We commend Paul for undertaking an investigation that concerns critical clinical and theoretical issues. This type of longitudinal developmental research is exactly what is needed to advance the scientific basis of our profession. We also respect Paul's attempt to construct a firm bridge between her findings and their clinical implications. The necessary and sufficient data that completely solve a clinical problem are rarely available. Because clinicians do not have the luxury of waiting until the best data imaginable are collected and analyzed before acting, it is helpful for researchers to generalize their results to the extent possible. However, because of its potential clinical, economic, and educational implications, we believe that a broad social policy like Paul's "watch and see" recommendation should be based on unambiguous evidence.
We have suggested that a number of the children in Paul's cohort may have been within the normal range in language development at the beginning of the study. Without individual data, it is impossible for us to know whether or not this was the case. To the extent that our suspicions hold true, Paul's study tells us that a number of children who function at the low end of the normal range of language development between 20 and 34 months stay within the normal range throughout the preschool and early school-age years.
Paul's suggestion of "watch and see" seems reasonable enough for the 74% of the children who tested within the normal range by kindergarten and first grade, but it may not have been sufficient for the 26% who did not. We believe children like those in this latter group would probably benefit from preschool language intervention and that very valuable language learning time could be lost if Paul's general "watch and see" policy were implemented.
It is possible that children with good outcomes and children with language delays that were significant and persistent had different profiles with respect to expressive vocabulary, receptive vocabulary, speech, and communicative intentions at the onset of the study. If so, one broad social policy may not be sufficient. We have asked Paul to provide additional data about the nature of the language difficulties exhibited by the children at the outset of her study, the predictors of continued language delay, and the results of language intervention efforts. It is our hope that Paul can provide the kinds of additional data and analyses we have requested in this discussion, and that this data can serve as the basis for refinements in definitions of early language delay, decisions about providing clinical services to very young children, and methods for analyzing intervention efficacy.