Technology-Based Writing: Qualitative Study of the Writing Process When Students With Dyslexia Use Speech-to-Text

This qualitative study examines the writing processes of four students (aged 10–12) with severe dyslexia in an educational context, using Hayes and Berninger's (2014) writing process model as the theoretical framework. The participants attended a 7-week instructional program focusing on speech-to-text (STT). Their writing processes were examined through screencasts of verbal protocols during writing with reading and writing technology (RWT), with a particular focus on STT. The transcribed screencasts were coded and analysed. Our analysis shows that although no conventional spelling errors were found, other technology-based errors occurred. The most frequent were misrecognition of single words and omission of words or parts of sentences. Three categories of technology-based errors were identified: homophonic or nearly homophonic words, incorrect suffix/morpheme errors and compound words. The analysis also reveals that the evaluation process consists of three parts: correction, low-level revision and high-level revision. This means that an additional correction process is required because of technology-based errors. Consequently, the recognition accuracy of STT technology appears to influence writing flow and working memory load during technology-based writing.