The Development of Informativeness from a Speaker's and a Comprehender's Perspective
We investigate the ability to produce and to evaluate informative utterances in 5-, 7-, 9-, and 11-year-old English-speaking children and adults. Experiment 1 shows (i) that informativeness develops with age, (ii) that there is a delay in development from the comprehenders perspective compared to the speakers perspective, and (iii) that the type of expressions used in an utterance affects both child and adult performance (but in different ways). With regards to (ii), the speaker-comprehender asymmetry, it is further shown that in a sentence-to- picture matching task (Experiment 2) and in a task where participants are asked to rate under-informative utterances on a five-point scale (Experiment 3) rather than categorically accepting or rejecting them (as in Experiment 1), young children are consistently informative comprehenders and the asymmetry disappears. These findings suggest that young children are sensitive to but also tolerant towards violations of informativeness: they appear to be under- informative comprehenders only in experimental tasks which require them to display their competence by categorically rejecting a pragmatically infelicitous utterance. We argue that pragmatic tolerance predicts the major findings of other studies in this area, and in experiment 4 we show that adults too exhibit such tolerance. In Experiment 5 a theory of mind component is embedded in the comprehension task of Experiment 1, and it is shown that both children and adults take into account their interlocutors epistemic state when evaluating informativeness. Overall, we propose that English-speaking children from age 5 onwards display all the aspects of competence that are predicted by pragmatic theory, both as speakers and comprehenders. However, young children in particular are tolerant towards pragmatic infelicity and, while their ability to derive informative interpretations is adult-like, their commitment on whether under-informative utterances ought to be rejected is not. The implications of this research for the methodology of language acquisition and the recent theoretical debate on the defaultness of pragmatic inferences are also discussed.