AI-based Assessment of Reading Proficiency
Reading is a crucial key skill to open up children's worlds of experience and establish the foundations for their entire educational journey. Reading is an interdisciplinary competence and a central medium for learning.
Being able to read securely and competently allows children to participate in society and plays an essential role in every lesson. The diagnosis of children's reading proficiency in primary school is a crucial prerequisite for individual support.
"Reading proficiency" consists of various subcomponents, all of which are significant for the overall development of this skill. Both high-level skills, such as reading comprehension, and low-level skills, such as reading fluency, are relevant to the development of reading proficiency. The latter includes reading speed, decoding ability, reading accuracy, and the prosody of the text.
Within the framework of the TALC project, software is being researched that can analyze and classify the reading proficiency of primary school children using machine learning.
Currently, capturing all the skills necessary for reading proficiency is only possible through the use of multiple different diagnostic methods. However, TALC-DIRA aims to enable the assessment of all reading sub-skills in one tool through the integration of text- and audio-based parameters.
By automating reading diagnostics through AI, teachers are relieved as children can independently carry out the procedure, and the software autonomously evaluates the results. The differentiated evaluation based on reading skills allows for individual lesson planning and support for children.
(Kopie 1)
What is TALC-DIRA about?
TALC-DIRA is intended to be available as a tablet-based app in the future. It aims to assess a child's reading skills, both aloud and silent reading, using various data sources such as audio, text, and possibly video. The tool is designed to be suitable for use in multilingual contexts and will be normed as a screening tool for children with reading disorders within the scope of a specific reading and writing disorder (LRS).
Project Status
Drawing on the experiences gained from automatic speech recognition of child spontaneous speech in the second TALC subproject (TALC-LSA), the TALC-DIRA software is currently in the development phase. A complementary team from the disciplines of special education, German didactics, speech therapy, and computer/electrical engineering is collaboratively developing and researching the necessary systems.
The piloting of an initial version is scheduled for the academic year 2023/24.
Publications
- Stelze, K.-M., Wallbaum, M., Ehlert, H. & Lüdtke, U. (2023). Schulische Anforderungen an den Einsatz künstlicher Intelligenz in der Lesediagnostik - eine Fragebogenerhebung mit Lehrkräften im Primarbereich. Jahrestagung der DGfE-Kommission Grundschulforschung und Pädagogik der Primarstufe.