What a spoken learner corpus tells us: construction and application of a pronunciation programme for English-language teachers

Abstract

The linguistic backgrounds of English learners in Hong Kong are highly complex and diverse. There are local Hong Kong learners who speak Cantonese as their native language, new immigrants and students from different dialectal regions in mainland China and non-Chinese learners from the Philippines and Pakistan, among others. To optimise the effectiveness of English pronunciation teaching with such a diverse student body, it is of paramount importance to provide Hong Kong teachers with a spoken learner corpus. This paper introduces a self-developed spoken learner corpus and shows how it can be used in the design of pedagogic applications. The establishment of a learner corpus can help teachers understand the recurrent pronunciation difficulties encountered by learners of particular language backgrounds so that they can customise their teaching materials. To promote pronunciation teaching using this corpus, a teacher training programme was designed and conducted. Pre-service and in-service teachers participated in three face-to-face and real-time streaming workshop sessions, two online sessions, and a lesson plan design competition. Most of the participants appreciated the authenticity of the learner speech data, the diversity of speaker language backgrounds and the effectiveness of using the corpus as a teaching tool. Nonetheless, some suggested that the diversity of reading materials for eliciting speech data could be improved and that more dialect subgroups could be included in the corpus.