The Probabilistic Turn in Semantics and Pragmatics

This article provides an overview of graded and probabilistic approaches in semantics pragmatics. These share a common set core research goals: ( a) concern with phenomena that are best described as graded, including vast lexicon words whose meanings adapt flexibly to the contexts which they used, well reasoning under uncertainty about interlocutors, their goals, strategies; b) need show representations learnable, i.e., listener can learn semantic pragmatic from data; c) emphasis on empirical evaluation against experimental data or corpus at scale; d) scaling up full size lexicon. The methods used sometimes explicitly not. Previously, there were assumed be clear boundaries among frameworks, classifiers machine learning, distributional approaches, but these have been blurred. Frameworks pragmatics use all three these, combination, address four questions above.