| Description |
High-vowel lenition is attested in various forms in a number of languages, including Shoshoni, Lezgian, East Cree, Andean Spanish, and Japanese, along with many others. It is also attested in the development of the various Romance languages from Proto-Romance. High-vowel deletion and devoicing are both attested in Quebec French, with some authors reporting devoicing but no deletion, and others reporting frequent deletion and devoicing. Research indicates that both surrounding consonantal context and sociolinguistic factors contribute to (non)lenition of Quebec French high vowels, with some authors treating deletion and devoicing as separate phenomena and others treating them as different manifestations of the same phenomenon. Few studies have investigated high-vowel lenition in other varieties of French. This study investigates deletion and devoicing of the high-vowel phonemes /i/, /y/, and /u/ in the French spoken in Quebec and Paris, and identifies which phonetic and social factors, including left and right context, vowel phoneme, provenance, gender, and style, best predict these phenomena. It also addressed whether high-vowel deletion and devoicing are different manifestations of a single phenomenon or two separate phenomena in these varieties of French. Data are from recordings of native French speakers from the Phonologie du Francais Contemporain (PFC) corpus project. Each speaker participated in two different interviews representing two levels of style. For each speaker, each interview type, and each high-vowel phoneme, twenty interconsonantal tokens were transcribed and coded as deleted or present, and as voiced or devoiced, along with the surrounding consonantal context. Tokens were subjected to statistical analysis. Despite most expectations, there are no statistical differences between the rates of deletion and devoicing in Quebec and Paris, and neither phenomenon is unique to Quebec French. The best predictors of deletion were place and manner of articulation of surrounding consonants, while the best predictor of devoicing was voiceless surrounding consonants. These results indicate that deletion and devoicing are separate processes. Although not significant at the aggregate level, sociolinguistic factors were significant predictors in more specific models. Deletion and devoicing of French high-vowels are both more complex and more widespread than previous studies have suggested. |
