soon, we won’t have the numbers to win it.”89 Tarrant mentioned “an invasion on a level never seen before in history”.90 He wrote that “we must return to replacement fertility levels, or it will kill us”.91 Roof painted the picture of a civil war, claiming that “white people are being murdered daily in the streets”92 and Baillet spoke of “suppressed Whites”.93 Similarly, Earnest was afraid of “the destruction of my race”94 and Crusius implied that his country was about to fall.95 Hitler wrote about “the Jewish menace” and warned that “if, through his Marxist faith, the Jew conquers the peoples of this world, his crown will be the death and destruction of all mankind”.96 Qutb’s blamed Western regimes for trying to “shake the foundations of Islamic beliefs and then gradually to demolish the structure of Muslim society“.97

Threat narratives also came in the form of beliefs in an out-group conspiracy and inevitable war. Eight out of nine violent self-sacrificial texts contained ideas about a conspiracy of the out-group, while among the ideologically extreme it was only one out of three and none within the group of moderate manifestos. Example language that was indicative of conspiracy myths were “betrayal”, “sold us”, “collude against”, “conspire”, “fake”, “fraud”, “corrupt”. Inevitable war narratives were explicitly mentioned in six out of nine violent self-sacrificial manifestos, while it did not play a role in nonviolent manifestos. The idea of an inevitable war was expressed in words such as “war”, “fight”, “battle” or “jihaad”, sometimes in combination with adjectives such as “inevitable”, “imminent” or “looming”.

Other Variables of Interest

Other key patterns observed in violent authors’ language mirror the fusion-violence moderated mediators identified in previous research and visualized in Figure 1. Out-group entitativity and violence condoning norms played a significant role in violent self-sacrificial texts. Figure 4 provides an overview of the presence of other variables of interest we identified in the manifestos.

Is There a Language of Terrorists? A Comparative Manifesto Analysis

Published online:

09 August 2022

Figure 4. Other variables of interest detected in analyzed manifestos.

Figure 4. Other variables of interest detected in analyzed manifestos.

Out-group entitativity manifested itself linguistically in the frequent use of third person plural pronouns combined with the use of derogatory, demonizing or dehumanizing vocabulary to denounce the out-group.

Dehumanization was detected in six out of nine violent self-sacrificial manifestos, while it did not occur in nonviolent manifestos. Words used to describe the out-group in a dehumanizing fashion included “animal”, “plague”, “dog”, “lower being”, “parasite”, “creature”, ““vermin” and “monster”. “Women are like a plague. They don’t deserve to have any rights,” Rodger wrote.98 He also called them “vicious, evil, barbaric animals” that “need to be treated as such”.99 Roof suggested that Black people are “lower beings” and “brute animals” and likened them to dogs.100 Earnest described Jews as a “squalid and parasitic race”,101 while Hitler labeled them “slimy creatures”, and compared them to “a parasite” that infests the world, a “wild beast” who tortures others and a “spider” that sucks the blood from the people.102

Slurs used to describe the out-group were uniquely found in the violent self-sacrificial manifestos. For example, Roof used offensive out-group labels such as “N****** are stupid and violent”.103 “Spics and n****** are useful puppets for the Jew in terms of replacing Whites”, Earnest wrote.104 Derogatory terms to refer to Jews such as “kikes” were prevalent in the anti-Semitic violent self-sacrificial manifestos. For example, Baillet wrote: “The only way to win is to cut of the head of ZOG, which are the kikes”.105 Hitler used the German equivalent of “kike” (Jüdlein”) as a hateful out-group slur: “If you carefully punctured this abscess with a knife, like a maggot in a rotten body who was blinded by the sudden influx of light, you would discover a Kike (ein Jüdlein)”.106

Demonization was not only detected in all violent self-sacrificial manifestos but could also be observed in all ideologically extreme nonviolent manifestos. Among the many demonizing terms used for the out-group were for instance “traitor”, “evil”, “enemy”, “vicious”, “barbaric”, “depraved” and “vile”. In contrast with the extreme manifestos, the moderate manifestos were characterized by a low occurrence of outgroup entitativity language. Martin Luther King Jr and Simone de Beauvoir explicitly denounced the othering of an out-group in their manifestos.

Violence condoning norms were identified in the form of justifications for the need to use violence, martyrdom narratives, violent role models and the perceived hopelessness of alternative solutions. Justifications of violence played a role in eight out of nine violent self-sacrificial manifestos, as opposed to just one of the ideologically extreme manifestos (Marx and Engels) and none of the moderate manifestos. Terms such as “preemptive”, “self-defense”, “protect” “forced to fight”, “need for jihaad” or “natural struggle” were used in reference to justifications of violence.

