I am sympathetic to most of what Joanna Walsh says about unconventional fiction in her recent Irish Times article, "The N Word: Against the Novel." Her question, "why not many novels, not all of which suit everyone; instead of novels that strive to create a world, why not novels that highlight their own artificiality, stretching the seams at which language is stitched to meaning, shuffling experience as it is shuffled in memory?" is entirely apropos, and her general call for works of fiction that break whatever rules they wish, that alter or invert any of the currently established practices of "literary fiction," is one I profoundly hope other writers hear.
Additionally, she is completely justified in asking why "so many contemporary novels continue to be encumbered by the realist demands of plot, character, place. . less able to make less room. . .for the truly 'novel' varieties of language" the novelist might employ. However, I think her answer to her own question is mistaken, or at least incomplete: "what many a novel most closely resembles is… another novel," she asserts.
It is indeed quite puzzling that, in American fiction at least, post-modernism and post-postmodernism, so many novels (and short stories) adhere to the familiar conventions of plot, character, and setting. Judging from the books that are most prominently reviewed and that win most of the prestigious awards, and from the kind of fiction published in most literary magazines (especially in print), it is almost as if Woolf, Joyce, and Faulkner, Beckett, Borges, and Barth never existed, the primary line of influence bypassing them and their alternative strategies in favor of the narrative conventions associated with "the great tradition" inaugurated in the 19th century. (There are exceptions of course. However, that they are usually explicitly noted by reviewers and readers as exceptional, out of the mainstream of establishment-authorized fiction, only reinforces the impression that the alternative tradition of literary experiment has largely been forgotten--to the extent it ever became familiar.)
But I don't think that the cause of such reflexive, and regressive, fealty to the old narrative verities is the kind of insular preoccupation with other novels that Walsh describes. I believe that the greatest influence on these kinds of novels is not fiction at all, but movies, which have never stopped favoring the 19th century narrative conventions. Writers both appropriate these conventions as borrowed by the more popular form and write stories they think could be adapted to film. (I do not include here such books as John Domini's Movieola or Robert Coover's A Night at the Movies, which parody, contort, and transform the plot conventions of movies to create an altered form of fiction.) This is the only conclusion I can come to when so many of the literary novels I try to read seem so obviously trying to be a movie in words, not so much the script for a movie, but a facsimile of one with words only, absent the equipment necessary to make an actual movie. I love movies, but in reading or choosing fiction I'm not really interested in a film simulacrum in place of an attempt to fashion a distinctive work of verbal art.
The novel is a "self-referential and self-affirming form" only in that novels always have some relation to the literary history that has made them identifiable as novels in the first place. Something so idiosyncratic or "new" that it doesn't apparently have a connection to that history would no longer be a novel but literally a new form. There's nothing wrong with that, but if this new form proves compelling enough, others will want to try it out as well, and a new kind of history begins to unfold, to which new works in the form will become unavoidably "self-referential." If it isn't compelling, it may eventually be considered just a curiosity.