This paper reports on a study of the process of perception and repeatability of Japanese pitch accent by native Tokyo and Kyoto-Osaka Japanese speakers. The results indicate that three factors interact: 1) the absolute (or relative) pitch or the "accentual pitch" of listeners, 2) the transfer of accentual pattern of listeners' native dialect, and 3) the physical features of sound production (in particular pitch). However, the transfer of native dialect is the primary factor: each subject categorically perceives his / her native dialect. These findings suggest that teachers of Japanese as a Second Language (JSL) can not correctly hear learners' interlanguage intonative patterns because of the transfer of native dialect, while learners learn Japanese (Tokyo) accentual patterns to acquire a Japanese accent. It is also important tor learners to acquire "accentual pitch" to discriminate pitch accents. However, each language / dialect seems to have its own discriminative point. I intend to clarify these points by perceptional experiments using synthetic sounds to develop hearing drills based on discriminative points.