陳浩然Hao-jan Chen賴淑麗Shu-li Lai2019-09-032010-7-232019-09-032010http://etds.lib.ntnu.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=id=%22GN0890210051%22.&%22.id.&http://rportal.lib.ntnu.edu.tw:80/handle/20.500.12235/97868許多的研究已經證實語料檢索工具（corpus tools）有助於外語寫作。然而，學生們在寫作過程中，會如何與這些工具互動，所知並不多。為了更了解這些過程，本研究探討台灣大學生使用線上語料檢索工具和字典時的行為，以及他們對這些寫作工具的看法。 有十四位非英語系大學生參與為期十七週的研究。本研究提供學生四種線上語料檢索系統，包含單語、雙語「單字檢索系統」（concordancer）以及單語、雙語「搭配詞檢索系統」（collocation retrieval system），同時也給予線上字典供學生寫作時使用。最初的兩週用來幫助學生熟悉這四個語料庫檢索工具，並提供實際上機練習。接下來的三個月，學生在課堂上寫三次計時寫作。所有的寫作過程，都由螢幕錄影軟體錄下。學生並在每次的寫作活動後，接受一對一的刺激回憶訪談（stimulus recall interviews）和半開放性訪談（semi-structured interviews）。四十二次的回憶訪談是本研究最主要的資料分析來源。其他資料來源包括寫作過程的影像檔，學生的工具使用日誌，作文，及研究者的觀察筆記。 本研究發現，學生最依賴雙語字典。在所提供的四個語料檢索工具中，學生認為雙語單字檢索系統幫助最大。從學生的查詢行為中發現，語料檢索工具和雙語字典的功能不同。當需要查詢字義（word meaning）和字的結構（word form）時，學生偏好雙語字典。當尋找有關字的用法（word usage），搭配詞（collocation information），或文法規則（grammar pattern）時，學生選用語料庫檢索工具的頻率高於雙語字典。此外，當字典提供的字義不清楚時，要查詢的中文是有關台灣當地的人、事、物（local Taiwanese referents）或是字串（word strings）時，以及當學生心中已經有想法，只是要做確認時，學生也會使用語料庫檢索工具查詢字義和字的結構。 研究結果也顯示，學生的語料檢索行為有其策略性，而且是目標導向的。學生在工具的選擇與使用、語料的分析、歸納、以及應用，都使用了策略。此外，學生們均認為作為寫作工具，這些語料檢索工具與傳統雙語字典有互補作用。語料庫檢索工具提供大量的例句，補足了雙語字典這方面的缺失。學生也提到單語檢索工具凸顯關鍵字的表現方式，讓他們有意外的學習（incidental learning）。有了這些工具的幫助，學生們對英文寫作變的比較有信心。 本研究同時也發現學生檢索語料時所遇到的困難，包括無法從索引句子中看出需要的資訊，做了錯誤的分析及歸納，以及錯誤的規則套用。資料顯示，這些問題多源自於語言能力的不足或受到先備知識（prior knowledge）的誤導。 透過實地了解學生的語料檢索行為，以及字典的使用，本研究提供了豐富的實證資料，幫助了解外語寫作者在寫作過程如何與這些工具互動，以及這些工具對外語寫作提供的協助為何。這些發現進而提供了語言教學以及未來相關領域研究的方向。Many studies have confirmed the benefits of using corpus tools in EFL writing but little is known concerning how EFL writers interact with these tools during the writing process. To better understand this question, this study aimed to investigate Taiwanese college students’ consultation behavior while using the corpus tools and the dictionaries during the writing process, and their perception of these tools as writing references. Fourteen non-English majors participated in this 17-week study. Four online corpus tools, including monolingual and bilingual concordancers and collocation retrieval systems, were provided along with two online dictionaries. The study began with two tool-training sections. In the following three months, the students performed three timed writing tasks online and received individual stimulus recall interviews after each writing task. The 42 recall interviews served as the main source of data for this study. Other data included video clips of the writing process, student tool logs, student writing samples, and the researcher’s notes. The results showed that the bilingual dictionary was the most relied-upon tool overall, and students considered the bilingual concordancer as the most helpful corpus tool. As their behaviors indicated, the students used corpus tools and the bilingual dictionary for different purposes. They tended to use a bilingual dictionary when information on word form and word meaning was needed. When searching for information related to word usage, collocation information, and grammar patterns, they chose corpus tools more often than the bilingual dictionary. Furthermore, they turned to corpus tools when the Yahoo dictionary failed to provide clear word meanings, when they needed English equivalents related to local Taiwanese referents or for a string of keywords, and when they needed to confirm an intuition regarding word form and word meaning. This study also found that the students’ use of corpus tools was goal-oriented and strategic, as could be seen in the process of selecting tools, analyzing concordance lines, generating rules, and transferring rules to the writing context. In general, the students perceived the corpus tools as a complement to a conventional bilingual dictionary. These corpus tools provided rich example sentences that compensated for the lack of such information in bilingual dictionaries. In addition, the students reported incidental learning as a consequence of the salient presentation of word patterns provided by the monolingual concordancer. With the support of corpus tools, the students also reported that they became more confident in their English writing. Various difficulties were also identified, such as failing to see word patterns in the concordance lines, generating incorrect rules, and making incorrect transfers. These problems were caused mostly by the students’ limited language proficiency and inappropriate prior knowledge of the foreign language. Overall, by examining the students’ consultation behavior in the three writing tasks, this study provides rich on-site data that helps to better understand how EFL writers interact with corpus tools and how these tools contribute to EFL writing. The findings also provide implications for instruction and direction to go for future research in relevant fields.語料檢索工具搭配詞檢索系統查詢行為刺激回憶訪談外語寫作corpus toolscollocation retrieval systemconsultation behaviorstimulus recall interviewEFL writing