摘要：Over the last thirty years the long range goal of the research project of the Speech Statistics (Sp.St.) international group is to develop and implement in computer simulation psycholinguistically realistic model of language behavior. This model, named the linguistic automaton /LA/, is a complex of hardware, software lingware /linguistic means of support/ and CALL-ware. The Sp.St. group was developing many programs to model different aspects of natural language /NL/ processing by man. The multilevel and multilingual LA-architecture integrates in a flexible way these programs and provides a control strategy capable of adapting itself to the requirements of a particular task. It permits a modular implementation of LA, making the modification of existing processes and the integration of new capacities in the LA-system easier.
