In the philosophical movement which became known as Logical Positivism, it was originally propounded that the meaning of any statement is given by its method of verification. What it means is this: "To say that proposition has meaning or, more strictly (as became evident in the discussion of the 1930's and 1940's), that it has factual or cognitive meaning, is to say that it is in principle verifiable, or at least probabilifiable, by reference to human experience." The main issue in the analysis of religious language, which has been given fresh sharpness and urgency by contemporary analytical philosophy, is whether religious statements are cognitive or noncognitive. In particular, do those religious statements refer to a special kind of fact-religious as distinguished from scientific fact? The issue is more concerned than ever because those statements have the form of factual assertions and as a matter of historical fact religious people have generally believed such statements as " God loves us as father loves his son" to be not only cognitive but also true. So we have the task of examining as sympathetically and as critically as possible whether those statements are genuinely true or not. One of the difficulties we have to face in examining religious language is a choice of a criterion by which to distinguish the factural from the nonfactual. But in view of a tendency of traditional theism to use the language of fact, the development of a criterion within contemporary analytical philosophy is directly relevant to the study of religious language. Therefore we will start with the verificational principle of meaning in the analysis of the language in question. Then we will shift the principle from the idea of verifiability to the complementary idea of falsifiability. The discussions of " Falsification and Theology " originally appeared in New Essays in Philosophical Theology will show the possibility of dialogue between traditional theism and contemporary philosophy. Now to analyze religious language by the above-mentioned criterion is necessarily equivalent to denying that religious statements are cognitive and/or factual. But this does not mean to rule out those religious statements as meaningless and/or nonsensical. Instead of ruling them out, we will examine as carefully as possible just how they are used to mean whatever they may mean. For they have their own function to fulfill in one way or another. In this respect we will attempt an entirely new set of analyses, that is, the later Wittgensteinian analysis of language : the meaning of any statement is given by the way in which it is used. This is no doubt a change of the verification principle into the use principle of meaning. Indeed the change of the principle in the more recent debate on the question will lead us to solve the problem of religious language in the way in which it does justice to the empiricist's demand that meaning must be tied to empirical use. What we intend to do is therefore to clarify empirical use of the language in question.