This article examines speakers' use of French chais pas ("dunno") when they find that their sequence-initial turn has not been responded to by their recipient (roughly: Speaker A: Maybe they're doing a master's degree; Speaker B: [no response]; Speaker A: Dunno). Two interactional workings are documented in this precise sequential location: Speakers use chais pas either for withdrawing their just-produced sequence-initial action, thereby canceling the relevance of a response or, on the contrary, for pursuing response while relaxing the preference for a precise type of response. Collection-based analysis shows that these uses differ in their embodied delivery, implementing distinct interactional workings with distinct sequential consequentialities. The findings add to our understanding of how grammar and the body interface in the course of the realtime production of turns and actions and provide evidence for the online malleability of action projection. Data are in French with English translations. Language is one among a range of resources that participants draw upon for the collaborative organization of social interaction (Goodwin, 1979; Ochs, Schegloff, & Thompson, 1996; Sacks, 1992). Linguistic resources are "positionally sensitive" (Schegloff, 1996, p. 63), that is, the workings of grammatical constructions can be differentiated by reference to position in turn and sequence. These workings are complexly intertwined with participants' embodied conduct (Goodwin, 1979). Based on this intertwinedness of multiple (semiotic) resources, participants produce and understand actions in accountable ways within situated social interaction: "the resources of the language, the body, the environment of the interaction, and position in the interaction [are] fashioned into conformations designed to be, and to be recognizable by recipients as, particular actions" (Schegloff, 2007, p. xiv). Therefore, a fine-grained understanding of action formation and ascription calls for close scrutiny of how grammar, placement, and co-occurring embodied conduct interface in the course of the real-time production of turns and actions (Keevallik, 2013, 2018; Mondada, 2014). In this article I am interested in the in-principle possibility of grammar, position within turn and sequence, and co-occurring embodied conduct to be interrelated in systematic ways. By systematic ways, I mean the recurrent occurrence of grammar-position-body constellations for specifiable interactional purposes (see Goodwin, 2007; Streeck, 2009, on multimodal action packages; see also Goodwin, 2013). Based on an understanding of grammar as a set of emergent and adaptive resources that are instrumental in participants' dealing with the unfolding of social interaction in real time (