How Machine Aided Spoken Language Evaluation (MASLE) can revolutionize speaking practice, testing and research
Professor Brian Teaman
Osaka Jogakuin Colleg, Japan
ELL Journal Room (Room NIE3-03-162, Level 3, Block 3), English Language & Literature Academic Group)
National Institute of Education, NTU
1 Nanyang Walk, Singapore 637616
11 March 2009/Wednesday
Speaking practice and testing has been neglected for many reasons. One of the main reasons is the logistical difficulty of evaluating spoken language. The MASLE system increases the options for oral practice and testing by allowing students to record their speech using a typical browser so that the speech can then be accessed by a teacher/rater and graded. The MASLE system has three main parts: a prompt delivery engine, a human-rater module and a machine-rater module. This presentation will review the different types of tests/practices available and the potential for each of these in learning to speak in a foreign language. Prompts range from those that are expected to generate totally predictable utterances to those that generate open-ended responses. The totally predictable utterances can be assessed by the machine or human rater while the open-ended questions can only be assessed by a human rater. The automatic assessment focuses on mainly the segmental aspects of pronunciation. The human rater can focus on any aspect of speaking such as fluency, accuracy, vocabulary or grammatical correctness. One example of the impact of the MASLE system will be shown as I discuss the results of a pilot test at Osaka Jogakuin College. This pilot test is the first step leading to the assessment of all students’ oral skills as a part of a placement exam which previously has had no oral component. This oral placement test blends machine and human scoring to produce an evaluation useful for placement within the college’s English curriculum. Although the development of MASLE has focused on evaluation, I hope to show that this additional tool will open up many rich opportunities for practicing and researching the spoken language.
Brian Teaman, a native of Lorain, Ohio (USA), is a professor at Osaka Jogakuin College in Japan where he teaches and researches pronunciation and speaking in a foreign language. He has worked in computer assisted language learning since 1982 with the advent of the first apple computer. He directed the development of acclaimed English materials for Hiroshima University’s Virtual University. When he is not doing research he enjoys making music and cycling.
The MASLE system is available at http://popear.org where you can try the system out using the classkey “NIE.” Please contact him if you are interested in using it for a class, school or want to see your test results.