Disambiguating Japanese compound verbs

The purpose of this study is to disambiguate Japanese compound verbs (JCV) based on two methods: (1) a statistical method which makes use of collocational or semantic information about different verb combinations, and (2) a manual rule-based method which utilises verbal and nominal semantic features. We also present a combined method where the output of the statistical method is fed into the rule-based method. In evaluation, we found that the pure rule-based method outperformed the statistical and combined method at 96% token-level accuracy, suggesting that fine-grained semantic analysis is an important component of JCV disambiguation. At the same time, the performance of the fully-automated statistical method was found to be surprisingly good at 86%, without making use of lexical semantics.