Can realism have anything to do with reality? I answer, yes, it can, but find that this innocent response now needs a careful defence. In his introduction to the eleventh issue of Communications, the issue devoted to vraisemblance, Tsvetan Todorov writes, ‘A work is described as having verisimilitude in so far as it tries to make us believe that it conforms to the real and not to its own laws; in other words, the vraisemblable is the mask in which the laws of the text are dressed up, a mask which we are supposed to take for relation to reality.’ We have here an assumption and a thesis. Todorov assumes that conformity to the real and conformity to literary laws are mutually exclusive; if you are doing the one you can’t be doing the other. And he asserts that artists in fact do the second; that, even when they claim to be addressing reality, they are really doing something quite different, namely following the laws of literature.

I submit that this assumption is wrong and that the thesis is by consequence also wrong. Reference to the real, so far from being at odds with adherence to convention, actually presupposes it. This is most obviously true at the level of language, which is endlessly formulaic, endlessly conventional. But if language cannot refer to reality we may as well all go home. It is fashionable to speak of the mysterious void which yawns between language and the world, a void constituted by the fatal heterogeneity of words and objects. But if words were indistinguishable from objects they could not be used to refer or designate. You don't point at a cat with a cat, you use your finger or a stick - or a word.