This qualitative study aimed to investigate adolescent English learners' multimodal communicative competence. Grounded upon a social semiotic approach with the concepts of affordances and transmediation, this research was carried out in an after-school programme at a secondary school in Korea. Three high school students participated in the multimodal composition project and produced narrative and argumentative multimodal texts. Multiple sources of data were gathered from interviews, video recordings with fieldnotes and student artefacts. The data were analysed using an inductive and deductive approach, along three orientations: modes as social action; modes as framing devices; and modes as agency representation. The students' multimodal texts were analysed based on multimodal text analysis. This study demonstrated that the participants personalised, avoided, modified and orchestrated semiotic modes to engage in multimodal composition. The study also revealed that the students' selection and use of semiotic modes were influenced by their audience awareness, genre knowledge and identities, and vice versa, all of which contributed to developing their multimodal communicative competence. The paper concludes with pedagogical implications for multimodal composition practices and suggestions for future multimodal composition studies.