Abstract
Abstract With the development of computer-assisted language learning (CALL), more and more second language (L2) instructors and learners have endorsed corpus-based approaches in classroom teaching and learning. Particularly for L2 productive language learning, data-driven learning (DDL) (Johns, 1991) has been highly recommended because the corpus and the concordancers, as a reference tool, provide authentic language models for L2 learners’ productive learning. To illustrate, a review of twelve studies focusing on the effectiveness of corpus-based activities on learning written language (Yoon, 2011) indicates that if proper training and assistance are provided, learner concordancing can be a viable tool for enhancing the linguistic aspects of L2 writing and even raising learners’ autonomy (Chang & Sun, 2009; Perez-Paredes, et al., 2011; Yoon, 2008). Recently researchers have started to carry out the relevant experiments by incorporating corpus-based teaching materials and activities in L2 writing. For example, viewing that written errors are not made arbitrarily, some researchers capitalized on the common word errors presumably caused by L1 interference in college-level English writing courses (Chen, 2011; Gao, 2011; Liao, 2011). They successfully integrated concordance consultation with writing practices and significantly increased EFL learners’ lexical-grammar abilities, like collocational knowledge by reducing negative language transfer on L2 vocabulary learning. Two important studies which targeted at analyzing Chinese-L1 English learner corpora (Liu, 2002; Yang, 2012) provided valuable insights into the common word errors appearing in learner language. That is, word errors often result from verb-noun miscollocations (Liu, 2002) and the influences of verb transitivity, semantic prosody, and near-synonyms (Yang, 2012). Nonetheless, these studies have not revealed precisely how effectively concordancing can reduce L1-influenced word errors; moreover, viewing that concordance users’ speed of data-mining and information-decoding still need to be accelerated, the present study aimed to investigate the effects of guided concordance use on EFL college students’ productive vocabulary learning and especially focused on its effectiveness on reducing word errors associated with cross-linguistic influence in L2 writing. Three research questions guided the current study:(a) Can EFL college learners’ productive vocabulary performance be enhanced after a concordancing training with guidance? If so, can the effect last after four weeks?(b) Can more L1-influenced errors (commonly caused by L1 interference) be reduced given the guided DDL, compared with errors unrelated with L1 influence?(c) What are the learners’ perceptions of the instructional design for their productivevocabulary learning? To answer these research questions, a single-group pretest-posttest design was carried out on fourteen English-major sophomores in a Reading & Writing course and then a delayed-posttest was held to ensure its sustainability four weeks later. Both the pretest and the posttest1 were identical in vocabulary measures, including 22 multiple-choice items, 10 proofreading items, and 10 Chinese-English translation items, all of which were selective, embedded, and context-dependent in assessing L2 learners’ vocabulary abilities (Read & Chapelle, 2001). Additionally, a paragraph making item was included that asked the subjects to choose any five tested phrases from the previous three parts. The delayed posttest contained half of the number of the posttest for not fatiguing the students. Items in all the three tests contained fifty-two target words with some related to L1-influenced errors and the other error types by consulting findings of the corpus studies (Chen, 2011; Liao, 2011; Liu, 2002; Yang, 2012). As for instructional design, our scaffolding prompts guided the participants to work through pedagogical exercises in class with an online concordancer, TANGO. The look-up procedures and the chosen sentences which contained target phrase items instructed the learners to notice the rules of word patterns, the constraints of English words, and the discrepancy between L1 and L2 from a database, British National Corpus. An evaluation questionnaire with 23 items of 5-point Likert scales and 8 items of open-ended questions was designed to understand the subjects’ perceptions of the effects of the concordancing process. The data collected from the pre-test and the post-test were analyzed by non-parametric Wilcoxon Signed Rank Sum Test to examine whether the students improved with guided concordance use in four weeks, and whether the effects lasted. In addition, the scores of L1-influenced words errors and non-L1-influenced word errors were compared. The participants’ perceptions of this guided concordance-based approach were presented by computing a percentage and means of each item in the evaluation questionnaire. The results showed that our concordance-based approach enhanced the college students’ overall vocabulary and writing abilities and the effects remained after four weeks. The students improved their productive vocabulary in writing by reducing word errors, especially for those influenced by L1. Most students held positive attitude toward the whole design and were willing to continue using it. We merged three features in the instruction design: concordancers as learning tools, scaffolding prompts as guidance, and pedagogical tasks as resources of teaching materials, which may contribute to the effects as shown in the treatment performance, extended assessment, and items related to L1 interference. In conclusion, our concordance-based approach has it merits of reducing EFL learners’ word errors, leading to more accurate and appropriate word use, and finally enhancing L2 learners’ writing quality. Future research may adopt a research design which involves a control group, as it was one limitation of the current study.