Addressing imagined future attackers: A corpus analysis of shared agency in the online manifestos of perpetrators of mass harm

The manifestos that frequently accompany mass shootings are usually freely available online within minutes of an attack taking place and remain so for several years afterwards. It is well documented that the writers of such texts reference past shooters or copy elements of previous attacks (e.g., Langman 2017 ; Kupper et al 2022 ). Studies predominantly focus on shared linguistic markers of psychological variables in the texts (e.g. Shrestha et al 2020 ) to try to predict attacks ahead of time. However, because the manifestos do not appear far enough in advance of the attacks for such approaches to be effective, this study instead examines how the language used in them inspires others who may carry out similar attacks in the future. This paper uses corpus analysis of keywords to identify the ways in which 15 perpetrators actively address imagined future attackers and anticipate them as an audience. Findings demonstrate that rather than a passive ‘contagion’ effect ( Kupper et al. 2022 ), writers of such texts use second person pronouns ambiguously to share agency and connect with future readers and instruct them, and that this varies depending on the ideology of the perpetrator. These findings have implications for the way in which the availability of such texts is viewed and suggest that the role of these texts in the perpetuation of violence should be taken more seriously by those responsible for disseminating them.