In research on children's language development, joint book reading is appreciated as being a situation beneficial for language learning. Motivated by this string of research, our aim was to explore whether the situation of joint book reading can be applied to a child–robot interaction. Before investigating whether and how this situation – when applied in child–robot interaction – can be used as a language learning scenario, the interactional requirements for a successful dialogue have to be studied. Our main aim in this paper is to present a study design for a child–robot interaction, in which a robot is introduced as a learner that acquires new color words, and the child is asked to teach the robot those words within a familiar interaction format of joint book reading. We then report the observations that we made in a single-case pilot study conducted with a 4;8-year-old child, and discuss these observations in terms of how the robot's interactional behavior needs to be shaped to successfully participate in the situation of joint book reading.