ISCA Workshop on Plasticity in Speech Perception (PSP2005)

Senate House, London, UK
June 15-17, 2005

The Role of Temporal Information in Vowel Perceptual Areas: A Cross-Linguistic Study

Christine Meunier, Robert Espesser, Cheryl Frenck-Mestre

Laboratoire Parole et Langage, Aix-en-Provence, France

Each language has its own more or less fixed inventory of phonemic units. Given this, each language differs as concerns the number of phonemes in its inventory. The study performed by Manuel & Krakow (1984) showed that the tolerance for variation in the production of a vowel is lesser in a language with a filled vocalic system. On the other hand, Maddieson & Wright (1991) observed very few variation in a low density system (three vowels). As concerns perception, the majority of studies have dealt with the more global problem of perceptual assimilation rather than with the specific issue of density. Flege and Munro (1994) argue in favor of a universal perceptual process, based upon a purely auditory component which is independent of any given phonological system. In recent studies (Meunier et al., 2003 & 2004), we observed that American English listeners had great difficulty in a vowel identification task, while French and Spanish had not. It has been suggested that English listeners could not find the cues they need to take their decision: French and Spanish system are mainly characterized by F1 and F2 values while, in English system, stress and duration are also relevant. In the present study, we compared the categorisation ability of English and French listeners. Three continua of F1 and F2 values have been synthesized (/i/-/a/, /a/-/u/, and /u/-/i/). Each stimulus was presented with two durations: short (200ms) and long (400ms). Stimuli were presented to French and English subjects which made and identification task (Reaction Times were measured). We observed that the perception areas of French and English are similar: a very large category for /a/ and smaller for /i/ and /u/. /i/ and /u/ showed abrupt boundaries while /a/ one were larger. For both languages, RT increased at the boundary of the categories, and were longer for /a/. This reflect that /a/ seem to be a vowel with fuzzy boundaries for both languages. For English subjects (but not for French ones) /a/ responses were different according as the vowel is short or long. To summarize, the size of English and French perceptual areas seem to be similar. We have to confirm, with a low density system, that this similarity can be due to density. The experiment have also pointed out that, for English listeners, duration change the boundary, and so the size, of vowel categories.

Bibliographic reference.  Meunier, Christine / Espesser, Robert / Frenck-Mestre, Cheryl (2005): "The role of temporal information in vowel perceptual areas: a cross-linguistic study", In PSP2005, 42 (abstract).