Speakers monitor auditory feedback for temporal alignment and linguistically relevant duration.

Recent altered auditory feedback studies suggest that speakers adapt to external perturbations to the duration of syllable nuclei and codas, but there is mixed evidence for adaptation of onsets. This study investigates this asymmetry, testing three hypotheses: (1) onsets adapt only if the perturbation produces a categorical error; (2) previously observed increases in vowel duration stem from feedback delays, rather than adaptation to durational perturbations; (3) gestural coordination between onsets and nuclei prevents independent adaptation of each segment. Word-initial consonant targets received shortening perturbations to approximate a different phoneme (cross-category; VOT of /t/ > /d/; duration of /s/ > /z/) or lengthening perturbations to generate a long version of the same phoneme (within-category; /k/ > [k hh ]; /ʃ/ > [ʃː]). Speakers adapted the duration of both consonants in the cross-category condition; in the within-category condition, only /k/ showed adaptive shortening. Speakers also lengthened all delayed segments while perturbation was active, even when segment duration was not perturbed. Finally, durational changes in syllable onsets and nuclei were not correlated, indicating that speakers can adjust each segment independently. The data suggest that speakers mainly attend to deviations from the predicted timing of motor states but do adjust for durational errors when linguistically relevant.