This study examined the relationship between variation in maternal language and variation in language development in a group of 12 children with expressive vocabulary delays. Mothers and their children participated in a parent-mediated intervention that adhered to the interactive model of language intervention. This intervention model arises out of social interactionist accounts of language acquisition and maintains that maternal language input has facilitatory effects on child development. The purpose of this study was to examine two compatible explanations for the facilitatory effects of maternal linguistic input in this intervention model: the responsivity hypothesis and the structural hypothesis. The responsivity hypothesis maintains that linguistic input that is semantically contingent on the child’s vocal or verbal utterances, or responsive to the child’s focus, facilitates language learning. The structural hypothesis posits that structural features of maternal language input that are just one step above the child’s abilities promote language learning. The results of this study indicated robust relationships between maternal use of imitation and expansion at Time 1 and measures of child language at Time 2. These results provided support for the effects of responsive language input on the language abilities of this sample of late talkers. These results have implications for social interaction theory and confirm the import of responsive input as viable intervention techniques for young children with expressive vocabulary delays.