earticle

영어

Readability is commonly defined as a statistical scale for measuring the ease of reading and understanding a text. In this paper, we propose that paragraph length, sentence length, and word length are applicable to readability metrics for Korean texts. A correlation between these readability metrics and text levels is statistically analyzed using correlation analysis and simple and multiple linear regression analyses based on data as found in Korean learning materials. Paragraph length and sentence length show a steady upward slope with increasingly higher learning levels, while word length has a downward sloping relationship with the elementary level only. This correlation can also be applied to materials evaluation focused on the arrangement of learning levels and chapters with respect to readability. This applicability is demonstrated by two statistical methods. First, cluster analysis is the assignment of a set of texts by level or by chapter into subsets sharing similar readability scales to evaluate materials arrangements in accordance with readability scales. Second, regression model with a breakpoint (Baayen 2008) divides a linear relation of readability scales into two models having a different slope so that we can explore a trend change of readability scales in materials.