I was driving back from a philosophy talk tonight and was thinking about Aristotle's Poetics. (The talk itself was a very interesting one on psychoanalysis of dreams; it was not what got me thinking about the Poetics.)
We tend to read the Poetics as literary criticism. It's very difficult for us not to do this, in fact; you pick up the book and it talks at great length about plot and character and diction. It's worth remembering, though, that this is not how it has always been read. Medieval commentators, for instance, read it as a book on (quite literally) logic -- the logic that deals with imaginative acceptance rather than probability or certainty, and that concerns discovery rather than justification or proof. It is very difficult for people to understand this reading -- however interesting, it just seems wrong.
But we are partial to ourselves, and there is an excellent argument to be made that if the medieval way of reading the Poetics is wrong, so is ours. The reason, historically, that we read the Poetics as a sort of literary criticism is that Renaissance humanists started using it as a manual to guide the writing of texts, and this, obviously, also became used in analyzing them. The reason we read the Poetics as a text on literary criticism seems to be due to nothing else than the fact that a large part of the vocabulary for literary criticism was lifted from the Poetics in the first place. But basically the same was true of poetics as a logical discipline in the Middle Ages.
We probably cannot know for certain how Aristotle and his immediate audience would have understood the Poetics, but if it were a matter in which the bet could be decided, I would bet a lot that they would have read it not as a text on literary criticism (they wouldn't even know what that was), nor as a text on logic, but as a text on political philosophy.
After all, why is the subject of the Poetics -- tragedy in the extant first book, comedy in the lost second book -- important enough to talk about? What were these things discussed? They were rituals in the civil religion of the Greeks, rituals that were of very great importance to their civic life, one of the most obvious and fundamental things that they did together as citizens and as Greeks. What Aristotle is doing in analyzing them is analyzing one of the major society-building events of his day. And Aristotle in general, of course, can be read as nothing other than the theory of Greek civilization. [ADDED LATER: And, of course, how could I have forgotten the most obvious argument in this direction. Take Plato's Republic, a book that explicitly discusses the nature of the polis, and compare the first part of Book III with the opening of the Poetics.]
So, in other words, if you want to at least read the Poetics in a way analogous to (you probably can never get closer) the way Aristotle and his immediate audience did, go home and read it as a treatise on political discourse that is not concerned with persuasion. (Political discourse concerned with persuasion belongs to the Rhetoric, of course.) We can find loose analogues even in our own day. There are lots of things citizen and politicans do that don't concern persuasion. In the United States, for instance, we have this amorphous but pervasive civil religion of In God We Trust and Truth, Justice, and the American Way, and veterans who, martyr-like, have died for us and for our Freedoms. When Americans celebrate the Fourth of July, the discourse is not about persuasion. So what's going on there? Well, Aristotle, looking at Greek civic rituals that were (it has to be allowed) much more structured, gives some answers.
I actually think that neither the medieval nor the modern way of reading the Poetics is wrong. They aren't inaccurate. The difference arises because we read the Poetics very narrowly, as if someone were to read Augustine's Confessions as nothing but an autobiography, while the medievals read the Poetics very abstractly, as if someone were to read George Eliot's Romola as a treatise in the philosophy of religion. These readings aren't wrong at all -- Augustine's Confessions are autobiographical and Eliot's Romola is structured by a philosophy of religion. Such readings are uncovering genuine elements of the text. The reading of the Poetics as a logic text really takes off with Muslim commentators; Avicenna, Averroes, and the like faced the problem of how to make sense of Aristotle's Poetics not in Greek culture nor in Greek-influenced Latin culture, but in the very different Arabic and Persian literary cultures. And how do you solve such a problem of cultural difference while staying true to the text, particularly if you are hampered in your ability to engage in close historical investigation? Read the text for its general themes and abstract ideas. And they did; very well, in fact.
On the other side, it is a failure of our modern reading that while the medievals could have made sense of the Poetics as a text on literary criticism (they didn't read it that way, but there are plenty of reasons and hints to suggest that they could have), we have so much difficulty making sense of the Poetics as a text on the logic of imaginative plausibility. But our failure is not a failure arising from inaccuracy of reading; it is a failure arising simply from being stuck in one mode of reading. The way of reading is not itself wrong.
But, again, if you want to read Aristotle's Poetics, try sometime to read it as a treatise on non-persuasive civil and political ritual and discourse; it will at least show you a side of it that you might not have noticed without deliberately looking for it.