THIS SOCIO-LITERARY CRITIC and structuralist philosophe, whom Frank Kermode has called "the most interesting, fertile and ambitious critic now writing," has always been a puzzle, even to other Frenchmen. During a writing career that has lasted over 25 years, he has avoided predictability to the extent that his readers have never had the slightest notion of what he would produce next. He has steadfastly refused to write "mere" criticism, and if the day ever comes when we decide to do away with the distinct borders which separate the literary text (which is "creative") from the critical commentary on that text (which is "analytic" and therefore somehow "anti-creative"), Barthes will be our exemplar.

His primary concern has always been the elucidation and development of semiology, the science of signs first sketched out by the French linguist Ferdinand de Saussure in the 19th century. This science, which Barthes now considers more or less identical with structuralism, is a methodology, a way of examining culture, rather than a philosophy itself. In essence, structuralism-/semiology is the study of how cultural phenomena are meaningful and by what hidden codes they speak to us. Language itself is obviously the chief model of a systematic cultural code, and structuralists simply apply, to a host of different disciplines, the knowledge modern linguistics has given us of how language codes work. So, for example, Claude Levi-Strauss, who is an anthropologist, has studied the behavior of primitive tribes in terms of the linguistic model. Jacques Lacan in psychology, Christian Metz in film, and Barthes in literature and contemporary social phenomena, have utilized virtually the same system.

Probably Barthes' most accessible work is Mythologies, which was originally published in 1957, but not translated in English until 1972. In this delightful book, he analyzes such diverse subjects as wrestling, striptease, toys, "the face of Garbo," "the brain of Einstein," and the Gallic passion for steak and french fries, regarding each one as a "text" to be interpreted. In so doing, he reveals the non-linquistic "languages" which operate through these phenomena to make them meaningful. He also shows how these codes are used to manipulate us.

In Systeme de la Mode. (1967), which has never been published in English, he attempted, with much less success, to apply this same method to the world of fashion, through the length of an entire book. Here the problem seems to be his attempt to find exact parallels with linguistics; the relatively simpler code of fashion is overwhelmed in the process. Much more successful was his study entitled Sade, Fourier, Loyola (1971, in English, 1976), in which he managed an outrageous comparison of the "languages of Sade's eroticism, Fourier's social utopianism, and Ignatius Loyola's Spiritual Exercise.

His best-known work is the recent S/Z. There, he amazed both admirers and detractors by taking as his text an entire Balzac short story ("Sarrasine), analyzing it passage by passage, and at the same time enumerating the five seperate codes he found therein. Serious omissions prevent the study from being as exhaustive as Barthes claims, but it is so studded with acute perceptions of how fiction really works that the reader is left astonished.

And now in Roland Barthes, the critic has at last turned toward himself as the text to be studied. The result is a highly unconventional "autobiography" (splendidly translated) which is brilliant and baffling by turns. Somewhat in the manner of his recent exercise, The Pleasure of the Text, this book is made up of hundreds of seemingly unrelated meditations (in addition to handwritten notes, a photo album, cartoons and doodles) where reason and intuition are allowed equal play. It cannot be read linearly, instead, the reader must dip in at random, savoring the flashes of poetic insight rather than seeking a sustained argument. The tone is at times confessional (his tuberculosis, his hinted-at homosexuality, his "panic boredom": "Might boredom be my form of hysteria?") and at other times distanced, as on the first page where he warns that "all this must be considered as if spoken by a character in a novel."

As in Mythologies, the range of topics is wide: migraine headaches, the nature of pleasure and friendships playing the piano; Chaplin; the benefits of perversion; likes and dislikes. In one luminous passage, for example, he analyzes the sudden shortcircuiting of a conversation he is having with a shopkeeper concening the weather when he comments on quality of light. He realizes that this sort of perception of what is not immediately obvious, as opposed say, to a gorgeous sunset, relates to a "class sensibility." "In short," he concludes, "nothing more cultural than the atmosphere, nothing more ideological than what the weather is doing."

Above all, his subject, as always, is language - commercial, political, literary - and its relation to the self. He refers often to his previous books, glossing their key terms, frankly outlining the stages of his intellectual development. The purpose of his most recent work, he explains, has been to introduce "the texture of desire, the claims of the body" into semiology/structuralism, which he feels has become too narrowly scientific, especially concerning phenomena, like literature, which are ultimately unquantifiable. The question he then poses himself is "where to go next?" In partial answer, this inner dialogue is scrawled on the final page: "And afterward?" "What to write now? Can you still write anything?" "One writes with one's desire, and I am not through desiring." Like those which came before, a difficult, astounding book.