Production and Detection of Speech Errors in Silent, Mouthed, Noise-Masked, and Normal Auditory Feedback Speech

Abstract

In this study subjects had to report their errors during the speeded production of tongue twister sentences in one of four speech conditions: silent, mouthed, noise-masked, and normal auditory feedback speech. In contrast to the other three conditions, silent speech comprises speech planning but no articulation. Error monitoring in the normal auditory feedback condition may occur both by means of an inner speech (prearticulatory) loop and by means of auditory feedback, whereas in the other conditions only the first channel is available. The results showed that reported error rates were roughly equal in the silent, mouthed, and noise-masked condition, with an increase in the normal auditory feedback condition. Significantly more phonemic-sized errors and disfluencies were reported with auditory feedback, whereas word errors were less frequent. Notwithstanding the differences with respect to error size, report rates for the individual error categories (e.g. anticipations, perseverations, substitutions, etc.) did not differ notably for the four conditions. Errors typically occurred at the same points across speech conditions. These results suggest that speech planning processes are similar in the four speech conditions. Moreover, actual motor execution (i.e. articulation) does not appear to be an important contributor to the error events under study. The main difference between conditions can be attributed to the available monitoring channels.