Ludics and Presupposition Projection ?

This work looks at the way a ludics-inspired approach to discourse deals with the discursive notions of projection and attachment of semantic content. An illustration of the case of presupposition is studied and discussed in the light of recent theories of presupposition projection. The case of the proviso problem and conditional sentences is also touched upon. It is then proposed to describe a specific use of but as an indicator of a meta-game, i.e. an invitation to revise a previous utterance which then grants access to previously inaccessible material such as presupposition. This paper deals with a fairly recent model of discourse interpretation proposed by Lecomte & Quatrini (2009,2010,2011) that rests on the ludics framework initiated by Girard (2001) as a way to rebuild the foundations of logic. The goal of this paper is to evaluate the ludics approach to discourse with regards to the question of the accessibility and attachment to discourse material conveyed by a discourse move. This question is prominently discussed in the literature on discourse interpretation, for example by Asher & Lascarides (2003) through the notion of right frontier, or more recently in works like Simons et al. (2010) about the notion of projective material, i.e. material that is conveyed by an utterance even though it is embedded under operators that are not usually veridical. It is thus worth to see in which way a ludics-based approach can deal with such facts. We first begin by outlining the ludics approach, after which we cover the empirical domain of projective material along with some recent proposals of analysis for this phenomenon. We then evaluate the way a ludics approach can handle the question of the projection of presupposed content as well as the impossibility to attach to presupposed material to establish a discourse relation. In a final part, we discuss the possibility to characterize the so-called monadic use of but as a marker of revision in the studied framework. 1 Ludics and Discourse Interpretation In this section, I briefly review the main assumptions and hypotheses about discourse interpretation in the framework of ludics. ? This work has been supported by ANR LOCI. I am grateful to Alain Lecomte and Myriam Quatrini for their interest and discussions on this topic. Ludics, as proposed by Girard (2001), offers a framework for proof theory that is built on the notion of interaction. As such, it can easily be interpreted in a dialogical perspective for natural language, and it is what has been proposed by Lecomte & Quatrini (2009,2010,2011). It is not our intention to repeat the formal and foundational definitions behind ludics, and the interested reader should consult either the presentation made by Lecomte & Quatrini (2011) or the foundational text by Girard (2001). One of the main ideas of the application of ludics to natural language semantics is to drop the usual semantic postulate that equates the meaning of a sentence to its truth conditions. Ludics rather considers that the meaning of a sentence is the set of its proofs. A proof is here intendend as an object which is intrisically rooted in interaction: the construction of a proof results from the interaction between the actions of two different individuals. Thus, when a speaker produces an utterance, his action can consist in both a positive action (the contribution he makes to the discourse) and a negative action that corresponds to the reactions the speaker expects regarding his move. These negative actions are the possible refutations on which one might ask the speaker to elaborate in order to prove or back up his assertion. They represent as many openings for the rest of the conversation, openings on which the speaker allows other participants to anchor their reactions. To illustrate the ludics approach and the way the speakers’ interaction is represented, the example (1) (adapted from Lecomte & Quatrini (2010) and taken from A. Dumas’ Count of Monte Cristo) is given a simplified and macroscopic treatment. Positive and negative actions are not represented at this stage, but are bundled behind the utterances of each participant. The point of this illustration is to introduce the way a dialogue is represented in the ludics approach. (1) a. E: I was to be captain of the Pharaoh; I was to marry a nice girl. b. F: Did someone had an interest in you not becoming captain? c. E: Only one man: Danglars. d. F: Now, tell me about the girl you were supposed to marry.