Photo: Aku-Matti Valkama
Heini Kallio tells us about her research on the phonetics of language learning, speech intelligibility and the automatic assessment of oral skills in a foreign language.
I am a university lecturer in phonetics at Tampere University, where I have worked since 2023. I defended my doctoral thesis in 2022 at the University of Helsinki, where I was part of the Phonetics and Speech Synthesis research group. I have been a visiting researcher at the Slovak Academy of Sciences and the University of Nitra, and I have contributed to several research projects related to language learning and language proficiency assessment. My academic background includes a major in phonetics as well as a broad range of studies in communication, psychology, languages, and speech sciences.
An endless source of inspiration in my research is the difficulty and ease of speaking in second and foreign languages. I explore how the acoustically measurable features of speech relate to people’s evaluations of a language learner’s proficiency or, for example, their experience of the speaker’s intelligibility. In the DigiTala project, I examined the Swedish spoken by Finnish high school students and the Finnish spoken by immigrants, with the aim of identifying acoustic parameters that could be used for the automatic assessment of oral language skills. In addition to the DigiTala data, my dissertation material also included English spoken by Czech, Slovak, Polish, and Hungarian learners, which I continued to study after completing my dissertation. Since then, I have also had the opportunity to investigate the prosodic features of Finnish learners of French.
In 2024–2025, I led my first research project, funded by Svenska Kulturfonden, on the pronunciation of Finland-Swedish among beginner learners from different language backgrounds. In the project, we studied the Swedish spoken by immigrants in Finland. We analyzed, among other things, the impact of language background on the acoustic features of Swedish speech and on listeners’ perceptions of beginner learners’ speech.
Before my current position, I worked as a researcher in the DigiTala project funded by the Research Council of Finland. In DigiTala, we developed the first tool for automatically assessing different areas of oral language proficiency in Finnish and Finland-Swedish. The end result of the project was the DigiTala Moodle plugin. This ambitious consortium project brought together experts in language didactics, language and speech technology, and phonetics, many of whom are continuing the work initiated in DigiTala by, for example, developing a mobile application for learners of Finnish (DigiTala in action) and researching the automatic assessment of oral interaction in the AASIS project.
Automatic assessment is based on speech recognition and machine learning models, which perform better the more extensive and relevant the training data is. Therefore, in DigiTala we designed and collected extensive corpora for research and development, including oral language proficiency tasks for learners of Finnish and Swedish, assessment criteria for different areas of oral language proficiency, and comprehensive sets of speech and assessment data. The speech and assessment corpora from DigiTala are available via Kielipankki for scientific research purposes. The oral tasks, assessment criteria, and questionnaires used in the research project can be found on the DigiTala Zenodo community. The speech data collected in DigiTala from learners of Finland-Swedish was also examined in the project “Pronunciation of Finland-Swedish in beginner learners with different native languages”.
Heinonen, H. & Kallio, H. (In press). Realiseringen av begriplighetsrelaterade uttalsdrag i finlandssvenska hos L2 talare från olika språkbakgrund. Svenskan i Finland 21.
Kallio, H. & Kaźmierski, K. (2024). Reduction of unstressed English vowels by EFL speakers with different language backgrounds. In International Symposium on Applied Phonetics (pp. 38 42). International Speech Communication Association (ISCA). doi: 10.21437/ISAPh.2024-8
Kallio, H. (2024). The contribution of speech timing, f0 change, and voice quality to perceived prosodic proficiency in L2: a cross lingual perspective. In Speech prosody (pp. 507 511). International Speech Communication Association (ISCA). doi: 10.21437/SpeechProsody.2024-103
von Zansen, A. & Kallio, H. (2024). DigiTala – Moodle sovellus suullisen kielitaidon automaattiseen arviointiin. (DigiTala – An online tool for automated assessment of oral language skills.) AFinLA teema, 17, pp. 91 116. doi: 10.30660/afinla.131465
Kallio, H., Kautonen.M. & Kuronen, M. (2023). Prosody and fluency of Finland Swedish as a second language: investigating global parameters for automated speaking assessment. Speech Communication, Volume 148, pp. 66 80, 2023. doi: 10.1016/j.specom.2023.02.003
Additional note: Unfortunately, the automatic assessment algorithms developed in the DigiTala project are not in general use due to the universities’ limited server capacity. In other words, the Moodle plugin can currently be used mainly to support oral practice and teacher-led assessment.
The FIN-CLARIN consortium consists of a group of Finnish universities along with CSC – IT Center for Science and the Institute for the Languages of Finland (Kotus). FIN-CLARIN helps the researchers of Social Sciences and Humanities to use, refine, preserve and share their language resources. The Language Bank of Finland is the collection of services that provides the language materials and tools for the research community.
All previously published Language Bank researcher interviews are stored in the Researcher of the Month archive. This article is also published on the website of the Faculty of Humanities of the University of Helsinki.
