Modelling the selective effects of slowed-down speech in pronou

Background. Up to the age of 6, children have been shown to experience difficulties in the interpretation of pronouns (but not reflexives) by incorrectly allowing the pronoun to corefer with the local subject about half the time (e.g., Chien & Wexler, 1990). Explanations for this Delay of Principle B-Effect (DPBE) range from a deficiency in pragmatic skills (Thornton & Wexler, 1999) and the effects of an unbalanced context (Conroy et al., ms), to limitations in processing resources (Reinhart, 2006; Hendriks, van Rijn & Valkenier, 2007). Hendriks et al. argue that the DPBE arises because children are unable to take into account the speaker’s perspective. In their computational ACT-R (Anderson et al., 2004) model of the development of pronoun comprehension, adult pronoun comprehension is simulated as two consecutive steps: selecting the optimal meaning for the (ambiguous) pronoun, and then checking whether a speaker would have expressed this meaning with the same form. If children’s speed of processing is not fast enough to perform both steps within a limited amount of time, pronouns remain ambiguous and a guessing pattern emerges. The ACT-R model thus predicts that children’s performance will improve if given sufficient time to construct an interpretation. Slowing down the speech rate is a way to give children more time. This prediction is tested by comparing pronoun comprehension at a normal rate with slow speech.