In this paper, we suggest methods to trace the flow of the use of filled pauses over the course of the conversation. Beyond using a normalized time with respect to each session's duration, we calculated the accumulative number of filled pauses and the accumulative number of words per speaker, whenever the speaker has expressed a filled pause or a new word. We then computed the ratio between these values at each such point in time, resulting in relative filled pauses use. The output produces a visualization of the global contour slopes that represents each speaker and the dynamics between the two speakers, in terms of the relative filled pauses use. The dialogues are taken from MaTaCop, the Hebrew map-task corpus, in which each speaker participated twice, once as a leader and once as a follower. Finding suggests that relative filled pauses use differs for the same speaker in different roles and that there are significant differences between participants who began as leaders versus those who began as followers. Moreover, the use of filled pauses shows convergence. These finding strengthen previous studies on the influence of sociolinguistic variables on the use of FPs.
Cite as: Silber-Varod, V., Amit, D., Lerner, A. (2020) Tracing changes over the course of the conversation: A case study on filled pauses rates. Proc. Speech Prosody 2020, 754-758, doi: 10.21437/SpeechProsody.2020-154
@inproceedings{silbervarod20_speechprosody,
author={Vered Silber-Varod and Daphna Amit and Anat Lerner},
title={{Tracing changes over the course of the conversation: A case study on filled pauses rates}},
year=2020,
booktitle={Proc. Speech Prosody 2020},
pages={754--758},
doi={10.21437/SpeechProsody.2020-154}
}