金庸与古龙都是新派武侠小说家中的代表人物，二人对于新派武侠小说的发展产生了巨大的影响。本文分别挑选金庸与古龙代表性的作品对金庸与古龙各自的风格以及二人的风格差异从计算风格学的角度分别进行了考察。以金庸全部十二部长篇小说为基础，分别从形符类符比、平均段落长度、词长、句长、词长离散度、句长离散度、标点符号、词类等特征对金庸十二部小说的风格进行了考察，并在此基础上使用三种聚类方法从总体上考察了词和词类N元文法、全体词类、高频词、实词、虚词、标点符号在不同小说中的差异，结果显示，金庸小说风格从总体上来说是比较稳定的，从计算风格学的角度无法对其进行明确的分期，但不同篇幅的小说的风格在某些特征上是有差异的：篇幅较长的小说形符类符比相对较低，词长离散度和句长离散度较大，并且从聚类来看篇幅较长的小说聚为一类，而篇幅较小的小说聚为一类。同时，其在词类、标点符号、实词、虚词等的使用上差异并不明显。根据古龙自己的说法将古龙创作生涯划分为前、中、后三个时期，前期挑选八部小说组成四个文本、中后期分别挑选四部小说来代表不同时期的作品，分别从平均段落长度、词长、句长、词长离散度、句长离散度、词类、标点符号等特征来对不同时期的小说风格进行了考察，同时，使用两种层次聚类方法利用词类整体、虚词词类、实词词类、高频词、词类的N元文法、词的N元文法、标点符号的N元文法以及反映文本可读性和语言变化程度的六个特征从总体上进行了考察，结果显示，古龙前、中、后三个时期的小说风格是有差异的，而且中后期相对比较接近，前期则与之相差较大，说明古龙小说有过一次较大的转型。在此基础上，分别挑选金庸与古龙各自六部小说作为代表，从平均段落长度、词长离散度、部分实词使用、部分虚词、词类、标点符号、高频词、文本从众性、句子破碎度、标点符号占比、词、词类标点符号的N元文法等角度，运用频率统计、主成分分析、假设检验、聚类、分类等方法系统地比较了二者的差异，结果发现：金庸小说平均段落长度要高于古龙，而词长变化程度较古龙要小，反映出金庸小说的可读性较古龙小说要弱，古龙小说用词更富变化；金庸更关注江湖人的家国责任，古龙更倾向于描写江湖人的个体感受；金庸文本从众性高，更喜欢详细的动作描写，运用更多书面性更强的词语。
Jin Yong and Gu Long are both greats of the new-style kung fu fiction. Their works have great influence on the development of it. This paper analyzes the style of kung fu fictions written by Jin Yong and Gu Long and the differences between the two authors in the way of stylometry.This paper selects all of the 12 novels written by Jin Yong. Ratio of token and type, the average lengths of paragraphs, the lengths of words and sentences, dispersions of word length and sentence length, punctuations and part of speech are analyzed. Thus we compare all the novels based on the features as follows by clustering: the N-gram of words and part of speech, all the content words, function words, part of speech and punctuation. All these features chosen show that Jin Yong’s fictions have a stable style, which cannot be divided into several periods in the way of stylometry. However, we can see some differences between fictions with different lengths: shorter fictions have lower ratios of token and type, greater dispersions of word length and sentence length; what’s more, the shorter fictions are clustered into one cluster while the longer ones clustered into the other. There are no significant differences among these fictions in the features as follows: part of speech, punctions, content words and function words.All the fictions written by Gu Long are divided into 3 periods according to the viewpoint of himself, thus 8 fictions of first period are chosen and combined into 4 texts with both 4 fictions chosen from the middle period and the last period, totally 12 texts are taken as the representatives of the 3 periods. Fictions of the 3 periods are compared with each other accoding to these features: average lengths of paragraphs, the lengths of words and sentences, dispersions of word length and sentence length, punctuations and part of speech. Moreover, all these 12 texts are clustered into 3 clusters respectively with 2 clustering methods based on part of speech, content words, function words, high-frequency words, N-gram of words, N-gram of part of speech, N-gram of punctuations and 6 features that can reflect the readability and variation of the fictions. The results show that there exist great differences among the first, the middle and the last periods of Gu Long, while the style of fictions in middle and last periods are much closer than that in first periods and other periods, which can prove the truth that great change has occurred in Gu Long’s fictions.On this basis, Jin Yong and Gu Long’s representative six novels are selected respectively. The analyzed features include average paragraph length, the dispersion of word length, some content words, functions words, high-frequency words, different parts of speech, punctuation, the conformity of the texts, sentence fragment, N-gram of words, N-gram of part of speech and N-gram of punctuations. We make a comparative study of their novels by means of frequency counting, hypothesis testing, hierarchical clustering, k-means clustering, principal component analysis and classification. We find that average paragraph length and the frequencies of comma, words of national responsibility, verb, noun of location, prepositions the conformity of the texts and sentence fragment in Jin Yong’s fictions are higher than those in Gu Long’s fictions. The dispersion of word length and the frequencies of question mark, colon, quotation mark, auxiliary word, adverb and conjunction in Jin Yong’s fictions are lower than those in Gu Long’s fictions. There are obvious differences between forty-nine function words and high-frequency words in Jin Yong’s novels and those in Gu long’s novels. Gu Long’s novels are more readable and changeable in word length than Jin Yong’s novels. Jin Yong pays more attention to national responsibility and Gu Long pays more attention to individual feeling.
