The UTEP-ICT Cross-Cultural Multiparty Multimodal Dialog Corpus

Abstract

To help answer questions about conversational control behaviors across cultures, a collaborative team from the University of Texas at El Paso and the Institute for Creative Technologies collected and partially coded approximately ten hours of audiovisual multiparty interactions in three different cultures and languages. Groups of four native speakers of Arabic, American English and Mexican Spanish completed five tasks and were recorded from six angles. Excerpts of four of the tasks were coded for proxemics, gaze, and turn-taking; interrater reliability had a Kappa score of about 0.8. Lessons learned from the multiparty corpus are being applied to the recording and annotation of a complementary dyadic corpus. 1.