Martyrdom narratives played a role in seven out of nine violent self-sacrificial manifestos, compared to just one of the ideologically extreme and none of the moderate nonviolent manifestos. Vocabulary used in connection with the glorification of self-sacrifice and martyrdom were for example “die in glory”, “sacrifice”, “knight”, “martyr”, “dying selflessly”, “immortal”, “act of preservation”, “my death” and “defending the work of the Lord”. “Equip yourself and arm up, for today you will become immortal,” Breivik wrote.107 “You will forever be celebrated by your people as a martyr for your country, protecting your culture and fighting for your kin and for Christendom”.108

Violent role models were mentioned by four of the nine authors of violent self-sacrificial manifestos. The authors used names of previous attackers, often in combination with the words “inspire”, “hero” or role model”. For example, Tarrant claimed that he was influenced by the “writings of Dylan Roof and many others, but only really took true inspiration from Knight Justiciar Breivik”.109 Earnest stated in his manifesto that Adolf Hitler, Robert Bowers, Brenton Tarrant inspired him,110 and Crusius wrote that he supported “the Christchurch shooter and his manifesto”.111

The perceived lack of alternative solutions, hence political or other peaceful solutions was observed among some violent self-sacrificial manifestos. Five out of the nine analyzed violent self-sacrificial texts denounced nonviolent solutions as impossible or ineffective. These narratives were often expressed with words such as “democratic”, “peaceful” “political”, “dialogue” or “passivity” in combination with “meaningless”, “failed”, “end”, “vanish”.

Calls to Violence

As the classification of manifestos in categories from violent to nonviolent was based on the authors’ actions following the publication (see above) rather than their words, the fusion-violence correlation could have been examined without an additional linguistic assessment of calls for violence and self-sacrifice. However, we considered it important to examine linguistic indicators of violence to provide further details on the relationship between fusion language and violent language. We anticipate that the lists of terms and phrases we identified that are used to call for violence will, when independently validated and refined, be of assistance for government security services and technology companies in their efforts to develop better early violence detection mechanisms that can help to effectively prevent attacks.

Eight out of nine of the analyzed violent self-sacrificial manifestos contained explicit calls to violence against the out-group. “All category A and B traitors who continue to oppose us will be executed.” Breivik wrote.112 He recommended that members of his in-group must “embrace and familiarize themselves “with the concept of killing women, even very attractive women”.113 “KILL ANGELA MERKEL, KILL ERDOGAN, KILL SADIQ KHAN“, Tarrant wrote in his manifesto.114 Earnest called on “every anon” reading his manifesto “to carry out attacks.”115 Equally, Hitler called for the use of violent force against Jews, declaring: “we need to use the most brutal weapons against a criminal group that is hostile to the state”.116 Qutb urged all Muslims to “fight against those among the People of the Book who do not believe in God and the Last Day”.117

Given the nature of the material examined it is no surprise that many of the terrorists explicitly state in their manifestos that they were about to carry out a violent attack. Tarrant noted: “It was there I decided to do something, it was there I decided to take action, to commit to force. To commit to violence. To take the fight to the invaders myself.”118 Rodger announced: “All of those popular people who live hedonistic lives of pleasure, I will destroy, because they never accepted me as one of them. I will kill them all and make them suffer, just as they have made me suffer.”119 In some instances, the violence indicators rather took the form of the announcement of self-sacrifice. Multiple terrorists suggested in their writings that they would fight and die for the in-group. “If I fail and die but kill a single jew, it was worth it,” Baillet wrote.120 Earnest stated: “I sacrifice this for the sake of my people. OUR people. I would die a thousand times over to prevent the doomed fate that the Jews have planned for my race”.121 “My death is likely inevitable,”, Crusius declared in his manifesto.122 Both the ideologically extreme and the moderate nonviolent manifestos lacked any indicators of violence. Additionally, many authors used language that reflected their rejection of violence.


There are several important limitations with the current study. We made efforts to select relevant comparison manifestos, including from contemporary high-profile cases, but our selection is by no means exhaustive and is restricted to English language material. Although previous studies have demonstrated the generalizability of the theoretical framework of the fusion-plus-threat model, this study only examines a small number of manifestos and is predominantly focused on far-right terrorist documents. Despite the absence of comparable pre-attack publications by jihadist and far-left perpetrators, we recognize the need to analyze more material, ideally with independent replications and preregistered studies, to confirm whether the patterns we identify here are reliable and valid. This is a crucial step that would need to be undertaken ahead of any application in regard to designing potential interventions that rely on the content categories provided.

