Argumentness and Probabilistic Case Structures

Argumentness and Probabilistic Case Structures Dan-Hee Yang Department of Computer Science, Pyongtaek Univ. 111 Yongyi-Dong, Pyongtaek, Kyungki-Do, 450-701, KOREA dhyang@ptuniv.ac.icr Ili-Hwan Lee Department of English, Yonsei University 134 Shinchon-Dong, Seodaemun-Gu, Seoul 120-749, KOREA ihlee@yonsei.ac.kr Abstract This paper proposes that the argument structures be stated in a way that uses probabilities derived from a corpus to replace a Boolean-value system of subcategorization. To do this, we make a cognitive model from a situation to an utterance to explain the phenomena of arguments' ellipsis, though the traditional term ellipsis is not suitable under our new concepts. We claim that the binary distinction is neither rational nor suitable for a real syntactic analysis. To solve this problem, we propose two new concepts argumentness and probabilistic Case structures by adapting the prototype theory. We believe that these concepts are effective in the syntactic analysis of NLP. 1� Introduction Researches on building Korean Case frames have been done by several organizations such as the National Academy of the Korean Language, Language Research Institute of Seoul National University, and the Center for Linguistic Informatics Research of Yonsei University, including the SEJONG 21 National Project. For English, there are several comprehensive Case frames such as FrameNet, COMLEX corpus, and LDOCE (Longman Dictionary of Contemporary English), etc. All these researches distinguish optional arguments from obligatory ones in the logic of black or white, assuming that arguments are the participants minimally involved in the activity or state expressed by the predicate, and they discuss only obligatory arguments (Dusan 1998, Hong 1997, Lee 1997, Song 1999). We note, however, that the term minimally is very vague. Furthermore, it is very dubious that such a simple binary distinction of obligatory vs. optional arguments is objectively well-grounded (Nam 1993). Concerning this issue, the present study recognizes that there is a significant difference in native speakers' intuition on whether or not an argument can be omitted. By establishing a cognitive model from a situation to an utterance, we explain why arguments' ellipsis occurs, though the traditional term ellipsis is not suitable under our new concepts. Here we devise two filters: an individual cognitive filter and an individual linguistic filter. Then, we claim that the binary distinction is not appropriate. Instead, we propose two types of new concepts, namely argumentness and probabilistic Case structures by adapting the prototype theory. Finally, we show that these concepts have several merits for NLP. 2� Cognitive Process of Arguments' Ellipsis 2.1 Ellipsis of Arguments It has been noted that there are two rather different kinds of ellipses of arguments: namely, syntactic ellipsis and pragmatic ellipsis. A syntactic ellipsis can occur when we consider the semantic property of the predicate. A pragmatic ellipsis can occur when the omitted constituent can be naturally recovered from the context. Therefore, in a given sentence, an obligatory argument has been understood to undergo the pragmatic rather than the syntactic ellipsis. On the other hand, an optional 447 argument has been understood to undergo the syntactic/pragmatic ellipsis. However, we observe some delicate differences in …

Content truncated to 500 words