Background

Early childhood language development is of central importance for children’s academic success and social participation. The assessment of language skills as part of the school entry examination is therefore of great importance. In Bavaria, the school entry examination is carried out by employees of the local health authority in the year before the child enters school. In order to record the language level in a standardized way throughout Bavaria, the Bavarian Ministry of Health and the Department of Speech and Language Therapy at the Ludwig-Maximilians-Universität in Munich developed a screening instrument called BESS for the assessment of the language level prior to school enrolment with the involvement of the Bavarian health authorities. BESS consists of 4 tasks to test grammar and phonological awareness. The child’s general language comprehension, articulation, voice and fluency are also assessed.

Goal

The aim of this research project is to support the BESS with methods of automatic speech and language processing. The integration of AI can help to enable objective diagnoses of language development disorders in children. The evaluation takes place using audio recordings taking into account the respective test. The main requirement for this is a reliable speech recognition system that can process the variable speech of children.