Abstract
The purpose of this study was to represent the phonological space of Japanese phonemes by means of multidimensional scaling and to identify each dimension not only linguistically but also psychologically. Ten subjects (graduate or undergraduate students) were required to read aloud each pair of katakana (a certain type of Japanese phonemic letters), and to rate the auditory similarity between the two items. The similarity responses were analyzed by ALSCAL, and the 3-dimensional-configuration was interpreted mainly on the basis of the rating scores on 20 semantic differential scales. The results revealed that Dimension 1, which correlated with several SD scales, separated seion and dakuon contrastively. On the other hand, Dimension 2, which didn't have significant correlations with any pair-of-adjectives scales, divided bilabials from other sounds, suggesting that this particular articulation might have some psychological reality affecting the auditory images of speech sounds. Furthermore, Dimension 3, which was not amenable to simple phonetic interpretations, proved to be a dimension reflecting psychological factors such as activity and tension.
