Large-Scale Portuguese Language Evaluations: A Research on the Complexity of Texts that Support Items that Assess Reading

Abstract

This article discusses the theme of the complexity of texts that support large-scale assessment items in Portuguese language, based on the results of a survey in which 763 texts used in large-scale assessment items in Portuguese were analyzed. The research methodology consisted of item analysis supported by the theories of cognition, textual linguistics, and statistical parameters of the items based on the Item Response Theory, a three-parameter model. Based on the research carried out, four levels of complexity of the texts were defined. The results point to the relevance of clear criteria to define the complexity of texts in the development of new items and to a more precise description of performance standards based on the reading proficiency presented by the evaluated students.