Fifth and Sixth Graders Speak More Richly than They Write: Benefits of Speech Input for Digitally Assessing Conceptual Understanding

Abstract Assessing conceptual understanding in mathematics education requires students to explain meanings and to make connections between concept elements explicit. However, many students struggle with written answers to open-ended questions in assessments, but are able to explain their thought processes orally. The growing availability of devices capable of automatic speech recognition (ASR) and the improved performance of ASR models calls for research on integrating speech input into digital formative assessments. This study explores how speech-input answers ( n = 354) from 53 Grades 5 and 6 students differed from typed answers ( n = 595) in a between-group comparison and analyses in detail which individual and item factors shaped the features of the speech-input answers. Results from multi-level models indicated that speech input answers were substantially longer, belonged more often to an everyday language register, but were richer, including more arguments and explanative discourse practices so that shallow understanding became more visible. Especially for low-achieving students, speech input seemed to be beneficial, independent from the chosen input format, for both dictation tools and voice messages.