Abstract
This study examines the deviation (Itsudatsu) through out conversations on the study introduction between non-Japanese speaking research students (NNS) and a native speaker (NS). The results indicate 2 adjustment types. One type is surface deviation in which it is easy to return to the main sequence; this results from errors such as the NNS's incorrect pronunciation and misuse of technical terms. The other type requires multiple interactions. The causes are the NNS's explanations without concrete examples or meta-language, and descriptions too specialized for the listener's requirements. The results also indicate the importance of using strategies of presenting key words and accessible examples, utilizing the NS's knowledge.
