Introduction

It has been frequently noted that head-final languages such as Japanese are a challenge for a left-to-right parser, in which information provided by the verb plays a very important role. Among other constructions, relative clauses in Japanese in particular pose a tough challenge because they potentially involve multiple kinds of temporary/global ambiguities at different stages of parsing. In order to correctly interpret a Japanese relative clause, the parser needs to know, first of all, that it is a relative clause construction, which isn't evident in the early stages of the left–right parsing. It is also necessary to determine at which point of the sentence the clause starts, and what grammatical and/or thematic role the head noun is associated with within the relative clause. Furthermore, there can also be ambiguity as to the identity of the head noun, when the first NP in the relative head position is a complex NP, which is part of an larger NP.

These problems come from the syntactic and morphological properties of the language that fail to provide sufficient information about the correct structure for the input; including head finality, word order, and availability of empty arguments, as well as the lack of overt relative pronouns. In what follows, I will relate these syntactic facts with the emergence of various kinds of temporary/global ambiguities that occur at multiple levels of syntactic parsing.