IRFP: Metalanguage Identification for Interactive Language Technologies

The International Research Fellowship Program enables U.S. scientists and engineers to conduct nine to twenty-four months of research abroad. The program's awards provide opportunities for joint research, and the use of unique or complementary facilities, expertise, and experimental conditions abroad.

This award will support a twenty-month research fellowship by Dr. Shomir Wilson to work with Prof. Jon Oberlander at the University of Edinburgh in Scotland and Prof. Alan Black at Carnegie Mellon University in Pittsburgh, Pennsylvania.

Metalanguage is a crucial linguistic mechanism which allows us to communicate information about language itself. It is essential for many language activities, including attributing statements, explaining meaning, assigning names, clarifying assumptions, and correcting misunderstandings. The ability to understand metalanguage is a skill that humans employ frequently in conversation and assume in fellow interlocutors, regardless of context or topic. However, in spite of its centrality to linguistic communication, metalanguage has received little attention in the development of interactive language technologies, leaving such technologies unable to cope with or exploit occurrences of the phenomenon in human-computer dialog.

This research is the first effort to study the use of computational methods to identify occurrences of metalanguage in informal English contexts. This project consists of three stages. The first is the creation of corpora of instances of metalanguage in informal contexts, particularly from spoken conversations and informal blog texts. The second is the detection and delineation of the phenomenon, using features of metalanguage that are peculiar to the phenomenon. Finally, the third stage is the practical evaluation of metalanguage detection and processing within the context of a spoken dialog system deployed for use by the general public.

The project will fill a long-standing and consequential research gap in artificial intelligence and computational linguistics. The practical applications of spoken dialog systems - such as voice menus, car GPS voice interfaces, and accessibility software for the blind - are expanding rapidly, as improvements are made to speech recognition, natural language processing, and other technologies that support conversational interaction with computers. Studying metalanguage will remedy a missing element in the design these of dialog systems, bringing them further in-line with users' expectations and thus making them more natural to use. Moreover, this research will lead to new international collaborations between the principal investigator and the two hosts' research groups, the Institute for Language, Cognition, and Computation (ILCC) at the University of Edinburgh and the Language Technologies Institute (LTI) at Carnegie Mellon University.