The present study concerns the multimodal nature of music education. How children (aged 4-8 years) respond when faced with the challenge of talking about what they hear in pieces of music is studied. The semiotic tools children and their teachers use in these situations and how they transduce between modalities (verbal, sound, colours and gestures) are analysed. Paradoxically, this analytical focus on multimodality results in an emphasis on the importance of verbal language in the learning situations studied. Some implications for music education are discussed.