The Stylometric Analysis Of The Harry Potter Series (Ⅰ-Ⅵ)

This thesis studies the Harry Potter series (I—VI), analyzes and compares the series' stylometric characteristics with corpus-based approach. With the help of Foxpro, computer programs are designed for data collection in terms of word length, sentence length, punctuation, type versus token ratio, vocabulary growth rate, and common word frequency, and frequency chunk method is widely used for data collection. Data are sorted out and tested by statistical tests to check whether there exists significant difference in the style among the six books. The tests include Kolmogorov-Smirnov test, analysis of variance test, Kruskal-Wallis test, and principle component analysis.The research shows that the mean word lengths, mean sentence lengths, mean type versus token ratios, and vocabulary growth rates increase in the six books' chronicle order. Although the tests show there are significant differences in the above four aspects, they are closely related with the novels' contents; thus, the results can not reflect the author's style. Punctuation and common word frequency, which are not influenced by contents, can reflect the author's style, and the tests shows there are not significant differences in the two aspects.There are few applications of stylometry in the analysis of current children's literature. This research is to accumulate experience and shed light on quantitative analysis of children's literature, automatic text categorization, as well as authorship attribution.