Capacity demands in short-term memory for synthetic and natural word lists

Abstract

Two experiments comparing recall for synthetic and natural lists of monosyllabic words were conducted to determine the locus of previously demonstrated perceptual difficulties for synthetic speech produced by rule [D. B. Pisoni and S. Hunnicutt, ICASSP, 572–575 (1980)]. If the perception of synthetic speech demands increased processing capacity in early encoding stages, recall differences between synthetic and natural speech should arise when short-term memory is differentially stressed. In the first experiment three presentation rates (1, 2, and 5 s per word) were used to manipulate the demands placed on short-term memory. Although recall was consistently poorer for the synthetic lists at all presentation rates, the decrement for synthetic stimuli did not increase with faster rates. A similar pattern of results was obtained in the second experiment in which strings of digits of varying length (0, 3, and 6 digits) were presented visually for retention prior to each spoken word list. However, the recall of the digits was considerably worse for the 6-digit list relative to the 3-digit list when the following word lists were synthetic. These results indicate that at least some of the difficulties observed in the perception and comprehension of continuous synthetic speech are due to increased processing demands in short-term memory. [Supported by grants from NIMH and NINCDS.]