Developmental data on young children’s verbalizations in naturally-occurring pretend play are used here to help understanding, on the one hand, the pretending status of children’s play activities and, on the other, children’s emerging abilities to take into account, on a practical level, the mental state of their interlocutors. In particular, it is the study of the relationship between children's meaningful actions and verbalizations that will be considered to throw some light on these issues. Though developmental changes in the use of substitute objects and in the organization of pretend play by young children have been extensively discussed, changes in aspects of pretend play that young children choose to put into language have rarely been the object of study. Our analysis shows that children may choose to talk about the ‘LITERAL’ aspects of the pretend situation and, in this case, their verbalizations do not refer to any aspect of pretend meaning (e.g. there, placing a baby doll into a toy crib in a context where she is apparently putting the doll to sleep). When children talk about pretend meanings, they may refer to those aspects of their play whose meanings are already sufficiently evident from the objects and actions performed: in this case verbalizations DUPLICATE or at most make more explicit these meanings (e.g. night-night, in the same situation as above). Sometimes, however, children’s verbalizations are determinant in CLARIFYING and/or in SPECIFYING pretend meanings (e.g. drop, for nose drops, while touching with her index the face of the doll just placed in the toy crib), or express meanings that have NO COUNTERPART in the activity part of the pretend play (e.g. cold, before placing a toy quilt on the baby lying in the toy crib). In this case children’s verbalizations provide to the onlookers, or to the potential participants, RELEVANT INFORMATION about the pretending aspect(s) of their play activities. The results obtained from the longitudinal studies of four children in the process of acquiring French (age range covered: 15-27 months) show developmental changes in the pattern of the different types of verbalizations described above, with an increase in the last two “informative” types of language expression. These results are related to those obtained in other studies of “informative” uses of language, in particular, making reference to past events and providing explanations/justifications. The comparison shows that these language uses appear at about the same time in the development of each of the four children. Given the diversity of the domains, and in line with other authors, we make the hypothesis that these new pragmatic behaviors reflect a first important developmental step in children’s ability to attribute internal states, intentional, emotional and mental, to their interlocutors, reflecting a beginning of a "know-how” about the mind.