IF, as Matthew Arnold suggested, literature is the criticism of life, then literary criticism must be a paltry, twice-removed thing indeed, hanging onto life by barely a thread. Arnold made his suggestion in 1883, in the middle of a hot debate about the relative places of literature and science in a gentleman's education. The debate was renewed, with even greater heat, when C. P. Snow proposed his notion of the "two cultures" in his 1959 Rede lecture. That controversy, if not quite dead, is now at least passe, but the nerve Sir Charles hit in those he dubbed the "literary intellectuals" has not yet been soothed.

Literary criticism is not so much under attack as it is increasingly ignored in a technological culture for which the language arts seem more and more unsuited. But, prone to shifts in fashion and forever self-renewing, literary critics have been at pains lately to assure us that their work is every bit as rigorous and scientific as quantum mechanics. Two of these three recent critical studies, Narrative Discourse and Allegories of Reading, present themselves as "systems" of inquiry into literary phenomena. The third, Tony Tanner's Adultery in the Novel, concerns itself less with methodology than with interpretation.

Tanner's fine book, a literary history of ideas that is a model of critical acuity, provides insightful readings of three major western European novels, Rousseau's La Nouvelle Heloise, Goethe's Die Wahlverwandtschaften (The Elective Affinities) and Flaubert's Madame Bovary. Along the way, we also get interpretations of Homer, Malory, Shakespeare, Richardson, George Eliot and Tolstoy, and Wonderful commentaries on Vico, Marx, Freud, Saussure, Levi-Strauss and Lacan.

Adultery in the Novel is immensely detailed and original. Tanner writes with an ease that makes the book hard to put down (something that can't be said for many critical studies these days). He defines adultery as "any compulsion to make illicit remixings or reorderings within the exiting framework," and he concludes that the novel, as the essential form for bourgeois literature, is committed to tracing the intimate history of the family structure, a structure in which "contracts and transgressions are inseparable, the one generating the other."

Tanner articulates a fascinating view of the connections and dissolutions implicit in the marriage bond. He understands marriage as the ultimate pattern of combination; adultery, then, represents a central and consuming threat of disintegration. Tanner proposes that "it is the unstable triangularity of adultery, rather than the static symmetry of marriage, that is the generative form of Western literature as we know it."

Tanner sees the novel as the same kind of deviant from literary norms that adultery is from social norms -- a descent into the unclassifiable and the inappropriate. He understands bourgeois society to concern itself, ultimately, with taxonomy. The networks and patterns of interrelationships he investigates in the novels of Rousseau, Goethe and Flaubert reveal, finally, the new way middle-class cultural bric-a-brac becomes the material of literature.

By way of preface, Tanner says that he wrote this book "blind," as "an exercise in reading." The study, unburdened by exhaustive cross-references to other critics, carries the weight of its learning lightly, but convinces with its strikingly original interpretations of well-worn texts. Tanner proposes a volume to follow which will take up the theoretical questions raised by his superb, razor-sharp critical study.

Narrative discourse, Gerard Genette's first full-length publication in English, is more abstract and, as a consequence, less accessible than Tanner's book. Genette incorporates and extends material from some of his earlier essay collections published in France under the title Figures.

This new book examines Marcel Proust's Remembrance of Things Past and offers a structural analysis of time, space and narrative activity in what is perhaps the most elusive, vertiginous and diabolical mass of studied prose (excepting, possible, Finnegans Wake ) the century has yet experienced. Genette's aim, he says, is to "restore this work to its sense of unfulfillment, to the shiver of the indefinite."

In his foreword to the book, Jonathan Culler claims that it represents "a systematic theory of narrative," and Genette himself refers to what he practices as the science of "narratology," a literary term of recent coinage. What we get is brilliant, pseudo-mathematical, ingenious gamesmanship and a game that seems always to take itself seriously. At one point, for example, Genette reduces a long passage to: A4 (B3) C5-D6 (e3) F6 (G3) (H1)

. . . And it makes sense, in a perverse sort of way. The passage lights up after he unravels for us the way Proust embeds temporality into his language.

