Colloquium co-organized with Prof. Stanislas Dehaene, Prof. Stéphane Mallat and Prof. Luigi Rizzi.
This colloquium focused on the convergence and divergence between computational, neuroscientific and formal approaches to language. The last decade has seen major advances in computational approaches to language, leading to practical applications. At the same time, cognitive neuroscience has made significant progress in understanding the mechanisms of language processing in the brain, and formal linguistics has continued to make steady progress in the structural description of language. The three approaches have followed largely independent paths, which is hardly surprising, given the substantial differences in their methodologies and aims. Nevertheless, one might have expected a higher level of integration, since they all deal with the same object, natural language. This has only happened to a limited extent. Specifically, while there are several examples of the integration of neuroscientific and linguistic approaches, computational approaches have largely been developed in isolation from the other two perspectives. Is this an inevitable consequence of differences in objectives and methods, or a temporary situation linked to the history of these fields? Is it possible to envisage, in the years to come, greater convergence towards a "standard model" of language and languages? We discussed these questions with a select group of linguists, psychologists, brain scientists and computer specialists.