Children make early commitments to sentence structure but often fail to
revise after late-arriving conflicts. To understand why, we engaged
cognitive-control systems by interleaving Stroop trials and active/passive
garden-path sentences. In contexts where children had to overcome an
agent-first bias, Cognitive-control engagement improved 5 year-olds’
interpretation of passives, relative to actives. In contexts without an
agent-first bias, cognitive-control engagement led to parallel effects
across active and passive constructions. These findings are inconsistent
with a resource-depletion account which predicts worse performance on
passive sentences after Incongruent Stroop trials. Instead, this suggests
that cognitive-control engagement actually helps children ignore unreliable
parsing heuristics, such as their agent-first bias.