期刊
COMPUTERS AND THE HUMANITIES
卷 38, 期 1, 页码 61-82出版社
SPRINGER
DOI: 10.1023/B:CHUM.0000009225.28847.77
关键词
agglutinative languages; authorship attribution; statistical analysis; stylochoronometry; stylometry; Turkish
This study investigates the writing style change of two Turkish authors, Cetin Altan and Yasar Kemal, in their old and new works using respectively their newspaper columns and novels. The style markers are the frequencies of word lengths in both text and vocabulary, and the rate of usage of most frequent words. For both authors, t-tests and logistic regressions show that the length of the words in new works is significantly longer than that of the old. The principal component analyses graphically illustrate the separation between old and new texts. The works are correctly categorized as old or new with 75 to 100% accuracy and 92% average accuracy using discriminant analysis-based cross validation. The results imply higher time gap may have positive impact in separation and categorization. For Altan a regression analysis demonstrates a decrease in average word length as the age of his column increases. One interesting observation is that for one word each author has similar preference changes over time.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据