Last updated 9 Jan 08
Abstract
Much of our human thought is not in a fully explicit public language, with all its paraphernalia of phonological form and grammatical structure markers. Non-human animals are capable of some degree of structured thought, but obviously not in a human language. Human language was built upon, and in its turn modified, pre-existing structures in animal minds for the representation of the external world. What are such non-linguistic, but nevertheless semantic, representations like? This talk works from two ends of the problem: (1) desiderata for representing the most basic meanings expressible in human languages; and (2) what we know about human and non-human neural processes. A simple 'box' notation is proposed for representing predicate-argument structure in such a way that it eliminates many of the separate traditional categories of conventional predicate logic and linguistic semantics. The most fundamental moves are the reduction of both individual constants and role markers (e.g. AGENT, PATIENT) to predicates, and a reduction of all predicates to one-place. Such moves have precedents in the semantic literature, but have never been combined in a single approach. The implicit psychological claims are justified by two main classes of psychological findings, relating to the distinction between global and local attention, and internal 'frame-of-reference' systems.
© 2008 James R. Hurford
Last updated 9 Jan 08