Biological And Psycholinguistic Influences On Architectures For Natural Language Processing

Systems for natural language processing (NLP) are based on some linguistic theory adapted to available computational technology. Since the earliest NLP systems, a variety of linguistic, philosophical, logical, and computational issues have created controversies about the design choices. Today, advances in technology have created new options, and insights from neuroscience can help guide the choices. Section 1 summarizes the controversies and the shifting design choices caused by different theories and technologies. Section 2 presents neural evidence about connections among language areas of the brain and their implications on the design. Section 3 summarizes psycholinguistic considerations on the design and implementation of conceptual graphs. Section 4 describes how the VivoMind Language Processor (VLP) implements these design options. In recent tests, they enabled VLP to get significantly better results than more traditional designs.