A Report on Stylometric Experiments with Pre-Modern Chinese Texts

Abstract

Although stylometry and related quantitative-linguistic approaches have long been in use for assessing text authorship or for addressing other textual issues in Western literature, I am not aware of any serious attempts in the field of study of pre-modern Chinese texts. It is quite surprising given the fact that the authorship and dating of particular ancient Chinese books or their components have belonged to the most fiercely discussed topics in pre-modern Chinese philology in the last few decades. This paper is therefore conceived as an experiment in this vein, centering on the problem of dating. On the basis of a set of detailed statistic data, involving e.g. occurrence frequencies of word classes, syntactic functions, special function words, constructions, or polysyllabic words, I tried to find out applicable parameters, which would be sensitive to date of composition. As the primary linguistic material is concerned, it consists of a series of stylistically variegated chapters of several texts a) clearly dated as Warring States texts, b) clearly dated as mediaeval texts, c) two chapters of the Yinwenzi, a text with disputed dating. In the article, the results of these experiments are reported, interpreted, and evaluated, and some recommendations for the future research are suggested.