This article addresses, experimentally, the question of how presuppositions are cognitively processed and retrieved in discourse. In the proposed research, we have administered tweets produced by Italian politicians to native speakers so as to assess how easily they could retrieve the presupposed content of two presupposition triggers (definite descriptions and change of state verbs), as opposed to their explicit paraphrase, by answering verification questions. Results showed that content presupposed by change of state verbs was likely to receive more attention than (...) content conveyed by definite descriptions; this could possibly be due to the greater effort involved in mentally representing the event taken for granted by the predicates. Definite descriptions, on the contrary, seem to instruct to a shallower processing modality, which means that their content is processed less attentively or in a ‘good-enough’ way. (shrink)