Capturing the quantity and location of adult wh-words in the preschool classroom using a sensing tool system

Observational approaches may limit researchers' ability to comprehensively capture preschool classroom conversations, including the use of wh-words. In the current proof-of-concept study, we present descriptive results using an automated speech recognition (ASR) system coupled with location sensors to quantify teachers' wh-words by preschool teachers in the literacy activity areas of a preschool classroom. Data from two children, one is 5.3 years old with attention-deficit/hyperactivity disorder (ADHD), and another is 5 years old without identified disabilities, along with teachers, were analyzed. We found that the ASR system is a viable solution for automatically quantifying the number of adult wh-words during interactions in preschool classrooms at different time points and locations. This paper reports how an ASR model, coupled with location sensors, quantifies the frequency of wh-words between two-time points and between a child with ADHD and a typically developing child. The results provide a proof of concept that an ASR model, including acoustic and language models, can automate the detection of wh-words in preschool teachers’ classroom speech. However, further research with larger and more diverse samples is required to explore the cost and time implications of scaling up across a variety of settings and populations to inform efficient classwide and individualized data-driven instructional practices.