We explore deception detection in interview dialogues. We analyze a set of linguistic features in both truthful and deceptive responses to interview questions. We also study the perception of deception, identifying characteristics of statements that are perceived as truthful or deceptive by interviewers. Our analysis show significant differences between truthful and deceptive question responses, as well as variations in deception patterns across gender and native language. This analysis motivated our selection of features for machine learning experiments aimed at classifying globally deceptive speech. Our best classification performance is 72.74 F1-Score (about 27% better than human performance), which is achieved using a combination of linguistic features and individual traits.

We analyze a set of acoustic-prosodic features in both truthful and deceptive responses to interview questions, identifying differences between truthful and deceptive speech. We also study the perception of deception, identifying acousticprosodic characteristics of speech that is perceived as truthful or deceptive by interviewers. In addition to studying differences across all speakers, we identify variations in deception production and perception across gender and native language. We conduct machine learning classification experiments aimed at distinguishing between truthful and deceptive speech, using acoustic-prosodic features. We also explore methods of leveraging individual traits for deception classification. Our results show that acoustic-prosodic features are highly effective at classifying deceptive speech. Our best classifier achieved an F1score of 72.77, well above both the random baseline and above human performance at this task. This work advances our understanding of deception production and perception, and has implications for automatic deception detection and the development of synthesized speech that is trustworthy.

Collecting spontaneous speech corpora that are open-ended, yet topically constrained, is increasingly popular for research in spoken dialogue systems and speaker state, inter alia. Typically, these corpora are labeled by human annotators, either in the lab or through crowdsourcing; however, this is cumbersome and time-consuming for large corpora. We present four different approaches to automatically tagging a corpus when general topics of the conversations are known. We develop these approaches on the Columbia X-Cultural Deception corpus and find accuracy that significantly exceeds the baseline. Finally, we conduct a cross-corpus evaluation by testing the best performing approach on the Columbia/SRI/Colorado corpus.