从“其”替换“之”看上古-中古汉语的兼语式

The So-called Pivotal Construction in Old and Middle Chinese: Viewed from the Replacement of zhi(之) by qi(其)

一般所说的兼语式,兼语位置上,先秦本用"之",但从战国末期始,"之"可以为"其"替换。本文意在从此现象入手,利用历史的资料以及历时的演变,对兼语式、控制结构、使役句等复杂问题进行探讨。对于这种替换,王力(1980,1989)暗示:中古"其"在兼语位置,是充当包孕句的主语;而魏培泉(2004)等认为仍作宾语。其实答案并非如此简单。本文视上古-中古汉语的兼语式为控制结构,在将其与其他类似结构——小句宾语、紧缩式、连动式、双宾语——进行区分之后,我们将上古-中古汉语中的兼语式分解为4类,并逐一说明各类的结构,以及兼语位置"其"替换"之"的情况。

In the pre-Qin period, zhi（之） with accusative case was used in the pivot position in the so-called pivotal construction in Chinese. However, by the end of the Warring States, it came to be replaced by qi（其） taking the genitive case. The replacement ofzhi by qi aroused Chinese linguists＇ interest in the function ofqi in the construction. Wang （ 1980, 1989） suggested that qi functioned as the subject of the embedded clause, while Wei （2004） regarded it as the object of the matrix verb. However, the actual situation might be more complex than expected as the so-called pivotal construction discussed in the literature was often used to label structures that were quite different in their syntactic configurations. In this paper, the pivotal construction in Old and Middle Chinese are defined as the control structure. It shows that there are distinctions between the pivotal construction and other structures similar to it, which includes those characterized by finite complement clauses, contracted clauses, serial verb constructions, and double object constructions. It is argued that in Old Chinese the pivotal construction was different from the finite complement clause of the verb which could take as complement the ＇ subject ＋ zhi ＋ predicate＇ structure. As to the contracted structure, it also differed from the pivotalconstructions as it was formed by two conjoined clauses with pro, rather than PRO, in the subject position of the second predicate. Ahhough serial verb constructions and pivotal constructions seemed to have similar structure （ as represented by a structure like NPl ＋V＋ NP2 ＋VP）, they differed in their choice of controllers for PRO. Syntactically, while subject control occurred more frequently than object control, the latter should not be considered as exceptional. It is argued that partial control might not exist in Old Chinese, since even in cases where the embedded predicate exhibited some kind of apparent collective or reciprocal meaning, only one agent should be selected for ＇ controller＇, and the other（s） should be regarded as dative. To sum up, pivotal constructions can be categorized into four types： Type 1 is implicit causatives, though the matrix verb per se carried no causative meaning. In this kind of construction, the pivotal element was always the object, and the replacement of zhi by qi never occurred. Type 2 is explicit causatives, which might be treated as not only a type of control construction, but also a type of double object construction. In this kind of construction, the pivotal element was always the object, regardless of the actual occurrence of the replacement of zhi by qi. Type 3 is designated as ＇pure causatives＇, which could be taken as ECM constructions. The pivotal element （except qi） acted as the subject, but carried the accusative case. VPs with the genitive case form qi occurring in the pivotal position in Old Chinese should be treated as a gerundive phrase, with qi as its genitive subject occurring in the attributive position. However, in Middle Chinese, V ＋ qi ＋ VP became ECM, and consequently qi should be analyzed into the subject. Type 4 is termed as ＇ causative resultatives＇, with the pivotal element as the subject, which could not be instantiated by zhi, though the use of qi was allowed. In addition, some pivotal construction verbs developed into the ones taking a subject- predicate phrase as the object, and the replacement of zhi by qi was associated with the change of the pivotal element from the object into the subject.