Because logical positivism is often taken to believe a statement of the following sort: "a proposition is meaningful iff it is empirically testable (in principle if not in practice) or an analytic tautology or contradiction". But this statement would seem to be neither empirically testable, nor a tautology, nor a contradiction. Therefore it is not meaningful. Hence self-defeating. You can change some of the words in the statement above but the idea is the same. Whether there is a way for positivists to escape this sort of argument, I don't know, I haven't studied it.