What is it about?
We use a large language model to analyze transcripts of student speech. Students responded using natural speech to describe a STEM vocabulary word or use it in a sentence. We scored student responses on a range of three (no understanding, some understanding, and mastered).
Featured Image
Photo by Kenny Eliason on Unsplash
Why is it important?
We find that a language model can score students near identically to a trained human rater. This can greatly increase the scalability of assessing students from verbal responses. This could have far reaching ramifications in a number of fields where assessment of speech is time consuming.
Perspectives
This publication builds on a growing trend in my lab and others where machine learning has the ability to reliably replicate instructor assessment. While much research needs to be completed, I am incredibly impressed with PhD student Zhongdi Wu and his diligence in evaluating the work.
Eric Larson
Southern Methodist University
Read the Original
This page is a summary of: Towards Scalable Vocabulary Acquisition Assessment with BERT, July 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3573051.3596170.
You can read the full text:
Contributors
The following have contributed to this page
