This paper will attempt to outline very briefly a recent approach to logic known as model theory. It will compare the aims and techniques of this approach with more traditional frameworks and then make a few remarks about possible applications of logic in computer systems dealing with natural language. Logic is concerned with distinguishing correct, or valid, reasoning from that which is incorrect, or invalid. To take an example from the classical Aristotelian syllogisms, we recognize intuitively an instance of valid inference (that is, if we accept its premises, we cannot escape its conclusion):