EFFECTS OF INPUT IN THE ACQUISITION OF FORMULAIC SEQUENCES BY EFL LEARNERS IN TAIWAN

• Main Authors: Hui-Ju Chen, 陳惠如
• Other Authors: David Wible
• Format: Others
• Language: en_US
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/21904397218007996793

博士 === 淡江大學 === 英文學系博士班 === 95 === With the advances in computer technology, the importance of formulaic sequences has recently been recognized in the field of language learning and acquisition. In particular, the application of corpus studies on quantities of natural text has reinforced the theory that much of our language output is created in the form of multi-word chunks, stored and produced as such. However, experimental studies in exploring the use of formulaic sequences by second/foreign language learners still remain few and unrefined and little pedagogical progress has been made when it comes to the application in the ESL/EFL classroom. This quasi-experimental study intends to investigate the acquisition of formulaic expressions by 182 EFL students in Taiwan through applying a web-based EFL language learning platform (IWiLL), with corpus-based data used as input. Based on the concept of data-driven learning, the corpus-based input was designed into randomly-arranged and arranged patterns. Treatment procedures included an English proficiency test, a pretest, an immediate posttest and a delayed posttest on Cloze and Translation tests. MANCOVA, ANOVA, Chi-square, and Multiple Linear Regression tests were the major statistical techniques used in getting the quantitative data. Qualitative data were the self-reported reflective feedback collected from the discussion boards on the online learning platform. Specifically, this study addresses three issues: (a) Will learners consulting corpus-based input lead to better understanding and use of formulaic sequences? (b) Will learners who receive input in a randomly arranged pattern perform differently from those who receive input in an arranged pattern (c) What linguistic, affective and cognitive factors may or will facilitate the acquisition and the production of formulaic chunks? The results show that: (a) Exposure to corpus-based input treatment, both randomly-arranged and arranged ones, had significant impacts on learners'' acquisition compared to those of the control group, immediately after the treatment and last for two months for Cloze test. (b) Processing corpus-based input through a data-driven learning task proved to be an efficient way of helping learners to internalize the target chunks. This has been validated by the results that learners under the randomly-arranged treatment constantly gain improvement on the translation test scores and spend decreasing time on the task. (c) The advantages of processing corpus-based input through the randomly-arranged treatment were more evident in the production of longer fixed lexical chunks, which suggests that learning through corpus data does not have to be arranged in a series and presented in a clear and neat concordancing patterns. (d) No significant association was found between English proficiency and the acquisition of formulaic sequences. Other variables like good control of computer literacy had small but positive effects on the acquisition of formulaic sequences through corpus-based input. The dissertation closes by exploring the implication of this study for discovery learning, and for teaching English formulaic sequences in EFL classroom settings.