Voicing assimilation in Catalan three-consonant clusters

Abstract

Electroglottographic and acoustic data for Catalan three-consonant clusters composed of a word- and syllable-final C1C2 sequence and an underlyingly voiced C3 in word- and syllable-initial position were collected in order to verify whether a regressive voicing assimilation process operates on all obstruents placed at the C2 and C1 sites. Data reveal the presence of low percentages of vocal fold vibration during final stops and fricatives, C3-dependent regressive voicing effects occurring to some extent during C2 but not during C1, and a more robust voicing interaction involving C1 and C2 and thus consonants placed within the same syllable and the same word. Moreover, except for perceptually and/or aerodynamically conditioned cases, voicing effects were found to be triggered by sonorants rather than by obstruents and the trill /r/ in agreement with the predictions of the Degree of Articulatory Constraint (DAC) model of coarticulation. Acoustic data for three-consonant cluster pairs differing in the underlying C3 voicing status show that speakers may use C2 duration and intensity (but not C1 or preceding vowel duration) as voicing cues. Taken together these results suggest that voicing assimilation in Catalan three-consonant cluster cannot be modeled as a long-distance regressive process, and is strongly dependent on syllable/word affiliation and on the manner and place of articulation characteristics of the consonants in the cluster.