Authors:

Mcqueen, J., Cutler, A. & NORRIS, D.

Reference:

Year of publication:

1999

CBU number:

3976

Abstract:

Dutch listeners categorized a /ta/-/da/ continuum before adaptation, and again during an adaptation phase in which they heard multiple repetitions of five different words. Dutch words, like English words [A. G. Samuel, Cog. Psych. 32, 97-127 (1997)], produced reliable adaptation: there were more /t/ decisions when each adapting word contained a /d/ (and no /t/'s) than when each word contained a /t/ (and no /d/'s). But when these phonemes were replaced with signal-correlated noise, such that the phonemic restoration illusion was induced, there was no adaptation, contrary to the English results. Thus, although listeners heard the replaced phonemes in the noise, these phonemic percepts did not cause adaptation. These results are consistent with the Merge model [Norris et al., Behav. Brain Sci. (in press)], in which the lexicon can influence phonemic decisions (and hence the restoration illusion), but, because there is no feedback from lexical to prelexical levels, cannot induce acoustic-phonetic adaptation. Samuel (1997) claimed on the basis of his results that the lexicon can produce potent phonemic percepts, that is, percepts which can modulate prelexical processing; the present results challenge this claim.