The pronunciation of Italian voiced obstruents by Swiss German learners

Abstract

This study examines the pronunciation of voiced obstruents in Italian as a second language of Swiss German learners. On the basis of a contrastive analysis and the predictions of current models of L2 phonology, we claim that Swiss German learners ‘assimilate’ the voiced obstruents of Italian to the ‘lax’ (but unvoiced) sounds of their L1, and that devoicing is particularly frequent in the case of /dz/, /dʒ/ and /z/+C clusters. These hypotheses are borne out by an acoustic analysis of a corpus of read speech gathered from 10 high school students in Zurich, from which we calculated the locally unvoiced frames of 644 tokens. We conclude that L2 pronunciation is determined both by the ‘perceptual filter’ of the L1 and by general patterns of markedness.