When is 'or not' required in an embedded polar question?

Abstract

Polar question embedding has diverse morphosyntactic expressions across languages; one strategy that comes up frequently is the insertion of a disjunctive particle 'or not.' For languages that use this disjunctive particle, a striking pattern emerges: those which have an interrogative complementizer and/or clause-typing marker in their embedded polar questions may optionally leave out 'or not,' whereas those which lack them require 'or not' across all syntactic embedding environments. Adopting a Hamblin-Karttunen approach to the semantics of questions, these patterns can be explained from the requirement that embedded polar interrogatives denote non-singleton propositional sets consisting of a positive and negative alternative. Where 'or not' is optional, an alternative strategy is available to coerce this bipolar question meaning, subject to language-specific locality constraints.