Temporal aspects of self-monitoring for speech errors

We report two four-word tongue twister experiments eliciting consonantal errors and their repairs, in word initial and medial positions, testing some predictions relating to temporal aspects of self-monitoring. Main findings: (1) After internal error detection interrupting the speaking process takes more time than speech in- itiation of the error form. This implies that “covert repairs” are rare. (2) Word onset-to-cutoff times are longer for medial than for initial errors. This implies that scanning internal word forms for errors takes time. (3) Cutoff-to-repair times of 0 ms are overrepresented. This shows that often repairs are available at interruption. (4) Cutoff- to-repair times are longer for medial than initial consonants. This shows that repairing takes more time for medial than for initial errors. (5) Detection rate decreases from early to late within word forms. Temporal aspects of self-monitoring suggest time-consuming scanning of internal word forms, strategic postponement of interruption, and variations of selective attention.