Speech errors and segment duration : an investigation of word-initial/sp, st, sk/-clusters under conditions of rapid repetition

Speech errors, or "slips of the tongue", have been studied in attempts to understand the speech production process, to investigate phonological units and rules, and to provide insights into historical linguistic change. The present study examines speech errors and their relation to segment durations in word-initial /sp, st, sk/-clusters produced under rapid repetition conditions by six adult native speakers of English. Fifty percent of the errors produced could be classified as repetition errors; these were examined for duration in the initial clusters, both error and corrected productions. General results following from analysis of the data were: (1) Error clusters and their component segments were consistently longer in duration than their subsequent and immediate corrections. (2) The clusters /sp/ and /sk/ are longer than /st/, which may be attributable to the faster moving, more highly innervated tongue tip musculature involved in the production of Is/ and It/, compared with the heterorganic clusters. (3) The stop consonant in a given cluster appears to determine the overall cluster duration, since the duration of /s/ remains fairly constant irrespective of context. In light of the results, it was speculated that the excessive duration of the cluster (or of its component parts) violated a timing constraint on the production of an utterance, necessitating recalibration and correction of the error. It was further inferred that feedback must be present in order for the system to recognize the duration error, to compare it with planned output, and finally to execute a correction. Two types of feedback were considered necessary for the adequate functioning of a speech production model, which would also allow for speech perception: (a) continuous auditory feedback, which is supplemented by (b) intermittent proprioceptive feedback, both of which are used in perceiving input ,:and manipulating output. Such a system provides a plausible account of speech error production as described in this study. The hypothesized variable servomonitor system advocated here (and in other studies) in general provides an efficient means for producing, monitoring and correcting speech production.Medicine, Faculty ofAudiology and Speech Sciences, School ofGraduat