In order to describe complex predicates of visual perception with a support verb, we focus, at the formal pole of the construction, on the lexical variation around the verbal slot, without losing sight of its combination with predicating nominal elements. We intend to analyze whether there are predicating nominal elements of visual perception more or less strongly linked to each verb in the construction. Therefore, we aim to verify possibilities for perspectivizing the visual perception event. To do so, we compared data from two varieties of Portuguese (from Brazil and from Portugal), in comparison to French and English. We adopted the theoretical approach of Construction Grammar (GOLDBERG, 1995; 2006), taking into account its articulation with Sociolinguistics, which results in a socio-constructionist bias (MACHADO VIEIRA; WIEDEMER, 2019). We collect construction incidents via the Sketch Engine platform and we use the Word Sketch tool to check collocations. After selecting only the data that interests us, we arrive at 967 occurrences for Portuguese and 750 occurrences for French and 769 occurrences for English. We analyzed the data using quantitative and qualitative methodology. Results (AUTHOR, YEAR) showed that, in the use of the construction in question, Portuguese and French are similar in the sense that they both present prototypical support verbs (dar/ ‘to give’ and jeter/ ‘to throw’) that reflect the focus of the action on something external. English, on the other hand, presents a strong tendency to use the prototypical support verb take in the construction in focus, so that movement or the action, even if figurative, is in the sense of the participant's field of reference. These results confirmed our hypotheses that the most frequent support verbs in the construction of complex predicates of visual perception in Portuguese, French and English would be, respectively, dar (‘to give’), jeter (‘to throw’) and take.