It is not simply disagreements among critics, but also the variations in the nature of the text, which deprive interpretation of many of the traditional assumptions by which it has claimed certainty. The diverse historical purposes of the text, from the mimetically transparent to the deliberately self-contradictory, make it difficult for the interpreter to claim that any critical language he may favour is given authority by virtue of its correspondence to the nature of nature, of literature, of literary convention, or even of language itself. The once confident relationship between critical language and its object has to be given up. But this confusion of literary and critical purposes tells us less about the ultimate nature of literature than about the varying philosophical assumptions with which we may approach the institution of literature and our talk about it.
One group of critics assumes or hopes that readers may come to agree about some relatively determinable meanings for texts, and their relationship to a stable world or world-like context. They have learnt enough from the Saussurean approach to linguistics to recognize that the ordering of nature in language may have been different, that a child for example may have learnt different distinctions between 'swan', 'duck', and 'goose' (or between epic elegy and ode at a later stage) had he belonged to a different community, i.e. that we impose orderings on nature and literature rather than vice versa. The similarities and differences which we learn and which give us control over the use of such concepts might have been exploited differently. Hence what we learn when we attempt to master the language with which we deal with reality, or the same language within the text, is, as we have insisted all along, a series of likeness-relations which are suitable for our purposes. They are
the preferred arrangement of some community, rather than something insisted on by nature itself. Nature does not mind how we make clusters from the vast array of similarities and differences we are able to discern in it; all that is required of such clusters is that they constitute a tolerable basis for further usage. The clusters are conventions; the similarity relations which concepts stand for are conventions.
None the less... in learning such conventions knowledge of nature is increased. At the end of his walk the child can identify swans in his environment, and hence expect of particular identified swans whatever his