| 摘要、提要註: |
This dissertation presents a new theory of the relationship between vagueness, context-sensitivity, and adjectival scale structure. Based on both new and well-known data, I show that the following subclasses of adjectival predicates are empirically distinguished in languages like English and French based on these three phenomena: Relative Adjectives (ex. tall, short, expensive, cheap, intelligent, stupid, narrow, wide ...), Total Absolute Adjectives (ex. bald, empty, full, clean, smooth, dry, straight, flat ...), Partial Absolute Adjectives (ex. dirty, bent, wet, curved, crooked, dangerous, awake ...), and Non-Scalar Adjectives (ex. atomic, geographical, pregnant, illegal, dead, hexagonal ...). The main goal of this work is to develop an formal account of both the semantic and pragmatic similarities and differences between the four subclasses of adjectives. I propose that the patterns concerning the behaviour of relative adjectives like tall vs absolute adjectives like straight and bent are reflexes of a single underlying difference in the semantics of these lexical items involving (a certain kind of) context-sensitivity. Moreover, I show that the data concerning both vagueness and scale structure can be derived from the interaction between (lack of) context-sensitivity and tolerance/indifference relations associated with general cognitive categorization processes. Building on insights into the connection between context-sensitivity and scalarity from the work of Klein (1980) (among others) and insights into the connection between tolerance relations and the Sorites paradox from the work of Cobreros et al. (2011a) (among others), I propose a new logical framework that captures the intimate and complex relationship between these three aspects of adjectival meaning. |