(Mis)alignment and (dis)affiliation within feedback sequences in face-to-face interactions

Using a videotaped corpus of semi-spontaneous interactions between French university students, this paper explores how alignment and affiliation may structurally and formally differ in the context of feedback, combining quantitative annotations with qualitative analyses. Our annotation scheme comprises several categories, ranging from modality, type, and value, following previous corpus-based models. Our findings show that alignment tends to be more frequent than affiliation, where listeners seem to prioritize comprehension signals over agreement or stance. The visual-gestural modality also plays an essential role, especially in disaffiliative feedback, while disalignment remains mainly verbal. Recurrent forms, such as nodding or the interjection ah point to the multifunctionality of feedback, whose meaning is largely context-bound. Overall, while alignment and affiliation often overlap, systematic annotations as well as micro-analyses of the data highlight subtle distinctions. This study on multimodal feedback offers insights into how listeners' contributions are essential for shaping interaction.