We are furthermore aware that the comparative ethnographic content analysis is inherently subjective and relied on the assessments of the lead author. The identified categories helped to provide a consistent structure for the assessments but invariably researcher’s degrees of freedom and subjective biases are likely to have influenced the content analysis. The reliability check conducted with experts and non-experts helps to assuage concerns about the coherence of the categories devised, but it does not directly validate the reliability of the full manifesto codings presented. Language tends to be subjective, contextual, and ambiguous. In the reliability checks, coders highlighted a few instances in the coding where the randomly selected phrases were long or dense enough for more than one or two narratives to be applicable. For example, “the only way to win is to cut of (sic!) the head of ZOG, which are the kikes”, could be read as a slur, a conspiracy theory reference, or a call to violence. Other featured sample phrases were too short or taken out of context. The phrase “My death is likely inevitable” caused many coders to choose “none” or “other” in the category drop-down list. Although within the context of the manifesto, this phrase would have been interpreted as “martyrdom narrative”, it did not fit into any categories as an isolated phrase. Likewise, the term “jihaad” could, without further context, be interpreted as either violent or nonviolent.

Similarly, we recognize that by using binary absent or present indicators that we are simplifying our analyses at the cost of a more higher resolution examination of the content. Our defense for this decision is that we regard the current research as providing the foundations for a more robust quantitative NLP analysis that we intend to carry out in a follow up study. Another limitation of the research is that manifestos cannot offer comprehensive insight into their authors and are written with particular audiences in mind. This does not impact the usefulness of the content for identifying linguistic cues of violence, however it does mean that we should be wary of taking declarations or descriptions of motives as being accurate. Similarly, the fact that a motive does not feature in a manifesto does not necessarily mean that it did not play a role in the author’s radicalization or that is not relevant regarding why they decided to conduct an attack. The important caveat here then is that the content and patterns identified should be approached critically. The categories identified are intended primarily to serve as helpful tools to identify common traits in the language and themes of violent extremists rather than claiming to offer a comprehensive psychological profile.


Psychological profiling has a role to play in the prevention of violent extremism and terrorism but only to the extent that it correctly identifies relevant predictors and a suitable method of detecting them. These are formidable challenges, requiring a well-substantiated theory of the motivators of terrorism and a feasible method of measuring them in noncompliant individuals. In this study we have suggested novel solutions to both problems. This paper provides the first evidence of the utility, validity, and reliability of a novel and theoretically informed codification schema that we anticipate being useful for researchers and other parties examining extremist literature and written communications, including manifestos, blog posts or chat messages.

Our method uses proxies for psychological constructs that provide a window into actors’ motivational states and their capacity for murderous forms of self-sacrifice. The link between fusion and pro-group action is well established but has previously been based on measuring fusion directly using pictorial and verbal scales, a procedure that would be impractical to adopt with uncooperative members of extremist groups. We have shown, however, that those bent on carrying out atrocities unknowingly reveal their levels of fusion by using language redolent of familial ties and psychological kinship. When this is coupled also with language expressing a strong conviction that the group is imperiled, and buttressed also by violence-condoning norms and the willingness to vilify, demonize, and dehumanize enemies, it robustly predicts a deadly mindset.

While our findings appear promising for the predictive power of the fusion-plus-threat model, they also highlight the relevance of other variables that might play a role in mediating pathways to violent extremism. Our study therefore also supports alternative models that can be reconciled or even combined with the fusion-plus-threat model, such Sternberg’s “Duplex Theory of Hate”,123 Bandura’s “Moral Disengagement Theory,124 Kruganski et al.’s “Quest for Significance Model”125 and Louis et al.’s research on the power of group norms.126

Our initial tests of coherence and reliability conducted with both experts and non-experts provided very promising results. Nonetheless, we recognize the need for a more fine-grained examination that provides more detail on the prevalence of the identified narrative categories in the content rather than a binary indicator of presence or absence. The next stage of this study is an NLP-based analysis in R to quantify the occurrence of linguistic markers for each category in the fifteen manifestos. We recommend working closely with counter-terrorism professionals when refining the fusion-based violence prediction model in the future.

Our qualitative ethnographic assessment of the fifteen manifestos also offers a much-needed contemporary update to earlier examinations of manifesto content given the expanded size of the genre. Although we have focused our efforts here on showing how to recognize the language of violent extremism in groups that are already known to be violent, the methods we propose are generalizable and could be used, via early detection, to prevent at risk individuals and groups from engaging in violence before it is too late. These methods are neither obvious nor intuitive insofar as they do not treat any of the core variables alone as predictors of violent extremism. Simply being fused with a group is not enough to motivate acts of terrorism, nor are violence-condoning norms, or even outgroup hatred. Indeed, these variables in isolation may be harmless, even if their expressions may be unpleasant or offensive to many. Just like some volatile chemicals, they only become dangerous when they are mixed together. Linguistic analysis can help us to identify when that is the case and act more swiftly to avert threats before they are tragically realized.

