Voicing assimilation in Catalan two-consonant clusters

Abstract

This paper reports electroglottographic (EGG) data for consonant sequences composed of a word final stop or fricative followed by a voiced consonant produced by eight speakers of a Romance language, i.e., Catalan, where these clusters undergo regressive voicing assimilation. Analysis results reveal considerable speaker- and consonant-dependent differences in the temporal period of vocal fold vibration during C1. In agreement with the degree of articulatory constraint (DAC) model of coarticulation, there appears to be a direct relationship between the extent to which consonants allow contextual voicing (voicing coarticulation resistance) and exert voicing effects on other consonants (voicing coarticulation aggressiveness) in a good number of cases; in other cases, however, this prediction does not hold, mainly in fricative+nasal, lateral sequences presumably due to the aerodynamic requirements involved. EGG and acoustic data for two-obstruent cluster pairs where C2 may be underlyingly voiced or voiceless but agrees in place and manner of articulation show that speakers may use not only the temporal extent of vocal fold vibration but also C1 and preceding vowel duration (as well as fricative noise intensity in clusters with C1=/s/) as voicing cues; in particular, segmental duration was found to stay more constant than vocal fold vibration across speakers. In view of this cooccurring relation, it is concluded that regressive voicing assimilation in Catalan may be signaled by vocal fold vibration and segmental duration and intensity acting interactively.