Journal Abstract
Advances in computational approaches to detect violent extremists have made it possible to reliably identify these individuals through their communication, with accuracy ranging from 87 to 92% (95% CI), depending on the model used. Of interest is whether various facets of their communication can be explained through existing empirical research, whose findings may point to opportunities for intervention. To explore the possibility of explanatory evidence, this study conducted a systematic review of studies that presented original data and findings at the individual level of extremism. Samples of the qualifying studies (n = 272) over the last 20 years were disaggregated by individual type (e.g., lone-actor terrorist, online extremist), which included affiliations with a range of extremist groups, organizations, and ideological classifications. Using a series of linguistic dimensions observed in extremist communication as an organizing framework, the findings of these studies were grouped together as possible factors behind the various fine-grained manifestations of extremist ideology observed by high-performing models capable of extremist detection (n = 28). The work serves practitioners interested in an organizing framework for extremist communication studied as big data or those who merely seek a state-of-the-art systematic review of Islamic-based extremism. Comparatively, this study is the largest of such reviews to date, as an exploratory effort to offer possible explanations for ideological communication by extremists, and draws on samples from multiple populations of empirical studies on fundamentalism, Islamism, jihadism, and radicalization into terrorism.