Journal of Mixed Methods Research, Ahead of Print.
This study investigates differences in torture-related witness statements during the Khmer Rouge Tribunal. It follows a three-phased sequential mixed methods design to identify disparities between testimonies of former detainees and interrogators and to examine how different methods complement each other for a comprehensive perspective on witness accounts. This includes training a natural language processing (NLP) model, sentiment analysis (SA), and qualitative content analysis (QCA). The qualitative and NLP-based analyses showed apparent differences between witness groups; a significant difference in sentiment values could not be detected. This study presents the first mixed methods approach based on court transcripts in genocide research. Its digital approach contributes to mixed methods research (MMR) by showing how NLP and data transformation can contribute to integration.