March 29, 2010
Predicting What They Say
New research findings in linguistics
An Australian-American team of investigators has made novel discoveries about the human ability to predict what other people are about to say. Their findings could have significant applications for educators, speech therapists, entrepreneurs, and many others interested in communication and comprehension.The study, "Predicting Syntax: Processing Dative Constructions in American and Australian Varieties of English," to be published in the March 2010 issue of the scholarly journal Language, is authored by Joan Bresnan and Marilyn Ford. A preprint version is available on line at: http://www.lsadc.org/info/documents/2010/language/0168-0213.pdf
Everyone is familiar with the practice of completing someone else's sentence"”essentially predicting what the other person is about to say. To a remarkable degree, people are quite accurate in their ability to make these predictions, not only in terms of the basic content of the message, but also in terms of the word choices and phrasing of the sentences. This ability to effectively predict the syntax of others in context comes from our knowledge of "linguistic probability." The human capacity for determining this probability is based on our day-to-day experience of the language. The greater the amount of experience that individuals have of a language, the greater their ability to predict. This is true of different dialects within a language. For example, Australian speakers of English and American speakers of English detect slightly different patterns of phrasing and usage among their respective fellow speakers, thus enabling them to more effectively predict the syntax that will be used in a variety of contexts.
This intrinsic ability to predict based on probability has implications for language comprehension. Educators engaged in foreign language instruction might effectively focus their initial efforts on the most probable sentence constructions. Entrepreneurs engaged in marketing their products or services might use the most probable phrases in preparing their advertising messages. These research findings on linguistic probability may also be helpful in making computerized language more natural. Another practical application would be in the refinement of tools used in profiling and diagnosing those with language disorders. As noted by the authors in an interview, "Linguistic patterns are important in predicting comprehension. If we can make better use of these patterns to enhance comprehension, then we can improve people's ability to understand one another."
On the Net: