|Budget Amount *help
¥11,000,000 (Direct Cost: ¥11,000,000)
Fiscal Year 2006: ¥2,700,000 (Direct Cost: ¥2,700,000)
Fiscal Year 2005: ¥2,800,000 (Direct Cost: ¥2,800,000)
Fiscal Year 2004: ¥2,800,000 (Direct Cost: ¥2,800,000)
Fiscal Year 2003: ¥2,700,000 (Direct Cost: ¥2,700,000)
In order to build a Japanese language processing system, we have to develop both the theory of how the information of sentences and discourse should be constructed, and its implementation on a computer. For the former part-at least in the sentence level---, we have seen the development by computational linguistic theory, such as Japanese Phrase Structure Grammar. It is widely accepted that it is very hard to obtain a reasonable interpretation of a sentence, or even a word, without context information in Japanese language processing, or natural language processing in general.
Based on Japanese Phrase Structure Grammar and Dynamic Syntax theory, we tackle the problem of how sentence and discourse can be interpreted in context, what interpretations are preferred, and what kind of logical (not statistical) framework we should take as the basis. In our study, grammar is redefined: in the narrow sense, it describes how the context independent meaning of words and sentence is calculated; in th
e wider sense, it is a set of principles that define how the interpretation of sentence and discourse is constructed from the context-independent meaning based on context information, especially discourse structure. This is very contrasted with the traditional view of grammar that syntactic rules and principles play the most part of roles. We have investigated the way of discourse understanding, and open a gate to the learning from discourse.
What we did is that describing several aspects of grammar from the view of Japanese discourse interpretation processing. To be concrete, we take Segmented Discourse Structure Theory and Dynamic Syntax theory as our central theory, we analyze the structure of speech act, discourse structure, and utterance context. As the specific topics, we take the interpretation process of Japanese ambiguous particles, zero anaphora and tense. Furthermore, we try to develop an language understanding model which covers not only written texts such as newspapers, but also oral discourse such as conversations and dialogue between a child and her caretaker. Finally we consider the characteristics as a language of Japanese contrasting English. As the result, we come to a conclusion that the main characteristic of Japanese is its flexibility, that is, the easiness of context shift with the multi-layered contexts Less