A fuzzy model of natural language acquisition and syntax recognition by humans

We describe a model of inductive learning of concepts under imprecision by a typoid automaton, which repeats the performance of an adaptive recognizer. We generate a fuzzy model of how children learn step by step from their relatives and teachers to separate - with growing precision - the syntactic elements of a sentence discriminating their function and understanding their semantics. We model the learning process with a fuzzy learning controller. The correct sequence or sentences are represented as matrices of fuzzy sequence in Markov processes. Hence, learning of a context-oriented grammar (instead of just terminology and pronunciation) by a child is represented as the growth of a fuzzy classification of word types (typoids) by their grammatical function inside the sentence and of fuzzy sequentiality matrices of these words.