Séminaire de Recherche en Linguistique

In this talk, we will look at Finnish yes-no questions, argued to show the four possible combinations of [±pronounced] and [±interpreted] negation. The data indicates that object case-marking covaries with licensing of polarity items across positive and negative yes-no questions: in a question with a bounded predicate, the presence of [+interpreted, ±pronounced] negation requires partitive case-marking and licenses negative polarity items, while [-interpreted, ±pronounced] negation requires accusative case-marking and licenses positive polarity items. Three properties specific to Finnish allow us to track the interpretation and pronunciation of negation: i) the expression of negation with a negative auxiliary that is inflected for person and number but not tense, originating between a tense projection and a projection with φ-features (Holmberg et al. 1993), ii) the movement of the highest visible head of IP to FocP in yes-no questions (Holmberg 2003, 2013), and iii) the semantically conditioned variation of structural object case, where the object of an unbounded VP always receives partitive case-marking, including objects of negated VPs (Kiparsky 1998). While the data is subtle when only object case-marking is taken into account, the use of polarity items clearly brings out the underlying values of the [±pronunciation] and [±interpretation] of negation. The proposed analysis is also supported by data from exclamatives using the yes-no question particle -kO ([-interpreted, +pronounced] negation) and rhetorical questions ([+interpreted, -pronounced] negation), and leads to a natural account of the observed answer pattern (in the sense of Farkas and Bruce 2009, Roelofsen and Farkas 2015).