Implementation of an automated grading tool for phonetic transcription training

ABSTRACT

Clinical phonetic transcription is regarded as a highly specialised skill requiring hours of practice for mastery. Although this skill is a critical part of students’ clinical preparation to become speech-language pathologists, students often report feeling unprepared to apply the skill in clinical practice. Previous studies suggest that increased opportunities for practice and timely feedback on transcriptions are needed in order to develop skill confidence. However, providing more opportunities for practice can be impeded by the limited resources to manage the grading of additional assignments. The purpose of this study is to show the implementation of a web-based learning management system (LMS) designed in our labs for phonetics instruction. The Automated Phonetic Transcription Grading Tool (APTgt LMS) was developed to provide a platform for assignment delivery and automated grading of transcription assignments. The APTgt LMS has three embedded IPA keyboards (basic, advanced, and full IPA) and an automated edit distance algorithm modified by phonetic alignment principles, which allows for individualised scoring and visual course-level feedback in an interactive online environment. For pilot testing, student confidence was queried before and after practice opportunities using APTgt. A concurrent mixed methods research design was used to analyse four Likert scale and three open-ended questions. Student confidence in transcribing disordered speech was found to significantly increase (p <0.001) following additional practice. Students reported concerns related to accurate transcription of disordered speech and that additional practice is still needed. Tools like APTgt can aid in facilitating student learning and increasing student confidence in applied transcription.

KEYWORDS:

• Phonetics
• phonetics teaching
• edit distance
• online learning

Acknowledgments

The Automated Phonetic Transcription Grading System (APTgt) was developed at Auburn University as a collaboration between the departments of Speech, Language, and Hearing Sciences and Computer Science and Software Engineering. We would like to give a special thank you to the numerous undergraduate and graduate students that served in our labs to develop the application: Sicheng Li, Guorui Li, Ishaan Mishra, Yaoxuan Luan, Jueting Liu, Yang Cao, Robertson Bassy, Chaohui Ren, Chang Ren, Abigail Bennett, and Jordan Robinson. Further we would like to acknowledge the funding support for this project by Auburn University’s Breeden Endowment Award for Teaching Excellence, the Auburn University College of Liberal Arts Stevens Endowed Scholar Award, and the National Science Foundation (NSF #2021755; NSF #1457855).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the NSF [NSF #2021755; NSF #1457855].