The Design of Semi-lexicality: Evidence from Case and Agreement in the Nominal Domain

This dissertation tackles the topic of semi-lexicality, a term used to describe elements which show a mix of lexical and functional properties (Corver and van Riemsdijk 2001). Often, semi-lexical elements have a surface similarity to some lexical category (noun, adjective, verb), but have certain properties, usually functional, which prevent them from being treated with that category. Numerals present a classic example. In many languages, numerals show a similar morphosyntax to nouns or adjectives (Corbett 1978), but also have a number of idiosyncrasies which make the label “noun” or “adjective” unsatisfactory. This has led to many debates on how to treat the category of a numeral within and between languages. Examples of semi-lexicality challenge the traditional division of categories into nouns, verbs, and adjectives, and it is the aim of this dissertation to understand how semi-lexicality arises. This dissertation adopts an exoskeletal skeletal approach to categories (Borer 2005) and assumes that lexical categories (nouns, verbs, adjectives) are built from a root and the set of functional projections associated with that category. It argues that semi-lexicality is the special case where some root is also lexically specified for syntactic features, termed the Semi-Lexicality Hypothesis. The syntactic features act as instructions for the syntax, constraining how syntactic structure is built above and around the root. This, in turn, has consequences for how processes such as case and agreement are computed, producing semi-lexical effects. Under the Semi-Lexicality Hypothesis, semi-lexicality is the not-so-special case in which a root does not behave how a “lexical” root is expected to behave. The approach is defended through three case studies: Polish numerals, English quantificational nouns (lot, ton, bunch, number) and English kind-words (kind, type, sort). These case studies address phenomena of case and agreement in the nominal domain with binominal constructions (N1 (of) N2, e.g. a lot of people, these kinds of trophies). These case studies provide evidence for three “ways” of being semi-lexical, depending on what the syntactic feature of the root requires. Some semi-lexicality constrains the functional structure of the root. Pluralia tantum nouns, for example, require the projection of a plural #P (Number Phrase), while Polish numeral 1000 requires the absence of a γP (Gender Phrase). This affects their morphosyntax. Another type of semi-lexicality constrains the contexts where the semi-lexical root and its functional structure can be inserted. The English quantificational nouns lot and bunch, for example, only surface in indefinite environments, showing an incompatibility with definiteness. This affects their distribution. The final type of semi-lexicality interacts with how other roots in the environment project. In the English kind-construction, the second noun is prevented from projecting further functional structure once the kind-word is inserted (similar to restructuring in the verbal domain). This thesis predicts that a study of other examples of semi-lexicality should find similar types of patterns to the ones identified here. Once we understand what the canonical structural of a noun, verb, or adjective would be, it will be possible to explore how the specification of a functional feature interacts with the syntax