Avaliação em larga escala da alfabetização: um estudo sobre a habilidade de ler frases da matriz de referência de língua portuguesa dos anos iniciais do ensino fundamental

Abstract

This work, developed within the scope of Professional Graduate Program in Education Management and Evaluation (PPGP) of Public Policy and Education Evaluation Center at the Federal University of Juiz de Fora (CAEd/UFJF), presented a reflection based on the possibility of (re)defining processes for the elaboration and revision of items that make up the large-scale educational assessment tests prepared by CAEd. The research investigated the items that assess the ability to read sentences, provided for in the Portuguese Language Reference Matrix – Literacy – CAEd, supported by a theoretical framework of Linguistics, Statistics and Evaluation. Theoretical discussions in the field of Linguistics supported the analysis of items on the ability to read sentences, those in Statistics helped in the definition of methodological criteria for selecting valid items, and those on the large-scale assessment of literacy contextualized the ability to read sentences as an important milestone in the development of reader training. Methodologically, this study carried out an incursion into the set of items supported by statistical data organized in a scale in which children's proficiency is calculated by the mathematical model of Item Response Theory (IRT) and pedagogical analysis. Thus, it sought to understand elements of complexity that explain the distribution of a set of items in the ability to read sentences at different points on the proficiency scale. Through this methodological route, the work sought to answer the guiding question of this study: what are the reasons that lead the items of the ability to read sentences to be anchored at different points of the scale? Therefore, the general objective was defined to understand how the accumulated production of selected response items, which assess the ability to read sentences, can provide information about the path implemented by CAEd in working with this ability in large-scale assessments focused on literacy. In the incursion to the data, we concluded that there are different complexities in the way the items are elaborated, generating different information about the development of the ability to read sentences, considered, in this dissertation, as important for the formation of the beginning reader. Based on the analysis of the items, it was possible to propose an action plan that aims to contribute to the CAEd Itens Bank, both in the process of elaboration and in the revision of items due to the detail presented and the proposal to contribute with a specific document to the work processes in order to guarantee the elaboration of increasingly accurate tests.