摘要：Although there are studies on multimodal elements for more than one decade ago in Linguistics studies, only in the past five years the Text Linguistics took into account non verbal semiotic resources in their discoursive/text analyzes. This paper aims to investigate how multimodal resources can also be treated as referring expressions that recategorizate discourse objects. For this, we investigated a corpus of ten multimodal texts published on social networking sites whose characteristic is the recategorization of the referent. The results show that the visual elements belonging to the composition of multimodal texts are also central to the construction of the meaning of the text, so that they are often responsible for the recategorization process.
