The central claim of this thesis is that the combinatorial properties of words are determined by their lexical structures. I argue that the lexicalist position follows from the Word Grammar (WG) analytical framework. This framework, developed by Hudson (1984, 1990, 1994, 2003), is the one that I adopt in the linguistic descriptions in the thesis, and the further development of the framework is a significant part of the work. I consider the work o f a number o f writers, some of whom take a similar lexicalist position. 1 show how the lexicalist assumption, together with the analytical properties of WG, provides for a meaningful and explanatory analysis of a number of grammatical patterns, including the interaction between verbs of motion and their satellites, the syntactic and semantic behaviour of resultative expressions and alternation between causative and unaccusative uses of verbs. A detailed analysis of the meanings of verbs and prepositions provides the means for predicting their syntactic and semantic behaviour in the relevant constructions. I concentrate on English chiefly because it is my own first language, but also because of the body of existing work, including corpuses, in or on English which provides a basis for the work in the second part. I do consider data from other languages when they show illuminating similarities or differences to the patterns found in English. I concentrate on the structure and behaviour of verbs since they often stand at the centre of grammatical structures, determining to a large extent the relationships among the elements that surround them. However, it is of course impossible to study the behaviour of verbs without also considering the properties o f those other elements. Considerable time is spent on the relationships between verbs and other argument-taking words, including prepositions, particles and adjectives.