Abstract

This paper discusses the effects and the theoretical backgrounds of "speech production training through reconstruction of a full message from a text skeleton" proposed by Someya (2011). A pilot study was conducted on 35 non-English-major university students for 15 weeks to examine if the method is applicable to "average" university students. After getting familiarized with texts through shadowing or repeating activities, students were asked to reconstruct the conceptual contents of the text orally with the help of "text skeleton," or a simplified version of an outline of the text, which consists of key-words and -phrases of the target text. Students were required to reconstruct the text making a full use of their meta-linguistic abilities. They were encouraged to paraphrase the original sentences and speak without preparation to promote their online processing. The pilot test results showed improvement in the sentence construction ability of most students. This implies, from the perspective of Levelt's speech production process, the method can be effective in promoting learners' grammatical encoding abilities. Also, the questionnaire conducted on the 15th week indicated a high sense of self-efficacy among students.