WHY ELICITED PRODUCTION? Since the early years of research on language development, the most common source of data used to infer young children’s linguistic knowledge has been their spontaneously produced utterances. The popularity of such data stems in part from the relative ease and naturalness of its collection: With many children who have begun to talk, the researcher need only create a comfortable and interesting environment and prepare to deal with an abundance of linguistic information. The recent availability of computerized databases of transcribed spontaneous utterances allows researchers to analyze this information without even interacting with children at all. What could be better?