Quantitative evaluation of writing styles based on text analysis: Methods and case study

Mathematical metrics indexing writing style of literature works based on a series of text classification techniques are introduced in this paper. Four different Chinese translation versions of the classical masterpiece of Maupassant´s Boule de Suif (Ball of Fat) are adopted as a case study to illustrate the inherent popularity, conformity and unique stylistic choices of translation language by different translators. Character frequency entropy (CFE) developed from modified Zipf-Mandelbrot principle is used here to evaluate the inherent popularity. The diction of phrasal materials and their clustering indices are then scrutinized with a critical parser of Chinese Word Segmentation (CWS) to evaluate writer´s conformity to conventional language. Sentence length and dispersion are calculated to reveal the habit of a loose or a compact syntax. The full analysis of sample texts from zi (character), ci (word) to ju (sentence) demonstrates a panorama of linguistic style of translators involved.