In this article, a learner-centered pedagogical process for scaffolding a deliberate use of MT is presented with the goal of promoting student agency and personal expression. By developing awareness that translations entail contextually-sensitive options, students learn to critically assess different forms while actively engaging with translation software. Grounded within SLA research on interaction and negotiation of meaning, our meta-translation feedback circuit supports form-function mappings whereby students analyze and potentially adjust machine-generated translations. Within this functional approach, each component involves a series of questions adaptable to varying proficiency levels and languages. The first set invites students to situate the speech activity within its sociopragmatic context and to make explicit connections with recently studied topics. The second set helps students investigate MT's output through a formal language analysis of referents within and across sentences. The third focuses on integration by checking for adequacy of fit between forms and situated meaning. The feedback circuit is illustrated in the context of a 3rd semester French course and is followed by pedagogical strategies applicable to any foreign language classroom. Embedding computer-aided translation into an otherwise traditional L2 task represents an opportunity to foster dialogue on MT, create a teacher-mediated metalinguistic analysis of MT output, connect with the language learners' 'toolbox', and support intentional engagement with the activity itself.