Abstract
The phonemic inventory of French coherently exploits the contrast between voiced and unvoiced obstruents. This opposition is distinctive in almost all phonotactic contexts, with two exceptions: i) /z/ does not occur word-initially; ii) the feature [±voice] may be assimilated in accordance with a following obstruent. Quite a different system is found in Swiss German dialects, where pairs of obstruents sharing the same place and manner of articulation are not differentiated through the presence/absence of laryngeal activity, but rather in terms of longer or shorter duration; such an opposition is maintained even word-finally, though some kind of postlexical fortition does occur if two obstruents follow each other. Therefore, one may predict that Swiss German learners will find major difficulties in realizing French voiced obstruents before other obstruents, whereas they might be more prone to achieve voicing in intervocalic contexts. All in all, these hypotheses are confirmed by the acoustic analysis of a corpus of read speech: the overall degree of voicing in the whole data only amounts to 46%; intervocalically, 68% of the obstruents were voiced, whereas in the prepausal position the subjects only obtained a degree of voicing of 6%.
Schmid, Stephan