Third International Conference on Spoken Language Processing (ICSLP 94)
To investigate the relation between a scene analytic process and a schema-based process in the perception of speech duration, the perceptual duration of restored speech segments whose duration was manipulated experimentally was measured in psychophysical experiments. In Experiment 1, the vowel /i/in the 3rd mora of the Japanese word, /shimağbiğru/f was modified and replaced by a noise burst. The duration of the noise-replaced speech was perceived to be significantly shorter than that of the intact speech. Besides the effect of acoustic interruption, a significant effect of the durational modification was observed. However, there was no clear evidence that these two main factors interacted with each other. This suggests that a scene analytic process responsible for handling acoustic discontinuity is independent of the speech module responsible for the anchoring to the speech template. In Experiments 2 and 3,. a wider continuum of the standard stimulus, covering the Japanese word, /da*i*chi/, and another word, /daĞi#iğchi/, was tested to investigate whether the cyclic pattern of the underestimation and overestimation could be observed owing to the double anchoring points. However, no clear cyclic pattern was found. This indicates that the anchoring effects by these two points, /daichi/ and /daiichi/, were not equivalent. Although no independent relation between the acoustic interruption and the durational modification was replicated because of the effect of presentation order, when the standard stimulus did precede the comparison stimulus, the noise replaced speech was perceived to be shorter than the intact one in most cases.
Bibliographic reference. Tsuzaki, Minoru / Kato, Hiroaki / Tanaka, Masako (1994): "Effects of acoustic discontinuity and phonemic deviation on the apparent duration of speech segments", In ICSLP-1994, 1971-1974.