Search

Final vowel devoicing (e.g. mais oui_hhh, je t'ai vu_hhh, pas du tout_hhh) is a phenomenon in which utterance-final vowels lose their voicing and produce intense, fricative-like whistles. Despite being readily observable in Continental French, little is actually known about what this variable means. Previous work has determined its structural, prosodic and phonological preferences among L1 French speakers (statement-final, low tone, open syllable, high vowel), but none has empirically studied its social or pragmatic predictors, such as Gender, Age, Speech Rate, Register or Affective Stance. This variable also appears to be highly indexical in nature, marking something about the speakers who use it. It has oft been labeled a feature of Parisian French, however, several studies have observed it in samples collected outside of Paris and among non-Parisian Continental Francophones living abroad. Moreover, since a devoiced vowel exhibits high-frequency energy that is easily detectible in the speech signal, it is also frequently remarked upon by L2 French users, many of whom incorporate it into their own French based on their L2 understanding of what it might mean. This presents an interesting point of comparison, since even among L1 French speakers, its value is murky at best. In an attempt to explore these aspects of devoicing, this dissertation will employ a three-part inquiry, calling on methods from phonetics, lab phonology, sociolinguistics and psychology to: (1) examine the linguistic, social and pragmatic predictors of rate and degree of production among L1 French speakers, (2) examine the linguistic, social and pragmatic predictors of the perception of devoicing among L1 speakers (3) examine how L1 hyper-devoicers describe themselves in terms of personality traits, and (4) extend the findings from (1)-(3) to a sample of advanced L2 French speakers. Results reveal different rates but similar degrees of production across speaker groups, influenced largely by locational, phonological and pragmatic factors. Perceptually, it is demonstrated that L1 speakers view this variable as indexing both positive and negative traits, while L2 speakers perceive of it as overwhelmingly positive. I argue that this misalignment occurs because L2 speakers have understood this sociophonetic variable to be a reflex of formal register.