ELSEVIER Journal of Pragmatics 24 (1995) 77-98 ~lirmll ol Even" How to make theories with a word Alexis Kalokerinos Department of Linguistics, University of Crete, Perivolia, GR-74100 Rehythmno, Greece Abstract Even is a keyword in the semantic and pragmatic formulation of theories. This paper is pri- marily conceived as a criticial approach to Gricean theories as applied to the treatment of even, and aims at uncovering some of their problematic aspects. The constructive part of the paper, which comes last, defends an argumentative view of the inferential phenomena that appear in discourse when the word even is used. 1. Preliminaries The inquiry into the semantics of even ~ opens a battle between two theories based on opposing epistemological assumptions. The first approach is based on the theory of argumentation as developed by Jean-Claude Anscombre and Oswald Ducrot. The second represents a constellation of approaches as developed by Gilles Fauconnier, Larry Horn and Paul Kay, which, for convenience, will be called 'pragmatic'. The issue centers on what seems to be a fundamental disagreement on the nature of meaning. This opposition has its source in contrasting views on the relation between linguistic meaning and logic. Anscombre and Ducrot (1983: ch. 3; 1986: 87-88, 92; Ducrot et al., 1980: 10; Anscombre, 1989: 18-28), support the theory that natural language and logic con- stitute incompatible orders of discourse. Natural language has its own purely argu- mentative rules that cannot be reduced to a truth-functional system. In fact, the very notion of truth, which is central to logic, is considered irrelevant to semantic descrip- * The author would like to thank Jean-Claude Anscombre, Oswald Ducrot and Pierre-Yves Raccah, who read earlier versions of this paper and provided insightful comments, and Marlene Dolitsky, who helped turn a text in interlanguage into an English one. l Even will be taken here as a functional equivalent of French m~me. This does not mean that the two words are inter-translatable in every context (even if one believes in the possibility of local translation, pace Quine). Here we are concerned only with the function of "even/m~me' in 'enhancement contexts' ('contextes d'ench6dssement'). For a tentative unified description of the various uses of mdme, see Martin (1975). 0378-2166/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0378-2166(94)00091