1 A more extensive descriptive explanation of each selected manifesto can be found in the Appendix A.

2 We reached this conclusion after having searched academic articles, news reports and primary source archives, and consulted the founder of the jihadist primary source website Jihadology Aaron Y. Zelin.

3 Although we could identify additional narratives (e.g. desire for personal revenge or masculinity crisis), we did not include any narratives that were not present in at least four violent self-sacrificial manifestos.

1 Rafal Zgryziewicz, Violent Extremism and Communications (Riga: NATO Strategic Communications Centre of Excellence, 2018); DHS, Digital Forum on Terrorism Prevention: A Report on Trends and Insights (Washington, DC: U.S. Department of Homeland Security, 2017) and Erin Saltman, Countering Terrorism and Violent Extremism at Facebook: Technology, Expertise and Partnerships (New Delhi: Observer Research Foundation, 27 August 2020).

2 W. B. Swann, M. Buhrmester, Á Gómez, J. Jetten, B. Bastian, and A. Vásquez, “What Makes a Group Worth Dying For? Identity Fusion Fosters Perception of Familial Ties, Promoting Self-Sacrifice,” Journal of Personality and Social Psychology 106, no. 6 (2014): 912–26.

3 Cf. Whitehouse, “Dying for the Group”.

4 See for example H. Whitehouse, B. McQuinn, M. Buhrmester, and W. B. Swann, “Brothers in Arms: Warriors bond like Family,” Proceedings of the National Academy of Sciences 111, no. 50 (2014): 17783–5; H. Whitehouse, J. Jong, M. D. Buhrmester, Á. Gómez, B. Bastian, C. M. Kavanagh, M. Newson, et al., “The Evolution of Extreme Cooperation via Shared Dysphoric Experiences,” Scientific Reports 7 (2017): 1–10 or C. C. Tossell, C. Chad, Angel Gómez, Ewart J. de Visser, Alexandra Vázquez, Bianca T. Donadio, Amanda Metcalfe, Charles Rogan, Richard Davis, and Scott Atran, “Spiritual Over Physical Formidability Determines Willingness to Fight and Sacrifice through Loyalty in Cross-Cultural Populations,” Proceedings of the National Academy of Sciences 119, no. 6 (2022).

5 Christopher Kavanagh, Susilo Wibisono, Rohan Kapitány, Whinda Yustisia, Idhamsyah Eka Putra, Any Rufaedah, and Harvey Whitehouse, “Exploring the Role of Identity Fusion and Group Identification in Predicting Parochialism amongst Indonesian Islamic groups,” PsyArXiv (2019).

6 M. D. Buhrmeister, D. Zeitlyn, and H. Whitehouse, “Ritual, Fusion, and Conflict: The Roots of Agro-Pastoral Violence in Rural Cameroon,” Group Processes & Intergroup Relations 25, no. 1 (2020): 298–311.

7 H. Whitehouse; J. Jong, M. D. Buhrmester, Á. Gómez, B. Bastian, C. M. Kavanagh, M. Newson, M. Matthews, J. A. Lanman, R. McKay, and S. Gavrilets, “The Evolution of Extreme Cooperation via Shared Dysphoric Experiences,” Scientific Reports 7, no. 44292 (2017): 1–10.

8 Martha Newson, “Football, Fan Violence, and Identity Fusion,” International Review for the Sociology of Sport 54, no. 4 (2019): 431–44 and Martha Newson, Tiago Bortolini, Michael Buhrmester, Silvio Ricardoda Silva, Jefferson Nicássio Queirogada Aquino, and Harvey Whitehouse, “Brazil’s Football Warriors: Social Bonding and Inter-Group Violence,” Evolution and Human Behaviour 39, no. 6 (2018): 675–83.

9 Harvey Whitehouse, “Dying for the Group: Towards a General Theory of Extreme Self-Sacrifice,” Behavioral and Brain Sciences 41 (2018): 1–62.

10 Harvey Whitehouse, “What Motivates Extreme Self-Sacrifice?,” Pacific Standard 9, no. 2 (2016): 26–31.

11 Harvey Whitehouse, “Four Things We Need to Know about Extreme Self-Sacrifice,” Behavioral and Brain Sciences 41 (2018).

