Fragment answers and double negation in strict negative concord languages
This paper revisits the phenomenon of negative concord (NC) as an instance of polarity sensitivity. We shed light on a new set of data regarding n-words as fragment answers to negative questions and show that we find unexpected double negation (DN) readings for fragment n-words in view of their behavior in non-elliptical constructions. To account for this pattern, we offer an updated version of the hypothesis that n-words are strong NPIs, making use of an alternative and exhaustification approach. We argue that the difference between n-words and other NPIs should be seen as the result of two parameters: (i) whether reconstruction of the polarity item is allowed, and (ii) whether the polarity item has the ability to license a covert negation operator. The result is an explanatory account of NC and DN readings in both non-elliptical and elliptical environments, which allows for an easier integration of n-words in the broader typology of polarity sensitive items.