The Licensing of Negative Sensitive Items in Jordanian Arabic

This study investigates the licensing conditions on Negative Sensitive Items (NSIs) in Jordanian Arabic (JA). JA exhibits both types of NSIs that are discussed in the literature: Negative Polarity Items (NPIs) and Negative Concord Items (NCIs). Although these two sets of items seem to form a natural class in the sense that they show certain sensitivity to negation, they display important distributional differences that call for different analyses. First, NCIs can sometimes express negation on their own as in fragment answers; whereas NPIs cannot do so. Second, the licensing of NCIs is clause-bound; whereas the licensing of NPIs is not. Third, NPIs are acceptable in a number of contexts that do not involve overt negation; whereas NCIs are acceptable in only a subset of these contexts, namely without-clauses and before-clauses. The licensing of NPIs and NCIs in JA is discussed in light of previous theories that are mainly based on the distribution of these items in English and European languages. The investigation of NPI licensing in JA shows that the distribution of these items can best be captured by the semantic notion of (Non-)veridicality (Giannakidou 1998, 1999, 2000, 2002, 2006, 2011). Data from JA show that NPIs in the language need to be in the c-command domain of a non-veridical function at LF as proposed by the (Non)-veridicality Approach. The investigation of NCI licensing in JA shows that none of the NCI licensing theories previously proposed in the literature extends to JA. Alternatively, an account is proposed that is basically a crucial modification of the Non-negative Indefinites Approach (Zeijlstra 2004, 2008; Penka 2007, 2011) which takes Negative Concord to be a manifestation of syntactic agreement between an NCI and a semantic negation in the clause, where syntactic agreement is defined in terms of feature checking following recent assumptions within Minimalism (Chomsky 1995, 1998, 2000, 2001). I argue that NCIs are non-negative indefinites that are endowed with an [uNEG]-feature that needs to be checked against an [iNEG]-feature of a semantic negation that can be either overt or abstract in the clause. I also propose that Spec-head agreement and Head complement agreement exist side by side with c-command as licensing configurations for NCIs. I further argue that the level of representation at which NCI licensing takes place is not the same among all NCIs: while some NCIs are licensed at LF, other NCIs are licensed in the surface syntax. I show that this alternative account can capture the distribution of NCIs in JA. I also show that this account extends to NCIs in other languages such as Moroccan Arabic, Polish, and Spanish and is thus supported cross-linguistically