S1 Data -

The aim of the study was to find whether certain meaningful moments in the learning process are noticeable through features of voice and how acoustic voice analyses can be utilized in learning research. The material consisted of recordings of nine university students as they were completing tasks concerning direct electric circuits as part of their course of teacher education in physics. Prosodic features of voice—fundamental frequency (F0), sound pressure level (SPL), acoustic voice quality measured by LTAS, and pausing—were investigated. The results showed that instances of confusion and understanding were manifested in acoustic parameters. F0 was significant in characterizing the both kind of learning instances. Confusion had lower SPL and alpha ratio, indicating that voice quality was softer than in understanding. Degree of voice pauses was lower in understanding, suggesting less hesitation or need for clarification for understanding compared to confusion. Voice research adds to the research of learning as speaker´s voice is affected by the different instances in the process of learning. This research approach can be used for identification of important instances of learning and directing these instances to closer analysis of content or interaction to further understand the learning processes. Therefore, this study is a novel contribution to the study of learning as it adds acoustic voice and speech analyses to the discipline.