How selective are 3-year-olds in imitating novel linguistic material?

In 3 studies we explored when 3-year-olds would imitate novel words in utterances produced by adult speakers. Child and experimenter took turns in requesting objects from a game master. The experimenter always went first and always preceded the object's familiar name with a novel adjective (e.g., "the dilsige duck"). In the first 2 experiments, we found that children were more likely to reproduce the adjective when there were 2 different instances of the same object present in the situation than when there was only 1 or when there were 2 objects of different types present. Thus, children seemed to be sensitive to the descriptive and contrastive function of the adjectives in determining which parts of the utterances to reproduce. Nonetheless, replication of even redundant material was over 50%, suggesting a strategy of somewhat blind copying. In the 3rd experiment, we found that children were less likely to reproduce a redundant adjective when the speaker indicated gesturally that he did not intend to produce it than when he clearly produced it intentionally. We distinguish insightful imitation (the copying of a speaker's goal and means when motivated by insight into why those particular means were chosen) and blind imitation (the copying of a speaker's goal and means with no awareness of why those specific means were chosen) from mimicry. We explore the roles that these modes of imitation might play in language development.