Modern Chinese Compound Words, Word Formation

This paper made an analysis and survey on 40,778 double syllables and polysyllables, except idioms, from Modern Chinese Grammar Information Dictionary (E-edition) compiled by Institute of Computational Linguistics in Chinese Department, Beijing University. Among the 40,778 items, 39, 454 are studied on their construction models, lexical categories and cores, 1,324 words like single-morpheme, the proper noun, transliterates and vague-structured words, are excluded. To the whole study, the principle of manual work combined with computer, qualitative way with quantitative one, and empirical study with material one is applied, to explore the influence of internal structure of Chinese compound word on the syntactic character of compounds.It is revealed that there is no direct relation between the construction of compound word and its part of speech. The head-feature assimilation theory is applied that the grammatical features of head-element of compounds assimilate the whole compound, thus determining their parts of speech. Statistics show that more than 90 percent of compounds comply with head-feature assimilation rule, with less than 10 percent of exception, which can be explained by the head-feature dissimilation rule that the grammatical-feature of head will change when assimilating a compound, so the part of speech of the whole compound becomes different from that of the head. We call the compounds in line with the assimilation rule centripetal words and the rest centrifugal words. It is pointed out that the influence of construction varies with the frequency. The more frequently the words are used, the less possibly they are assimilated; hence in Chinese the word frequency and the core attribute seep the inverse ratio. It is necessary to pay great attention to the vocabulary control and choice in foreign Chinese glossary teaching.In addition, 2,568 new derivatives with the suffix of "Hua"(920), "Xing"(808) and "Shi" (840) are studied on their grammatical features, with 89.13%, 100% and 99.06% in accordance with the assimilation rule respectively. The results further indicate that major Chinese new derivatives are centripetal words in compliance with assimilation rule. This conclusion to some degree offers some assistance to the Chinese information processing project.