Transition and Parsing State and Incrementality in Dynamic Syntax

Transition and Parsing State and Incrementality in Dynamic Syntax∗ Masahiro Kobayashi a and Kei Yoshimoto b aUniversity Education Center, Tottori University, 4-101 Koyama-cho Minami, Tottori, 680-8550 Japan, kobayashi@uec.tottori-u.ac.jp and bCenter for the Advancement of Higher Education, Tohoku University, Kawauchi 41, Aobaku, Sendai, 980-8576 Japan, kyoshimoto@mail.tains.tohoku.ac.jp Abstract. This paper presents an implementation of a gramar of Dynamic Syntax for Japanese. Dynamic Syntax is a grammar formalism which enables a parser to process a sentence in an incremental fashion, establishing the semantic representation. Currently the application of lexical rules and transition rules in Dynamic Syntax is carried out arbitrarily and this leads to inefficient parsing. This paper provides an algorithm of rule application and partitioned parsing state for efficient parsing with special reference to processing Japanese, which is one of head-final languages. At the present stage the parser is still small but can parse scrambled sentences, relative clause constructions, and embedded clauses. The parser is written in Prolog and this paper shows that the parser can process null arguments in a complex sentence in Japanese. Keywords: Japanese, Dynamic Syntax, implementation, rule application, null argument 1. Introduction Incremental processing of a sentence as it is inputted from left to right has been taken as most accurately simulating human sentence processing. When we attempt theoretically to account for this incrementality of sentence processing, however, we notice that there is a difference between head-initial languages and head-final languages. What we need to take into account is that in head- initial languages, like English, parsers can look ahead to syntactic structures to be established to some extent as a word is consumed. By contrast, this is not the case in head-final languages such as Japanese, because in these languages the syntactic positions of NPs remain unfixed until the matrix verb is inputted finally in the sentence. The Dynamic Syntax approach (Kempson, Meyer-Viol, and Gabbay, 2001; Cann, Kempson, and Marten, 2005), which allows the parser to process a sentence in an incremental fashion, has settled this difficulty by adopting structural underspecification. However, Kempson, Meyer-Viol, and Gabbay (2001) and Cann, Kempson, and Marten (2005) do not provide an explicit algorithm for implementation of the framework as formally as Moot (1999) presents Grail, the toolkit for a parser of Categorial Grammar Logics. This paper will delineate the basic idea of a parser for Japanese based on the Dynamic Syntax (hereafter, DS) framework, providing the algorithm of the application of lexical rules and transition ∗We would like to thank two anonymous reviewers for valuable comments. 0 Copyright 2007 by Masahiro Kobayashi and Kei Yoshimoto rules. This paper will also show that the parser is able to cope with null arguments or empty pronouns in embedded clauses. Currently the parser is implemented in Prolog. The outline of this paper is as follows: the subsequent section will provide a brief introduction to the DS formalism and address issues of parsing inefficiency and the algorithm of rule appli- cation, discussing the previous studies. Section 3 will illustrate the implementation of the DS formalism …

Content truncated to 500 words