The CALL system developed in our project enables ESL/EFL learners to enhance their skills of shadowing and reading aloud. Learners are required to record their shadowing and reading aloud into the computer while listening to passages read by a native speaker of English. After recording, they can listen to their voices and observe the sound waves of their own recording and the model one. Through auditory and visual comparison of the two recordings, they can understand the shortcomings of their performances and where they should practice more. Especially in shadowing practice, learners' shadowed speech is automatically analyzed and evaluated by the computer using speech information processing technology like GOP (goodness of pronunciation). Their English proficiency levels measured by TOEIC (Test of English as International Communication) are also predicted and presented. Based on the results of automatic scoring, the learners can understand how well they have conducted shadowing objectively and also grasp their own proficiency levels. From the viewpoint of material development, this CALL system enables instructors to choose any speech data obtained from CDs, DVDs, Web sites, etc, and use them as practice materials. For instance, if both audio and text files of President Barack Obama's inauguration address are available, the learners can practice shadowing and reading aloud using his famous speech. Thus instructors' selection of speech data suitable for the learners' interests and proficiency levels can increase student motivation and continuous use of this system, hence improving both aural and oral skills.
Dean Luo, Nobuaki Minematsu, Yutaka Yamauchi and Keikichi Hirose. "Analysis and comparison of automatic language proficiency assessment between shadowed sentences and read sentences," Proc. SLaTE 200