Predicting Scalar Implicature Interpretations From Lexical Knowledge

We measure typically developing monolingual child Spanish speakers' lexical development with a range of standard expressive and receptive tests. We also measure their comprehension of sentences with the existential quantifier algunos "some" to determine their abilities to generate "some, but not all" scalar implicatures or pragmatically enriched quantifier interpretations. We then determine the degree to which lexical development predicts implicature interpretations.

We fit regression models with lexical measures as predictor variables and implicature interpretations as the outcome variable. We then divide the child sample into implicature generators (50/61) and implicature nongenerators (11/61) and test the usefulness of the four lexical measures in a linear discriminant function analysis to separate children into these two categories.

Results show significant correlations between each lexical measure and the outcome variable and, in a regression, that three of four lexical measures account for unique variance. Furthermore, the linear discriminant function analysis separates children into implicature nongenerators with 100% accuracy (11/11) and implicature generators with 88% accuracy (44/50).

The Quantity Scale, or set of quantity-expressing determiners, proposed by Horn and Grice, develops as a function of the links among its quantifiers. We speculate that children's lexicons refract approximate number system representations in language- and morpheme-specific ways. These quantified noun phrases (NPs) are then merged into sentences interpreted pragmatically with conversationally computed implicatures, using higher order reasoning.