DIACRITICS: One source of continuity in your critical and theoretical work over the past twenty years has been your unwavering attention to the performance of the reader. In your recent work, however, one also senses that the framework within which you envisage the reading process has been shifting, that it is more clearly predicated on a model of the relation between a text and the understanding that this text engenders. It is as if the text contains a paradigm of the reader's transformative praxis. While you still insist on the way in which a knowledgeable reader will understand certain elements in a text (ranging from lexical items through cliches, quotations, and sound combinations to syntactic structures), you no longer speak of the super-reader; and rather than staking claims on the potential normative value of a set of reader responses, you insist on the over-determination of the reading process by the text. We would like to know how, in your own mind, you have come to your current emphasis on the process of mimesis cancellation, on the praxis of semiosis, and so forth. Do you envisage your recent books as a significant break with your earlier work, or simply a revision or theoretical extension? M. RIFFATERRE: My present work is a continuation of my earlier research, without any break I am aware of. At first I concentrated on the heuristic reading stage (the microcontextual determination of the points that fix the reader's attention). What I called the super-reader was never anything like a real reader or his substitute, nor was it a reading norm. I meant a technique for spotting the segments of the text that drew most reader response. My next step was naturally to investigate how the network of these activated segments guides and orients the hermeneutic reading stage. The constraints imposed on the reader's possible interpretations soon looked to me like being thwarted, regularly though temporarily, by indirection of meaning, by inappropriateness, and in general by ungrammaticalities I classify as catachresis. I recognized that these blocks (especially poetic obscurity) were affecting the mimesis in many distinct ways, but that looked at from another angle they all served the same function. This single, all-transforming function had to be the semiosis. My breakthrough was the concept of descriptive system: it explains how metonyms are changed into metaphors, and how one marking can by itself transform all the system's components simultaneously into a code for the representation of something other than its natural referents. DIACRITICS: What value or function do you ascribe to the distinction of theory from practice in your work? Would you claim that your theory, just as much as your reading praxis, is imposed by the poetic text? M. RIFFATERRE: Certainly. My theory is indeed imposed by the text. Theory must propose a model that accounts as economically as possible for