论文部分内容阅读
本项研究旨在应用写作测量指标对二语写作评分员及评分结果进行探索性的研究,探索目的为:1)是否能以此发现评分员在评分中所表现出来的具体差异;2)能否对不同评分员的综合评分方式进行比较直接的观察以获得相对较好的综合评分;3)由测量指标所获得的预测分与评分员的评分能在多大程度上相匹配。在探索性的二语写作测量指标体系的基础之上所进行的研究分析结果显示评分员之间的确存在差异,具体表现为对某些语言特点各有侧重。同时也发现评分员之间在主要的语言特征方面也存在共同点或相通之处。对照三种不同的综合评分方式的结果发现本次研究中以权重方式进行的综合评分比“纯粹平均”的综合评分更接近实际分数。几项重要指标比如:R平方及DW值也显示以权重方式所得到的结果甚至稍胜于多层面分析(FACET)所得出的结果。以测量指标为基础所获得的预测分与评分员之间的评分匹配达到百分之五十以上。最后,本文结合测量指标的应用以及测量指标体系本身的特点进行了总结和归纳。本项研究所具有的方法上、理论上和实际应用方面的作用也在各相关部分有所阐述。
The purpose of this study is to explore the use of writing measurement indicators for second-language writing scorers and scoring results to explore the purpose of: 1) whether it can be found in the grading scoring showed in the specific differences; 2) No direct comparison of the overall score of different scorers to obtain a relatively good overall score; and 3) the extent to which the forecast score obtained from the measures and the scorer’s score match. The results of research and analysis conducted on the basis of the explorative indexing system for second language writing show that there are indeed differences among the scorers, which are manifested in the emphasis on certain linguistic features. At the same time, it has also been found that there are commonalities or commonalities between graders in the main linguistic features. The results of the three different composite scoring methods found that the composite score weighted by this method was closer to the actual score than the composite score of “pure average”. Several important indicators such as: R-squared and DW values also show that the results obtained by weighting are even slightly better than those obtained by the multi-level analysis (FACET). The score matching between the forecast score and the scorer based on the measurement index reached more than 50%. Finally, this article summarizes and summarizes the application of measurement indicators and the characteristics of the measurement index system itself. The methodological, theoretical and practical aspects of this study are also described in the relevant sections.