4. Language and Complexity: Neurolinguistic Perspectives

In the previous chapter we dealt with the problem of ambiguity and how simulation of input-driven, psycholinguistic processes enables a semantico-syntactic translation model to deal effectively with ambiguity. In the present chapter we deal with the issue of complexity, focusing in particular on the constraining effect that cognitive complexity has on MT development as it attempts to cope with the ambiguities of langauge. We define cognitive complexity as the difficulty developers experience in maintaining complex systems. We show how the associative nature of Logos Model’s neural-like translation paradigm allows it to deal more effectively with cognitive complexity than is possible with rule-based technology, or indeed any other MT paradigm. We attribute the reasons for this to Logos Model’s serendipitous correspondence to findings of neuroscience on the brain’s processing of language, citing the brain’s evident freedom from complexity in processing language as a motivation for this direction in Logos Model design. We focus on two regions of the brain that are involved with language: (1) the prefrontal temporal cortex designated as the Broca area, commonly connected with rule-based processes, and (2) the hippocampus, a well-defined reticulum in the medial temporal lobe distinguished for its declarative, associative memory processes, and whose connection with language processes has only recently been proposed by neuroscientists. We provide illustrations and examples of how the associative processes of the hippocampus have been simulated in Logos Model, and how Logos Model has benefited from this simulation.