The empty category principle : antecedent - government

Abstract

본 논문의 목적은 지배이론(Government theory)과 경계이론(Bounding theory)을 통합하는 데 있어서 기본이 되는 장벽(Barrier)의 개념을 재정비하고 하접조건(Subjacency Condition)을 완화함으로써, 하접도 1 (1-Subjacency)까지 이동(movement)에 있어서나 지배, 특히 공범주 원리(Empty Category Principle)에 있어서 문법적으로 허용하며, 공범주 원리를 선행사 지배(Antecedent-government)로 확립하는 데 있어서의 문제점들을 해결하여 공범주 원리는 이동으로 인한 공범주만을 대상으로 의미부(LF component)에서 적용되며 어휘 지배(Lexical-government)는 보편문법(Universal Grammar)에서 제거됨을 보이는 데 있다. 제 2장에서는 먼저 본 논문에서 이용하고 있는 X-bar이론을 소개한 후, Chomsky(1986b)에서 처음으로 시도된 지배이론과 경계이론을 통합하기 위한 분석이 내포하고 있는 문제점들을 지적하고, 이러한 문제점들이 Chomsky(1987)에서 어떻게 해결되었으며 그 해결방법이 또 다른 각도에서 어떤 문제점들을 보유하고 있는지 살펴본다. 제 3장은 공범주 원리에 관한 연구로서 고유지배(proper government)에서 지배자의 자격으로 어휘 범주 (lexical category)와 최대 범주(maximal category)를 허용하는 불일치를 해결하기 위하여 Stowell(1981), Kayne(1981b), Bouchard(1984), Aoun(1985 and 1986)등이 어떠한 분석방법을 제시하였는가 알아 본 후 각각의 분석방법에 있어서의 문제점들을 지적하고, Chomsky(1987)에 이르러 공범주 원리가 선행사 지배로 확립되면서 그 지배 과정에서 파생되는 문제점들을 알아본다. 제 4장에서는 2장과 3장에서 제시되었던 문제점들을 해결하기 위하여 지배와 장벽의 개념들을 수정하며, 이동에 있어서나 선행사 지배관계에 있어서 하접도 1까지 문법적으로 허용하도록 함으로써 that-흔적 여과 (that-trace filter)를 음운부(PF component)에 설정한다. 제 5장에서는 명사구로부터의 이동에 있어서 해결되지 않은 문제점들을 Bowers(1987)가 제시한 지정사구 (Determiner Phrase)를 부분적으로 채택함으로써 설명하고, 공범주 원리로부터 분리해 나온 어휘 지배가 궁극적으로는 불필요함을 보임으로써 결국 모든 공범주는 선행사 지배에 의하여 그 문법성이 결정됨을 보인다. ; The purpose of this dissertation is two fold: first, to explore a more appropriate way to unify the theories of government and bounding, revising the notion of government and barriers and weakening the well-formedness condition on Subjaccency and second, to establish the Empty Category Principle (ECP) as antecedent-government, eliminating Lexical-government in universal grammar (UG). The purpose of Chomsky s Barriers is also to explore the possibility of a unified approach to theories of government and bounding. There seem to me to be three major problems in his framework. First, Chomsky allows two conceptions in determining what constitutes a barrier: In one conception, any maximal projection might be a barrier and a second conception involves "minimality". However, barriers in the sense of the Minimality Condition are relevant only to the theory of government, whereas barriers in the first sense are relevant to the theory of government and movement. Therefore, we cannot regard it as a unified approach to these topics. Second, Chomsky regards IP as a "defective" category; it can only become a barrier by inheritance, not simply by being a "Blocking Category"(BC) itself. However, this concept of the inherited barrier is not intuitive at all. Lastly, Chomsky suggests the possibility that proper government can be reduced to antecedent-government by chain coindexing. However we cannot explain the antecedent-government of the trace in NP-movement within NP because N s is a barrier to government (i.e., proper government). Moreover we cannot assume SPEC-head agreement within NP as opposed to NP-movements within a clause which contains AGR. Chomsky has revised his idea in Tokyo lectures in 1987. In his revised framework, three problems mentioned above do not exit any more. However, there arise different problems. Here, he makes an ad hoc device, C -deletion which explains the RES (NIC) but not PRO theorem. And he establishes the ECP as antecedent-government. He assumes that the ECP is a principle of the links which are formed not only by movement but also by agreement in order to explain the antecedent-government in NP-movements. Therefore, the ECP is no longer a principle of empty categories and in NP-movements, there is a case in which the antecedent-government cannot be satisfied even though it is grammatical. Chomsky explains the specificity condition by allowing the SPEC of NP to be an escape hatch in case that there is non-specific determiner. However, this leads to a contradiction that the SEPC position of NP can be either A-position or A -position. Finally, Chomsky s subjaccency does not show the difference in the degree of grammatically between WH-movements from Complex NPs (between relative clause construction and noun-complement construction). In this dissertation, I propose that there is an "Obstructing Category" (OC) which is a similar concept to a barrier proposed in Tokyo lectures. Here, a maximal category that includes two OCs constitutes a Barrier to government and movement. And I revise the definitions of government and subjaccency using the above notions. Hence, 1-subjacency is allowed not only in grammatical movement but also in the ECP(i.e., antecedent-government). And I assume the that-trace filter as a PF filter. This assumption eliminates the ad hoc C -deletion and makes the ECP to be applied at LF, irrespective of the types of ECs (argument traces as well as adjunct traces). And I establish the pure antecedent-government even in NP-movements. In other words, I eliminate the possibility of a verb to be an antecedent of NP. Therefore, the ECP is a principle of empty categories, not a principle of the links by agreement. Besides, I explain the specificity condition and the difference in the degree of grammatically between WH-movements from Complex NP s by adopting Determiner Phrase (DP). And I eliminate Lexical-government which is separated from the ECP. It follows that ECs are subject only to antecedent-government, not to Lexical-government.