Discriminating spoken words in french: the role of the syllable and the CV phonological skeleton

Many studies on the role of phonological structure in speech segmentation have used the sequence detection task. Yet, this task may be influenced by post-lexical processing or intentional segmentation. In order to avoid this possible bias, we used a new task requiring the same-different classification of two words. All the experiments were run in French. Experiment 1 manipulated syllabic congruency by comparing first syllable matching and non-matching pairs of words (e.g. ba.lai-ba.llon vs. bal.sa-ba.llon). Consistent with the syllable hypothesis, participants were slower to discriminate between words sharing the first syllable. However, only the fastest responders displayed this effect. In Experiment 2, in which first syllable matching words (e.g. pa.lette-pa.tron) had different C-V structure (CVCVC-CVCCV) while first syllable non-matching words (pal.mé-pa.tron) had identical C-V structure (CVCCV), there was no effect at all. In Experiments 3 and 4 (this latter providing a cross-splicing control for possible acoustic artifacts), the procedure pushed listeners to answer rapidly. In both experiments, the fastest responders showed an advantage for pairs sharing CV structure over those sharing their first syllable. The whole set of results suggests that, in the discrimination of spoken words, different types of phonological structure play a role at temporally distinct stages of processing: first, a frame specifying consonant-vowel slots, similar to the phonological skeleton proposed by autosegmental linguistic theory, and, at a later stage, the syllabic structure.SCOPUS: ar.j