About

Semantic research in the Montagovian tradition has taken an experimental turn in the last decade, as the empirical goals of semanticists have gotten more ambitious. Large-scale inference datasets—and the statistical models they require—have become a more standard part of the semanticist’s toolkit. pds is an approach to developing models of inference datasets that involves seamlessly deriving them from semantic grammar fragments.

As a library, pds encodes the ideas described in Probabilistic Dynamic Semantics by Julian Grove and Aaron Steven White. Applications of the framework can also be found in Grove and White (2025a) and Grove and White (2025b).