The framework and development of SERDA: Speech enabled reading fluency assessment for Dutch

Abstract
The importance of reading for educational, vocational and societal life cannot be understated. Nonetheless, recent large-scale studies reveal that the reading comprehension of students has declined globally, and specifically in the Netherlands. Developing fluent reading skills allows children to read quickly, accurately and with proper expression, which is fundamental to become a good reader. To monitor this development, teachers need to assess fluency on a regular basis. However, fluency assessment is currently time-consuming for teachers, provides limited information, and neglects prosody assessment. This chapter presents a framework for, and the development of, a digital automatic fluency assessment tool for early primary education that overcomes current issues through incorporating Automatic Speech Recognition (ASR): the Speech Enabled Reading Diagnostics App (SERDA). Three word- and passage reading tasks were developed based on popular pen-and-paper instruments, and administered to 653 primary school children. The results provide usability, validity and reliability evidence for SERDA’s speed and accuracy measures. Furthermore, SERDA reduces the testing burden placed on teachers, increases the information gained, and facilitates prosody assessment.

In: CIDREE Yearbook 2024, p. 100