Construction as a Theoretical Entity : An Argument Based on Mandarin Existential Sentences

The role that constructions play in a linguistic theory has changed throughout the evolution of generative theories. Construction specific rules are common when transformations are envisioned as tree to tree operations in classical TG. On the other extreme, constructions, as well as all structural properties, are regarded as derived linguistic properties predictable from various principles in recent GB theories. Since whether a construction is an autonomous linguistic entity or not has great implications for either a formal or a computational linguistic theory, we will examine the status of Mandarin existential construction based on the theory of Construction Grammar [1,2]. We will show that the Mandarin existential construction represents an unique structure-meaning pair that cannot be captured in a grammar unless the pairing is regarded as a theoretical entity in linguistics ([3]). Since constructions are shown to exist in Mandarin Chinese, we support the theoretical claims of Construction Grammar as well as the position that constructions must be taken into account in NLP. I. Existential constructions and its two sub-constructions The existential sentences in Mandarin Chinese, such as (1) and (2), have been traditionally considered to involve movement or lexical rule in a verb-centered paradigm. (1)Zhuo-shang fang LE yi ben shu. table on put ASP a CL book 'There is a book on the table' (2) Chuang-shang tang ZHE yi ge ren. bed on lie ASP a CL person 'There is a person lying on the bed' The surface structure of these existential sentences can be schematized as follows: (3) [ Locative V Theme ] The apparent structural uniformity represented by (3) gives the verb-centered accounts their strongest motivation. This is because a verb-centered account can easily predict the structure by the cross-the-board nature of the movement rule or the lexical projection rule. However, there are two facts involving the existential sentences that would prove difficult for verbcentered accounts but could be easily predicted by a constructional account. First, the existential meaning is manifested only when the structure schema of (3) is employed. In other words, if the existential sense is attributed to the verbs, the grammar will be overloaded with unnecessary ambiguities in structures other than (3). The verb-centered