‘The meaning of a word is what the explanation of the meaning explains.’ I have called this dictum of Wittgenstein's the fundamental principle of analytical philosophy; and we have so far seen no reason for not continuing to adhere to it. (In the last lecture it receded into the background inasmuch as it could not form the basis of the behaviouristic version of the thesis that the meaning of an expression is determined by a conditional rule. For precisely this reason however I stressed that one can understand the conditional rule not only as a causal rule but also as a rule which the person who uses the sign follows and which one can therefore explain. My critique of the conditional-rule theory was not specially directed against its causal version and can therefore also be understood as a critique of a theory which takes Wittgenstein's principle as its basis.)

Furthermore it still seems trivial to say that to explain the meaning of an expression can only be to explain the rule of its employment.

I began the analysis of the meaning of predicative sentences with the enquiry into the meaning of predicates. The sequence was as follows: in analysing the predicative sentence it seemed natural to start with predicates because it was this part of the predicative sentence on which the object-orientated conception of the meaning of predicative sentences foundered. In the meantime a second reason for this order of proceeding, and one more closely related to the matter in hand, has become apparent. If unlike the object-orientated philosopher one starts not with representations but with modes of behaviour and if furthermore one takes into account languages more primitive than sentence-language, then it becomes clear that characterization-expressions are the linguistic expressions which are the easiest to understand. Predicates have a pre-form which I have called quasi-predicates and which already function as independent expressions, hence do not require to be supplemented by singular terms. Their explanation is unproblematic for it is achieved by means of examples by simple assignment to particular circumstances (perceived by the sign-user or to be brought about by the recipient). In the last lecture I did not directly refer to signals as quasi-predicates, for I wished to avoid confusion in the examination of the question of whether predicative sentences can be construed as signals. But subsequently there is no reason for not also calling signals quasi-predicates.