Abstract: Based on large scale ChinescEnglish parallel corpus, this paper described a systematic experiment of statistical analysis for bilingual verb subcategorization. Firstly, with lexical and grammatical compatibility as heuristics, probabilistic distributions of 654 bilingual subcategorization frames were estimated by means of a two-fold MI_E filtering method. Then,linguistic classification of the frames was determined according to Chinese and English syntax Finally,linguistic classes for each frame were labeled via SVM on the basis of their supporting corpus.

Key words: Chinese-English verb subcatcgorization, Statistical analysis, SVM