影響中文文本閱讀難易度的多層次語言特徵分析

A multilevel analysis of the linguistic features affecting Chinese text readability

本文研究文本理解的特性，論述採用詞彙類、語意類、句法類和文章凝聚性四個主要層次的32 個語言特徵，以探討中文文本閱讀難易度，克服前人僅採用一些表面語言特徵的限制。研究的實驗材料取自臺灣三個版本的小學一年級到六年級的課本共386 篇文本，從低年級到高年級的課文當作是文本從簡易到相對繁難的安排，這32 個語言特性的閱讀難易度以統計學的趨向分析依文本在年級的高低分布計算出來。過去華語為二語的文本分析大都採用學生偏誤的語料，本研究則以正規的文本為依據，做出中文文本閱讀難易度，希望研究成果能對華語為二語教學有所貢獻。

The purpose of the present study is to investigate the characteristics of text comprehension. We use an approach that includes 32 linguistic features from four main levels (word, semantic, syntactic, and cohesive) that relate to Chinese text readability. The aim is to overcome the restrictions that resulted from adopting only surface linguistic features in investigating text readability in the past. In the present study, the research materials include 386 texts from Chinese textbooks for Taiwan elementary school students in six grades. The texts are reasonably assumed to increase in complexity of readability from grade 1 to grade 6. The readability of each of the 32 features is derived from the statistical trend analysis with respect to the grades. It is hoped that the results of this study can be of significant use in teaching Chinese reading in Chinese as a second language.