Speech Prosody 2008

Campinas, Brazil
May 6-9, 2008

A Learner Corpus - ESCCL

Hua Chen (1), Qiufang Wen (2), Aijun Li (3)

(1) School of Foreign Studies, Nantong University;
(2) National Research Center of Foreign Language Education, BFSU;
(3) Institute of Linguistics, Chinese Academy of Social Sciences; China

This paper describes the reason, the method and the implication of compiling a learner corpus - ESCCL (English Speech Corpus of Chinese Learners). As for the main reason, the existing spoken corpora of Chinese EFL learners in China are completely text-based, and not suitable for phonetic analysis because of the poor quality of recording. The subjects at 4 different educational backgrounds were asked to fulfill 2 tasks reading-aloud and topic-based spontaneous dialogue. The recordings were collected from different parts of China and dialectal areas. The annotation system employed in the corpus combines the British system and the American one. The corpus-based research findings have important implications for China's EFL pedagogy, and will be helpful for the improvement of rating rubrics for China's oral English tests.

Full Paper

Bibliographic reference.  Chen, Hua / Wen, Qiufang / Li, Aijun (2008): "A learner corpus - ESCCL", In SP-2008, 155-158.