Manipulating cognitive demands on second language (L2) tasks, along with the provision of recasts and its effects on L2 development, has motivated recent inquiry within task-based research. However, empirical evidence remains inconclusive as to the impact of task complexity, and it is unknown how it may affect heritage language (HL) development. To address this issue, this study tested 81 adult HL and L2 learners of Spanish. Participants in the experimental conditions completed either a simple or a complex version of a monologic computerized task that delivered written recasts as corrective feedback but differed according to intentional reasoning demands. Participants completed three oral and written assessment tasks to measure development of the Spanish subjunctive in adjectival clauses. Results revealed that the simple group demonstrated greater gains, especially in written production. L2 learners and the HL simple group benefitted more from task-based instruction in comparison to the HL complex group. Findings have implications for the role of prior language experience and task outcomes.