Abstract
Recently much of the effort of child-language investigators has been invested in writing child grammars. It is argued that there is much diversity between children and that too much leeway is left in formalizing the grammar of even a single child. As a consequence, writing a child grammar is fruitless, unless it is part of a developmental theory of child language. A methodological implication of such an approach is identified. The final part of the article stresses the need for an investigation of crosslinguistic differences in language acquisition.