ABSTRACT
This study was a field trial rather than a true experiment with controls. For example, we did not compare the robot with an ordinary computer English teaching game. A detailed experiment might offer more precise and reliable results on the teaching of English to Japanese students. However, our main goal was not to teach English optimally but to learn how to create partnership in a robot. We believe that field trials in a frontier research area (e.g., partner robotics) are essential for developing the discipline. Afield trial provides us with valuable information on the deficiencies in our approach, which is helpful to inspire future technological developments. We would be pleased if this work inspired rigorous research in the social aspects of human-robot interaction.We did not associate videotaped interactions with tag data from each child. We believe this kind of fine-grained analysis would be particularly useful, for example, in checking the number of utterances of each child and in observing how each child initiated interaction with the robot. In future research, it would be very helpful to code the videotapes and thereby combine qualitative observations with tag data, which lack detailed information.
