Lars Christian Jensen, University of Southern Denmark, DK

Social robots are employed in many classrooms and have been shown to aid learning. However, studies show that while schools intend for these robots to be social actors, they are not treated as such by the students. As the social factor is crucial for interactional engagement, this talk discusses how students’ engagement with a social robot can be systematically investigated and evaluated using Wittgenstein’s metaphor of “language games” in a conversation analytical (CA) framework. CA works from an understanding that human conversations are made up of adjacency pairs, a turn-taking procedure, in which two utterances produced by different speakers are functionally dependent on each other. The pair parts of an adjacency pair exemplify very well Wittgenstein’s concept of language games, in that there are interactional rules for them that are beyond mere grammar. Specifically, analyses of adjacency pairs show the extent to which ``language games'' initiated by one party is taken up by the other, while analyses of repair initiations and uptake show what happens in the interaction when this happens. In order to expose how such an analysis can proceed, I present a small user study in which a robot plays a word formation game with single human participants, in which engagement is determined by means of an analysis of the ”language games” played with the robot.