Increasingly, learning in science and mathematics is considered in terms of induction into the multimodal language practices of the disciplinary community. A strong strand of research in this tradition has involved students being challenged to invent multimodal language forms, and their ideas refined through structured guidance. Often, however, research into these inquiry processes focuses on student learning, such that detailed descriptions of teaching strategies are not well represented in the literature. In this paper, we explore the nature of teacher orchestration of student learning in a guided-inquiry classroom sequence on the multimodal language of data modelling. The paper describes a classroom sequence in which nine-year-old students measure, invent, compare, and revise data displays to explore ideas about data variation related to their teacher's arm span. Data included video capture of the teacher's orchestration of tasks and classroom discussion, student artefacts, and teacher and student interviews. We examine the classroom processes, including task construction and individual, group and class discursive production, through which the teacher orchestrates students' invention and revision of data displays. These patterns of discursive moves offer fresh insights into the ways in which teachers can support deeper levels of student learning of foundational multimodal science and mathematics literacy practices.