The re-conceptualisation of texts over the last 20 years, as well as the development of a multimodal understanding of communication and representation of knowledge, has profound consequences for the reading and understanding of multimodal texts, not least in educational contexts. However, if teachers and students are given tools to "unwrap" multimodal texts, they can develop a deeper understanding of texts, information structures, and the textual organisation of knowledge. This article presents a model for working with multimodal texts in education with the intention to highlight mutual multimodal text analysis in relation to the subject content. Examples are taken from a Singaporean science textbook as well as a Chilean science textbook, in order to demonstrate that the framework is versatile and applicable across different cultural contexts. The model takes into account the following aspects of texts: the general structure, how different semiotic resources operate, the ways in which different resources are combined (including coherence), the use of figurative language, and explicit/implicit values. Since learning operates on different dimensions -- such as social and affective dimensions besides the cognitive ones -- our inclusion of figurative language and values as components for textual analysis is a contribution to multimodal text analysis for learning.