The relation between Negative Polarity and Negative Concord

Abstract

Negative Polarity and Negative Concord tend to be viewed rather differently. While Negative Polarity is widely analyzed in semantic terms (e.g. Ladusaw, 1980), an influential analysis of Negative Concord views it as an instance of syntactic agreement (e.g. Zeijlstra, 2004). This paper interrogates this division of labor and rehearses arguments for saying that Negative Polarity and Negative Concord are closely related (e.g. Herburger, 2024). Being an NPI on this view amounts to carrying a meaningless feature that needs to be checked in a downward entailing environment (e.g. Dowty, 1994). Semantically predisposed expressions can come to be parsed as carrying this feature but don’t have to. Negative Concord results when an NPI comes to have a semantically negative homophone with different (but not necessarily disjoint) distribution. Comparing Spanish to French and French to English, I argue that the change from Negative Concord to a Double Negation system ensues when the [uneg] versions have been crowded out by semantically negative versions as a result of process that starts gradually. An important step in this process, I argue, relates to how sentential negation comes to be realized postverbally as part of the Jespersen Cycle.