Abstract
A critical review of EFL studies (some conducted on Taiwanese learners) indicated English verb-noun (V-N) miscollocation was a dominant lexical weakness among intermediate learners, and explicit teaching plus practice could be useful. Yet, few of the previous studies have tried CALL approaches. With the development of new tools such as a concordancer, web-based technologies may promise to promote EFL learners’ learning of V-N collocation. Five web-based hourly practice sessions were thus designed based on the analysis of major V-N miscollocation data by local learners with two significant features: a sample of four important V-N miscollocated types, and the use of an online bilingual concordancing tool, TotalRecall (developed by a research team named CANDLE-Corpora And NLP for Digital Learning of English). Four types of V-N lexical collocations included synonymous verb pairs, hypernymy and troponymy verb pairs, de-lexicalized verbs, V-N collocations that are non-congruent between Chinese and English. Both deductive and inductive teaching methods were incorporated into the design based on the nature of particular V-N types. Within the CANDLE environment, a web-based tracker was designed to record students’ online learning history. Then, 32 college freshman students of non-English major were recruited to participate in the empirical evaluation part with pretest, posttest, and delayed posttest measures of 36 purposefully sampled items of blank-filling in a sentence context. Additionally, a background questionnaire and an evaluation questionnaire were used to elicit participants’ data and perception about the effectiveness of the CALL approaches. Results of the pretest measures were used to categorize learners into different groups. The online practice lasted for about five weeks. Research questions address the impact of factors such as learners’ prior collocation knowledge, different types of V-N collocation, and different online teaching methods on the effectiveness of collocation learning, besides the overall effect and retention (after two and a half months), and learners’ perception about TotalRecall and the design of the practice sessions.It was found that learners did improve in V-N lexical collocations as demonstrated in the immediate posttest scores. In particular, students at lower collocation levels benefited more from the online practice. Although students regressed significantly after two and a half months, their residual of collocation learning sustained as they performed better in the delayed posttest than in the pretest. Besides, students performed differently in the 4 V-N types as indicated in the three collocation tests. Among four commonly misused V-N collocations, collocations with de-lexicalized verbs and in lack of direct translations in Chinese and English were more receptive to collocation instruction via CALL as can be seen students’ significant progress of these two VN types in the immediate posttest. The residual effects indicated that students performed better in collocations with de-lexicalized verbs and were weakest in collocations involving hypernyms and troponyms. Students’ residual of collocations in lack of direct translations in two languages was slight better than collocations involving hypernyms and troponyms and synonyms though not reaching significant level. With regard to different teaching approaches, the inductive method was relatively more effective in assisting collocation acquisition than the deductive method. Learners progressed significantly with the support of the inductive method based on their performance in the immediate posttest. The inductive method also contributed to learners’ residuals after two and a half months though students regressed significantly more in collocations taught inductively. Students’ responses in the evaluation questionnaire pointed out they perceived collocation instruction via the CALL approaches positively and they wanted to use similar online practice design in future English collocations learning. Although students were used to the deductive method, analysis of three collocation tests showed that they gained more with the inductive method incorporating the use of the online bilingual concordancer.The results of the current study supported that a web-based collocation instruction module is able to assist EFL college students’ collocation acquisition and students endorsed the usefulness and convenience of it. Since students’ performance was different in terms of different collocation types, instructors might need to pay different attention to different V-N collocations in material design. The findings also implied that the inductive approach with a concordancer has the potential to scaffold collocation learning. It is, therefore, crucial to equip learners with appropriate concordancing skills to smooth the pattern-inducing process. Meanwhile, teachers’ proper guidance and carefully planned activities would facilitate inductive learning. Combining new computer tools and online material design with sound pedagogy would benefit language learning. More participants should be included and long-term studies are suggested in order to verify the effects of collocation instruction in future research.