In an age of rapid social and technological change, digital trace analysis is crucial for law enforcement and intelligence agencies. The massive volume, variety, and velocity of digital data pose a significant challenge in identifying relevant and unsuitable content. This study addresses these challenges by employing a lexicon-based classification to distinguish violent jihadist discourses from non-violent Salafist discourses. Using open-source transcriptions of jihadist and Salafist videos originally produced, translated or subtitled in English, the study identifies linguistic themes through the LIWC Grievance dictionary. Key findings indicate that Impostor, Soldier, and Weaponry themes are statistically discriminant for all violent jihadist discourses. Furthermore, group-specific markers emerged, such as Deadline for al-Qaeda, and Suicide, Surveillance, and Violence for the Islamic State. While these tools offer significant potential for large-scale data processing, the study emphasises the necessity of qualitative contextual analysis to mitigate algorithmic biases and false positives.
Photo credit: federico stevanin/Shutterstock.com
