The present study analysed formerly unexamined indirect routes for relationships between time 1 maternal speech and later child language development. Ten normally developing children in Brown's early stage 1 and their mothers were the subjects. For each dyad, two free-play sessions occurring five months apart were videotaped in the subjects' homes. Mothers' pragmatic language use was coded from time 1 sessions. Child language level was coded at both sessions. Even though time 1 scores of the outcome were controlled, seven of the ten relationships involving mother speech and child language development were indirectly related through one of two time 1 child language measures. The results indicate that a mother-driven, direct influence model may be inappropriate for many mother speech-child language development relationships. We argue that child-driven and mother-driven explanatory models for the indirect relationships are equally feasible.