Jackendoff's analysis of lexical semantic structure could be extremely useful to the computational linguist interested in creating a highly constrained lexical framework which aims at capturing as many generalities as possible about linguistic and semantic relationships among words. Jackendoff does not attempt to characterise the distinction between many verbs within a class, e.g. verbs of locomotion -- run, swim, fly, etc. -- within his proposed lexical structure, unless the distinction is syntactically relevant or seems to be associated with a general pattern of inference. This means that the information to be lexically represented does not attempt to model the full range of meaning of words, but rather is restricted to capturing generalisations.
The adoption of a semantic ontology with respect to which all lexical semantic structure is developed, a notion of thematic roles which allows certain syntactic phenomena to be elegantly modeled, and the identification of function-argument structures which reflect distinctions in both syntactic use and semantic inference among words together form a framework for lexical semantics which is constrained, extendable, and implementable. Jackendoff's approach seems to be precisely the kind of approach to lexical semantics which is necessary for maintaining a distinction between lexical and world knowledge, and for adequately modeling the ways in which words can be used.
Jackendoff's work concentrates on capturing generalisations about word use. It must also, however, be emphasised that the lexicon is a repository of word-specific information which should include idiosyncrasies of the use of individual lexical items. A balance must be struck between being stipulative and being explanatory in lexical representations; generalisations must be reflected in order to predict word use and interpretation and to address the generativity of the lexicon, while apparent ``accidents'' -- behaviour or interpretation which does not seem to follow from general (semantic) principles -- must also be encoded to account for the full range of use and interpretation of individual words. Lexicon theory therefore differs from syntactic and pragmatic theories, which aim to be fully explanatory. Not everything that a person knows about words can be motivated; there are idiosyncrasies which cannot be explained. We will see this clearly in Chapter 5 with respect to the phenomenon of logical metonymy.