L1-L2 Interference: The Case of Final Devoicing of French Voiced Fricatives in Final Position by German Learners

This work is dealing with a case of L1-L2 interference in language learning. The Germans learning French as a second language frequently produce unvoiced fricatives in word-final position instead of the expected voiced fricatives. We investigated the production of French fricatives for 16 non-native (8 beginnerand 8 advanced-learners) and 8 native speakers, and designed auditory feedback to help them realize the right voicing feature. The productions of all speakers were categorized either as voiced or unvoiced by experts. The same fricatives were also evaluated by non-experts in a perception experiment targeting VCs. We compare the ratings by experts and non-experts with the feature-based analysis. The ratio of locally unvoiced frames in the consonantal segment and also the ratio between consonantal duration and V1 duration were measured. The acoustic cues of neighboring sounds and pitch-based features play a significant role in the voicing judgment. As expected, we found that beginners face more difficulties to produce voiced fricatives than advanced learners. Also, the production becomes easier for the learners, especially for the beginners, if they practice repetition after a native speaker. We use these findings to design and develop feedback via speech analysis/synthesis technique TD-PSOLA using the learner’s own voice.