Retour vers liste

Détail de la contribution

Auteur: Seiko FUJII

Non-predictive conditionals

Abstract/Résumé: Studies in logic on the operator IF have largely focused on the predictive functions of conditionals – the kind of conditionals where if P, then Q gives rise to the implicature that if not P, not Q. Work in pragmatics, starting with Van der Auwera (1986), and recent work in mental spaces theory (Sweetser 1990, Dancygier 1998, Dancygier and Sweetser 2005, etc.), on the other hand, has also analyzed non-predictive conditionals. E.g., If you need any help, my name is Chris (ibid) does not implicate that the speaker’s name is not Chris if the addressee does not need help. Speech-act, metalinguistic, and epistemic conditionals are all most frequently non-predictive in function, unlike content domain conditionals (ibid). Cross-linguistic questions arise from these observations: Are there parallels in how languages constructionally mark non-predictive conditionality, or how (or whether) they mark particular categories of non-predictive conditionals? The general purpose of this paper is to lay out these questions for crosslinguistic studies of the semantics and pragmatics of conditional constructions. The specific purpose is to explore non-predictive conditionals, focusing on epistemic and speech-act conditionals in Japanese and English, with brief pointers to relevant phenomena in other languages. Japanese frequently conveys epistemic conditionality with overt epistemic modal marking on the consequent (Fujii 1992, 1993); English does not exhibit such a strong correlation, although epistemic modals do readily occur in epistemic conditional Q-clauses. Second, the antecedent of epistemic conditionals is also typically marked by a specific linker that constructs a tensed clause – nominalized P; or if not nominalized, by NARA rather than other linkers; or by a complex linker which transparently sets up an epistemic mental-space. Unlike other conditional linkers, NARA permits free choice of tense within P independently of Q; it does not require P to be perfective or temporally ordered prior to Q. Nominalizing P makes P disjoint from Q, conveying that the speaker takes P for granted. Based on these observations, it further attempts to clarify the nature of non-predictive conditionals. Whereas predictive conditionals are strictly subject to the requirements of concordance (Fillmore 1990; epistemic coherence Dancygier & Sweetser 2005) between P and Q, non-predictive conditionals are not. In this way, P and Q in non-predictive conditionals can be disjoint formally or semantically. Though the grammatical devices differ, Japanese non-predictive conditionals are also conveyed by constructions that accommodate disjoint P- and Q-clauses, where the construal of temporal and causal links between the two events in the content domain is not required.