The effects of noise masking and required accuracy on speech errors, disfluencies, and self-repairs.

The covert repair hypothesis views disfluencies as by-products of covert self-repairs applied to internal speech errors. To test this hypothesis we examined effects of noise masking and accuracy emphasis on speech error, disfluency, and self-repair rates. Noise reduced the numbers of disfluencies and self-repairs but did not affect speech error rates significantly. With accuracy emphasis, speech error rates decreased considerably, but disfluency and self-repair rates did not. With respect to these findings, it is argued that subjects monitor errors with less scrutiny under noise and when accuracy of speaking is unimportant. Consequently, covert and overt repair tendencies drop, a fact that is reflected by changes in disfluency and self-repair rates relative to speech error rates. Self-repair occurrence may be additionally reduced under noise because the information available for error detection--that is, the auditory signal--has also decreased. A qualitative analysis of self-repair patterns revealed that phonemic errors were usually repaired immediately after their intrusion.