While there exists an academic consensus stressing the importance of extremist words that sharply delineate, reify and polarise in‐ and out‐group identities, much research remains to be done on the precise qualitative difference between the structures and linguistic markers that are evident in violent and non‐violent conspiratorial language, especially on the extreme far right, and how this encourages an individual to violent action.
What we have found in this report is both striking and, in some cases, unexpected. Looking through the qualitative analysis of the manifestos, we can find that the common denominator in all manifestos is the common conspiratorial narrative that the white race is becoming extinct and replaced by non‐whites – though the timeline for action and the call to action is brought forwards and obviously tilted in a more violent direction for RWLA manifestos. In terms of linguistic features, there are considerable differences within the RWLA and non‐violent manifestos as to the targeted out‐group, format and solutions prescribed by the authors. For example, while Hispanics are the target of Crusius’ ire, so black people are the main out‐group for Roof and the Islamic population fulfils this role for Tarrant and Rathjen. What is also interesting is the differing levels of conspiratorial language used between these manifestos; Rathjen is the outlier in his focus on the paranoid conspiracy of a secret organisation monitoring his every movement.
Using the Grievance Dictionary, what is evident is how similarities overshadow differences between the violent and non‐violent manifestos. On the whole, we found a greater percentage of violent and threat‐based language in four of the six violent manifestos when compared with that of those cleaving to non‐violent manifestos. Worryingly, however, we see in the cases of Roof and Rathjen something equal to the threat‐based language of the violent manifestos. In the case of the Der Dritte Weg and Traditionalist Worker Party manifestos, the language is perhaps more violent and threat‐based than the violent manifestos.
In sum, then, what we have found is a more complex issue than expected. Both violent and non‐violent manifestos use dehumanising terms to define their out‐groups and draw the horizon for people to take action (violent or not) in such a way that defined out‐groups appear an existential threat. We hope the charts and tables in this report aid tech companies, policymakers and practitioners to appreciate this overlap but also to use the findings herein in terms of the structure, patterns and themes to identify the functionality of dehumanising rhetoric, how it works and how it can be used to create dangerous ideological ecosystems that encourage individuals to action in aid of a conspiratorial, exclusionary cause. In the interest of academic humility, we do not suggest any predictive nature of the modelling stipulated but instead throw the gauntlet down to other practitioners and researchers to use the Grievance Dictionary and Baele’s archetypes of 2019 to analyse violent, conspiratorial language in other online extremist communities.