In the literature of the generative grammar, de is treated as a case-marker (Huang, 1980; Li, 1989; Tang, 1990; Tsai, 1994).
And it is also analyzed as the C head (Huang, 1982; Bao, 1987; Cheng, 1986; Ning, 1993). However, these studies can only capture some of the characteristics of the item de. It seems that so far we have not had a unified account of de and de-constructions. Then in the spirit of the Minimalist Program, Ning (1995, 1996) maintains that de should be regarded as a functional head, and gives the description of the mechanism for the derivation of de-constructions involving feature raising and checking. However, the mechanism proposed by Ning can only account for some of the de-constructions similar to the English relative clauses. Based on the assumptions made by Ning (1995, 1996), the Chinese particle de is analyzed under our present discussion as a special kind of
functional category which is responsible for the generation of all the so-called de-constructions in the language. De has no semantic meaning or descriptive content, which is one of the most important characteristics of the functional category. The categorial feature of de is De. We assume that de has a strong nominal feature which is uninterpretable, and all the derivations of de-constructions are motivated by the checking and erasing of this strong uninterpretable nominal feature. The formal feature of de is checked either by the matching feature of the null NP raised from within the IP which de selects as its complement in the case of the simple or bare de-construction, or by the feature of a
nominal phrase (NP or DP) taken from the numeration, and put into the spec position by pure merge in the case of the complex de-construction. The features are checked against each other in the Head-Spec relationship. The category status of the de-construction derived is DeP. The simple de-construction is termed DeP1 or the simple DeP; and the complex de-construction is called DeP2 or the complex DeP. In regard to referential meaning, some DePs are NP, and others are DP in nature. DePs interact with other nominal phrases forming even more complicated nominal phrases. NP or DP can adjoin to DeP1, and the result is a compound DeP which is different from the complex DeP in inner structures; a D head can choose a DeP as its complement, forming a complex DP. I maintain that the only possible mechanism involved in the formation is the costless merge: substitution merge and adjunction merge, and no movement of any kind is needed here. Substitution merge involves a minimal projection, and a maximal projection; adjunction merge involves two maximal projections.The suggestions, proposals and arguments presented in this paper are tentative and open to criticism and discussion. Much has left to be desired in the further minimalist inquiries.