Human languages are thoroughly contextual in nature. This fact can be seen plainly by considering how the meaning of a sound, word, sentence, or passage can change drastically when taken out of context, or placed in a different context. The contextual nature of language has proven to be a major challenge for developing computer algorithms that can understand or produce language, yet humans process language in context so easily that we are usually not even aware of it. Linguistic and cognitive scientists have been studying how humans process context for decades, often with respect to individual words in the context of other words. For instance, it is well established that a person's ability to recognize a word like "dog" is enhanced when it appears shortly after a word like "cat" that is related in meaning. Results like these are referred to as "semantic priming", and our knowledge of the mechanisms that underlie semantic priming and other kinds of contextual effects is mostly based on research using small handfuls of words chosen to make very targeted theoretical points.
With support of the National Science Foundation, Drs. Hutchison and Balota will take a complementary approach to studying the mechanisms of semantic priming. Rather than restricting their scope to small handfuls of words, the investigators will develop a large database of semantic priming results for over 3,000 words of English. The database will include a range of statistics about the words and their usages, and the data will be made available to the research community via a powerful web interface. The database will help researchers throughout the world to advance theories and computational models of the processes that allow humans to not only overcome the challenge of context, but actually use context to their advantage in the comprehension and production of language.
