In the development of a Chinese-English machine translation system, one of the kev Droblems is how to generate different morphologies of words(such as，the posses-sive and Dlural forms of noun，the past participial form of verb etc．)according to current context situation．In this paper，a SC grammar-based generation algorithm ior Chinese-English machine translation is proposed．In which，by designing a generation-driven lin-guistic feature description scheme and taking a unified language analysis approach of incor-porating the generation with parsing，through using various knowledge which is produced and used in parsing, it is possible to generate proper morphologic feature oi each compo-nent and to efficiently generate corresponding auxiliary verb and articles in sentence.