The discourse of school shooters: Computational linguistics analysis of radicalization markers
Media Linguistics · 2026 · 13 (1) · 170-201
DOI: 10.21638/spbu22.2026.109 ↗Kurz gefasst
Fachartikel in der Zeitschrift Media Linguistics, 2026. Verfasst von Anna Yu. Karpova und 3 weiteren Autor:innen.
Abstract (Originalsprache)
One of the current areas of research in the field of school shooting is the study of imitation mechanisms. This interdisciplinary subject attracts the attention of specialists from different fields of knowledge: sociologists, psychologists, linguists, criminologists and data scientists. The study of warning behavior to successfully prevent an attack is inseparable from the study of the imitation mechanism. Such warning behavior signals are contained in written messages left by school shooters in both online and offline communication. In this paper, we focus on an experiment with case studies of written messages from known school shooters. The aim is to identify linguistic markers of warning behavior using computational linguistic techniques. The study identified signs of cognitive bias, various emotions, intentions and named entities using open source pre-trained natural language analysis models. In order to separate meaningful markers of warning behavior from written communication common features, open text datasets were also processed, including social and news media texts, collections of essays, blogosphere posts and literary works. The study revealed the following characteristic features inherent to the texts of school shooters: (1) the predominant presence of cognitive distortion ‘personalization’ signs in comparison with the corpus of texts not directly related to the topic of school shooters; (2) the presence of an exceptional number of anger signs and a significant number of sadness signs; (3) in addition to asocial and destructive intentions, also intentions with explicitly positive connotations occupy a significant place in the texts of shooters; (4) texts contain statistically significantly higher proportions of all toxicity categories. Despite a number of used methodology limitations, we note that the results obtained should be taken into account when developing tools for the prevention of school shooting attacks.
Zitieren
APA
Karpova, A. Y., Savelev, A. O., Mishankina, N. A., & Tretyakov, D. A. (2026). The discourse of school shooters: Computational linguistics analysis of radicalization markers. Media Linguistics, 13(1), 170-201. https://doi.org/10.21638/spbu22.2026.109
Vancouver
Karpova AY, Savelev AO, Mishankina NA, Tretyakov DA. The discourse of school shooters: Computational linguistics analysis of radicalization markers. Media Linguistics. 2026;13(1):170-201. doi:10.21638/spbu22.2026.109
BibTeX
@article{Karpova2026,
author = {Karpova, Anna Yu. and Savelev, Aleksei O. and Mishankina, Natalia A. and Tretyakov, Dmitri A.},
title = {The discourse of school shooters: Computational linguistics analysis of radicalization markers},
journal = {Media Linguistics},
year = {2026},
volume = {13},
number = {1},
pages = {170--201},
doi = {10.21638/spbu22.2026.109}
}
EUVIMED