Parallels between self-monitoring for speech errors and identification of the misspoken segments

This paper investigates self-monitoring for speech errors by means of consonant identification in speech fragments excised from speech errors and their correct controls, as obtained in earlier experiments eliciting spoonerisms. Upon elicitation, segmental speech errors had been either not detected, or early detected or late detected and repaired by the speakers. Results show that misidentifications are rare but more frequent for speech errors than for control fragments. Early detected errors have fewer misidentifications than late detected errors. Reaction times for correct identifications betray effects of varying perceptual ambiguity. Early detected errors result in reaction times that are even faster than those of correct controls, while late detected errors have the longest reaction times. We speculate that in early detected errors speech is initiated before conflict with the correct target arises, and that in both early and late detected errors conflict between competing segments has led to detection.