12 Cliodhna O’Connor and Helene Joffe, “Intercoder Reliability in Qualitative Research: Debates and Practical Guidelines,” International Journal of Qualitative Methods 19 (2020): 1–13.

13 See for example Arun Kundnani, A Decade Lost: Rethinking Radicalisation and Extremism, (London: Claystone, 2015) and Lauren Powell, “Counter-Productive Counter-Terrorism. How Is the Dysfunctional Discourse of Prevent Failing to Restrain Radicalisation?” Journal for Deradicalisation 6 (2016).

14 Peter R. Neumann, “Countering Violent Extremism and Radicalisation That Lead to Terrorism: Ideas, Recommendations, and Good Practices from the OSCE Region,” OSCE Report, 2017 and Pail Gill, John Horgan, and P. Deckert, “Bombing Alone: Tracing the Motivations and Antecedent Behaviors of Lone-Actor Terrorists,” Journal of Forensic Science 59, no. 2 (2014).

15 Quintan Wiktorowicz, Radical Islam Rising: Muslim Extremism in the West (London: Rowman & Littlefield Publishers, 2005).

16 Marc Sageman, “A Strategy for Fighting International Islamist Terrorists,” The Annals of the American Academy of Political and Social Science 618 (2008): 223–31.

17 Magnus Ranstorp, Understanding Violent Radicalisation: Terrorist and Jihadist Movements in Europe (London: Routledge, 2010).

18 Whitehouse, “Dying for the Group”.

19 See Institute for Strategic Dialogue’s website: https://www.isdglobal.org/extremism/.

20 See fir example Mattias Gardell, “Crusader Dreams: Oslo 22/7, Islamophobia, and the Quest for a Monocultural Europe,” Terrorism and Political Violence 26 (2014): 129–155; Brett A. Barnett, “20 Years Later: A Look Back at the Unabomber Manifesto,” Perspectives on Terrorism 9, no. 6 (2015): 60–71; Alejandro Beutel, “The New Zealand Terrorist’s Manifesto: A Look at Some of the Key Narratives, Beliefs and Tropes” (College Park, MD: START, April 2019) and E. Chebrolu, “The Radical Lens of Dylann Roof: Racial Anxiety and White Nationalist Rhetoric on New Media,” Review of Communication 20, no. 1 (2020): 47–86.

21 Lisa Kaati, Amendra Shrestha, and Katie Cohen, “Linguistic Analysis of Lone Offender Manifestos” (Uppsala: Swedish Defence Research Agency, 2016).

22 Kaati et al., “Linguistic Analysis”.

23 Alain Van Hiel and Ivan Mervielde, “The Measurement of Cognitive Complexity and Its Relationship with Political Extremism, Political Psychology 24, no. 4 (2003): 781–801.

24 Jacob Ware, “Testament to Murder: The Violent Far-Right’s Increasing Use of Terrorist Manifestos” (The Hague: ICCT Policy Brief, March 2020).

25 Ware, “Testament to Murder.”

26 H. Whitehouse, “Dying for the Group: Towards a General Theory of Extreme Self-Sacrifice,” Behavioral and Brain Sciences 41 (2018): 1–62;

27 W. B. Swann, M. Buhrmester, Á Gómez, J. Jetten, B. Bastian, and A. Vásquez,. “What Makes a Group Worth Dying For? Identity Fusion Fosters Perception of Familial Ties, Promoting self-Sacrifice,” Journal of Personality and Social Psychology 106, no. 6 (2014): 912–26.

28 Alexandra Vázquez, Juan R. Ordoñana, Harvey Whitehouse, and Ángel Gómez, “Why Die for My Sibling? The Positive Association between Identity Fusion and Imagined Loss With endorsement of Self-Sacrifice,” International Journal of Social Psychology 34, no. 3 (2019): 413–38 and H. Whitehouse, J. Jong, M. D. Buhrmester, Á. Gómez, B. Bastian, C. M. Kavanagh, M. Newson, et al., “The Evolution of Extreme Cooperation Via Shared Dysphoric Experiences,” Scientific Reports 7 (2017): 1–10.

29 M. D. Buhrmester, W. T. Fraser, J. A. Lanman, H. Whitehouse, and W. B. Swann, “When Terror Hits Home: Identity Fused Americans Who Saw Boston Bombing Victims as ‘Family’ Provided Aid,” Self & Identity 14, no. 3 (2014): 253–70.

30 Harvey Whitehouse and Jonathan A. Lanman, “The Ties That Bind Us: Ritual, Fusion, and Identification,” Current Anthropology 55, no. 6 (2014): 674–95.

