The early structuralist Barthes still trusts to the possibility of a 'science' of literature [...] Such a scientific criticism would in some sense aim to know its object 'as it really was'; but does this not run counter to Barthes' hostility to the neutral sign? The critic, after all, has to use language too, in order to analyse the literary text, and there is no reason to believe that this language will escape the strictures which Barthes has made about representational discourse in general. What is the relation between the discourse of criticism and the discourse of the literary text? For the structuralist, criticism is a form of 'metalanguage'--a language about another language--which rises above its object to a point from which it can peer down and disinterestedly examine it. But as Barthes recognizes in Système de a mode, there can be no ultimate metalanguage: another critic can always come along and take your criticism as his object of study, and so on in an infinite regress. [...]