Mmmm

Shifting Standards: Children’s Understanding of Gradable Adjectives*
Kristen Syrett, Evan Bradley, Christopher Kennedy, Jeffrey Lidz Northwestern University

Two studies demonstrate that children have knowledge of scalar distinctions between three sub-classes of gradable adjectives: relative (big, long), absolute with a maximal standard (full), and absolute with a minimal standard (spotted).Performance on these adjectives is compared with controls (shape, color, mood). Children appropriately shift the standard of comparison with context-dependent, relative gradable adjectives, and do not do so for the others. Reasons for non-adult-like performance with full are discussed. Evidence is presented that children know about the presuppositions of singular definite descriptions, suggestingthat children as young as three have an accurate semantic representation of the.

1. Introduction Children’s acquisition of adjectives has received a great deal of attention in the field of psychology. The vast majority of this work has focused on a core set of topics: children’s ability to distinguish properties from kinds of objects (cf. Gelman and Markman 1985, Waxman and Booth 2001); theinfluence of exemplars of or reference to object category on children’s ability to extend adjectives to properties (cf. Klibanoff and Waxman 2000, Mintz and Gleitman 2002); the role of comparison and contrast in adjective learning (cf. Clark 1972 and 1973, Ehri 1976, Gentner and Rattermann 1998, Ryalls 2000); and the distinctions children make between adjectives of color, dimension, and property (cf.Bartlett 1976, Nelson and Benedict 1974, Sandhofer and Smith 1999). The current investigation adds to this body of research by presenting findings regarding children’s comprehension of the semantic distinctions within the class of gradable adjectives.

*Portions of this work were presented at the LSA Annual Meeting in Oakland, CA, 2005. This study was funded in part from an NSF Grant(BCS-0094263) to Kennedy, an NSF Grant (BCS-0418309) to Lidz, and an NIH grant (HD30410) to Sandra Waxman for the Project on Child Development at Northwestern. Special thanks go to the children and staff of Chiaravalle Montessori School and Warren W. Cherry Preschool in Evanston, IL, and to Erin Leddon and Josh Viau for help running the experiments. We are grateful to members of the Northwestern AcquisitionLab Group for helpful comments.

2. Linguistic Background Gradable adjectives (often referred to as scalar, dimensional, size, or spatial) are adjectives such as big or tall, which describe properties of objects that hold to different degrees. For this reason, gradable adjectives can appear in comparative constructions, while non-gradable adjectives cannot. This observation is illustrated inexample (1). (1) Neanderthal man is [more primitive than / *more extinct than] Homo sapiens.

A standard semantic analysis of gradable adjectives is that they include as a core part of their meaning a function that takes as its input an object and returns as its output a measure of the extent to which that object possesses a property denoted by the adjective (see Klein 1991). This measurement canbe formally represented as values, or degrees, in an ordered set, or scale. The values of the scale correspond to the dimension labeled by the adjective (e.g., height, weight, age, etc.) (see Kennedy 1999, Kennedy and McNally 2005). On this view, sentences constructed out of gradable adjectives express orderings between degrees. This is most clearly illustrated by comparatives like (1), whichexpress explicit orderings between degrees on a scale. Although the positive (unmarked) form of a gradable adjective lacks overt comparative morphology, it, too, describes a relation between degrees on a scale, albeit an implicit one (cf. Sapir 1944). In this case, the degree to which an object possesses some property is related to a context-dependent standard of comparison, which represents the...