31 Whitehouse “Dying for the Group”.

32 M. Rosenberg, “Depersonalisation: The Loss of Personal Identity,” in Self and Identity: Perspectives Across the Lifespan, ed. T. Honess and K. M. Yardley (Abingdon: Routledge, 1987), 193–206.

33 See for example W. B. Swann and M. D. Buhrmeister, “Identity Fusion,” Current Directions in Psychological Science 24, no. 1 (2015): 52–7; M. A. Hogg and Scott A. Reid, “Social Identity, Self-Categorization, and the Communication of Group Norms,” Communication Theory 16, no. 1 (2006): 7–30, M. A. Hogg and J. C. Turner, “Intergroup Behaviour, Self-Stereotyping and the Salience of Social Categories,” British Journal of Social Psychology 26, no. 4 (1987): 325–40.

34 Whitehouse, “Dying for the Group”; W. B. Swann, A. Gómez, C. D. Seyle, J. F. Morales, and C. Huici, “Identity Fusion: The Interplay of Personal and Social Identities in Extreme Group Behavior,” Journal of Personality and Social Psychology 96 (2009): 995–1011 and Á. Gómez, M. L. Brooks, M. D. Buhrmester, A. Vázquez, J. Jetten, and W. B. Swann, “On the Nature of Identity Fusion: Insights into the Construct and a New Measure,” Journal of Personality and Social Psychology 100, no. 5 (2011): 918–33.

35 W. B. Swann, A. Gómez, J. Dovidio, S. Hart, and J. Jetten, “Dying and Killing for One’s Group: Identity Fusion Moderates Responses to Intergroup Versions of the Trolley Problem,” Psychological Science 2, no. 8 (2010): 1176–83.

36 Whitehouse “Dying for the Group,” Swann et al., “What Makes a Group Worth Dying for?” and Whitehouse et al., “The Evolution of Extreme Cooperation.”

37 A. Lankford, “Précis of the Myth of Martyrdom: What Really Drives Suicide Bombers, Rampage Shooters, and Other Self-Destructive Killers,” Behavioral and Brain Sciences 37 (2014): 351–93.

I. S. Sheehan, “Are Suicide Terrorists Suicidal? A Critical Assessment of the Evidence,” Innovations in Clinical Neuroscience 11, nos. 9–10 (2014): 81 and A. Lankford, “Could Suicide Terrorists Actually be Suicidal?,” Studies in Conflict & Terrorism 34, no. 4 (2011): 337–66.

38 Whitehouse, “Dying for the Group”.

39 V. K. Choi, M. J. Gelfand, and J. C. Jackson, “The Role of Entitativity in Perpetuating Cycles of Violence,” Behavioral and Brain Sciences 41 (2018); B. Lickel, N. Miller, D. M. Stenstrom, T. F. Denson, and T. Schmader, “Vicarious Retribution: The Role of Collective Blaming in Intergroup Aggression,” Personality and Social Psychology Review 10, no. 4 (2006): 372–90.

40 J. M. Berger, “Extremist Construction of Identity: How Escalating Demands for Legitimacy Shape and Define In-Group and Out-Group Dynamics,” ICCT Research Paper, 2017.

41 R. Giner-Sorolla, B. Leidner, and E. Castano, “Dehumanization, Demonization, and Morality Shifting,” in Extremism and the Psychology of Uncertainty, ed. M. A. Hogg and D. L. Blaylock (Hoboken, NJ: Wiley-Blackwell, 2011), 165–82.

42 Whitehouse, “Dying for the Group”.

43 Gilad Hirschberger, Tsachi Ein-Dor, Bernard Leidner, and Tamar Saguy, “How Is Existential Threat Related to Intergroup Conflict? Introducing the Multidimensional Existential Threat (MET) Model,” Frontiers in Psychology 7 (2016): 1877.

44 Ali Mashuri and Esti Zaduqisti, “The Effect of Intergroup Threat and Social Identity Salience on the Belief in Conspiracy Theories over Terrorism in Indonesia: Collective Angst as a Mediator,” International Journal of Psychological Research 8, no. 1 (2015): 24–35.

45 Julia Ebner, The Rage: The Vicious Circle of Islamist and Far-Right Extremism (London: Bloomsbury Academic, 2017).

46 W. R. Louis, C. McGarty, E. F. Thomas, C. E. Amiot, and F. M. Moghaddam, “The Power of Norms to Sway Fused Group Members,” Behavioral And Brain Sciences 41 (2018): 29–30.

