报告题目：Is the way we talk similar to the way we think: A comparison of computational linguistics data and neurophysiology data.
报告人：Sabine Ploux Ph. D. (CNRS-Lyon University)
Sabine Ploux Ph. D., Research Fellow, Chargée de recherche première classe CNRS at L2C2 CNRS-University of Lyon 1.
Head of the Team "Computational and Mathematical Models for Language
Creator of the Semantic Atlases Project.
The neuroscience literature has devoted many studies to the semantic processing of words. However, the research remains specific to certain categories, studied separately, and does not address the lexicon as a system. I will present a method of automatic language processing aimed at studying concepts and their processing, not in a category-by-category fashion, but as a system taken as a whole. The notion of system is a key feature in linguistics in order to describe the semantic of words. The method is based on correspondence analysis (CA), a data-analysis method commonly used in computational linguistics. Here, it was applied to the electrophysiological signals. This method permits: (a) to find the neural spatio-temporal networks that participated to the processing of the stimuli-words and that characterize their semantic similarities and differences; (b) to give a big picture of the structural organization of this processing; (c) to compare the organization revealed by the two processes: word-usage and neurophysiological word-processing. The method is general and could be applied in other neurophysiological experiments and with other types of data (fMRI, MEG,..).