Abstract
Detection time for word-initial phonemes is faster when the phoneme is followed by a vowel (as in band) than when the phoneme is part of a cluster (as in brand or bland). We investigated this effect as a function of the model which subjects were given for the target phoneme (e.g., "ßbß as in blue"). Subjects listened for word-initial target phonemes in continuous utterances. The targets occurred in single-phoneme, ßrß cluster or ßlß cluster onsets. Each type of word produced faster responses when the target had been modeled with the same onset. No such effects were found with consonant-vowel onsets; the "matching model" effect is specific to syllabic onsets, and suggests that word-initial clusters are perceived as integral units.
