Abstract
The term 'logical' in Carnap's 'Elimination of Metaphysics through Logical Analysis of Language' refers either to formal logic or to applied logic, i.e. epistemology. Corresponding to these two senses of logic we can distinguish two kinds of pseudo-statements. A statement is called a pseudo-statement if it violates either the rules of logical syntax or the epistemological criterion of empirical significance. According to Carnap both kinds of pseudo-statements we meet with in metaphysics. Unlike epistemology ('applied logic') formal logic is characterized by topic-neutrality, by impartiality towards the subject matter to which it is applied. Formal logic abstracts from (the significance of) the subject matter of a statement. On the other hand, this subject matter of a statement, its empirical significance, is epistemology's primary concern. If a metaphysical statement has to be rejected on formal logical grounds it is rejected because it violates the rules of logical syntax, not because its subject matter is metaphysical. If, on the contrary, a metaphysical statement has to be rejected on epistemological grounds, it is rejected because it is metaphysical, that is to say, at odds with the empiricist criterion of empirical significance. Such a rejection, however, cannot properly be termed 'logical' in the strict sense of the word. So, as far as Carnap's argument against metaphysics is strictly logical it leaves metaphysics untouched and cannot result in an elimination of metaphysics. As far as his argument is epistemological it directly touches upon metaphysics, but in that case it cannot properly be classified as 'logical' ; as an epistemological argument it lacks the neutrality peculiar to formal logic. We conclude that Carnap is far from having demonstrated that 'logical analysis' forces us to give up metaphysics