Oral language skills are an essential part of the communicative conception of language skills based on the Finnish National Core Curriculum for General Upper Secondary Schools. However, a section to measure speech is missing at language tests in the Finnish Matriculation Examination because of practical challenges and expenses. Automatical speech recognition would make automatic assessment of students' oral performances possible cost-effectively and reliably. Automatic assessment will be used together with human ratings and it will enhance and standardize language education.
The main goal for DigiTala research project is to develop a digital tool based on automatic speech recognition for assessing oral language skills. Target languages are Swedish and Finnish as a second national language. When the tool has learned a sufficient qualitative model for assessment from humans, it will adapt for independent training for students from different language backgrounds regardless of time and place.
The meaning of the digital tool developed in the project is to make the assessment of oral language skills possible in big and important language tests. Furthermore, students can train their pronunciation and speech production in foreign languages independently outside the school or without the teacher’s guidance at language classes. Read more:
The research project is financed by Academy of Finland in 2019–2023 and combines expertise in speech and language processing, language education, and phonetics at the University of Helsinki (grant number 322619), Aalto University (grant number 322625), and the University of Jyväskylä (grant number 322965). Here you can visit the websites of University of Helsinki, Aalto University and University of Jyväskylä. In the beginning, the target languages in the research project are Swedish and Finnish as a second national language. We work in co-operation with the Finnish Matriculation Examination Board and Finnish National Agency of Education.
At the first stage of DigiTala project (2015–2017) a demo version of automatic assessment of Swedish pronunciation by using speech recognition was developed. The project was financed by the Svenska folkskolans vänner foundation.