Closing the gap between oral lexicons and sight vocabulary: Examining speech recognition technologies

Abstract

Hey Siri, Alexa, Google, can you help me learn to read? Speech recognition apps can take dictation. Elementary, middle, high school students, and adults, who struggle to read can dictate to speech recognition apps and see their oral vernacular become written words. However, at the time of this study, speech recognition apps remained notoriously inaccurate. The purpose of this study was to examine whether first-graders who struggled to read would develop sight vocabulary for the words they dictated to speech recognition apps. If so, how did their speech recognition-generated words compare with words they might encounter in other texts? Conversely, did they use vocabulary represented on high-frequency word lists? Derived from a broader ethnographic study, for 4 months, students attended a classroom writing centre and used speech recognition apps to compose. Findings indicated that speech recognition inaccuracies were inconsequential to participants' development of corresponding sight vocabulary. Within the timeframe and writing process used in the study, the students spontaneously generated a relatively low percentage of words they might encounter in other texts. However, speech recognition appeared to support students' abilities to embrace personally meaningful oral language and transfer their linguistic and culturally diverse oral vernacular to sight vocabulary. Findings raise issues of ecological validity and the origins of sight vocabulary curricula.