We want to pass on a big warm congratulations to Reuben Cohn-Gordon who successfully defended his dissertation "Probabilistic Models of Pragmatics for Natural Language" on Apriil 16! We also wanted to give an extra heartfelt shout-out to all the staff and faculty who helped make this all-video thesis oral possible! Amazing to see the adaptations we're taking in these times.

The abstract for Reuben's thesis can be found below:

When conveying or receiving information via language, humans make use not only of shared conventions about what linguistic units mean (semantics), but also draw inferences based on the context they are in and the intentions of their interlocutor (pragmatics). In recent years, probabilistic models of nested reasoning have emerged as a promising approach for understanding natural language pragmatics, in conjunction with existing logical approaches to semantics. However, these models are often idealized, applying only to very simple domains. I describe my dissertation research, in which I fuse statistical representations of semantic meaning with probabilistic models of pragmatic reasoning to build systems for interpreting and producing natural language, and discuss both their practical value and theoretical significance.