Twenty children (10 from a Day Care Center and 10 from a Head Start Center) were administered a 28-item, parallel form language comprehension task. The method utilized concrete materials (i.e., puppets and other familiar objects, spoon, flower, ball) which subjects manipulated when presented with sentences of 7 different grammatical constructions. There were 2 main effects, that of sentence type (active versus passive voice) and that of different preschool groups (Head Start Center versus Day Care Center), with no significant interaction. The reliability coefficient of Form A versus Form B was significant at the .001 level. (Author)