Abstract

This dissertation asks how words have meaning. It acknowledges that when we inquire into what words mean, we find an irreducible complexity, one which insists on recognizing the uniqueness of each occasion of a word's use. It rejects the idea that words have determinate meanings and the idea that definitions have a role to play in theories of meaning. Support for this stand is found in the theory of definition, in the problems facing current proposals which posit determinate meanings, and in our public conceptions of what words mean, as revealed by their commonly accepted definitions. It advocates, instead, providing an account of how we use words meaningfully--without, however, identifying anything functioning as the meaning of a word even for each person individually. A retreat into total sceptism--where everyone speaks her own ever-changing language--is averted by letting procedure executions characterize utterances. With each occasion of a word's use demanding its own uniquely determined computations for speaker and hearer, a picture emerges of private procedures constantly modulating under pressures of usage. This approach has to recognize, however, that there are principles operating to constrain our use of words; consequently, in the computational account each procedure for a word is understood to meet the constraints provided by the public conception of what that word means. These constraints circumscribe the procedures' operations, but still permit inexhaustible variations at the level of fine detail. An account of how we learn words will be concerned with how newly created procedures comply with these constraints. This dissertation offers an illustration, worked out in detail with supporting computer programs, of procedures for the word 'cadence' being synthesized from already existing procedures for the word 'melody'. The illustration is given in support of the computational account of semantic development, and as a first step towards a theory of word acquisition