ABSTRACT
Can a machine understand human language? A generation ago, this would have been regarded as an absurd question. Today, it would likely evoke a variety of answers from different people. At one end of the scale, one might object that even the most sophisticated computer could never be capable of such a complex mental feat (see Turing, 1963 for a landmark discussion). At the other extreme, a person might, naively perhaps, take it for granted that computers already understand language. After all, we are regularly bombarded with fictional human-computer dialogues, in contexts ranging from science fiction novels to an awardwinning TV commercial in which a person discusses with a computer the supposedly insoluble question of whether "chunky soup" should be eaten with a spoon or a fork. Where exactly does the truth lie?
