Use of gradient anticipatory nasal coarticulatory cues for lexical perception in French

Abstract

The current study examines anticipatory nasal coarticulation in French, a language which is known to exhibit a) quality differences between phonemically oral and nasal vowels and b) relatively low amounts of nasal coarticulation in CVN contexts. In a production study, thirty native Northern Metropolitan (Parisian) French speakers produced seven sets of CṼ-CVC-CVN words (e.g., [sɛd] cède, [sɛn] scènes, [sæ̃] saint). Consistent with previous studies, results indicated quality differences between the vowels in CṼ versus CVC and CVN words, and also that acoustic nasalization in CVN contexts was relatively small; nevertheless, it was still significantly greater than in CVC contexts, and variable across speakers. In a perception study, the CV portions of the production recordings were played to fifty French listeners, who identified the corresponding word in a forced-choice task. Results showed that stimuli from CVN contexts were highly confusable with CVC items, but not with CṼ items. Most importantly, increased degree of acoustic nasalization on individual CVN stimuli significantly correlated with accuracy. We conclude that, despite the overall relative weakness of coarticulatory cues to nasality, French listeners can nevertheless employ these cues when they are stronger.