Text readability analysis with Natural Language Processing Tools: assessment of the “Literatura para Todos” Collection

Abstract

This paper presents results of a research on readability of the Literatura para Todos 1 Collection, published by MEC/SECAD (2006) and distributed to newly literate youth and adults. The investigation of text readability combined assumptions from Psycholinguistics and natural language processing tools. We used readability criteria referred in psycholinguistics literature and tried to evaluate objectively the degree of linguistic complexity of the books by means of computational tools: the morphosyntactic analyzer PALAVRAS and the program Coh-Metrix Port. Our results suggest that these books are complex for the intended audience. Assuming that new readers are in the initial stage of literacy process, it is clear that reading the books of the Collection takes an effort which is beyond their ability.