In general, I am interested in how meaning can be represented in memory at the cognitive, neural, and computational levels. Consequently, my research, supported by an award from the National Science Foundation, explores the memory mechanisms that underlie language comprehension, the processing of semantic and syntactic ambiguity, as well as figurative language. From a neurolinguistic perspective, my work investigates the complementary roles of the two cerebral hemispheres in activating information in memory using both normal, computational and brain-damaged subject populations.
Much of my recent work has been concerned with how lexical, semantic, syntactic, and discourse information can be used by the language processor in order to produce a competent reader. Some of our current research investigates how various factors contribute to potential reading problems when ambiguity is encountered.
An ongoing area of research in the lab involves characterizing the contribution the cerebral hemispheres in language comprehension. With respect to lexical/semantic processing, we have gathered considerable evidence that the two hemispheres differ in the rate of meaning activation, but do not differ much at a representational level. Our goal is to fit these findings in the context of our basic psycholinguistic understandings of how we understand ongoing discourse. At the word recognition level, we have developed a computational implementation of cerebral asymmetries that makes use of our Hyperspace Analogue to Language model.
A major project has been the development of the Hyperspace Analogue to Language (or HAL) model which is a computer simulation of human memory. HAL has a lexicon of 70,000 items and uses a 300 million word corpus as input and a global co-occurrence learning mechanism to develop cognitively plausible high-dimensional meaning representations without the need for human judgements on stimuli. HAL exploit the context in which words and sentences appear by allowing meaning (semantic or grammatical) representations to naturally emerge as a product of the systems experience with its environment and offers an explicit account of how environmental input is transduced to representational information. Word meanings (for 70,000 items) are represented in 140,000 dimensional space (thus, Hyperspace Analogue to Language). The model accounts for a wide range of semantic, language, grammatical, and syntactic phenomena. I am interested in how the model can be used at a theoretical level, and also how to develop software tools based on human cognitive ability.
Look more at my lab's web pages for more information on my research, what my students are up to, photos and Quicktime movies of lab facilities, and a HAL demo.
Back to my lab homepage ...