行政院教育改革審議委員會建議在國小以羅馬拼音取代現行的國語注音符號教授 國語，以便利學生同時學習國語及英語。此項提議，引起不少爭論。為響應此構 想，台北縣於八十七學年度國小高年級學生試辦羅馬拼音教學。本研究以參與此 項試辦教學的學生為研究對象，以個案研究方式，配合訪談、問卷調查、能力測 驗及學習成就測驗，探討國小學生學習羅馬拼音所遭遇的困難、策略、成效等議 題。本研究發現已學過注音符號之國小六年級學生學羅馬拼音會受到注音符號干 擾，因此先學注音符號再學羅馬拼音的折衷制並不理想。也發現羅馬拼音學習成 效與其英文能力並無因果的柑關性。本研究建議：（1）國小以折衷制教授羅馬拼 音宜詳細比較注音符號與羅馬拼音兩種拼音符號系統的異同，以避免學生可能產 生的錯誤。（2）加強英文拼音觀念，應自英語教學法著手，如教授自然發音法或 講解拼字與發音的對應關係等。（3）解決注音符號或羅馬拼音的爭議，須做進一 步實證性的研究來加以探討。未來研究方向，可自國小一年級即實驗以羅馬拼音 取代注音，以探知直接以羅馬拼音學習國語的學習成效，或在國小中年級或國中 以後實驗羅馬拼音教學，以探知注音符號下擾羅馬拼音學習及羅馬拼音與英文兩 種羅馬字母拼音系統的混淆是否會因受測年齡的不同而有所改變。

　　　　　The Education Innovation Council has suggested that English education start from as early as the primary school level. On the other hand. Dr. Li Yuan-che, who is in charge of Academia Sinica, proposed in Oct. 1996 that the current transcription system Zhuying be replaced by romanization such that learners can be free of the burden of memorizing two transcription systems. This consideration can be argued to facilitate learners' acquisition of English. Given the fact that Zhuying has been put into use in the Republic of China for more than 60 Years, Dr. Li's suggestion has aroused heated discussion. Since primary schools in Taipei county offered courses of the romanization skill in the 97 academic year, this project picked up students who took this course as subjects to study the issues involved in learning romanization, such as the learning difficulties, strategies and effectiveness. This study also finds out that learning romanization does not have a cause-result type of correlation with English abilities. The result of this study indicates that students' background of Zhuying interferes with their learning romanization, and thus suggests that a compromised way of learning romanization, i.e. learning Zhuying before romanization is in adequate. This study also finds out that learning romanization does not have a cause-result type of correlation with English abilities. Suggestions on romanization teaching, improvement of spelling concept, and directions of future research are made accordingly.