Key points are not available for this paper at this time.
Forty-three first-grade children who had received no formal instruction in addition and subtraction were individually administered 10 verbal problems. These problems were selected to represent the following semantic types: Joining, Separating, Part-Part-Whole, Comparison, and Equalizing. In spite of the lack of formal instruction, most children successfully solved the problems. In general, children's solution processes were consistent with a predictive model proposing that solution processes would directly represent the action or relationship described in individual problems.
Carpenter et al. (Thu,) studied this question.