This thesis focuses on smiles in thematic transitions of conversations. These moments of negotiation between two themes particularly involve the collaboration of the participants. We have thus studied the role of smiles as ressources of this collaboration. Smiles are very frequent facial expressions but have not been studied in their relation to the organization of the interaction. In parallel, we studied the effect of the interactants’ relationship on this collaboration. To explore these issues, 40 interactions from two corpora (CHEESE! and PACO) were studied. In the first corpus, the participants know each other well, while in the second they met for the first time on the day of the recording. Smiles of the interactants were annotated according to the Smiling Intensity Scale. Thanks to the automatic smile detection tool SMAD, developed for the purpose of this thesis, we conducted a fine-grained analysis of the deployment of the three smile intensities. Our analyses consisted in articulating two complementary approaches: those of Conversational Analysis and Interactional Linguistics via sequential analyses as well as the approach of Corpus Linguistics which consisted in quantitatively interrogating our data. This methodology allowed us to identify three main results concerning transitions in relation to random moments in the conversations studied. (1) The phases of transitions are characterized by specific processes, some of which are invariant regardless of the relationship of the participants. (2) When the speaker initiates a new topic, a suppression of the smile is more frequently observed. (3) When the addressee accepts the thematic proposal, he is more inclined to increase his smile intensity. This thesis shows that the smile is a resource mobilized by the interactants when they make thematic transitions. For this reason, this thesis is in line with the work arguing for the inclusion of the smile as a resource to be taken into account in the multimodal analysis of interactions.
