Charlie Siu

A well-known problem facing Lasersohn's theory of pragmatic halos is that it fails to predict that the loose contents of negated maximal standard absolute adjectives (e.g. ``not straight'') and of minimal standard absolute adjectives (e.g. ``bent'') are stronger than their literal contents. Recently, Dinges argues that Klecha's optimality-based theory is faced with the same strengthening problem and proposes to solve it by using Hoek's theory of conversational exculpature. This paper argues that the strengthening problem can be better solved by combining Lasersohn's and Klecha's theories after revising each theory slightly: Pace Lasersohn, we do not identify the loose content of a sentence with the disjunction of the values in the sentence's halo. Pace Klecha, we do not hold that the question under discussion (QUD) of a conversation has any sentence content coarsening function. Once these modifications are made to Lasersohn's and Klecha's theories, the relevant loose content of a sentence can be determined based on an optimality-based reasoning about the halos of the speaker's utterance and its alternatives.

Carter (Noûs 55(1):171–198, 2021) argued that while most simple positive numerical sentences are literally false, they can communicate true contents because relevance has a weakening effect on their literal contents. This paper presents a challenge for his account by considering entailments between the imprecise contents of numerical sentences and the imprecise contents of comparatives. I argue that while Carter's weakening mechanism can generate the imprecise contents of plain comparatives such as `A is taller than B', it cannot generate the imprecise contents of comparatives that quantify the difference between their arguments, such as `A is exactly n times as tall as B'. I then propose an alternative theory on which intervals serve as both thick degrees on a scale and denotations of numerical expressions. I argue that the alternative theory can account for the imprecise contents of both forms of comparatives as well as the data that motivate Carter's theory.

It is a consequence of both Kennedy and McNally’s typology of the scale structures of gradable adjectives and Kennedy’s economy principle that an object is clean just in case its degree of cleanness is maximal. So they jointly predict that the sentence ‘Both towels are clean, but the red one is cleaner than the blue one’ is a contradiction. Surely, one can account for the sentence’s assertability by saying that the first instance of ‘clean’ is used loosely: since ‘clean’ pragmatically conveys the property of being close to maximally clean rather than the property of being maximally clean, the sentence as a whole conveys a consistent proposition. I challenge this semantics–pragmatics package by considering the sentence ‘Mary believes that both towels are clean but that the red one is cleaner than the blue one’. We can certainly use this sentence to attribute a coherent belief to Mary: One of its readings says that she believes that the towels are clean by a contextually salient standard ; the other says that she believes that the towels are clean by her own standard. I argue that Kennedy’s semantics–pragmatics package can’t deliver those readings, and propose that we drop the economy principle and account for those readings semantically by assigning to the belief sentence two distinct truth conditions. I consider two ways to deliver those truth-conditions. The first one posits world-variables in the sentence’s logical form and analyzes those truth-conditions as resulting from two binding possibilities of those variables. The second one proposes that the threshold function introduced by the phonologically null morpheme pos is shiftable in belief contexts.