Acoustic-Prosodic Entrainment and Social Behavior

2012 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 11–19, Montr´eal, Canada, June 3-8, 2012. c⃝2012 Association for Computational Linguistics Acoustic-Prosodic Entrainment and Social Behavior Rivka Levitan1, Agust´ın Gravano2, Laura Willson1, ˘Stefan Be˘nu˘s3, Julia Hirschberg1, Ani Nenkova4 1 Dept. of Computer Science, Columbia University, New York, NY 10027, USA 2 Departamento de Computaci´on (FCEyN), Universidad de Buenos Aires, Argentina 3 Constantine the Philosopher University & Institute of Informatics, Slovak Academy of Sciences, Slovakia 4 Dept. of Computer and Information Science, University of Pennsylvania, Philadelphia, PA 19104, USA rlevitan@cs.columbia.edu, gravano@dc.uba.ar, law2142@barnard.edu, sbenus@ukf.sk, julia@cs.columbia.edu, nenkova@seas.upenn.edu Abstract In conversation, speakers have been shown to entrain, or become more similar to each other, in various ways. We measure entrain- ment on eight acoustic features extracted from the speech of subjects playing a cooperative computer game and associate the degree of en- trainment with a number of manually-labeled social variables acquired using Amazon Me- chanical Turk, as well as objective measures of dialogue success. We find that male-female pairs entrain on all features, while male-male pairs entrain only on particular acoustic fea- tures (intensity mean, intensity maximum and syllables per second). We further determine that entrainment is more important to the per- ception of female-male social behavior than it is for same-gender pairs, and it is more impor- tant to the smoothness and flow of male-male dialogue than it is for female-female or mixed- gender pairs. Finally, we find that entrainment is more pronounced when intensity or speak- ing rate is especially high or low. 1 Introduction Entrainment, also termed alignment, adaptation, priming or coordination, is the phenomenon of conversational partners becoming more similar to each other in what they say, how they say it, and other behavioral phenomena. Entrainment has been shown to occur for numerous aspects of spo- ken language, including speakers’ choice of re- ferring expressions (Brennan & Clark, 1996); lin- guistic style (Niederhoffer & Pennebaker, 2002; Danescu-Niculescu-Mizil et al., 2011); syntactic structure (Reitter et al., 2006); speaking rate (Lev- itan & Hirschberg, 2011); acoustic/prosodic fea- tures such as fundamental frequency, intensity, voice quality (Levitan & Hirschberg, 2011); and phonet- ics (Pardo, 2006). Entrainment in many of these dimensions has also been associated with different measures of dialogue success. For example, Chartrand and Bargh (1999) demonstrated that mimicry of posture and behavior led to increased liking between the dialogue par- ticipants as well as a smoother interaction. They also found that naturally empathetic individuals ex- hibited a greater degree of mimicry than did oth- ers. Nenkova et al. (2008) found that entrainment on high-frequency words was correlated with nat- uralness, task success, and coordinated turn-taking behavior. Natale (1975) showed that an individ- ual’s social desirability, or “propensity to act in a social manner,” can predict the degree to which that individual will match her partner’s vocal inten- sity. Levitan et al. (2011) showed that entrainment on backchannel-preceding cues is correlated with shorter latency between turns, fewer interruptions, and a higher degree of task success. …

Content truncated to 500 words