Linguistic Predictors of Taking Violent Action: A Comparative Content Analysis of Targeted Violence Manifestos and Extremist Calls to Action
Date
2024-05-28
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
0009-0001-2183-937X
Type
Thesis
Degree Level
Masters
Abstract
Increasing media and scholarly attention is currently on the rising threat of ‘lone wolf violence’, mass violence planned and executed by a single person without the direction of an external command structure. These lone attackers frequently publish their intentions, autobiographies, and ideologies online as manifestos intended to radicalize readers and inspire further violence. However, huge volumes of similar violent rhetoric are posted online every day by people who never attempt deadly violence, making it extremely difficult to identify serious online threats before violence occurs. Luckily, psychological traits of language users can be revealed by examining automatic linguistic patterns – elements of language produced without conscious awareness – even when the explicit meaning of given writings is similar. We gathered a novel sample of each group’s writings – lone attackers and non-attackers – and searched for intergroup differences in these automatic patterns with a two-stage analysis using Linguistic Inquiry and Word Count (LIWC) software. Stage 1 identifies group differences in the frequency of use of different types of language and stage 2 uses word cooccurrence analysis to examine differences in how frequently these types of language cooccur. Stage 1 found that samples differed significantly in their use of Agency, ‘I’ language, Anxiety, Sadness, Friendship, and Positive Emotion language, with lone attackers using significantly more of every language type. Stage 2 only found relationships between these linguistic types in lone attacker writings. ‘I’ language was related positively to Friendship, sadness, and anxiety language, and negatively to Agency and Positive Emotion language. It also found negative relationships between Sadness and Positive Emotion, Agency and Friend language, and Agency and Sadness language. The intergroup linguistic differences identified in these analyses can be used as linguistic correlates of violent behaviour.
Description
Keywords
Extremism, word count analysis, manifesto(s), online, hate speech
Citation
Degree
Master of Arts (M.A.)
Department
Psychology
Program
Applied Social