Structure mapping and semantic integration in a construction-based neurolinguistic model of sentence processing.

摘要

The current research provides a theoretical, computational and neurophysiological framework in which particular aspects of sentence comprehension and non-linguistic sequence transformation processing are implemented by a common neural mechanism for structure mapping. The theoretical context is derived from construction grammar theory in which language is considered in terms of a structured inventory of form to meaning mappings. Computationally, the construction grammar concept is implemented in a hybrid neural network model that is derived from functional neuroanatomical studies. In particular, based on data from Hoen et al. (2006, this issue), the generalized structure mapping capability is attributed to a local cortical network that includes Brodmann's area (BA) 44, while the integration of semantic structure into this transformation mechanism relies on BA 45.