View article

By modeling difficult sources of linguistic variability in speech and language, we can design interfaces that transparently guide human input to match system processing capabilities. Such work will yield more user centered and robust interfaces for next generation spoken language and multimodal systems.