寫作評量之研發與應用：以國小六年級記敘文為例

The Development and Application of the Sixth Grade Student’s Narrative Writing Assessment: A Study at the Middle of Taiwan Area

本研究根據現行九年一貫課程綱要重新審視國小六年級學生的各級寫作能力，研發國小階段記敘文寫作評量。本研究是以國語文第二階段寫作能力指標內容為基礎（修訂前），研發包含語句、段落及全篇寫作三層題型之小型測驗；並導入現代測驗理論之概念，透過定錨不等組設計（Non-Equivalent groups with Anchor Test design， NEAT）建立國小六年級學生寫作評量的分數量尺。同時，將全篇寫作具體分成內容表達、文字修辭、敘寫技巧三個向度進行評閱，期待能透過資料所衍生的訊息，提供研究者更多影響學生寫作表現的因素或現象。最後，本研究藉由全篇寫作和段落、語句寫作的高相關，期望能發展更多元、完備的語句或段落測驗，藉此取代部分全篇測驗，以減少人工評閱等問題，並掌握學習偏誤、落實補救教學。

The purpose of this study is to develop a small-scale writing assessment of different writing content that making sentence， paragraph， and composition. In this study， we re-survey the writing ability among the grade 6 students based on the grade 1-9 integrated curriculum guidelines and construct a scoring scale for a writing assessment by equating design. The results are listed as below: 1. The narrative writing ability among the grade 6 students have no significant difference between different areas; 2. Among the different class scale， some significant differences are found in different scoring categories of composition. It is found that in the average test scores of students， the schools with over 60 classes are better than those with 30-60 classes; 3. The different narrative writing content of “describing scenery”， “describing objects” and “describing characters”. The result of the analysis show that there are significant distinctions in “describing scenery”， “describing objects” and “describing characters”. In the average test scores of narrative writing content， “describing scenery” is better than others， and “describing objects” is better than “describing characters”; 4. The different writing content of making sentence， paragraph， and composition. It is found that in the average test scores of students， making sentence is better then others. Since the results of high correlation among making sentence， paragraph， and composition， we will develop composition， paragraph， and making sentence to solve the problems that it costs a lot of testing time and manual grading.