Abstract

This qualitative study proposes a multimodal framework for modelling subordinate constructions in spontaneous conversation, based on their action on interpretative frames. Subordinate constructions have long been described in linguistics as dependent elements elaborating upon some primary features. However, Cognitive Grammar has challenged this view in showing that syntactic embedding often only reflects the starting point speakers choose to convey their message. Subordinate constructions are practices in interaction that offer an interpretative reconstruction of discourse. The different syntactic types of subordinate constructions refer to different interpretative frames in the speaker’s experience. The selection of these interpretative frames is expressed by the different amount of prosodic and gestural boundary cues produced in co-occurrence with each syntactic type of subordinate construction.