Individual Regularities and Cognitive Automata

This article proposes a new formal model for the analysis of texts, particularly adapted to the study of verbalizations. The model has been built to characterize individual regularities between texts written by the same author. Its purpose is to provide a precise definition of the notions of linguistic and cognitive styles. It includes two levels : at the enunciative level, lexical and syntactical regularities in the expressions of an identical intention are represented by « enunciative patterns » ; at the discursive level, the global organization of discourses is taken into account and common points between various verbalizations of a same task are captured into « cognitive patterns ». The regularities extracted are represented in a symbolic way and the structure of « Cognitive automata » is defined to sum-up all information collected from both analyses. The definitions are illustrated by examples from a real corpus. This model is a formal approach of what language tells us about the one that tells it.