KANSEI Information Processing for Perception of Vowel Sounds

Chandrajith Ashuboda Marasinghe, Ajith P.Madurapperuama, Stephen G. Lambacher, William L. Martens, Michael Cohen and Susantha Herath


In this paper we investigate the specifications of the KANSEI information processing in perception of American English vowel sounds. A common perceptual space for 10 American English vowel sounds was derived for two groups of listeners, a group of native speakers of the Japanese language, and a group of native speakers of Sinhala, a language of Sri Lanka. The stimuli used in the experiment were the ten vowel sounds synthesized using the often utilized formant frequency values published by Peterson and Barney in 1952. Subsets of these two groups made ratings on 12 KANSEI bipolar adjective scales for the same set of sounds, each of the two groups using anchoring adjectives taken from their native language. Though there was no evidence of any difference between the two groups' in their INDSCAL-derived perceptual dimensions for these vowel sounds, the adjectives were used differently in describing those same perceptual dimensions by the two groups. The results of semantic differential analysis (SDA) support the conclusion the two groups' ratings on 12 KANSEI bipolar adjective scales related somewhat differently to the dimensions of their shared perceptual space, Though a few of the adjectives were used to describe similar perceptual variations, one implication of this investigation is that caution be exercised in generalizing semantic differential ratings obtained in one language, especially when those ratings are intended to aid in the interpretation of data from listeners speaking a different native language.