Abstract
This article is concerned with the application of corpus-derived materials to improve Taiwanese students’ English writing in verbs, sentences and compositions. This study is divided into two parts. Since the previous studies have fewer discussions on verb voice errors and verb usage errors, the first part of this study investigates and analyzes learners’ verb errors and accordingly proposes using large bilingual parallel corpora, Sinorama and Hong Kong Laws, to provide EFL learners with complete bilingual authentic examples, to help them to know how a word is used within real contexts so as to reduce the errors. The second part of this study explores the application of a monolingual corpus of our own compilation, the 20,409 abstracts collected from the web source CiteSeer. Verb BASE ranks as the top two high-frequency verb (occurring 8,768 times) in the corpus. BASE is one of the English verbs predominantly used in the passive voice. However, our survey showed that our participants including both EFL graduates and undergraduates at beginning and low-intermediate levels of proficiency can hardly recognize and produce it. On the other hand, a survey revealed a shortage of verb usage of five lexical verb classes and digressive structures in students’ in-class free compositions. It proposes using the monolingual abstract corpus to help EFL learners improve near-synonym distinction, make sentences with target words, expand EFL learners’ range of accessible academic verbs and improve their writing proficiency. Empirical results have proved that the methods proposed are effective. Besides, it found that different cultural expressing conventions can be self-learned via bilingual parallel corpora data. Dictionary definitions assist in near-synonym distinction, and a writing organizational structure note helps students’ compositions to be more well-organized. The findings of this study have potential to spread to teaching materials and especially textbooks for foreign language writing.