Words as Warnings: Does the Language of Grievance-Fueled Lone Mass Violent Offenders Differ Significantly from the Language of Non-Offender Radicals?

Abstract

Law enforcement and security services worldwide carry a heavy burden in trying to detect and disrupt mass violent attacks before they can be carried out. These kinds of attacks have taken many forms - in the last few years alone, a racially-motivated shooter killed 10 people in a Buffalo, NY supermarket, four people were shot and killed at a Georgia high school by a teenager known to have posted extremist content on Discord, and authorities arrested a teenage religious extremist who was gathering explosive materials in preparation for an attack in Las Vegas. Unlike members of terrorist cells, lone violent actors are particularly difficult to detect due to their isolation and their tendency to display fewer observable radicalization indicators. Online communications, however, offer a potential point of exposure, as radicalized individuals sometimes reveal their violent intentions in virtual communities created for sharing extremist content. This study analyzed the online language of individuals aligned with involuntary celibate (“incel”) ideology to compare terminology used by those who committed or planned misogynist grievance-fueled mass violent attacks with that of the broader online incel community. Using Linguistic Inquiry and Word Count (LIWC) word frequency analysis, offender texts were found to differ systematically from non-offender peers. They contained higher levels of absolutist, anger-related, self-oriented, and violence-related language, more references to infamous grievance-fueled attackers and aspects of daily life, and fewer ideology-specific slang and other-oriented language. These findings indicate that linguistic markers in online discourse may serve as meaningful indicators for threat assessment and risk profiling. Keywords: radicalization, incel, misogynistic terrorism, grievance-based violence, linguistic analysis, threat assessment