The purpose of this highly quantitative study is to manifest the direct effects of communicative learning in the context of public high school teaching in Japan. A small-scale (n=12) experiment was conducted to reveal intergroup differences. The experiment was based on the assumption that there are clear differences in the amount of input from and kinds of interaction with native speakers of English in a communications-based learning environment and a traditional, regular learning environment. Preliminary assessment of the academic and attitudinal levels of the subjects enabled the selection of six matched pairs from the two learning environments. The subjects' semi-spontaneous interactions with the native speakers in six quasi-realistic contexts were videotaped and transcribed. The data analysis focused on three aspects of their communicative behaviors: discourse-related, functional, and formal. It was found that communicative learning does enhance learner performance but not in the way it has been generally assumed or expected. Rather than making the learners more fluent, as might be expected, the direct effects identified were the following: more active involvement in interactions, provision of richer information, greater degree of encoding, and more extensive use of clause-containing and minimal/phrasal acts. (Contains 17 references.) (KFT)