Learning Simple Spatial Terms: Core and More.

How do children learn the meanings of simple spatial prepositions like in and on? In this paper, I argue that children come to spatial term learning with an a priori conceptual distinction between core versus non-core concepts of containment and support, and that they learn how language maps onto this distinction by considering both the simple prepositions and the company they keep-that is, the distributions of their co-occurrences with particular verbs. Core types of containment and support are largely expressed by in/on together with the light verb BE; non-core types are expressed by lexical verbs such as insert, hang, stick, and so on, which represent the specific mechanical means by which containment or support is achieved. These latter types arguably depend on extensive learning about the particular mechanisms of containment and support, many of which are invented by humans, as well as learning the specific lexical verbs that encode these mechanisms. The core versus non-core distinction is reflected in young children's and adults' linguistic descriptions of different spatial configurations, via different distributions of expression types across different configurations. Differences between children and adults are not likely to be rooted in either conceptual or semantic differences, but rather, in the probabilistic nature of available expressions, along with early limits on children's vocabulary of lexical verbs that express complex mechanical relationships between objects.