47 Louis et al., “The Power of Norms”.

48 Jacob Davey and Julia Ebner, “‘The Great Replacement’: The Violent Consequences of Mainstreamed Extremism” (London: Institute for Strategic Dialogue (ISD), 2019) and Katie Cohen, Fredrik Johansson, Lisa Kaati, and Jonas Clausen Mork, “Detecting Linguistic Markers for Radical Violence in Social Media,” Terrorism and Political Violence 26, no 1 (2014): 246–56.

49 E. F. Thomas and W. R. Louis, “When Will Collective Action Be Effective? Violent and Non-Violent Protests Differentially Influence Perceptions of Legitimacy and Efficacy Amongst Supporters,” Personality and Social Psychology Bulletin 40 (2014): 263–76.

50 See for example, the UK’s Counter-Terrorism and Security Act 2015, https://www.legislation.gov.uk/ukpga/2015/6/contents/enacted.

51 Florence Keen, Blyth Crawford and Guillermo Suarez-Tangil, “Memetic Irony and The Promotion of Violence Within Chan Cultures,” CREST (2020) and Julia Ebner, Going Dark: The Secret Social Lives of Extremists (Bloomsbury, 2019).

52 Ware, “Testament to Murder.”

53 Bjorn Ihler, “On the Terrorist Attack in Buffalo,” Khalifa Ihler Institute, 15 May 2022, Online: https://www.khalifaihler.org/newsb/2022/5/15/on-the-terrorist-attack-in-buffalo-ny.

54 E. A. Bäck, H. Bäck, M. G. Sendén, and S. Sikström, “From I to We: Group Formation and Linguistic Adaption in an Online Xenophobic Forum,” Journal of Social and Political Psychology 6, no. 1 (2018): 76–91.

55 Whitehouse and Lanman, “The Ties That Bind Us”.

56 Björn Technau, “Going Beyond Hate Speech: The Pragmatics of Ethnic Slur Terms,” Lodz Papers in Pragmatics 14, no. 1 (2018): 25–43.

57 L. Anderson and E. Lepore, “Slurring Words,” Nous 47 (2013): 25–48.

58 Robin Jeshion, “Slurs and Stereotypes,” Analytic Philosophy 54, no 3 (2013): 314–29.

59 Nahi Alon and Haim Omer, The Psychology of Demonization: Promoting Acceptance and Reducing Conflict (London: Routledge, 2005).

60 Elihu D. Richter, Dror Kris Markus, and Casey Tait, “Incitement, Genocide, Genocidal Terror, and the Upstream Role of Indoctrination: Can Epidemiologic Models Predict and Prevent?,” Public Health Review 39, no. 1 (2018): 1–22.

61 Monika Schwarz-Friesel. “Using Language as a Weapon: Verbal Manifestations of Contemporary Anti-Semitism,” in The Exercise of Power in Communication , ed. Rainer Schulze and Hanna Pishwa (London: Palgrave Macmillan, 2015), 161–83.

62 Jeroen Vaes, Jacques-Philippe Leyens, Maria Paola Paladino, and Mariana Pires Miranda, “We Are Human, They Are Not: Driving Forces Behind Outgroup Dehumanisation and the Humanisation of the Ingroup,” European Review of Social Psychology 2, no. 1 (2012): 64–106.

63 G. T. Viki, L. Winchester, L. Titshall, and T. Chisango, “Beyond Secondary Emotions: The infra-Humanisation of Groups Using Human-Related and Animal-Related Words,” Social Cognition 24 (2006): 753–75.

64 Cf. Hirschberger et al, “How is Existential Threat Related to Intergroup Conflict?”

65 Cynthia Miller-Idriss, “White Supremacist Extremism and the Far Right in the U.S.” (Farmington Hills, MI: Gale Primary Sources/Cengage, 2021).

66 J. M., Bale, “Political Paranoia v. Political Realism: On Distinguishing between Bogus conspiracy Theories and Genuine Conspiratorial Politics,” Patterns of Prejudice 41 (2007):45–60.

67 C. Klein,, P. Clutton, and A. G. Dunn, “Pathways to conspiracy: The Social and Linguistic Precursors of Involvement in Reddit’s Conspiracy Theory Forum,” PLoS ONE 14, no. 11 (2019).

68 Matthew Kriner, Meghan Conroy, and Yasmine Ashwal, “Understanding Accelerationist Narratives: ‘There Is No Political Solution’” (London: GNET Report, Kings College, 2021).

69 Winnifred R. Louis, Donald M. Taylor, and Rebecca L. Douglas, “Normative Influence and Rational Conflict Decisions: Group Norms and Cost-Benefit Analyses for Intergroup Behavior,” Group Processes & Intergroup Relations 8, no. 4 (2005): 355–74 and Dario Spini, Guy Elcheroth, and Rachel Fasel, “The Impact of Group Norms and Generalization of Risks across Groups on Judgments of War Behavior,” Political Psychology 29, no. 6 (2008): 920–41.

