Date of Degree


Document Type


Degree Name





Rivka Levitan

Subject Categories

Computational Linguistics


personality recognition, speech, intergroup variability, deceptive speech, self-reported personality, computational linguistics


Automatic Identification of personality in conversational speech has many applications in natural language processing such as leader identification in a meeting, adaptive dialogue systems, and dating websites. However, the widespread acceptance of automatic personality recognition through lexical and vocal characteristics is limited by the variability of error rate in a general purpose model among speakers from different demographic groups. While other work reports accuracy, we explored error rates of automatic personality recognition task using classification models for different genders and native language groups (L1). We also present a statistical experiment showing the influence of gender and L1 on the relation between acoustic-prosodic features and NEO- FFI self-reported personality traits. Our results show the impact of demographic differences on error rate varies considerably while predicting “Big Five” personality traits from speaker’s utterances. This impact can also be observed through differences in the statistical relationship of voice characteristics with each personality inventory. These findings can be used to calibrate existing personality recognition models or to develop new models that are robust to intergroup variability.