Abstract
This study investigated the acquisition of #sC and non-sC clusters by 45 Dutch children between two and three of age. The main focus of the study was to test whether the Sonority Sequencing Principle can explain correct realisations and reduction strategies in seven different #sC clusters. Target clusters were elicited in meaningful words using a picture-naming game. Results show that non-sC clusters are more accurate than #sC clusters. Within the #sC clusters, /sl/ is produced correctly most often whereas /sk/ is the most difficult cluster. In /s/ + obstruent clusters, rendition patterns obey the Sonority Sequencing Principle. This is not the case for /s/ + sononant clusters, in which the more sonorous element is retained. The rendition patterns are best explained by an interaction between sonority and the prosodic head of the cluster. An exception is /sx/, in which /s/ is retained, a pattern that follows neither Sonority nor Headedness principles.
Keywords: