The purpose of this paper is to observe how speech with gradual changes in rate is perceived. I created 27 stimuli in Japanese to solve this question. They consisted of three kinds of speech rates (fast, normal, and slow) and three kinds of ways of reading (even, acceleration, and deceleration). Utilizing these stimuli, I conducted two experiments on 10 Japanese native speakers.

In the first experiment, the styles of reading were found to affect judgments of speech rate on the stimuli.

In the other, the judgments were found not to be fixed, when the subjects listened to the stimuli with changes in rate.

The purpose of this study is a research of time-feature of HAIKU by acoustic analysis. Analysis data was measured each phrase time length and each pause time length by waveform and sound spectrogram. It is thought a pause is inserted between each phrase of the 5-7-5 syllable HAIKU. However, 14 Japanese speaker's data clarified a pause insertion is a tendency for which it depends on the second phrase's clause structure and individual difference. Moreover, if the pause time length just behind each phrase is also included in each phrase time length, a common feature appears between speakers; in order to stabilize the total time length of the first phrase (include the next pause) and the second phrase (include the next pause), they become longer or shorter.

This study was carried out to determine the characteristics of the formant frequency of vowels used in the recitation of Kyogi Karuta (Competitive karuta). All vowels within the first unit of Ogura Hyakunin Isshu (One hundred poets, one poem each) were considered as the sample and subjected to acoustic phonetic analysis when delivered by a reciter at All Japan Karuta Association. As a result, we confirmed that F1 for the recitation of Kyogi Karuta was grouped at the centre of the formant diagram compared to normal speech. Moreover, we confirmed that F1 was grouped even closer to the centre of the formant diagram after the second unit. This research showed that as compared to normal speech, the formant frequency of vowels in the recitation of Kyogi Karuta is centralized, which was demonstrated by F1 and F3 more clearly than F2.