Detail of contribution
Auteur: Laurence HORN
Titre:
Revisiting the licensing question: some negative (and positive) results
Abstract/Résumé: There are two leading families of programs addressing the "licensing problem" for the distribution of negative polarity items: • Ladusaw’s downward entailment theory, refined and extended by Heim, Kadmon & Landman, Krifka, von Fintel, Hoeksema and others, on which NPI licensing is governed by monotonicity and other semantic principles expressible as progressively stricter Boolean conditions • Linebarger’s syntactico-pragmatic account invoking direct licensing by overt negation and indirect licensing by negative implicature, the latter reconstructed more recently as a supplementary “rescue” operation supplementing the non-veridicality theory of Giannakidou and her colleagues. I argue that no strictly formal approach can do justice to the empirical domain, especially given the reliance on entailment underlying mainstream approaches. By surveying a range of cases that demonstrate the empirical inadequacy of entailment and implicature based approaches and extending the database to a range of cases in which NPIs are ill-formed despite the presence of an accessible discourse-salient but non-asserted negative proposition, I will argue that NPI licensing in languages that work like English depends not on what is ENTAILED or what is IMPLICATED but on what is ASSERTED, or more generally on the illocutionary point of the utterance, what is at issue. Among the question to be addressed in this presentation are: (i) What can the medieval theory of vertical inference tell us about monotonicity? (ii) How do the formal definitions of downward entailment and anti-additivity allow for the existence of anti-additive but non-downward entailing operators, and how does this relate to the distribution of free choice items? (iii) How might the notion of assertoric inertia help us capture both the semantic symmetry and pragmatic asymmetry of a class of “exponible” non-downward entailing NPI licensers including exclusives ("only"), exceptives ("nobody/nothing but"), "at most n", and "barely"? (iv) How can we sort out the conflicting empirical claims about licensing in the scope of free squatitives (e.g. "know squat about"), "could care less about", and other expressively negative environments lacking formal negation? (v) What are the implications of the distribution of NPIs in sarcastic utterances ("Like I’d ever eat any of that"), conditionals, polar questions (with and without interrogative syntax), clefts, and the restrictors of definite DPs? (vi) What do the empirical problems are encountered by Giannakidou's “rescue”-based analysis for NPIs in veridical contexts tell us about the nature of polarity licensing?
Titre:
Revisiting the licensing question: some negative (and positive) results
Abstract/Résumé: There are two leading families of programs addressing the "licensing problem" for the distribution of negative polarity items: • Ladusaw’s downward entailment theory, refined and extended by Heim, Kadmon & Landman, Krifka, von Fintel, Hoeksema and others, on which NPI licensing is governed by monotonicity and other semantic principles expressible as progressively stricter Boolean conditions • Linebarger’s syntactico-pragmatic account invoking direct licensing by overt negation and indirect licensing by negative implicature, the latter reconstructed more recently as a supplementary “rescue” operation supplementing the non-veridicality theory of Giannakidou and her colleagues. I argue that no strictly formal approach can do justice to the empirical domain, especially given the reliance on entailment underlying mainstream approaches. By surveying a range of cases that demonstrate the empirical inadequacy of entailment and implicature based approaches and extending the database to a range of cases in which NPIs are ill-formed despite the presence of an accessible discourse-salient but non-asserted negative proposition, I will argue that NPI licensing in languages that work like English depends not on what is ENTAILED or what is IMPLICATED but on what is ASSERTED, or more generally on the illocutionary point of the utterance, what is at issue. Among the question to be addressed in this presentation are: (i) What can the medieval theory of vertical inference tell us about monotonicity? (ii) How do the formal definitions of downward entailment and anti-additivity allow for the existence of anti-additive but non-downward entailing operators, and how does this relate to the distribution of free choice items? (iii) How might the notion of assertoric inertia help us capture both the semantic symmetry and pragmatic asymmetry of a class of “exponible” non-downward entailing NPI licensers including exclusives ("only"), exceptives ("nobody/nothing but"), "at most n", and "barely"? (iv) How can we sort out the conflicting empirical claims about licensing in the scope of free squatitives (e.g. "know squat about"), "could care less about", and other expressively negative environments lacking formal negation? (v) What are the implications of the distribution of NPIs in sarcastic utterances ("Like I’d ever eat any of that"), conditionals, polar questions (with and without interrogative syntax), clefts, and the restrictors of definite DPs? (vi) What do the empirical problems are encountered by Giannakidou's “rescue”-based analysis for NPIs in veridical contexts tell us about the nature of polarity licensing?