Cory Bill (Leibniz-ZAS Berlin)
Investigating children's interpretations of ‘Every … some’ sentences
Sentences containing the scalar quantifier ‘some’ embedded under the universal quantifier ‘every’, such as Every pig carried some of his rocks (EverySome sentences) have been associated with two scalar inferences: the NotEvery inference that Not every pig carried all of his rocks, and the stronger None inference that None of the pigs carried all of his rocks (Chemla and Spector 2011, among others). Previous research has reported that adults derive both these inferences to some extent, with a preference for the NotEvery inference (Geurts and Pouscoulous 2009; Chemla and Spector 2011, among others). While previous developmental studies have reported that children often struggle to derive scalar inferences from unembedded ‘some’ sentences (Noveck 2001; Chierchia et al. 2001), no previous work has investigated children’s interpretations of embedded EverySome sentences. While no theories seem to make explicit predictions regarding children’s interpretations of such sentences, certain proposals raise the possibility that such sentences may improve children’s performance with scalar inference derivation. For example, the Alternatives-based approach (Reinhart 2006; Barner et al. 2011, among others) proposes that children’s typical difficulties deriving these inferences is a result of limitations in their ability to generate alternatives. It is plausible that the presentation of the stronger scalar alternative ‘every’ in EverySome sentences may help children compute the relevant inferences. There is also the question of which inference children will prefer. The Subset principle (Crain et al., 1994) proposes that, for learnability reasons, children prefer stronger interpretations. If this principle also applies to interpretations derived through inferences, then we might expect children to prefer the stronger None inference over the weaker NotEvery inference. I will present two experiments revealing that children derive inference-based interpretations of EverySome sentences at an adult-like rate, and moreover that they prefer the stronger None inference. I will discuss the implications of these results.