In this paper, we will introduce a hypothetical cognitive model of Japanese speech and try to examine the hypothetical model by a cognitive experiment. In the model, a cognitive unit is segmented from acoustic signal using only prosody and then the segmented unit is matched against the dictionary in the listener's brain. We performed a cognitive experiment where the subjects are asked to segment meaningless Japanese speech which is pronounced imitating meaningful speech. The experimental result suggested that segmented units are almost the same as accentual phrases. From the result, it is suggested that the cognitive unit can be detected by only prosody and it could be the unit for matching against the dictionary. This observation is consistent with our hypothetical model.
展开▼