Inferring priors

This research project concerns how understanding language relies on background beliefs of the listener and whether these beliefs reveal themselves in the interpretations. I approach this topic from both experimental and modeling perspectives. Together with Gregory Scontras, Christian Stegemann-Philipps, Johannes Lohmann, and Martin Butz we developed a Rational Speech Act formalism to simulate the inference of priors upon observing a choice of an object in a signaling game.

We further investigate how shared beliefs can be discovered in the process of chooosing and interpreting ambiguous utterances. Here, we investigate the phenomenon of indiretness and show how indirect utterances allow speakers to navigate complex communication goals. This project aims to explore how communication contributes to maintaing of social bonds

Surprising anaphoric references

Speakers take into account a number of linguistic and cognitive factors when they choose how to refer to a previously mentioned entity. In this project, we investigate how speakers describe surprising and non-surprising events. To quantify surprise as a cognitive category, we first evaluate what expectations speakers have about possible event progressions. We can further manipulate prior expectations by training the participants possible interactions between novel characters. We then ask participants to create simple narratives and observe how they refer to different characters.

We demonstrate that speakers choose more informative references when they describe surprising events. Is this caution warranted? What happens when listeners perceive such event descriptions in noisy environment? In a new collaboration project with Marjolein van Os, Vera Demberg, and Martin Butz we explore how participants weigh their prior expectations and the linguistic signal in the process of utterance comprehension.

Williams Syndrome

Individuals with Williams Syndrome display a rare cognitive profile: it combines severe spatial and numerical deficits and grammatically complex, fluent language. The nature of pragmatic skills displayed by speakers with Williams Syndrome remains debated. In this project, we investigate how these speakers understand utterances with quantifiers and other scalar terms. Our goal is to investigate whether individuals with WS have abstract semantic and pragmatic representations. This project is done in collaboration with Barbara Landau (Johns Hopkins University), Rennie Pasquinelli (Johns Hopkins University), Julien Musolino (Rutgers University), and Martin Butz (University of Tübingen).