Drawing on findings from dialect acquisition, historical linguistics, and variationist sociolinguistics, this talk addresses one of the most long-standing and central questions in the study of language change: How does change advance across successive generations of speakers? Past research has engaged in post-hoc theorizing about the continuous advancement of change but it has never addressed it directly. Because children must speak differently from their parents for any change to both survive and progress, only real time observation of the same speakers can provide answers to this question. I introduce a project that explicitly sets out to observe the onset and early progression of change in order to track the diachronic evolution of specific linguistic features, zeroing in on the period when children begin to participate in change by shifting their language model along an apparently pre-set direction of change.