Project leader: Mikko Kurimo, Aalto University
Project duration: 2021-2026
Participating countries: Finland, Sweden and Norway
Funding from NordForsk: 14,850,007 NOK
Project website: https://teflon.aalto.fi/
The Nordic countries accept many immigrants – not least because of the war in Ukraine. This means that there is considerable demand for learning one of the Nordic languages as a second language. Learning the language, and especially oral language skills, is crucial for the immigrants to become part of the Nordic societies. The TEFLON-project focused on how to create digital games to teach immigrant children to speak Finnish, Norwegian and Swedish.
Impact story
TEFLON has developed games because they increase children’s motivation to learn languages. Children enjoy playing, and digital language-learning games can be particularly attractive to immigrant children. The games can supplement classroom-based language teaching, because they can be used individually, and they can be used regardless of where you’re located or the time zone you are in.
They have developed a speech technology component that can be added into various fun games instead of building completely new games for language learning. Also, native children who have various speech sound disorders can use the games and learn to speak fluently and correctly.
They created new fruitful collaboration between experts in speech recognition, AI, neuroscience, speech therapy, and education from Finland, Sweden and Norway.
To build impact and trust to the stakeholders, they engaged directly with:
Cities and school directors
Teachers, who generously adjusted their schedules and trusted them with their classrooms.Bridged the TEFLON researchers with parents.
Parents, who signed forms, asked thoughtful questions, and shared concerns about screen time and learning. Spent their time and efforts to let children play.
Children, who laughed, played, struggled, and learnt. Whose openness made all the difference.
Key Findings
- We collected and annotated new children’s speech data and used them to develop five versions of TEFLON mobile game applications called Pop2Talk_Nordic that were implemented, tested and used for learning experiments: SwedishSSD, Finnish-R, SwedishL2, NorwegianL2 and EnglishL2.
- Several articles were published about the data collection, mobile game application and our new methods for automatic pronunciation rating.
- Human learning experiments were also performed on the effectiveness of the game on immigrant children’s language learning, comparing children’s reading scores in their native language before and after gaming intervention and evaluation of the speech training game for children with speech sound disorders.
Key outputs
Yaroslav Getman, Nhan Phan, Ragheb Al-Ghezi, Ekaterina Voskoboinik, Mittul Singh, Tamás Grósz, Mikko Kurimo, Giampiero Salvi, Torbjørn Svendsen, Sofia Strömbergsson, Anna Smolander and Sari Ylinen. Developing an AI-assisted Low-resource Spoken Language Learning App for Children. IEEE Access, vol. 11, pp. 86025-86037, 2023.
Strömbergsson, S., Fröjdh, M., Pettersson, M., Grósz, T., Getman, Y., & Kurimo, M. Listening like a speech-training app: Expert and non-expert listeners’ goodness ratings of children’s speech. Clinical Linguistics & Phonetics, 39(2), 144-165, 2025.
Anne Marte Olstad, Anna Smolander, Sofia Strömbergsson, Sari Ylinen, Minna Lehtonen, Mikko Kurimo, Yaroslav Getman, Tamás Grósz, Xinwei Cao, Torbjörn Svendsen and Giampiero Salvi. Collecting Linguistic Resources for Assessing Children’s Pronunciation of Nordic Languages. Proc. Language resources and evaluation conference (LREC-COLING) 2024, May 2024.
Yaroslav Getman, Tamás Grósz, Mikko Kurimo. NON-NATIVE CHILDREN’S AUTOMATIC SPEECH ASSESSMENT CHALLENGE (NOCASA). Proc. IEEE Workshop on Machine learning for signal processing (MLSP) 2025, September 2025.