A novel problem for the likelihood-based semantics of "even"
Semantics and Pragmatics
This paper describes and examines novel data which is problematic for the popular view that the scale in the scalar presupposition of even is based on comparative (un)likelihood or probability. Several theories in the past attacked this characterization of the scale by pointing out examples where even is felicitous although the likelihood-based presupposition does not seem to be met, indicating that the comparative unlikelihood of p is not a necessary condition for a felicitous use of even p.
... us use of even p. In contrast to these theories, this paper examines novel data where even is infelicitous although the likelihood-based presupposition is met, indicating that the comparative unlikelihood of p is not a sufficient condition for a felicitous use of even p. In particular, I look at the unexpected infelicity of even p in the context of a salient p or q alternative. I examine, and reject, several potential objections to the claim that such sentences pose a problem for the likelihood-based approach to even and conclude that the novel data constitute a real challenge for this popular approach to even.