ABSTRACT
Hewlett's proposals to overcome some of the difficulties noted by Hawkins, and Harris and Cottam, involve the positing of a three-level model of speech control. In introducing this he describes three different types of speech error. Adopting a speaker-oriented approach he offers an analysis of errors 'in terms of the process by which the error was created' (Hewlett, 1985, p. 158). The first type concerned a speaker choosing to utter a sound unit other than that of the target utterance; the second involves the choice of the correct sound unit by the speaker, but the actual realisation is slightly distorted as compared with the target sound. The third type again involves the speaker choosing the correct unit, but the phonetic production involves greater distortion such that the resultant sound is identical (or virtually so) to another phonemic unit.
