The Influence of Comma- and Period-pause Duration on the Listener’s Impression of Speeches Made in Mandarin Chinese


The influence of pause duration at commas and periods on listeners’ impressions of speeches made in Mandarin Chinese was investigated. Spoken excerpts of speeches from textbooks were presented to native Chinese listeners (n=20). In the first experiment, the pause durations of both commas and periods in the speeches were manipulated together, in 8 steps from 0 - 4.8 s. The listeners were asked to rate the speeches on 23 categories on a rating scale, including categories regarding the tempo, quality, and continuity of the speeches. Factor analysis (based on principal component analysis) over the rating data showed that out of four extracted factors, two factors prominently appeared. These two factors were interpreted as reflecting speech naturalness and speech rate. The speech rate impressions increased as the comma- and period-pause durations decreased. The speech naturalness was the highest when the pause duration was 0.6 s. In a following experiment, comma- and period-pause durations were manipulated separately, varying from 0.15 - 2.4 s (original speech and speech without pauses were included as control conditions). Factor analysis over the rating data (n=20) again showed speech naturalness and speech rate as the main two factors. Both experiments convincingly indicated that speeches with a comma-pause duration of 0.6 s, along with a period-pause duration of 0.6 s or 1.2 s, are heard as having the highest speech naturalness, i.e., close to that of the original speech. Thus, controlling and limiting pause durations when delivering a speech should be important, and possibly can be practiced.

Author Information
Mingji Lin, Kyushu University, Japan
Yoshitaka Nakajima, Kyushu University, Japan
Shimeng Liu, Kyushu University, Japan
Kazuo Ueda, Kyushu University, Japan
Gerard Bastiaan Remijn, Kyushu University, Japan

Paper Information
Conference: ACL2021
Stream: Language and Communication

This paper is part of the ACL2021 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window

Posted by amp21