EUROCALL 2014

Full Program »

Development of an automatic judging system for determining the difficulty levels of English audio materials

In order to overcome the unskillfulness of Japanese listeners to correctly comprehend even simple daily conversations by native English speakers, we have developed an e-learning program “your Personal LIstening MAnager(PLIMA)” which focuses on improving poor phonological analysis and the inability to hear liaison or unstressed sounds and so on. With PLIMA, practical research was conducted to increase the share of bottom-up listening comprehension skills by increasing learners’ knowledge about language, rather than a top-down approach which involves comprehension based on vocabulary recognition and understanding overall context, and we confirmed previous results of experiments indicating the effectiveness of our English listening system.
But the fact remains there are some certain types of liaison sounds that learners do not catch, and some types they do. That means if we specify what type, in other words, what level of liaison sounds they cannot catch, that will help PLIMA offer more effective learning to learners who lack certain listening skills.
To automatically determine the difficulty levels of the English audio materials, the text evaluation technique using lexical databases has been employed in many cases so far. For example, in the case of "SpeaK!" an application that supports learners in reading aloud and aural comprehension, an algorithm called “The Lexile Framework for Reading” is used to measure the difficulty of the text. However, automatic judging systems on the difficulty levels of the English audio materials itself, not using lexical databases, are almost unseen.
In this poster, we are going to propose a listening difficulty level determination system, to overcome such a challenging task as above, that applies a processing technique to extract the acoustic feature quantities that is employed in speech recognition engines. Speech recognition systems, in general, perform recognition process by matching the acoustic feature quantities of the unknown speech signal with the acoustic feature quantities in a database corresponding to many voices and the feature quantities obtained from them. Matching is implemented by selecting the minimum "distance" between the acoustic feature quantities of the unknown speech signal and the acoustic feature quantities in a database.

Author(s):

Hironobu Okazaki    
Research and Education Center for Comprehensive Science
Akita Prefectural University
Japan

Hironobu OKAZAKI is an associate professor in Research and Education Center for Comprehensive Science at Akita Prefectural University in Japan. He received his MA in English literature from Soka University, Tokyo. In addition to teaching, he has authored several books for Japanese learners of English.

Kanji Watanabe    
Department of Electronics and Information Systems
Akita Prefectural University
Japan

Kanji WATANABE is an assistant professor at the Faculty of Systems Science and Technology, Akita Prefectural University. He received his Ph. D. from Tohoku University in 2005. His research interests include spatial hearing, digital signal processing of acoustic signals. He is a member of the Acoustical Society of Japan, IEICE, VRSJ.

Shinichi Hashimoto    
Faculty of Informatics and Engineering
The University of Electro-Communications
Japan

Shinichi HASHIMOTO is currently a part-time lecturer at The University of Electro-Communications in Tokyo. Born in Japan, but living mostly in the United States, he received his MA in TESOL from Soka University of America in 1997. He has authored and edited several books related to English study and usage for a Japanese audience.

Mitsuko Suzuki    
World Language Center
Soka University
Japan

Mitsuko Suzuki is an Assistant Lecturer in the World Langugae Center in Soka Univeristy She received her MA in International Language Education: TESOL from Soka University in 2011. Her research interests include language learners’ motivation, professional career development, and CALL.

Eri Fukuda    
World Language Center
Soka University
Japan

Eri Fukuda is an Assistant Lecturer in the World Language Center at Soka University in Japan. She earned her MA in Education from Soka University in 2011. Her research interests include second language writing pedagogy, writing process, and CALL.

Kazuhiko Kido    
Department of Educational Administration
International Pacific University
Japan

Kazuhiko KIDO is an associate professor at International Pacific University in Japan. He received his Master of engineering from Soka University, Tokyo. His specialized field is “soft computing”.

 

Powered by OpenConf®
Copyright ©2002-2013 Zakon Group LLC