本研究探討間接評語(indirect feedback)之明確度對英語學習者搭配字(collocations)習得的成效。此研究之範圍以動詞-名詞搭配字(verb-noun collocations)為主。 研究目的主要在探討各類評語對學習者寫作上的影響，另外，也針對學生評語理解 (comprehensibility)與否與其立即錯誤訂正(immediate error revision)的相關性做一探討。因此，研究問題有四：(1)就詞彙學習而言，批改學生錯誤對寫作是否有幫助? (2)不同的評語內容對學生搭配字的使用情況是否造成差異? (3)評語內容的明確度對學生立即的錯誤訂正是否有影響? (4)學生對評語的理解與其立即的錯誤訂正是否有相關性(correlation)? 為探討以上研究問題，此研究落實以下幾項數位機制：(1)文法檢查工具(grammar checker): 學生在IWiLL繳交作文後，此工具可以自動偵測學生文章中動詞-名詞搭配字錯誤； (2)教師評語庫(teacher comment bank): 針對偵測到的錯誤，提供自動化評語。此研究將評語分成三類，並以內容明確度不同作區別。第一類評語的內容指出學生的錯誤位置與錯誤類別；第二類評語內容，除包含錯誤位置與錯誤類別外，另增加了錯誤類別的定義；第三類評語內容，除包含錯誤位置、錯誤類別與錯誤類別的定義外，另提供五個相關例句；(3) 學生立即回饋機制(immediate student feedback mechanism): 用以引導學生表示其是否了解教師評語，並追蹤學生立即錯誤訂正是否正確。 研究結果顯示，學生使用搭配字的成效，以收到第三類評語的組別表現最為明顯。此外，另外兩組，並無明顯差異。研究結果也顯示，在評語裡提供相關例句對學生立即的錯誤訂正也有明顯成效。本研究並根據上述結果，在英語教學上提供了相關的建議作為未來更進一步的研究方向。 This thesis is an investigation of the effects of degrees of explicitness of indirect feedback on English learners’ acquisition on collocations, particularly on verb-noun collocations (Liu, 2002). The purposes of the study are to investigate the effects of various types of feedback on learners’ writing performance and to investigate the comprehensibility of the teachers’ feedback and its impact on learners’ success in immediate collocation error revision. Thus, the study is guided by the four research questions: (1) Is error correction beneficial to the improvement of learners’ English writing at the lexical level? (2) Are there differences across feedback conditions in students’ writing performance in terms of collocation improvement? (3) Does the level of feedback explicitness also have an impact on students’ immediate error revision? (4) What are the effects of comprehensibility of the teachers’ feedback and how is the comprehensibility of the teachers’ feedback related to learners’ immediate error revision? To address these questions, I implement several digital tools in an existing online English learning platform called IWiLL and utilize two instruments, a guided composition and a sentence translation task, to elicit students’ collocation production. The tools include: (1) grammar checker that automatically detects collocation errors in the writing that learners submit to their teachers over IWiLL; (2) a teacher comment bank to consistently provide the same feedback content to the learners concerning the detected miscollocations. In total, there are three types of feedback, varying in feedback content. The simplest one indicates error location and error type, the next feedback types offers metalinguistic feedback, and the most explicit one provides linguistic input in the form of example sentences; (3) an immediate student feedback mechanism. The tool elicits the learners’ understanding of the teacher feedback and their immediate self-correction on the detected miscollocaiton as a means of determining whether the feedback was noticed and understood. Each of these steps supported by these tools was recorded and available for analysis. Results of the study indicate that there is substantial, highly significant improvement on both subjects and immediate error revision and their subsequent collocation production in the group receiving feedback containing positive linguistic input. On the other hand, there were no significant differences in editing success between the group that received simple feedback and the group received metalinguistic feedback, both without positive example sentences. Moreover, the findings suggested that providing adequate linguistic input for learners benefits the development of students’ collocational competence. The study provides suggestions and some directions for future research to enhance English teaching.