Indices prosodiques et phonotactiques pour l'identification de sa langue natale.

International audienceContribution of segmental and prosodic aspects are tested on the identification of French sentences by French listeners. For this purpose synthesis by concatenation of Diphones (“Mbrola”) is used to synthesise sentences of 8 languages (American English, German, Portuguese, Spanish, Chinese, Arabic, Italian and French) with a French set of diphones. We proceed in two steps. First segmental modifications were applied to the sentences so as to progressively reduce the number of phonemes used for synthesis. Results show that listeners are able to identify above the level of chance their own language with less than 10 phonemes used for synthesis. In the second part prosodic degradations were applied to the same sentences in order to neutralize the fo and duration variations' influence of each language. Results were very similar to those of the previous experiment, suggesting that listeners may use identification strategies not relying on prosodic cues