Speech perception of temporally reversed syllables by normally hearing adults

The general purpose of this study was to determine the effects of phonological knowledge on performance on a simple task of speech discrimination. If a test can be devised that is independent of the subject's phonological knowledge then it should be possible to measure auditory speech perception capacity in young deaf children before they have developed spoken language skills. In the present study, temporal reversal was used to limit normal subjects' use of phonological knowledge without removing acoustic information.

An auditory, 3-interval, forced-choice (oddity) test of phonetic contrast perception was administered to 5 normally hearing adult speakers of English. Perception of place of articulation was measured in stops and fricatives--both voiced and unvoiced. The test was presented both with natural syllables and with the same syllables reversed in time. In addition, the test was administered with a fixed context and a mixed context (i.e., context changing from trial to trial). The subject's task was always to identify the odd-man-out in each presentation of 3 syllables.

Overall, error rate was significantly higher for temporally-reversed than for natural syllables (6.7% versus 1.7%). The effect was, however, present only in stops (11.7% versus 1.7%) and not in fricatives (0.8% versus 0.8%). Changing from a fixed to a mixed context did not influence the effect of reversal. There were, however, significant differences among subjects in the temporally-reversed condition.

The main effect of reversal supports the hypothesis that performance falls, even on a simple task, when phonetic processing is impeded. The fact that the effect is present only for stops adds further support, because temporal reversal does not affect the essential acoustic properties of fricatives. The presence of inter-subject differences suggests that the ability to switch to an acoustic strategy when phonetic processing is impeded may be affected by aptitude and experience. The fact that error rate was low, even for reversed stops, suggests that a test of this type could be used for assessing speech perception capacity in deaf children with impaired phonology. The procedure is not, however, completely independent of the effects of phonetic processing.