Shell games

A paragraph that completely blew my mind, when I first read it during my freshman year in a borrowed copy of Genette's Figures II (the aesthetic properties of those Editions du Seuil volumes are very strongly imprinted on my imagination) - I give it here in the translation of Alan Sheridan as published in Figures of Literary Discourse ("The Frontiers of Narrative"):

Direct imitation, as it functions on the stage, consists of gestures and speech. Insofar as it consists of gestures, it can obviously represent actions, but at this point it escapes from the linguistic plane, which is that in which the specific activity of the poet is practised. Insofar as it consists of words, discourse spoken by characters (and it goes without saying that in a narrative work the role of direct imitation is reduced to that), it is not strictly speaking representative, since it is confined to reproducing a real or fictitious discourse as such. It can be said that verses 12 to 16 of the Iliad, quoted above, give us a verbal representation of Chryses' actions, but the same cannot be said of the next five lines; they do not represent Chryses' speech: if this is a speech, actually spoken, they repeat it, literally, and if it is a fictitious speech, they constitute it, just as literally. In both cases, the work of representation is nil; in both cases, Homer's five lines are strictly identical with Chryses' speech: this is obviously not so in the case of the five narrative lines preceding it, which are in no way identical with Chryses' actions: "The word 'dog' does not bite," William James remarked. If we call poetic imitation the fact of representing by verbal means a non-verbal reality and, in exceptional circumstances, a verbal reality (as one calls pictorial imitation the fact of representing in pictorial means non-pictorial reality and, in exceptional circumstances, a pictorial reality), it must be admitted that imitation is to be found in the five narrative lines and not at all in the five dramatic lines, which consist simply in the interpolation, in the middle of a text representing events, of another text directly taken from those events: as if a seventeenth-century Dutch painter, anticipating certain modern methods, had placed in the middle of a still life, not the painting of an oyster shell, but a real oyster shell. I make this simplistic comparison in order to point out the profoundly heterogeneous character of a mode of expression to which we are so used that we do not perceive its most sudden changes of register.