Proceedings of the Annual Meeting of the Cognitive Science Society

The assessment of how a deceptive message is produced in dif-ferent languages has received little attention, with the majorityof studies focused on the English language. Moreover, thereis no agreement about the stability of linguistic clues of deceitacross different languages. In this paper, we address this issueby analysing both theory-driven linguistic markers of decep-tion (cognitive load hypothesis) and standard text categorisa-tion features. After compiling a multilingual corpus of bothhonest and deceitful first-person opinions regarding five differ-ent topics, we assessed the cross-language applicability of fourdifferent features sets in within-topic, cross-topic and cross-language binary classification experiments. Results showedpromising classification performances in all the three experi-ments with few exceptions. Interestingly, linguistic markersof deceit linked to the cognitive load hypothesis exhibited thesame trend in the two languages under investigation and thecross-language evaluation highlighted their usefulness in spot-ting deceit between different languages.