Finding rules and words in a speech stream
Dr. Luca L. Bonatti
Tuesday, November 07, 2006, 01:00pm - 02:00pm
Laboratory of Cognitive Development, SISSA/ISAS, Trieste, Italy and University of Paris VIII at St. Denis, France
How do infants and adults extract words and rules from a continuous speech stream? Recent work in artificial grammar has revealed the existence of powerful statistical mechanisms that, it has been argued, may be used to solve both these problems. I will show that statistical computations in language learning are in fact limited. I will provide evidence suggesting that different mechanisms are responsible for segmenting a continuum and identifying the structural properties of the segmented items. One of them is based on extensive statistical sampling, while the other is not. I will show that language constrains in several ways the representations open to statistical inspection.
Note: If you would like to receive email announcements about the colloquium series, please contact the Business Office to have your name added to our announce lists at email@example.com.