70 Mohammed M. Hafez, “Martyrdom Mythology in Iraq: How Jihadists Frame Suicide Terrorism in Videos and Biographies,” Terrorism and Political Violence 19, no. 1 (2007): 95–115 and Thomas Fraise, “Shades of Jihad: Variation of Military Ethics between ISIS and alQaeda,” (Paris: Sciences Po Kuwait Program, Research Paper, 2017).

71 Cf. Rola el-Husseini, “Resistance, Jihad, and Martyrdom in Contemporary Lebanese Shi’a Discours,” Middle East Journal 62, no. 3 (2008): 399–414; Hafez, “Martyrdom Mythology in Iraq” and Amélie Blom, “Do Jihadist ‘Martyrs’ Really Want to Die? An Emic Approach to Self-Sacrificial Radicalization in Pakistan,” Revue française de science politique 61 (2011): 867–91.

72 Richter, Markus and Tait, “Incitement, Genocide, Genocidal Terror”.

73 Davey and Ebner, “The Great Replacement’” and Cohen et al., “Detecting Linguistic Markers for Radical Violence.”

74 Thomas and Louis, “When Will Collective Action be Effective?” and R. Spears “Group Rationale, Collective Sense: Beyond Intergroup Bias,” British Journal of Social Psychology 49 (2010): 1–20.

75 Cohen et al., “Detecting Linguistic Markers for Radical Violence”.

76 Breivik, “2083,” 1281.

77 Ibid., 1031.

78 Ibid., 1161.

79 Earnest, “An Open Letter,” 11.

80 Qutb, Milestones, 99.

81 Roof, “Manifesto,” 4.

82 Earnest, “An Open Letter,” 6 and 1.

83 King Jr, “I Have a Dream,” 3 and 5.

84 al-Qaradawi, The Lawful and Prohibited in Islam, 184.

85 Thunberg, “Our House,” 3.

86 Ibid., 2.

87 Breivik, “2083,” 74 and 663.

88 Breivik, “2083,” 1030.

89 Earnest, “An Open Letter,” 9.

90 Tarrant, “The Great Replacement,” 3.

91 Ibid., 20.

92 Roof, “Manifesto,” 4.

93 Baillet, “A Short Pre-Action Report,” 13.

94 Earnest, “An Open Letter,” 9.

95 Crusius, “The Inconvenient Truth,” 4.

96 Hitler, Mein Kampf, 45.

97 Ibid., 95.

98 Rodger, “My Twisted World,” 136.

99 Ibid., 136.

100 Roof, “Manifesto,” 2.

101 Earnest “An Open Letter,” 11.

102 Hitler, Mein Kampf, 40, 100, 201, 124 and 125.

103 Roof, “Manifesto,” 1.

104 Earnest “An Open Letter,” 9.

105 Baillet, “A Short Pre-Action Report,” 15.

106 Hitler Mein Kampf, 40.

107 Breivik, “2083,” 946.

108 Ibid., 947.

109 Tarrant, “The Great Replacement,” 19.

110 Earnest, “An Open Letter,” 10.

111 Crusius, “The Inconvenient Truth,” 1.

112 Breivik, “2083,” 797.

113 Ibid., 941.

114 Tarrant “The Great Replacement,” 40

115 Earnest “An Open Letter,” 6

116 Hitler, Mein Kampf, 23.

117 Qutb, Milestones, 49.

118 Tarrant “The Great Replacement,” 9.

119 Rodger, “My Twisted World,” 137.

120 Baillet, “A Short Pre-Action Report,” 15.

121 Earnest “An Open Letter,” 2

122 Crusius, “The Inconvenient Truth,” 4.

123 R. Sternberg, “A Duplex Theory of Hate: Development and Application to Terrorism, Massacres, and Genocide,” Review of General Psychology 7 (2003): 303.

124 Albert Bandura, “The Origins and Consequences of Moral Disengagement: A Social Learning Perspective,” in Understanding Terrorism: Psychosocial Roots, Consequences, and Interventions, ed. F. M. Moghaddam and A. J. Marsella (Washington, DC: American Psychological Association, 2004), 121.

125 Arie Kruglanski, Katarzyna Jasko, and David Webber, “The Making of Violent Extremists,” Review of General Psychology 22, no. 1, (2018).

126 W. R. Louis, C. McGarty, E. F. Thomas, C. E. Amiot, and F. M. Moghaddam, “The Power of Norms to Sway Fused Group Members,” Behavioral and Brain Sciences 41 (2018): 29–30.