Twentieth-century writers, scholars and critics have formulated explicitly and illustrated systematically the doctrine that literary language is autonomous. The aim of our thesis is to analyse the scope and implications of this now widespread view of literary language, which we examine here with particular reference to twentieth-century French literature. Our study follows three main lines of inquiry: In the first part we analyse how-the belief in the autonomy of literary language came to be formulated. We then try to envisage such an autonomous literary language in practical terms. Finally we outline some of its broader implications for our understanding of literature. An analysis of the scattered doctrines of the Tel Quel group, dedicated to bringing about the semiotic autonomy of the literary text," reveals the existence of a philosophy of literature based on the premise that, in its very essence, the literary text is not meant to communicate or to transmit an intelligible message and is, in fact, not dependent on any pre-established norms. Since full consideration of a theoretical view involves as well an appreciation of its applicability to the real world, we examine in the second part of our study the theory of an autonomous literary language from a pragmatic point of view. Linguists will readily admit that, in a theoretical sense, every individual is free to create his own independent linguistic code; but they will also maintain that, in realistic terms, the utilization of such a freedom is "absurd," since the process of communication per se can exist only at a socialized level. One must therefore conclude that any theory which propounds the practice of an autonomous literary language is not meant to be taken literally but calls for a metaphorical interpretation. We have searched out this metaphorical meaning by attempting to determine at what precise level, the literary proponents of this doctrine depart, in their own writings, from what linguists have defined as the normal process of communication; in other words, at what level do their writings become incomprehensible to, for example, the educated French reader? It becomes clear in the process that if such literary works in fact need to be "deciphered" it is not because they possess their own private phonological, lexical and syntactical systems but rather because they do not rely on the conventional coherence of traditional narrative. Instead of deriving from a subordination of events in a hierarchy of causal relations, this "new coherence" presents the reader with no more than a simple juxtaposition of semantically related narrative elements. The subordination of narrative units in a cause/effect hierarchy is not essential to the reader's understanding of the linguistic code of a particular novel or work. It is, however, clearly an essential element of the process of communication for without it the reader cannot objectively perceive the context to which the work refers. But what we have labelled a "new coherence" implies the very absence of such a hierarchy essential to the process of communication. Consequently we are forced to the conclusion that the doctrine of an autonomous literary language is significant enough to require new definitions of some fundamental aspects of narration, as well as a radical change in our perception of the historical development of literature and in our cognition of the individual literary text.