Séminaire de Recherche en Linguistique

Language processing is subject to complexity variation. Many works have shown that certain phenomena are source of complexification (e.g. the depth of the syntactic structure, the amount of dependencies between two elements, the discourse structure, etc.). More recently, other works have shown that there also exists elements of facilitation (e.g. the frequency of the cooccurrence between words). Several experiments in neurolinguistics lead to the same conclusion, showing at the same time that in the case of facilitation, a specific processing seems to be at play. 

We will present in this talk a model of language processing integrating these two mechanisms. It relies on the hypothesis that facilitation phenomena are easily detected, where difficulties requires more complex mechanisms. The model proposes a two-level organization, gathering shallow and deep parsing for implementing the two kinds of processing.