Retour vers liste

Détail de la contribution

Auteur: Shevaun LEWIS

Co-Auteur(s): Kaitlyn Harrigan, University of Maryland, United States Jeffrey Lidz, University of Maryland, United States

3-year-olds compute relevance inferences in indirect requests

Abstract/Résumé: Adults assume that all utterances are intended as relevant and make linking inferences to accommodate indirectly relevant utterances. Previous studies have suggested that children under 6 years are unable to make such inferences. In three experiments using more age-appropriate methodology, we show that preschoolers can in fact make relevance inferences. In each trial a puppet (Froggy) is offered a choice of two objects (e.g. some broccoli and a cookie). Froggy makes a statement related to one object without naming it. The child’s task is to give Froggy “the one he wants”. Experiment 1 tested whether 3-year-olds would interpret Froggy’s statement as an indirect request for the related object when the statement was Positive (1) and the unrelated object when the statement was Negative (2). Children chose the appropriate object in both conditions (Positive: 85%, Negative: 78%). (1) Healthy things make me big and strong. [target = broccoli] (2) Healthy things taste yucky. [target = cookie] Experiment 2 tested whether 4-year-olds could distinguish indirect requests (Positive or Negative) from Irrelevant statements (3) which did not express a positive or negative attitude. Children were instructed to give Froggy a sticker if his statement wasn’t “helpful” and they couldn’t tell what he wanted. (3) Sweet things are made with sugar. [target = sticker] When the statement was Positive, children chose the related object (91%); when Negative, the unrelated object (67%) or the sticker (24%). The sticker response is appropriate for Negative statements, which need not be interpreted as indirect requests. For Irrelevant statements children chose the related object less often (56%) than for Positive, but only rarely chose the sticker (24%), suggesting a limited ability to reject irrelevant statements. Experiment 3 tested whether 3-year-olds could distinguish relevant and irrelevant information presented side-by-side. Froggy delivered two statements in each trial: Positive/Negative, Positive/Irrelevant, or Negative/Irrelevant. Children were highly accurate in the Positive/Negative (89%) and Negative/Irrelevant (81%) conditions. Their performance in the Positive/Irrelevant condition was better than chance (67%), but suggests a limited ability to identify irrelevant information. The overall pattern of results suggests that children have an expectation for relevance, but have difficulty judging what can reasonably be considered relevant. When unsure they err on the side of accommodating the speaker. They are confident of speakers’ general intention to be relevant, but less confident of their own ability to detect relevant information.