Levels and descriptors of textual complexity for adults with low literacy: a benchmark from the iRead4Skills project

Abstract

This paper presents the iRead4Skills benchmark of textual complexity, a novel approach to text complexity difficulty tailored to low-literacy adult native speakers. Despite the existence of various instruments and guidelines related to reading proficiency (CEFR, PISA, etc.), little attention is given to adult reading difficulties and how these manifest linguistically in texts. Recognizing the pivotal role of literacy in personal and professional development, this tool addresses the challenges of identifying and adapting texts for this demographic. Based on existent benchmarks (ANQEP, CEFR, PIAAC, ALTE), this article builds upon them and proposes the delineation of levels of textual complexity for European Portuguese, pertaining to low-literacy adults. Integrating lexical, syntactic, and cohesion features, the framework defines three levels of textual complexity—Very Easy, Easy, and Plain— roughly aligned with the CEFR levels A1, A2, and B1. Descriptors of complexity are also advanced to detail each level. Alongside this proposal, the levels and descriptors were tested against real texts, resulting in the compilation of a corpus, graded by levels of complexity. With the aim of testing and validating this proposal, both the levels and descriptors were evaluated and discussed by experts in Adult Learning and Vocational Education and Training. The iRead4Skills benchmark of textual complexity aims at bridging theoretical research and practical application by combining quantitative and qualitative measures, ensuring a robust and contextually relevant analysis of textual complexity. Key contributions include the framework’s alignment with authentic texts and its validation by experts in Adult Learning and Vocational Education and Training. Practical applications of this framework encompass innovative tools such as a text complexity evaluation system and a writing assistant for text adaptation, addressing gaps in suitable reading materials for low-literacy adults. This research offers valuable insights for cross-linguistic studies on text complexity, paving the way for future exploration in literacy and text complexity frameworks.