Abstract
The studies reported in this thesis address a number of questions concerning semantic interpretation of adjective-noun combinations for which traditional models (Hampton, 1979a, Murphy, 1990, Smith et al., 1988} offer relatively weak accounts. In Chapter 1, it was argued that the sense enumeration hypothesis might be inadequate for a large class of adjectives which do not seem to represent a clear property (e.g., 'interesting', 'nice', 'good'). For these kinds of adjectives the meaning computation hypothesis might be more plausible. This hypothesis was tested in the study reported in Chapter 2 by comparing the processing of homonymous and polysemous adjectives. In this study, only a weak support was found for the meaning computation hypothesis. It was concluded that the degree of meaning relatedness (homonymy/polysemy) may not be the sole determinant of the degree in which adjectival meanings are computed. The experiments reported in Chapter 3 showed that the noun-related factor concreteness plays an important role in semantic interpretation of combinations involving polysemous adjectives. In the experiments reported in Chapter 4, evidence was obtained that both the complexity of the adjective as well the salience of the noun properties affect the semantic interpretation of adjective-noun combinations. Finally, the findings reported in Chapter 5 suggest that adjectival formal type (intersective, subsective) largely determines the level of adjectival noun-dependence in combinatorial semantic interpretation
