On the hypothesis that all natural or artificial languages of interest to uscan be given transformational grammars of a certain not-very-special sort, it becomes possible to give very simple general answers to the questions:

(1)  What sort of thing is a meaning?

(2)  What is the form of the semantic rules whereby meanings of compounds are built up from the meanings of their constituent parts?

It is not my plan to make any strong empirical claim about language. To the contrary: I want to propose a convenient format for semantics general enough to work for a great variety of logically possible languages.This paper therefore belongs not to empirical linguistic theory but to the philosophy thereof.