Since 1968 I have taught a course called “Early Language Development” at Teachers College, Columbia University. The subject matter has changed over the years, but one prevailing theme has not. The theme is what I call the accountability of evidence in child language research. We need to hold whatever evidence we obtain accountable to the conclusions we draw. Talk is cheap: Young children are talking all the time, and it is easy enough to write down or record what they say. The ultimate value children's talk may have as data and evidence depends upon the methods we use to process the talk and the interpretations we give it. The value of talk as data depends upon the methods we use. The value of data as evidence depends upon the interpretations we make. Both theory and method inform one another in the transformation Talk ⇒ data ⇒ evidence, and both change over time with advances in technology, in what we learn about language and its acquisition, and in what we learn about children and their development.

The research reported here began with several basic theoretical and methodological assumptions. The first theoretical assumption was that infants learn language in their endeavor to express and articulate some aspect of a mental meaning – some representation of the way the world appears to be or the way they want the world to be in their beliefs, desires, or feelings.