To represent the complexity of a visually perceived event, viewers need to attend selectively to different aspects of the event and its associated entities. Spoken descriptions of such complex events must encode the corresponding perceptual properties. This review discusses how the speaker’s attentional focus on one of the referents in a given event influences the structural choice in languages with different degrees of word order flexibility. First, we will discuss whether English speakers prefer to map visually salient referents onto a prominent grammatical role (e.g., Subject) or to a prominent linear position in the sentence (e.g., the sentential starting point). Comparison of this evidence with research in free word-order languages (Russian and Finnish) suggests the existence of a mapping mechanism wherein perceptual salience predominantly affects grammatical-role assignment and, to a lesser extent, assignment of linear positions.