We have developed an automatic interpretation system running on a mobile PC that helps oral communication between Japanese and English speakers in various situations during their travel abroad. In order to allow a wide range of expressions and topics in the applied domain, we adopted an approach which utilizes the general linguistic knowledge as well as the domainspecific linguistic knowledge. Speech recognition module performs speaker-independent large-vocabulary (50,000 Japanese words and 10,000 English words) continuous speech recognition. Translation module performs syntax directed translation based on a new lexicalized grammar rule formalism.