Abstract
In this study, I aim to clarify the feature of the genre by analyzing statistically the use of vocabulary of the Japanese modern literatures. It seems that the juvenile and the fantasy are different genre, have different readers, and therefore have different style. I select as juveniles Gongitsune 'Gon, the little fox'，Tebukuro wo Kaini' Buying Mittens' both by NIIMI Nankichi, Kumonoito 'The Spider's Thread' by AKUTAGAWA Ryunosuke, and Tyumonno-oi Ryoriten 'The Restaurant of Many Orders' by MIYAZAWA Kenji, and as fantasies Koya-hijiri 'The Saint of Mt. Koya', Yashaga-ike 'Demon Lake' both by IZUMI Kyoka, Ningen-isu 'The Human Chair' by EDOGAWA Ranpo, and Sangetsuki 'The Moon above the Mountains' by NAKAJIMA Atsushi. 300 pieces quoted from beginning of each target works in Aozora-bunnko library texts are analyzed by morphological analysis tool Web-chamame, and from the result got by the analysis above, the indicator which represent the usages of vocabulary is calculated and the basic tendency of the target works can be grasped.
I compose a new parameter which can sum up the information of data, perform the principal component analysis (PCA) to grasp the characteristic of the target works. Furthermore, I apply the cluster analysis using the new variable provided by PCA and by classifying target works in some unities (cluster），try to consider what characteristic of the target works can be seen, or what kind of difference can be studied by a genre from a quantitative point of view.
As a result of basic analysis, I can point out that the rate of Japanese words or that of Sino-Japanese written by hiragana characters show the statistical significant difference by a genre. By the cluster analysis, I recognized the similarity between the same genre, but also found that there are some works which have the different character from which genre is generally said.
