In his dissertation Heini Kallio seeks connections between acoustic properties of speech and human language proficiency assessments. The results can be applied to the development of oral language teaching and their automated assessment.
In 2017, the Executive Committee established by the Ministry of Education and Culture made a proposal to add an oral section to the language tests for the Finnish matriculation examination.
“Speech is the most common form of communication between people, but both language teaching and testing focus on writing, also in Finland,” says Heini Kallio. “In addition, the phonetic perspective has remained a margin in the assessment of Finnish language skills.”
In addition to the prerequisites for the matriculation examination, the practical challenges of speech evaluation are also the resources required for testing. DigiTala project Funded by the Academy of Finland) is researching the possibilities of automated speech assessment and developing tools that can be used, for example, in future matriculation examinations.
Initially, Kallio studied Swedish spoken by Finnish high school students together with the DigiTala project. Later, the material expanded to include English spoken by people whose mother tongue was Czech, Slovak, Hungarian, or Polish. The material that transcends language boundaries makes it possible to study the predictive features of universal language proficiency in speech.
– The dissertation showed that, for example, the frequency of articulation is a language-independent property that can be used to measure the speaker’s oral level, Kallio says.
In addition, Kallio analyzes stress production in relation to the characteristics of a language learner’s mother tongue. Without practicing the prosody of a foreign language, language learners can transfer the emphasis patterns of their mother tongue to the language being learned. This leads to an atypical speech rhythm, which makes it difficult to understand.
Wavelet transforms provide new insights into speech features
However, converting acoustic stress realizations to numerical parameters has been more difficult for researchers than, for example, measuring articulation velocity. Kallio measured acoustic stress implementations with a signal processing technique based on wavelet transforms that had not previously been applied to the speech of language students. The results of the analysis of the read speech were promising: the acoustic stress parameter made it possible to distinguish between beginners and intermediate speakers.
The wavelet conversion technique also revealed differences between groups depending on the speaker’s native language. In fact, the impact of mother tongue on the production of stress in foreign languages should be studied more comprehensively. A broader understanding of stress patterns – and thus of speech rhythm – will help to develop assessment methods and tools in addition to language teaching.
About the dissertation
Heini Kallio, M.Sc., defends her dissertation on phonetics “The Prosody Behind Spoken Language Proficiency: A Multilingual Study of Fluency and Syllable Emphasis in Other Languages 9.2.2022 at 14.00 in the Faculty of Humanities of the University of Helsinki.
The audience can be followed online at https://helsinki.zoom.us/j/63205819965. The opponent is Professor Emeritus Dafydd Gibbon from the University of Bielefeld and the custos is Professor Martti Vainio.
Contact information for the doctoral candidate
[email protected], +358 46 9234 917
Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of the bulletins sent to EurekAlert! through the participating institutions or through the EurekAlert system.