On the Modes and Mechanisms of Perception of Speech Sounds
Hiroya Fujisaki, Takako Kawashima · The Journal of the Acoustical Society of America · 1970
Previous studies indicated the existence of categorical and continuous modes in the perception of speech sounds. It has been shown, however, that these two modes are not dichotomous, but the degree of categorical response depends largely on various factors such as acoustical properties of stimuli, context, and test procedures [H. Fujisaki and T. Kawashima, Int. Congr. Acoust., 6th, Tokyo (1968), paper B-3-6]. A method is described for the precise measurement and quantitative expression of a subject's ability of discriminating between two speech sounds. The perception of various categories of speech sounds including vowels, semivowels, liquids, voiced stops, and fricatives is studied experimentally and analyzed using stimuli compiled by a digital computer. A mathematical model of the process of speech perception is then presented that relates discrimination to identification, taking into account the reliability of the short-time memories involved in the task of discrimination. The latter is shown to be affected by various stimulus attributes and test conditions, which explains the variability in the degree of categorical response to various speech sounds.