It has been known since at least Szabolcsi 2002 that the Russian plain disjunction marker ili behaves as a positive polarity item (PPI) when occurring under clausemate negation, which manifests itself in a violation of De Morgan’s laws. Among properties which the Russian plain disjunction marker ili shares with conventional PPIs such as
some are the locality of anti-licensing and the ability to be rescued by a second negation.
The Russian plain disjunction marker ili displays variable behaviour with respect to PPI-hood when interacting with Ā-movement. When the disjunction phrase appears to the left of the sentential negation marker ne ‘not’, the narrow-scope reading is strongly preferred. Exhaustification-based analyses of scalar inferences (Chierchia, Fox & Spector 2012, Spector 2014, Nicolae 2017) derive the basic pattern effortlessly but struggle with accounting for the observed obviation effects under syntactic fronting, as it is not clear if the account should be semantic or syntactic. Potential arguments for the syntactic approach would be the involvement of syntactic movement and the plausibility of the sentential negation marker ne being merely a negative concord item/NPI rather than the fully-fledged negation (Zeijlstra 2004). On this view, moving the disjunction phrase past the negation marker and closer to the negation operator Op¬ in the left periphery would enable the disjunction to scope under negation. Movement is thus the cause of the obviation effect but it is unclear what mechanism gives rise to the anti-licensing effect in the first place.
This paper sketches, instead, a meaning-based account of both the positive-polarity behaviour of the plain disjunction marker ili and the circumstances of obviation. To account for the general pattern of the inability of ili to scope under clausemate negation, I follow Nicolae (2017) in claiming that the scalar inference accompanying disjunction arises due to the interaction of a covert exhaustification operator Exh and a covert necessity modal, which renders the otherwise non-innocently excludable alternatives in upward-entailing contexts innocently excludable. In the case of plain disjunction (as opposed to complex disjunction, Spector 2014), the exhaustifier Exh only manipulates domain alternatives. The incompatibility with negation results, on this view, from a ban on vacuous exhaustification: if an LF containing Exh does not yield a strengthened meaning, it is ruled inadmissible.
Two sets of facts speak in favour of a meaning-based analysis of PPI obviation effects when the disjunction phrase undergoes Ā-movement: (i) obviation is sensitive to the interpretation of the Ā-moved item (topic; *focus); (ii) when multiple disjunction phrases are fronted, only one, rather than all, of them can scope under negation. Topicalisation would create additional alternatives alongside the already defined domain alternatives, and exhaustification would not be vacuous. This is in line with Nicolae’s (2017) suggestion that PPI-disjunctions take into account both the truth-conditional and non-truth-conditional meaning. On this view, fronting will not be the cause of obviation effect but rather a byproduct of topicality, given the fact that Russian topics are typically fronted.
You can view the slides here
