CLASP
The Centre for Linguistic Theory and Studies in Probability

Algebraic effects in Montague semantics

Within the last two decades, research into the multifaceted nature of linguistic meaning from the perspective of dynamic semantics has gained insight from the theoretical-computer-science notion of computational side effect. The side effects of a computer program may include, for example, reading or writing to the environment, computing non-deterministically, accepting input, or producing output. Theories of side effects have provided useful design patterns in linguistic semantics that allow traditional tools, like the simply typed lambda-calculus, to deal with otherwise unruly dynamic phenomena like indefiniteness, binding, anaphora, and presupposition, as well as apparently non-compositional phenomena, like quantification, in a systematic, elegant, and compositional way.

More recently, the theory of algebraic effects has provided an approach to side effects in programming languages that allows very different notions of effect (for example, input/output and non-determinism) to be combined in a relatively seamless way. In this talk, I will show how algebraic effects may be used to combine analyses of linguistic side effects, focusing on anaphora and quantification. The approach lends itself to a simple encoding using traditional tools, as well as fairly conservative analyses of the individual phenomena at play. In addition, it leads to some interesting empirical predictions about how quantification and anaphora interact.

The talk is accompanied by a blog post, which is available at: https://juliangrove.github.io/blog/algebraic_effects_montague.html