Priming effects in speech and nonspeech modes of perception

Abstract

A phonemic perception was assumed to be a process in which representation of speech signals is matched to phonemic representation in long-term memory. Since stimulation is supposed to activate relevant memory representation, priming effects were compared between speech sounds and their sinewave analogs, either in the speech mode or nonspeech mode of perception. An analog target imitating syllables /ba/ or /ga/ in Session 1 or their original speech target in Session 2 was preceded by one of 5 primers, which were these 4 target sounds, and a 1,000 Hz pure tone, with one of 6 interstimulus intervals (ISIs). In Session 1, half of the participants were asked to identify analogs as ill-synthesized syllables, "BA" or "GA," and the other half were asked to label them as computer sounds, named "A" or "B." In Session 2, all of them identified original syllables. Regardless of target types and ISIs, subjects given speech instructions showed faster identification of targets preceded by phonemically identical primers, while subjects given nonspeech instructions identified targets faster only when targets were preceded by physically identical primers. These results support that speech sounds and their analogs are accessible to common memory representation.