Dissertation "The compositionality of adjective noun constructions" (2024)
Goethe Universität Frankfurt, submitted 09/2023, defended 05/2024, published 12/2024
Supervisor: Prof. Dr. Ede Zimmermann
Supervisor: Dr. Cécile Meier
This dissertation explores the compositionality of adjective noun constructions. It investigates the question to what extent local modification – to be distinguished from non-local modification – can be captured with an intersective analysis.
Chapter 1 introduces the topic of the dissertation and outlines how the chapters of the dissertation are organized.
Chapter 2 is concerned with the theoretical background and introduces the two general approaches that have been suggested for the analyses of adjectives: the predicate and the modifier analysis. Moreover, a further variety of adjectival modification is addressed: non-local modification. Despite being syntactically located in an adjective noun construction, in this reading the adjective takes material outside its syntactic domain in its semantic scope. Non-local readings of adnominal adjectives can therefore not emerge through the standard analyses for local adjectives and instead call for more radical measures. With this distinction in mind, focus is put on local adjectives.
In chapter 3, the semantics of local adjectives like blonde, tall and skillful is in the focus. An adjective like blonde can straightforwardly be analyzed as a predicate. Other local adjectives first need to be “made” expressing a property. For many adjectives, such as small and tall, adding a mute parameter – a comparison class – will do. For other adjectives, such as skillful, famous and talented, apart from a comparison class an additional parameter is needed: a comparison property. It is argued that the context-sensitivity of the parameters enables a predicate analysis of these adjectives.
In chapter 4, it is investigated how the context dependence of the comparison property can be implemented in the theory. This is discussed in the form of two possible theories: theory A, which assumes noun and context dependence, and theory B, which assumes only context dependence. Since it turns out that theory A actually overgenerates, the decision is eventually made in favour of theory B. As a result, also an adjective like skillful can receive a predicate analysis. Cases where the modified noun serves as the basis to determine the parameter(s) are captured by the “Preference principle for the pragmatic specification of free variables” (Maienborn 2020).
In chapter 5, it is investigated how the comparison property is derived from the modified noun to obtain the default reading. In the first part of the chapter, the ability of a noun to serve as the basis for the determination of the comparison property is captured by the assumption that these nouns have an event argument in their denotation (Rapp 2015). Importantly, both deverbal and non-deverbal nouns can have an event argument. According to the approach suggested in this dissertation, the event argument is not accessed compositionally by the adjective. Rather, the pragmatic reasoning behind the determination of the comparison property is assumed to be sensitive to the presence of an event argument. In the second part of the chapter, event denoting nouns like teaching and performance are addressed. Since it turns out that context sensitivity of the adjective can also be observed in the presence of an event denoting noun, the claim that an intersective analysis suffices for local modification can be maintained.
In chapter 6, further local adjectives are addressed. While for seemingly non-subsective adjectives like fake even an intersective analysis seems to be possible, temporal and modal adjectives such as former and alleged are more problematic. These adjectives behave different than “ordinary” local adjectives in crucial aspects. Thus the question emerges how temporal and modal adjectives affect the claim that all local modification can be captured by means of an intersective analysis. Eventually, an adverbial analysis is assumed for temporal and modal adjectives (Zimmermann 2022). This allows to reduce the complexity of the lexicon and reflects the meaning of these adjectives in an intuitively more appropriate way. As a consequence of this approach, it turns out that the modifier analysis is not needed to capture the meaning of adjective noun constructions.
Finally, in chapter 7 the dissertation is recapitulated, and the main question discussed in this dissertation is addressed once again. As a result of the investigation, it turns out that adjectives are either predicates or (sentence) adverbs – but not modifiers. While non-local readings of adnominal adjectives require more complex analyses, a local adjective can either be analyzed as a (possibly complex) predicate (in the case of blonde, tall and skillful), or underlyingly it is actually not an adjective, but rather a sentence adverb (in the case of former and alleged). The theory about the semantics of adjectives put forward in this dissertation therefore does not employ a modifier analysis. Concluding remarks offer some suggestions for future research working with this point of view that embraces the context sensitivity of adjectives.