After acknowledging a gap in the field of audiovisual translation (AVT) regarding the theoretical approach of the multiplicity of semiotic elements that compose audiovisual products, this thesis searches for a theoretical path to discuss the interpretation of visual images (images that have a concrete visible form), articulating the concepts of multimodal discourse and interpretive communities. The premise is that texts are composed by different semiotic resources besides verbal language such as images, music, dance etc., which interact without having a hierarchical relationship. The focus is cast on the uses of images in social and discursive practices, the different formats in which they occur and the ways they circulate in society, highlighting the central role of vision. The discussion on visual images is based on the distinction between vision as a physical/physiological operation and visuality as social and discursive determinations that regulate that operation. Such distinction does not imply a dichotomic relationship, but a difference between the mechanism and data of vision and the socially shared rules for their interpretation.