This article provides a microscopic view of learners' first encounters with multimodal texts in their primary EFL classrooms. It is argued that multimodal texts create opportunities for language development in the primary EFL classroom: they offer different access points for comprehension, invite participation, and motivate repeated practice so that samples of natural language are memorized and can become part of the learners' own language repertoire. Audio-recordings of three sessions in different primary classes were analysed to gain a better understanding of young learners' engagement with different types of multimodal texts, such as an action story and video clips of one song and one chant. The eight- to ten-year-old learners responded to the different modes of the input, i.e. movement, sound, image, and speech, by repeatedly joining in with accompanying actions and sound effects, and by imitating language unprompted. With every encounter with the multimodal text, their verbal contributions grew.