One factor that enables language to be used creatively is the fact that, within a given language, formal patterns correlate strongly with the meaning of the utterance in which they appear. For example, speakers easily infer that the sentence "She mooped him something" means that she transferred something to him, even though they have had no previous experience with the nonsense verb "moop." Until recently, there has been virtually no research investigating exactly how children learn to associate a phrasal formal pattern with a novel meaning. This proposal offers a new training paradigm that is designed to investigate various factors involved in learning these correspondences. Initial results suggest that language learning is akin to the learning of other sorts of categories: it involves an impressive ability to generalize quickly on the basis of few examples. Results from this research will have a potentially broad impact in the fields of first and second language acquisition, language pathology, and language pedagogy since the studies focus on factors that aid or inhibit successful language learning. For example, previous research has shown that skewed input that emphasizes one type of example facilitates generalization beyond the training data; this implies that it may help second language learners or people with language disorders to be instructed extensively on a particular type of example initially, before the full range of possible extensions are introduced. The work promises to yield insights into the nature of our knowledge of form-meaning pairings (constructions) and into the development of this knowledge.
