If a picture paints a thousand words, then it cannot be just an utterance. Then pictures are more like texts, expressed in a code in a context. They are discourse events when they are produced or seen, and discourse documents when they are analysed as structure and function. This means that they tell stories, present arguments, or explain complex ideas. Just like extended metaphors and allegories, pictures therefore also work at the level of discourse more than language use. This also holds when they are part of verbal text, so that the whole text becomes multimodal.

For DMT, the approach to visual metaphor focuses on the role of the source domain. Visuals typically become metaphorical because many or most of their elements usually work as source-domain referents. They place the source domain within the situation model. As a result, most visual metaphors would also typically be deliberate. We collected and annotated several hundreds of examples in the Vismet corpus (VisMet).  DMT can do worse than revisit these materials for analysis and further research.