Chomsky's is a name to conjure with both in computer science and in linguistics. But he presents a different face according to which group you view him from. To linguists he implies transformational grammar, and they polarize for and against this development. To computer scientists it is his earlier work which has become the foundation of various developments by other workers, and it is suggested that it is among these that a more satisfactory theory of grammar is likely to be found. In particular, many features which cannot be explained by context‐free (i.e. phrase structure) grammars can be explained if a carefully controlled transient memory is included in the model. Other features appear to involve considerations which depend not merely on semantics but on knowledge and experience—the equivalent computer programs, must rely on data structures which are not part of the grammar—and this provides an indication of the limits on the proper role of grammar.