By JOHN O'BRIEN "W

HAT CONSTITUTES the novelist's strength is

precisely that he invents, that he invents quite freely, without a model. The remarkable thing about modern fiction is that it asserts this characteristic quite deliberately, to such a degree that invention and imagination become, at the limit, the very subject of the book." This remark by Alain Robbe-Grillet accurately represents, not just his own fiction, but that of other French "new novelists" (Queneau, Pinget, Butor, Simon, Sarraute, Sollers, Roche), as well as that of such other contemporaries as Beckett, Borges, Cortazar, Calvino, Goytisolo, Sorrentino, and Hawkes. That so many of these writers still strike reviewers and critics as odd, difficult, and "experimental" (i.e., "anti-novelists"), rather than as constituting the mainstream of fiction for at least the past 30 or 40 years, proves the point of For a New Novel (1953), that our view of the novel continues to be derived from our somewhat misguided conception of 19th-century narrative: "life-like" characters with believable motivations and psychological substance; familiar settings; and stories that observe the "logic" of time and place. The 19th- century novel is, in fact, no more realistic than one by Robbe-Grillet; each is as artificial and invented as the other.

Robbe-Grillet's Djinn is framed by the prologue and epilogue of a seemingly transparent narrator who finds a mysterious manuscript (The Rendez-vous) which he then reproduces. This Poe-like analyst limits himself in the prologue to providing the scanty and cryptic information available about the manuscript's author, Simon Lecoeur, and to vague speculations about its intent: "the ratio of probability of the reported event is almost always too low, in relation to the laws of traditional realism. Thus, it is not ruled out to see a mere guise in this pretense of a pedagogical intent. Behind that guise, something else must be concealed. But what?" The hand of this narrator is one possible answer to his question.

The first five chapters of The Rendez-vous record in first-person, present tense, Lecoeur's (or someone else's) response to a newspaper ad, which results in his taking an unspecified job with a clandestine organization that requires him to disguise himself as a blind man. He relates his bizarre experiences, culminating with a whack on the head when he lifts his opaque glasses to find himself in a room filled with others dressed identically, each apparently believing that he has been chosen for the same secret mission.

That whack leads to the last three chapters, multiple shifts in both person and tense, and to the structural game that Robbe-Grillet creates in this novel. A third- person narrator talks about Lecoeur in the past tense; then back to the first person, though not quite the same first person as before. Then other "I"s enter. With each shift in person and tense, the "facts" of the story are reshuffled and altered. When the original narrator re-enters in the epilogue, he reshuffles the "facts" once more and incorporates them into his explanation of events, and thus the novel ever so gently slips back into retelling itself. Its variations and rearrangements are limitless, requiring no more than a change of tense, person, or fact to set in motion another possible arrangement.

Now, for any of this to work, Robbe-Grillet must know at exactly what point to place a wrinkle in the narrative, so that having made the story function on one plane, he can then cause that plane to intersect with another. As soon as the facts are known in one way, the introduction of a new fact, or the slight alteration of a previous one, will call into question all that has been understood up to that point. Robbe-Grillet invites the reader to inspect the devices of his composition, the devices which the naturalistic author keeps secret, the machinery quietly working its effects. Perhaps more than with the naturalistic writer, Robbe-Grillet cannot afford mistakes, because the very act of his invention is on display. He makes no mistakes. Like the magician, he explains his trick, only to perform another one that he doesn't explain.

The Coover novel is a failed attempt to employ the methods of the nouveau roman; the repetitions, the variations upon images, the structural loops, the shifts in perspective, all seem wearily imitative, forced, and pretentious. Each morning a maid enters her employer's bedroom, and each morning she is spanked for her failures. As in Robbe-Grillet, there is an implicit invitation to see how the book is constructed. Here, however, the machinery creaks, sputters, and grinds; the tricks are telegraphed, even to the ending in which the employer and maid exchange roles. Finally, I began to suspect that some grand metaphor was rearing its ugly head. Or a fable: the man and his maid are supposed to represent the relationship between man and woman, between husband and wife, children and parents; or between artist and society, or artist and critic. No matter how well the artist does some things, so the fable might go, the critic will spank him for not doing others.

Spanking the Maid can be seen as new and inventive only if one forgets a dozen or so French novelists of the past 30 years. It is a simplification of the techniques of the French writers, and should not be viewed as much more. As James said, the distinction should be drawn, not between the old and the new, but between that which is executed well and that which isn't. Djinn is well executed, Spanking the Maid is not.