Computerized text-analysis of offenders of mass shootings: an investigation of moral foundations using linguistic inquiry and word count

The present study used a computational linguistic approach to examine moral foundations, emotionality, and personal concerns in the written communications of mass shooters. Writings by mass shooters (N = 36) were harvested from an online database and coded for writing type using five writing categories: manifestos, blog posts, social media, private letters, and journal entries. They were then submitted to linguistic analyses using Linguistic Inquiry and Word Count (LIWC). Shooters’ writings were then compared to a sample of prisoners (N = 35) who were convicted of violent crime along with the normative data available in LIWC. The Moral Foundations Dictionary (MFD) was also used to predict differences between the samples, although no significant differences were found between shooters and prisoners on the five moral foundations. Overall, results of these analyses indicated that mass shooters primarily use high rates of negative emotion and swear words. Unlike previous studies, however, mass shooters were comparatively low on cognitive processes and complexity relative to the prisoner sample. Exploratory analyses were then conducted using the entire LIWC2015 and MFD dictionaries to identify the set of word categories that maximally predicts differences between groups across writing types.