Genette wants to produce a systematic, scientific account of how fictional discourse operates, how stories get told and retold. He gives us lots of new categories for narrative analysis -- analepses, focalizations, paralipses, and the metadiegetic, just to name a few -- and he calls these formulae (which Culler apty dubs "teminological exuberance") a "technology . . . a packaging, the detritus of Poetics."

There is no doubt that Genette produces a useful and revealing methodology for analyzing fictional narrative discourse and locating its strategies. As methodology, the book breaks new ground and opens up a great work to new interpretations and to an enlarged understanding of Proust's genius. But Culler's claim that this is a "systematic theory," a claim that appears nowhere in Genette's own text, would argue that the techniques and categories developed here are more than a method for examining literary prose. Developing new terms for literary structures or rhetorical tricks does not constitute "theory." The distinction between theory and methodology has blurred at the same time that the beleaguered literary scholar has been made to pretend to be as scientific as the scientist.

Still, the theoretical difficulties Culler raises for Narrative Discourse do not overshadow the lucidity and intelligence Genette brings to bear on Proust's unclassifiable magnum opus.

Paul de Man also practices a style of literary criticism imported from Paris. This style employs the controversial term "decomstruction" to designate its central activity. Where other interpreters look for referential meaning, substance, values, truth -- the standard debris we have come to expect culture to toss up to us -- deconstructors see the traditional quest for meaning as a naive search for the nonexistent. "Deconstruction," de Man writes, "always has for its target to reveal the existence of hidden articulations and fragmentations" -- bits and pieces of linguistic wreckage -- "within assumedly monadic totalities."

Through elaborate and elegant, often difficult, close readings of poems by Rilke, Proust's Remembrance, Nietzsche's philosophical writings and the major works of Rousseau, de Man concludes that all writing concerns itself with its own activity as language, and language, he says, is always unreliable, slippery, impossible. "A narrative," de Man tells us, "endlessly tells the story of its own denominational aberration, and it can only repeat this aberration on various levels of rhetorical complexity." In other words, literary narrative, because it must rely on language, tells the story of its own inability to tell a story. Subject, narrator and reader are all illusions generated by the text.

Rilke's poetic skills, de Man finds, are "directed towards the rhetorical potentialities of the signifier"; Proust's epic novel is about the impossibility of understanding itself since it is "the allegorical narrative of its own deconstruction"; Nietzche's philosophy is "an endless reflection on its own destruction at the hands of literature"; and Rousseau's writings are "allegories of unreadability . . . and of (non)signification." De Man demonstrates, beautifully and convincingly, that language turns back on itself, that rhetoric is untrustworthy. Any reading, then, based solely on the linguistic texture of literature is bound to explode in the face of its reader.

De Man prefaces his book with a remarkable account of how he came to write it. He says that it began as an historical study and turned into a theory of rhetoric and reading. "This shift," he writes, "which is typical of my generation, is of more interest in its results than in its causes." The results here are a theory about the impossibility of theory, about the endless circularity of language in its eternal betrayals, indeed, about the impossibility of critical reading itself.

So maybe the causes are more interesting than the results. We has a whole generation of critics abandoned history and turned to the linguistic material of literature, as though poetic language were a set of mathematical symbols to be decoded?

Allegories of Reading, its author says, is "the elaboration and the undoing of a system of tropological transformations." That is, de Man presents a system of figurative language and literary structure. The, as systemically, he unravels his own system. The very circularity of that procedure makes it suspect; de Man's point is, apparently, that his enterprise was doomed from the start. Objectified language, removed from its historical context, turns out to yield nothing.

All three of these studies establish a theoretical apparatus which is then applied to a selection of literary works. Tanner uses a framework of social history and anthropology to elaborate a hypothesis about attitudes toward adultery in bourgeois society and about the development of the novel as a literary genre. Both Genette and de Man use theoretical constructs from post-structuralist poetics to anchor their readings. Genette's book, modest in tone and honest about its enterprise, achieves a clarity and excitement in its unravelling of the vicissitudes of Proustian prose. De Man, both more assured and more turgid in his insistence that the word is the thing, seems at times to lose sight of that goal in favor of the process.