Mention-some readings of questions: Complexities from number marking

Despite the well-known fact that ◇-questions systematically accept mention-some answers, number marking on wh-complements adds two constraints to the distributional pattern of mention-some. First, a singular question admits a unique true answer and hence disallows mention-some. Second, a plural question disallows a mention-some answer that names only a singularity. I argue that the former constraint comes from a closure requirement on short answers, and the latter from the anti-presupposition of the plural morpheme.