ABSTRACT
Early work in natural language understanding (e.g., Woods, 1970, Winograd, 1972, Schank, 1972) was concerned almost exclusively with individual sentences. Recently there has been a shift of attention towards whole texts (e.g.,Wilks 1973, Rumelhart 1975, Schank 1975). On the surface this shift seems simple enough. If one can understand individual sentences, then to understand a text, all one need do is treat it as a set of individual sentences and apply the same methods. Our work would be much simplified if this were the case, but sadly it is not. The meaning of a text is more than the sum of the meanings of the individual sentences that comprise it.
