Based on findings from a variety of research, Shin and Miller (2022) propose a 4-step process that children go through as they learn sociolinguistic variation. Their proposal raises many interesting questions that should inspire future research. Here, I discuss their Step 1 -- the stage in which, according to their proposal, children produce only a single variant, and so appear to be (temporarily) regularizing -- and suggest this may not be the right characterization for all children.