Previous research has shown that speakers can adapt their speech in a flexible manner as a function of a variety of contextual and task factors. While it is known that speech tasks may play a role in speech motor behavior, it remains to be explored if the manner in which the speaking action is initiated can modify low-level, automatic control of vocal motor action. In this study, the nature (linguistic vs non-linguistic) and modality (auditory vs visual) of the go signal (i.e., the prompts) was manipulated in an otherwise identical vocal production task. Participants were instructed to produce the word "head" when prompted, and the auditory feedback they were receiving was altered by systematically changing the first formants of the vowel /ε/ in real time using a custom signal processing system. Linguistic prompts induced greater corrective behaviors to the acoustic perturbations than non-linguistic prompts. This suggests that the accepted variance for the intended speech sound decreases when external linguistic templates are provided to the speaker. Overall, this result shows that the automatic correction of vocal errors is influenced by flexible, context-dependant mechanisms.