華語文寫作測驗近年來成爲主要大型華語文能力測驗項目之一，其重要性日鉅。本研究以美國外語教學委員會（American Council on the Teaching of Foreign Languages，簡稱ACTFL）所制定之語言分級能力標準為框架，制定出華語文寫作評分規準，並進一步以實證研究來探討ACTFL評分規準所遇到的問題。研究語料為國立臺灣師範大學國語教學中心華語文學習者所寫之同題目作文402篇，專業評分者受培訓後根據ACTFL評分規準評分，研究一方面能為瞭解ACTFL華語文寫作測驗的信度之實徵研究，將評分結果透過百分比一致性與類推性理論分析，檢測測驗之信度。另一方面分析ACTFL之評分架構，就評分規準在分數等級上、計分方式上、分向構面上之三主要研究議題來進行深究與探討，將對寫作測驗有實徵之回饋。

he demand for reliable and valid writing assessing methods of Chinese as a foreign language has grown significantly in recent years. This article reports the construction and the subsequent evaluation of the analytic rating scale based on ACTFL (American Council on the Teaching of Foreign Languages) writing proficiency guidelines. A corpus of 402 writing scripts from different levels at the Mandarin Training Center at National Taiwan Normal University, but all in response to the same writing task, was collected and graded by five professional raters. Each writing test is independently rated by two raters. The validity of the scale is rigidly controlled throughout the construction and raters training process. The grading results provide qualitative analysis of two main points. On the one hand, the reliability of the assessment is examined in terms of the percentage of agreement between raters as well as generalizability and decision studies. On the other, a number of major issues involved in the development of a rating scale are explored for the purpose of this paper, especially on the examination of analytic rating scale against holistic rating scale, as well as the exploration of rating criteria.