Date of Award

2026

Document Type

Dissertation

Degree Name

Doctor of Education (EdD)

Department

School of Education

First Advisor

Linda Samek, EdD

Second Advisor

Scot Headley, PhD

Abstract

The purpose of this study was to examine the accuracy rate of Automatic Speech Recognition (ASR) live captions in an in-person synchronous classroom setting and the perceptions of participants experiencing them. The reduction in the workforce of Communication Access Realtime Translation (CART) captioners, improvements in ASR, and previous studies suggesting benefits for all learners have left educators, institutions, and policymakers wondering if ASR live captions can ethically meet accessibility requirements under the Americans with Disabilities Act (ADA) and also benefit typically hearing and English Language Learners (ELL). Two general undergraduate education classes were selected to evaluate the accuracy rate of ASR live captions using the Number, Error, and Recognition (NER) model. Participants in each class were given a Likert survey with three scales and an open-ended question assessing perceived usefulness, perceived ease of use, and perceived reliance. Findings show ASR live captions meet the 98% accuracy threshold suggested by the NER model, however, there was an increase in serious errors compared to previous studies. Participants had a neutral to slightly positive perception of the ASR live captions when used in this setting with females favoring it more than males in the scale of perceived usefulness. These findings contribute to the literature on ethical best practices and practical considerations in using ASR live captions in an in-person synchronous classroom setting to promote inclusive learning environments.

Download

Included in

Education Commons

Share

COinS