1 The development of pragmatic abilities Myrto Grigoroglou & Anna Papafragou 1. Introduction It is widely assumed that human communication in adults relies on a complex species of mind reading that involves inferentially reconstructing the meaning that the speaker had in mind and wanted to convey (Grice, 1975; Sperber & Wilson, 1986/1995). Children’s pragmatic abilities, however, present a paradox. Experimental work has shown that remarkable competence in important preconditions for pragmatic reasoning is present from infancy (see Baillargeon, Scott, & Bian, 2016 for a recent review); nevertheless, research on a variety of linguistic phenomena (e.g., reference, implicature, metaphor, irony) has demonstrated that children’s pragmatic abilities are fragile and task-dependent even until late in childhood (e.g., Bucciarelli, Colle, & Bara, 2003; Filippova & Astington, 2008; Noveck, 2001; Waggoner & Palermo, 1989; Winner & Leekam, 1991, a.o.). Sometimes, the paradox surfaces even within the realm of a single phenomenon. For instance, in the domain of referential communication, some experimental evidence shows that even 2-year-old children can successfully integrate another person’s perspective in both their comprehension and production of referring expressions (Morisseau, Davies, & Matthews, 2013; Nadig & Sedivy, 2002; Nilsen & Graham, 2009; O’Neill, 1996), yet other evidence shows that children frequently ignore their interlocutor’s perspective as late as age 8 or 9 (e.g., Deutsch & Pechmann, 1982; Epley, Morewedge, & Keysar, 2004; Girbau, 2001). For other phenomena, there is debate in the developmental literature about whether they truly rely on rich pragmatic computations about others’ intentions or might involve non-pragmatic mechanisms. In word