In the philosophical inquiry adopted by logical empiricists, analysis of scientific language becomes something similar to a metaphysical endeavor which is meant to establish the bounds of sense, and this stance may be easily traced back to Wittgenstein’s Tractatus Logico-Philosophicus. On the other hand, the analytic tradition transferred this conception to the analysis of ordinary language, and this move, eventually, was able to restore the confidence of many philosophers in their own work. After all they were doing something important and worthwhile, that is to say, something no one else was doing, since linguists are certainly concerned with language, but from quite a different point of view. At this point we may well ask ourselves: What is wrong with this kind of approach, given the present crisis of the analytic tradition and the growing success of the so-called postanalytic thought? At first sight it looks perfectly legitimate and, moreover, it produced important results, as anybody can verify just reading the masterpieces of contemporary analytic philosophy. To answer the question: What is wrong?, we must first of all take into account language itself and check what it is meant to be within the analytic tradition. This will give our question a clear answer. We have to verify, furthermore, what kind of knowledge philosophy needs to be equipped with if it wants to preserve its autonomy. The logical positivists clearly claimed in their program that there is no synthetic a priori knowledge such as the one envisioned by Immanuel Kant. There is, however, an analytic and a priori knowledge which is supplied by mathematics and logic alone. Within this field, the techniques of contemporary formal logic are exalted because they allow us to build artificial languages which - at least theoretically - eliminate the ambiguities of everyday speech.