Analysis of phone confusion matrices in a manually annotated French-German learner corpus

This paper presents an analysis of the non-native and native pronunciations observed in a phonetically annotated bilingual French-German corpus. After a forced-choice automatic annotation a large part of the corpus was checked and corrected manually on the phone level which allows a detailed comparison of the realized sounds with the expected sounds. The analysis is reported in terms of phone confusion matrices for selected error-prone classes of sounds. It revealed that German learners of French have most problems with obstruents in word-final position whereas French learners of German show complex interferences with the vowel contrasts for length and quality. Finally, the correct pronunciation rate of the sounds, for several phonetic classes, is analyzed with respect to the learner’s level, and compared to native pronunciations. One outcome is that different sound classes show different correct rates over the proficiency levels. For the German data the frequently occurring syllabic [=n] is a prime indicator of the proficiency level.