Doctoral Meeting: 'Improving Reading Skills: From Corpora to Automatic Classifiers in Galician and Spanish'

In this talk, we will discuss how to automatically measure and predict the reading difficulty of Galician and Spanish texts. To this end, new readability corpora have been created specifically for adult readers, bringing together real texts from a wide variety of genres (messages, news, legal texts, fiction, etc.), which have been carefully classified by level of difficulty. Using this data, linguistic features that make a text easier or more complex (e.g., length, vocabulary type, and syntactic structures) were analyzed, machine learning and deep learning models were trained to predict readability levels, and an automatic annotator was developed to identify complex linguistic phenomena. This presentation will cover the created resources, model results, and a classification and annotation demo, as well as their potential applications in language teaching and designing materials for different reading proficiency levels.

• Supervisor: Marcos Garcia González
• Moderator: Álvaro López Paredes