Ontology Construction and Natural Language

The main thrust of this paper will be to illucidate and promote discussion on the relationship between ontologies and natural language. The value, power, and, I will claim, necessity of basing ontology construction on an understanding of natural language is recognized generally neither within AI nor within formal ontology. This is primarily because of a rather impoverished view of language, of what language is, how it functions, and of what it is doing when it`functions' | in short, the very role of language for modes of human Being is (again!) being underestimated. Language is misconstrued as anòn the surface of things' technique of building descriptions of the world. This obscures the reasons for using an understanding of language in ontology theory and practise. The paper will brieey set out these reasons, both theoretically and practically | theoretically, by grounding ontology construction in the understanding of language ooered by systemic-functional linguistics and relating this style of`doing linguistics' to its correspondences in the philosophies of later Wittgenstein and Hei-degger | and practically, by relating the discussion to our ongoing attempts to instantiate systemic-functional theory computationally. The most concrete result of which to date in the area of ontology construction is the Penman`Upper Model'. It is often not recognized that the Upper Model is`theoretically grounded' in this way, and its intended status is consequently commonly misunderstood.