Discussing Opportunities to Improve Educational NLP for Second Language Learners at the 17th BEA
Lei Chen, Lead Research Scientist at RIT, presented his latest research at the 17th Workshop on Innovative Use of NLP for Building Educational Applications. His co-written paper, Automatically Detecting Reduced-formed English Pronunciations by Using Deep Learning, was one of 31 papers accepted to this year’s workshop. Other authors of the paper include Chenglin Jiang, Yiwei Gu, Yang Liu, and Jiahong Yuan, respectively.
Learning a second language can be greatly challenging. Lei’s research examines NLP use to enhance second language acquisition focusing on the pronunciation of reduced forms, which are words that are not written in English but that are frequently used by native speakers. Lei and his co-authors experiment with a deep learning-based method using a convolution neural network (CNN) to evaluate reduced form pronunciations. They find that their CNN model demonstrates improved performance compared to other methods.
Automatically Detecting Reduced-formed English Pronunciations by Using Deep Learning has been made publicly available and can be read in its full length here.
About the BEA:
The BEA Workshop is a leading venue for NLP innovation in the context of educational applications. It is one of the largest one-day workshops in the ACL community with over 100 registered attendees in the past several years. The growing interest in educational applications and a diverse community of researchers involved resulted in the creation of the Special Interest Group in Educational Applications (SIGEDU) in 2017, which currently has 240 members. This year’s BEA was held in Seattle, Washington on July 15.