David B. Pisoni
Speech Res. Lab., Dept. of Psychol., Indiana Univ., Bloomington, IN 47405
A review of previous research on the perceptual evaluation of synthetic speech produced by rule reveals that most studies have focused on behavioral measures of segmental intelligibility of isolated words. For the most part, except for a few isolated studies, research on sentence processing and comprehension of long passages of fluent speech has received very little attention. This is not surprising because these are areas of research in spoken language processing that are difficult to study without the investment of time and energy in developing new methodologies that go well beyond phoneme perception and isolated word recognition. Studies of prosody and the role of naturalness in intelligibility and comprehension are needed at this time to quantify the contribution of these sources of knowledge to spoken language processing. In this presentation, some of the recent studies from this laboratory that have attempted to measure these more complex processes will be reviewed. Studies using a sentence verification task and several on-line measures of comprehension will be described. Other methods need to be developed in the future to study intelligibility, comprehension, and retention of synthetic speech. [Work supported by NIH Research Grant No. DC00111-17.]