The purpose of the study is to analyse young children's (Grade 1, ages 7-8) descriptive texts through a social semiotic lens, and to propose a metalinguistic model for analysing children's compositions in terms of intermodality. Previous research highlights the importance of teachers valuing all modes that children use to express their knowledge, identities, and emotions. This study introduces a framework grounded in Systemic Functional Linguistics to address the intermodality present in young children's compositions, focusing on alphabetic writing and drawing. The analysis identifies four categories of intermodal relations: Concurrency, Complementarity, Divergence, and Inactivity. The results of the analysis show that the intermodal relations in children's compositions are primarily word-driven. However, there are a significant number of compositions that challenge this primacy. This variety displays a variety of multimodal competences that were deemed functional by the students: Some compositions were driven by words, some by images and some by both. The framework provides valuable insights into how children use the modes of drawing and alphabetic writing to construct meaning in different ways. By developing a metalanguage for multimodal compositions, teachers can better scaffold a broad range of communicative competencies in young children, fostering richer and more inclusive classroom practices.