Abstract
Either out of inadequate technology or for the sake of convenience, most language proficiency tests tend to oversimplify students’ diverse needs and provide one size fit all solutions. Take Oxford Young Learners Placement Test for example. One part of this test combines grammar, vocabulary and functional language use. A student gets a final score and a corresponding CEFR level at the end of the test. However, such generic result cannot provide individualized solution for learners. Based on a sample of 156,221 young English learners aged 4 to 12, we find young ELLs have various language acquisition paths both in language knowledge (vocabulary and grammar) and skills (listening, speaking, reading and writing). For instance, learners’ grammar knowledge is not always in sync with their lexical range. For any individual, each aspect of the test results may fall into different categorizes, sometimes with huge gap. This indicates the report provided by many test developers might be either too general or unreliable because learners’ vocabulary and grammar levels are usually underrepresented or assumed identical to their four language skills. The clear insight into all aspects of language development can help us design a language proficiency test that provides tailored solutions for each individual in developing different domains of competence. With the big data analysis and computer-assisted technology, we also find some benchmarks that can predict what a learner know and doesn’t know so as to shorten test time by 5.4 times.
Author Information
Peter Shih, Hangzhou Panda Education and Technology Ltd. Co., China
Qiwen Zhou, Hangzhou Panda Education and Technology Ltd. Co., China
Paper Information
Conference: ACE2019
Stream: Assessment Theories & Methodologies
This paper is part of the ACE2019 Conference Proceedings (View)
Full Paper
View / Download the full paper in a new tab/window
Comments
Powered by WP LinkPress