<snip> What I meant was maybe something closer to "the field of possible authorial intentions." Something like: if I posit "carnivore" the response "oh, a zebra" is apparently silly. We don't know the author's precise intention, assuming that the intention was ever singular or precise, and there are fuzzy areas around the boundary, but we do know what the author couldn't possibly have intended, what's clearly the reader's fantasy unsupported by the text. <snip> What Relevance Theory calls *cognitive environment* may be useful here. Someone's CE is what is manifest to them. So to some degree a shared CE is needed (between author and reader) in order that some meaning should take place. On the other hand, the *intention* of the author in communicating (this is a less fuzzy definition of *intention*, I think) is to alter the reader's CE. As well as decoding what is in the text, a reader creates context as a (very) small subset from all those manifest assumptions comprising his CE and proceeds inferentially. For a good reader the inferences made are tentative, defeasible, and the interpretation(s) reached may be in flux. Over time and between different readers what is *shared* will vary. Someone reading badly, on the other hand, is attempting (vainly) to alter the author's CE. CW _______________________________________________ That which is the future here, when read from right to left, has already happened. (Giorgio Manganelli)