近年來，隨著資訊工程之自然語言處理技術的進步，多項不同語言之單語乃至雙語或多語語料庫檢索系統已被開發，也能在網路上隨時供學習者檢索以促進外語學習成效。同時，外語單字學習研究也由單一字詞教學朝多字單元方向發展（如搭配詞），因為要能提升外語程度確實需要超越逐字學習階段，語料庫檢索恰能結合數位學習及多字單元學習兩者之利基。本研究探討應用網路「中英雙語語料庫檢索」對大學生學習英文搭配詞中「形容詞─名詞以及動詞─名詞」組合，並運用於寫作上的學習成效。針對前人以中文為母語學習者之英文作文分析結果，本研究選擇64個搭配詞，採用「前測─實驗─後測─延遲後測」設計，以16位修讀大二寫作課程之學生為研究對象。前測後，經過5週使用語料庫TANGO檢索系統教學，並施予引導訓練。教學後立即給予後測，以及2週後的延遲後測。研究發現，學生使用搭配詞的表現在後測及寫作方面和前測相較均有顯著的進步，尤其在低分組最為明顯；同時在延遲後測並無忘失退步。不受中文母語影響之搭配詞，以及動─名之搭配詞顯示進步較多。本研究肯定前人研究中所發現的語料檢索之優點，但建議英語搭配詞及學習者本身因素（如母語或自身搭配詞程度），都值得未來深入探討。

Recently, more corpus tools with monolingual or multilingual data have been developed as digital learning and natural language processing technologies advance. Pedagogy which harnesses the power of concordance programs, or data-driven learning (DDL), promises to encourage collocation learning. Recent vocabulary researchers emphasize that English learners need not only more words but also more phrasal knowledge such as collocations (words combined due to habitual use) to upgrade their proficiency. The current study investigated the application of integrating a Chinese- English concordance program, TANGO, with lookups of adjective-noun and verb-noun collocations into English writing instruction and examined its impact of such invention on college students in Taiwan. The target collocations were chosen from previous studies on L1-Chinese learners’ misuse and relevant literature on collocation lists. With the singlegroup pre-/posttest design, the participants were given a collocation test and a paragraph writing task before the concordancing-enhanced instruction was given to increase their awareness and consolidate productive knowledge of including the target collocations in their writing. Such DDL was found to improve L2 students’ writing performance and test scores through guidance of authentic usage in concordance. The positive effects were maintained two weeks after the treatment. While similar benefi ts that prior research has found on using DDL for L2 writing may explain our results, we found interlingual and intralingual factors as well as learners’ prior collocation performance to exert varying influence on learners’ distinct collocation performance. Based on their pretest performance, the low scorers made more progress. Verb-noun combinations and non-L1 influenced collocations showed more improvement in the posttest, unlike findings in prior research. Limitations, implications, and future research are discussed. We argue that the study contributes useful findings to our existing knowledge to researchers and teachers by confi rming the effectiveness of integrating concordancers for learning English collocations and using them in writing. We also suggest that future research examine differential effects of collocation types (the role of Chinese-English difference and the intralingual factor of the English collocation system itself) on learner improvement.