abstract

Event Categorization in Language and Cognition: The Case of “Cutting and Breaking”

To be able to talk about their experiences, speakers have to parse the ongoing perceptual flow into units and categorize these units as instances of recurrent event types such as “running” or “breaking something”. Do the verbs of different languages group and distinguish events in largely the same way, perhaps due to biases in human nonlinguistic event cognition or to clusterings of event properties in the real world? Or do languages partition events in an infinite number of crosscutting ways? How do children arrive at the event categories they need for language? These questions can be explored by asking speakers of a range of genetically and areally diverse languages to describe the same sets of events, and then comparing their responses with the help of multivariate statistics in search of universal, language-specific, and age-related patterns in the semantic structuring of the domain.

This approach is illustrated with results from two Max Planck group projects on the categorization of events that have played an important role in lexical semantic theorizing: “separation in material integrity” (“cutting and breaking”, for short). The first project (adult language structure) shows that event categorization is constrained, but by no means universal: languages organize their cutting and breaking event categories according to a shared dimensional structure or “semantic space”, but they differ strikingly in the number of categories they distinguish and in the placement of category boundaries. The resulting fine-grained cross-linguistic variability in category structure raises difficult questions for the semantic representation of predicate meanings.

The second project (acquisition) shows that children approximate the “cutting and breaking” categories of their target language by as early as 2 years of age, although they are still adjusting and fine-tuning category boundaries even as late as 6 years. Learners' remarkable sensitivity to the categorization of events in the language they hear challenges a widespread developmentalist assumption: that early verb learning is simply a process of mapping words to a basic set of cognitively pre-established event types. The evidence suggests instead that event categories are formed, to a large extent, through experience with the semantic structure of a particular language.