Polar questions in nɬeʔkepmxcín: monopolar, bipolar, and exhaustive

There is debate about whether polar questions (PQs) have bipolar semantics (e.g., denoting a set of propositions {p, ¬p}), monopolar semantics (a singleton set {p}), or both. The issue is difficult to settle using English data alone. In this paper I bring new data to bear on the debate from nɬe ʔkepmxcín (Salish). I argue that natural language has both bipolar and monopolar questions, and that nɬe ʔkepmxcín morphosyntactically distinguishes the two. I further argue that bipolar questions come in two types, which are also morphosyntactically distinguished in nɬe ʔkepmxcín: exhaustive (presupposing that p and ¬p are the only two answer options), and non-exhaustive (allowing answers beyond p and ¬p). I thus argue that nɬe ʔkepmxcín’s three-way morphosyntactic contrast in polar question forms reflects a three-way semantic contrast. The nɬe ʔkepmxcín data have implications for the analysis of other languages. I argue against existing analyses of English plain PQs as either uniformly bipolar or monopolar, and in favour of an ambiguity analysis. The nɬe ʔkepmxcín data further support a distinction in at least some languages between so-called inquisitive and assertive declarative questions (DQs), rather than a unified analysis of these. Finally, nɬe ʔkepmxcín provides evidence that declarative-question-like or monopolar questions cross-linguistically need not be non-canonical, and their properties should therefore not be derived via markedness.