English for space animals – research team develop a machine-learning-based language game for children
‘So many colours, and colours make me happy!’
‘It was like so good, I got so many stars! Next time I need to get 20 stars!’
‘It was fun and I enjoyed it, because you need to say things and repeat and it’s funny.’
If you ask for a child's’ opinion, the most fun way to learn a new language is by playing games, and especially ones where you get to speak. Researchers from the University of Helsinki and Aalto University agree: the team has developed a new kind of pronunciation game for children based on speech recognition technology.
‘We have tested our game with hundreds of children, and we can see how much they enjoy talking.’ says Maria Sipilä. ‘Brain studies carried out at the University of Helsinki have also shown that children learn best through playing games.’
Stars motivating learners to improve
Behind the game’s development is many years of research. When the brain researchers and pedagogical professionals from the University of Helsinki combined their strengths with the speech recognition technology experts from Aalto University, Pop2Talk was born.
In the game, the children teach English to animals that have arrived on earth from space, and at the same time, they learn English themselves. The game uses machine learning through which the children are given stars depending on the quality of their pronunciation, and are therefore motivated to improve their performance. While the game is technologically ahead of its competitors, Maria would rather talk about motivation than technology.
‘It's truly wonderful how we can use technology to influence such a subtle factor as motivation. I believe that machine learning and speech recognition will bring us a lot of new learning tools in the future. However, more important than the technology itself is how it is used – and that’s where it’s essential to understand how learning happens.’
Societies lose billions of euros each year due to lack of language skills.
Language skills for better societies
The team behind the game believes that good language and communication skills can directly impact the economies of different societies and the number of children speaking English as a second language is constantly increasing due to worldwide migration. Deficient language skills affect school attendance for children and employment prospects for adults.
‘Societies lose billions of euros each year due to lack of language skills’, Sipilä adds.
According to Sipilä, there are approximately 1.5 billion English language learners worldwide, so the market is certainly large enough, and in the future, Pop2Talk could be used to learn other languages as well. The game is based on language-independent algorithms, which means that the child's native language can be anything.
From research to business
The project is currently run with funding from Business Finland’s New Business from Research Ideas (TUTLI) fund, and the goal is to set up a company next year to take the product forward. The team is aiming to export the game worldwide directly.
‘We are now preparing the investment cycle and looking at business models, pricing and scaling while building a partnership network through which the game could be spread to schools,’ Sipilä explains.
According to Sipilä, the essential part of commercialisation is always the end user.
‘What I have learnt from the past is that the end user's perspective must be preserved at every stage of product development. For us, the child is the most important critic, and understanding the child's world determines whether we succeed or not.’