The syntactic difficulty of Japanese sentences

This thesis investigates the various syntactic sources of difficulty or ease in processing Japanese sentences. The investigation utilises the theory of Word Grammar (Hudson 1984, 1990), within which dependency distance has been developed in order to measure syntactic difficulty. My main contention is that dependency distance is just one source of difficulty, so I argue against the common idea that syntactic difficulty equals syntactic complexity and that difficulty has a single syntactic explanation. After presenting the nature of syntactic difficulty and complexity and techniques of measuring syntactic difficulty in the first chapter, in the second chapter recent theories accounting for syntactic difficulty are reviewed. The third chapter outlines a model of a parser based on Word Grammar along with dependency distance. In the fourth chapter the status of 'words' in Japanese is discussed within the framework of Word Grammar, which seeks to express syntactic knowledge in terms of direct relationships between words, so as to resolve the special problem of function words that Japanese raises. The following two chapters provide preliminary experimental data. In the fifth chapter the first experiment tentatively shows serial order effect, word-length effect, chunking effect and the contribution of function words to working memory load. The experiments in the sixth chapter tentatively show that dependency distance has a positive relationship with memorability on immediate recall tasks and "chaining" and "overload" effects. The data in the seventh chapter gives experimental evidence tentatively showing the appropriateness of measuring dependency distance for Japanese in terms of morpheme as well as "head" effects and effects of word order. Spoken English and Japanese texts are analysed in the eighth chapter to make a comparison of the syntactic difficulty amongst the texts, in which the results tentatively show that in terms of dependency distance Japanese is no more difficult for syntactic processing than English