E EDUCTIONISM has become a word of abuse such that to call an analysis reductionist is tantamount to calling it mistaken. Given the kind of linguistic pluralism we have learned to acknowledge and accept from a study of the later Wittgenstein, we (and indeed justifiably) have come to entertain a lively suspicion of all reductionisms: they fail to recognize both the variegated uses that language actually has and the fact (putative fact) that every assertion has its own logic. Moreover, in analyzing some conceptual areas, reductionisms make use of the very distinctions reductionists would deny.' Since my analysis has been and no doubt will continue to be called reductionistic,2 1 want to examine what is at issue here. In discussing Wittgensteinian Fideism, I have already examined some of the relevant issues, but others are new. It is indeed true that in a way I make the very distinctions I deny. I have a participant's grasp of God-talk and some knowledge of the theological chatter about it. I know which utterances are deviant and which are not, and I have some understanding of the conditions associated with God. I naturally utilize this understanding in talking about God-talk and I do not deny that these distinctions can and should be made. In fact, if they could not, there would be nothing to analyze in analyzing God-talk. What I have done in laying out the structure of religious discourse and in specifying some of the crucial expectations believers have about it is to display its internal incoherence. When I go on to external criticism, as with my theory of ideological statements, I only need to establish that the believer himself claims