Iterability is the possibility of removing a mark from its context and placing it in another. Derrida believes that iterability is something that fundamentally structures all marks (written, oral or otherwise). Thus, every mark has the inherent possibility of being ‘cited’, ‘quoted’ or ‘grafted’ (to use Derrida’s terms). Derrida states that:
…communication must be repeatable —iterable— in the absolute absence of the receiver or of any determinable collectivity of receievers. Such iterability—(iter, again, probably comes from itara, other in Sanskrit, and everything that follows can be read as the working out of the logic that ties repetition to alterity) structures the mark of writing itself, no matter what particular type of writing is involved… A writing that is not structurally readable—iterable—beyond the death of the addressee would not be writing (SEC 7).
So for Derrida the possibility of iteration is a (if not the) defining characteristic of any mark. If it is not iterable (repeatable, even in the radical absence of any receiver) it is not a written mark.
This is all very well and good, but it doesn't really seem that problematic yet. And Derrida isn't Derrida without some pressing problematic. We discover the problems that this iterability contains when Derrida applies it to John Austin's version of speech act theory, specifically the version exhibited in Austin's How To Do Things With Words.
So just what problems does iterability pose (if any) for Austin’s speech act theory? Well, after taking iterability as an inherent structure or quality of any mark, Derrida applies this structure to Austin’s speech-act theory and asks:
Could a performative utterance succeed if its formulation did not repeat a “coded” or iterable utterance, or in other words, if the formula I pronounce in conforming with an iterable model, if it were not then identifiable in some way as a “citation”? (SEC 18).
Just before the above quotation, Derrida has illustrated Austin’s exclusion from his analysis of speech acts precisely the sort of speech act that Derrida is talking about: those made in plays, in jest, in poetry, or in fiction (ie: those that are obviously iterated/repeated). What Derrida wants to ask us is: does this sort of exclusion ignore (or discount) the very structure of the thing that it sets out to analyze? Or: Does the exclusion of these ‘parasitic’ speech acts (which are cited/repeated/iterated speech acts) ignore/discount the fact that all marks are fundamentally iterable. Derrida will argue that it does, and that such an exclusion is, for that reason, a metaphysical and dangerous one.
Since, unlike my friend frankdeluxe above, I've read and own a copy of How To Do Things With Words I think that I will do a writeup on its basic principles. Its rather helpful in understanding what Derrida is getting at with all this. And its also helpful in being critical of Derrida's reading of Austin...