Research Group
in Analytic Philosophy

Sobel Sequences as a Test Case for Static vs. Dynamic Semantics

    Adam Sennet (UC Davis)

Date: 20 March 2019

Time: 15:00

Place: Seminari de Filosofia UB

Abstract

Static semantics treats sentences (at contexts) as expressing truth conditions/propositions/information and constructs a theory of conversational update as a matter of pragmatics. Dynamic semantics treats sentences (at contexts) as expressing functions from contexts to contexts (so sentences affect contexts and have their truth conditions affected by contexts). Justifying one framework over the other has been a difficult matter but modals, counterfactuals in particular, exhibit conversational dynamics that has promised to support one over the other. The classic cases for consideration are Sobel Sequences (SSs) and Reverse Sobel Sequences (RSSs). The most promising representative of the dynamic approach is von Fintel’s dynamic semantics for subjunctive conditionals.

We argue that overall considerations, despite initial appearances, favour the static approach. More specifically, we will argue as follows:

 

(a) Von Fintel's semantics validate a principle that is supposed to explain the infelicity of Reverse Sobel Sequences. But there are counterexamples to the alleged entailment.

(b) Some RSSs are comprised of sentences that are jointly true regardless of their order of utterance. But von Fintel's theory entails that the second sentence of any RSS is inconsistent with the first.

(c) Von Fintel's theory utilizes only a limited range of tools to explain the predicted infelicity of Reverse Sobel Sequences. To substantially improve the empirical coverage of theories like his, one would need pragmatic elements of the sort posited by static semantic approaches.

(d) The dynamic semantics explanation of negative polarity licensing in subjunctive conditional antecedents over-generates.

 

We conclude that one of the more promising cases for a dynamic approach to modals fails and that pragmatic approaches to the promising data favour static approaches.