Classificação de linguagem simples: uma abordagem baseada em leiturabilidade e legibilidade

Abstract

The efforts of government agencies to adopt Plain Language (PL) are undeniable. However, there is still a clear need to extend these efforts to other spheres since PL is fundamental to enabling a clear understanding of information and promoting the inclusion of citizens who are not sufficiently literate. In the literature, there are several studies dedicated to measuring PL and textual complexity, some of which use automatic methods using machine learning, which lack explanations of how it is possible to improve the text; others use semi-automatic methods based on human evaluation, which impairs its application on large data sets. We investigated the use of analytical methods for PL classification in the light of Readability and Legibility. The results show that using Readability and Legibility measures to classify Simple Language is possible. The work contributes to the state of the art by studying six textual complexity metrics for classifying Plain Language. For the state of the practice, the work contributes with inputs for developing and deploying Plain Language classification systems, indicating aspects of improvement for the end user.