This paper deals with the treatment of figures of speech in Perelman’s and Olbrechts-Tyteca’s Treatise on Argumentation (TA), and, more broadly, with the place of figures in argumentation theory. The contrast between two conceptions (or two domains)\n of rhetoric, “a rhetoric of figures” and “a rhetoric of argument” can be traced back to Ramus, and it has been revived in\n the seventies through the perception of an incommensurability between Perelman’s “New Rhetoric” and the École de Liège’s “General\n Rhetoric”. Modern theories (...) of argumentation, oriented towards the characterization and denunciation of fallacious discourse,\n emphasize the gap between sound argumentative discourse and rhetorical discourse seen as a “powerful instrument of error and\n deceit” (Locke). This concept of argument presupposes a vision of language as ideal/transparent—a revised language that is\n not the language of ordinary argumentation. Figures are not basically “decorative”; they are manifestations of the complex\n process of language structuring in speech. Thus rejecting figures amounts to a negation of discourse as such. The second section\n turns to a somewhat neglected aspect of the TA, its both decisive and somewhat cavalier theory of figures of speech, and its extended use and re-definition of a complex set of figures. We argue that the TA, in its quest for descriptive adequacy, breaks with the traditional and comfortable concept of figures as useless fallacious\n “ornaments” and provides us with the first description of what could be characterized as the semantic level of ordinary argumentative\n discourse. This will be shown on the case of “figures of choice, presence and communion”, and could be extended to the discursive\n construction of objects and participants, including the speaker and her emotions. (shrink)