Prose

Scalar Representations in Natural Language Semantics

NSF CAREER BCS-009423
Chris Kennedy, PI


The ability to establish orderings among objects and make comparisons between them according to the amount or degree to which they possess some gradable property (such as height, brightness, beauty, etc.) is a basic component of human cognition. Natural languages reflect this fact: all languages have syntactic categories that express gradable concepts, and all languages have constructions that are used to make comparisons between objects. ``Gradable'' adjectives like tall, bright, beautiful, and so forth are canonical examples of such expressions, and have formed the central empirical domain for studies of grading in natural language.

A central conclusion of this body of work is that the semantics of grading must make reference to abstract representations of measurement, or ``scales''. A question that has not been seriously addressed, however, is whether scalar representations underlie the semantic properties of categories other than gradable adjectives. This question is important because gradability is not just a property of adjectives, but of nouns, verbs, and prepositions as well. This point is made quite clearly in several important early studies of grading (in particular in work by Sapir and Bolinger), but it has not been investigated in depth or detail by contemporary work in semantics. Although some analyses of non-adjectival categories have incorporated elements of grading and scalar representations (in particular, work on verbal aspect and directional prepositions), there have been no systematic investigations of the deeper role of scalar representations in natural language.

The purpose of this study is to conduct just such an investigation, through a comprehensive examination of the role of scalar representations across categories. The primary focus will be on scalar representations in adjectives and verbs, both because the core semantic properties of these categories are best understood individually, and because preliminary research points towards a regular relation between scale structure in adjectives and event structure in verbs. The project will begin with a close examination of the role of scalar representations within each category, and will then investigate the relations between the two, focusing on derivational relations between verbs and adjectives. The study will conclude with a preliminary examination of the role of scalar representations in the prepositional and nominal domains.

The educational component of this project will involve course development, graduate and undergraduate mentorship, and direct access to research. The empirical domain of the project will support the development of an integrated three-quarter sequence in lexical semantics, formal semantics, and pragmatics. In addition, the topic of scalar representations in language provides a foundation for an exploration of linguistic, cognitive, and philosophical questions about meaning in a one-quarter seminar targeted specifically at freshmen and designed to emphasize discussion, intellectual challenge, and the development of critical thinking skills. The project will also support one graduate research assistant per year and two undergraduate researchers per summer, providing an opportunity for supervised independent research in an empirical domain that allows for a range of variation in research focus. Finally, the project will introduce students and colleagues directly to research in the field by bringing in two researchers each year to give guest lectures in class and present colloquia, and through a workshop on the role of scalar representations in natural language to be held at the end of the five year period.