The Language of Trauma: A Comparative Analysis Using Natural Language Processing and Traditional Text Analysis Methods

Post traumatic stress disorder (PTSD) is associated with alterations in the recall ofautobiographical memories (AMs) regarding the traumatic event, but methodology of studies investigating traumatic and non-traumatic narratives in the context of PTSD vary. There is little consensus regarding the linguistic differences between different narrative valences and the influence of PTSD-symptomatology outside of clinical populations. We collected 450 narratives regarding traumatic, negative non-traumatic and neutral AMs of 150 healthy participants and their post-traumatic stress measured by the PTSD checklist for DSM-5 (PCL-5). For a comprehensive analysis of differences in linguistic patterns, we examined syntactic, lexical, semantic, and discourse-level features of the narratives. Compared to neutral narratives, traumatic narratives were linguistically richer, more PTSD-congruent in semantic content, less coherent, and were subjectively experienced as more emotionally intense and vivid. Trauma and negative narratives were largely similar in their linguistic and structural features, with differences confined to word count, death-related language, coherence, and semantic alignment with hyperarousal and intrusion symptom domains. Post-traumatic stress did not moderate any of the association between narrative categories and linguistic features, but higher PCL-5 scores predicted use of death-related words and how temporarily close participants felt to the traumatic event.Although future research should replicate these findings in clinical comparison samples, the present study provides a comprehensive attempt to characterize the language of trauma narratives.