The locus of the lexical shift in phoneme identification.

The lexical identification shift is used as a measure of speech processing in the phoneme identification task (W. F. Ganong, 1980). Interactive (bottom-up and top-down) models of word recognition account for the shift by claiming that lexical knowledge feeds back to a prelexical level and aids speech processing. Autonomous models (bottom-up only) maintain that the shift arises by other means and at later stages of processing. The locus of the lexical shift was investigated by using detection theory analysis procedures to measure perceptual changes in phoneme processing. Lexical status (word-nonword) of the utterance was varied in Experiments 1 and 3 and was found to influence phoneme processing. In Experiment 2 the effects of a postperceptual manipulation, monetary payoff, did not show up in the detection theory analysis. Implications of the results for both classes of models are discussed.