Erroneous articulatory routines: A performance-based model of speech production.

Abstract

Young children who learn a second language (L2) are able to attain native pronunciation norms. However, L2 learners beyond childhood rarely rid themselves of foreign accent. Various hypotheses and models have been offered to explain such age-related differences. Few of these explanations have addressed the issue of perception/production asymmetries. Neufeld's research has demonstrated that some older learners evidence native-like knowledge of phonological distinctions at the perceptual level, while unable to reproduce these distinctions in output. This asymmetry led him to propose his performance-based Pre- and Post-articulatory Verification model. This model assumes that, although native-like phonological representations may exist in the learner's L2 system, last-second morphophonological and phonetic adjustment may not take place because of a developmentally induced shift in focus from low- to high-level linguistic processing, i.e. to content and form. This thesis elaborates upon Neufeld's ideas by centering on articulatory realization of phonetic specifications derived in the ultimate stage of sentence planning. It is suggested that, in order to meet real-time constraints, frequent and well-practiced articulatory sequences are eventually encoded as rapidly accessible routines. These routines are packaged instructions which translate phonetic representations into articulatory goals. This extension of Neufeld's model seeks to explain much of foreign accent in adolescent and adult L2 learners as the result of entrenched erroneous motor routines.