2010 Volume 33 Issue 4 Pages 449-458
The present study aimed to build a readability formula to measure levels of school grade 1-9 based on school texts for the Japanese language. Five predicting variables were selected to determine readability as defined by school grades: (1) average number of characters in a sentence, (2) average number of phrasal units in a sentence, (3) average number of predicates in a sentence, (4) ratio of Chinese-originated words in a text, and (5) ratio of hiragana characters in a text. After excluding deviated texts and variables showing multicollinearity, a linear multiple regression analysis indicated two significant predictors of 'average number of predicates in a sentence,' and 'ratio of hiragana characters in a text', showing a high predicting power (R^2=0.791).