Writing together

A central topic of my dissertation:

I've wondered whether one general reason for some of the hostility toward the book is simply the fact that there are two writers, because people want you to disagree about things, and take different positions. So they try to disentangle inseparable elements and identify who did what. But since each of us, like anyone else, is already various people, it gets rather crowded.
-Gilles Deleuze, "Letter to a Harsh Critic," in Negotiations

Together, Felix and I would have made a good Sumo wrestler.
-Deleuze again, "Letter to Uno: How Félix and I Worked Together," in Two Regimes of Madness

What is a text that has been written together? In other words, neither a text that has been joint-authored, nor a text that has is a communal text, authored by the community? The status of this object seems elusive, at least from the way that we normally go about citing things. Witness even Derrida, in the video of his course that I posted a while ago, who is always as scrupulous as possible about names when there are any out there (a fact that, by the way, contradicts much that is stupidly said about him being the author of the death of the author--he in fact questions this complacent notion very thoroughly throughout his career, in particular in Dissemination). Here cites Deleuze within the joint texts (and I know of perhaps no better examples) of Anti-Oedipus and Thousand Plateaus. That is, he cites Deleuze alone within a work that is written together with Félix Guattari. Obviously, this makes sense: he wants to link what is said in those texts with what Deleuze alone writes in Difference and Repetition. He wants to look at Deleuze's share in this text. But of course the question remains (and I think the comment on the sumo wrestler above also shows it remains for Deleuze as well, though differently) whether these texts are shared out in this particular manner, taking place as they do more than one author (or rather, writer-function, emission), and whether in fact they can ever be linked with the texts of a single or singular name. (Derrida risks saying yes, and for good reasons, I think [but what were they exactly?]. But these reasons are specific and inadequate in the area that they perhaps make possible: the area we are interrogating here.)
The task then is to account for a text that is more than one, that is joint-authored and yet, also, at the same time, disjoins any act of joining (any togetherness) that is related to authorship, to the emitting source.