Complexidade semântica e habilidade de decodificação: um modelo quantitativo da compreensão de textos denotativos em língua portuguesa baseado na teoria da informação

Abstract

Based on the principles of cognitive neuroscience and information theory, with emphasis on the work of Claude Shannon, a statistical analysis of 33,101 words was done from the collection of scientific texts of the Brazilian Digital Library Of Thesis And Dissertation and the Digital Library of USP, using the C # programming language and Microsoft Visual Studio 2012 as a code complement, including SQL Server Management Studio 2012 for database management, for the development of the information processing program titled CalcuLetra, with the purpose of measuring the difficulty of textual comprehension in Portuguese Language. From the premises that the learning of the meanings of letters, words and other symbols provides the development of the central nervous system of humans; that the reader's metacognitive behavior allows the response to stimuli coming from the reading process; And that the words of greatest occurrence in the bank represent those best known by their authors, the algorithm thus determines the degree of familiarity of the words according to the mathematical and statistical parameters of the bank. Therefore, when comparing non-literary or denotative texts with the probabilistic values found, it shows how comprehensible is the text inserted in the program, considering neurotypical readers and that its content has the necessary elements of textual cohesion, according to the grammatical rules of the language. Our results reveal groups of words that cause misunderstanding or make reading easier. Additionally, we show gaps in vocabulary and dictionary usage. Despite the preliminary findings, this study was more a proof of concept for the method employed, and demonstrated its potential for future research. The quantification model’s methodology presented in the present study can be adapted to other languages, and we hope it contributes to the development of objective evaluation of behavioral disorders (e.g., dyslexia), by quantitative classification of written incomprehension; and helps in the analysis of dissertations of vestibular, ENEM and other public examinations, whose evaluation is still of a subjective form.