Meaning-driven Unacceptability, Modulated Logical Forms and the "Spontaneous Logicality of Language''

Please jlg363 [at] stanford.edu (e-mail) for Zoom link.

Abstract: This talk will present a specific implementation of the `Logicality of language' hypothesis, according to which the human language system includes not just a recursive syntax and compositional semantics, as standardly assumed, but also a computational system that identifies and filters out expressions that are informationally trivial. This hypothesis is motivated by a class of trivial expressions which are syntactically well-formed, have no semantic type mismatches, and are not more complex than comparable expressions which are easily processed, yet are reliably judged to be strictly unacceptable in a way that is indistinguishable from typical ungrammaticality judgments. Triviality-driven unacceptability restricts the distribution of various kinds of logical/functional terms and phrases, including determiners, modals, attitude verbs, prepositions, and overt and covert exhaustifiers. An important challenges for proponents of Logicality is to implement it in a way that doesn't over-generate strict unacceptability assignments for various kinds of superficial tautologies and contradictions. I will propose one way of solving this problem, based on a contextualist construal of logical form according to which content-based/open-class terms are subject to generalized meaning modulation operations. I will also argue that this approach is superior to various competing pragmatic accounts of the target cases of meaning-driven unacceptability.