Published online by Cambridge University Press: 23 March 2015
Children who have normal language development are aware of the distinction between closed-class and open-class words at a very early age. In order to test to what extent children know the closed class to be, in fact, closed, 104 children aged 3 to 5 years participated in a sentence repetition task. Each sentence contained a nonsense word that fulfilled either an open-class or a closed-class function. Children were more likely to repeat sentences correctly when the nonsense words functioned in open-class, rather than in closed-class, contexts. In addition, older children correctly repeated more sentences containing nonsense words that functioned in closed-class contexts than younger children. This last result shows a mechanism by which children may acquire new closed-class words. The theoretical implications of the results are also discussed relative to children with specific language impairments, especially in terms of their reliance on semantic value in word acquisition.