This study investigates how children learn not to overgeneralize about grammatical forms and how to reformulate hypotheses about the grammar of their language even when receiving little or no explicit feedback. Two proposals were looked at: (1) input monitoring theory stating that certain overgeneralizations are eliminated from production because children do not hear them in the input, and (2) a perspective that views much of the impetus for linguistic reorganization as coming from within the child rather than through input. The degree to which each of these perspectives explains children's reformulation of erroneous hypotheses was examined with 35 native English speakers aged 3.9 to 9.8. The children were given one production and two grammaticality judgment tasks. In one of the judgment tasks, the children were asked to judge a verb form in a single sentence, and in the other they were asked to both find nonsensical concepts and judge verb grammaticality. Results gave further evidence that children eliminate errors by monitoring the discrepancy between their production and their experience or input. However, when the experimental situation was modified so children were required to process information about both form and content, children did not acquire language to match an adult standard. (MSE)