An Implemented Linearization Grammar for Japanese

Free word order languages with rich case systems and discontinuous constituents pose a serious challenge to the task of constructing an efficient natural language parse system. As Japanese not only allows considerable freedom in word order but pervasive constituent ellipsis, it poses unique difficulties for efficient broad coverage parsing. We have yet to see a broad coverage Japanese grammar/parser that generates linguistically well-motivated syntactic representations for scrambled structures and interprets elided arguments. My Ph.D. project is intended to develop a linguistically grounded grammar that sustains parsing efficiency and covers this range of phenomena. I have already implemented a prototype version of the grammar/parser, the details of which are documented in [31] and I will discuss in Section 5. For this prototype I have adopted design principles and parse mechanisms which I believe are computationally and linguistically preferable over the existent Japanese parsing systems. My Ph.D. research project will consist of extending this prototype and completing a full-scale parsing system, by elaborating on the current mechanisms and overcoming a number of outstanding issues concerning word order, discontinuity and argument ellipsis. I will also make sure to incorporate research results on other languages with similar phenomena and make my parser as portable as possible, so that my study will contribute to the future parser building projects not just for Japanese but for free word order, free pro-drop languages in general.