VISUALIZING DISCURSIVE CONFLICT THROUGH PHONETIC EXPRESSION: A STUDY BASED ON TWO CONVERSATIONS

Abstract

The current analysis examines phonetic variation in conflictive conversations and, drawing from two spontaneous conversations from the ESPRINT_VLC corpus, utilizing pitch and the direction of exchanges as indicators, presents two visualization methods. Scatter plots with Z-Scores allow for tracking phonetic anomalies over time, identifying clusters of high or low tones and correlating them with significant conversational events. Network diagrams illustrate the interaction among speakers, including the tendency of each participant to modify their pitch and the duration of their turns, providing a "radiography" of the conversations. Thus, the proposed methodology has applications in conflict therapy and intercultural communication, as it enables the detection of conversational and interactive patterns, such as 'pursuer' and 'withdrawer', which are related to negative interactional dynamics.