Language and Speech Technology

We research and develop computational methods for the comprehension and generation of text and speech. We evaluate these methods intrinsically, and test their utility in real-world applications (such as automatic speech recognition in realistic environments, and text mining in historical collections and social media). We also connect these computational models to models in theoretical and applied linguistics, psycholinguistics, and communication studies. The group works with ‘big data' and methods that learn computational models from data, while at the same time the proper development of information systems demands a hybridization of data-driven learning and ‘rich priors' from the domain or context.