Methodological issues in language acquisition research with very young children

It is suggested that the dominance of observational methods in child language research with very young children has limited progress in understanding the language acquisition process. Evidence is presented that, contrary to popular belief, naturalistic methods do not guarantee ecological validity, nor are they free of task biases. Furthermore, observational studies do not necessarily provide better measures of linguistic competence than do more structured methods. Instead, it is argued that new insights from cognitive developmental research are applicable in studying language production in children younger than 3 years. A number of structured techniques are reviewed that have been used or could be used to study language production. It is concluded that a combination of methods is necessary in order to disentangle and control the many variables that enter into the language acquisition process.