Developing a diagnostic measure of linguistic competence in the English language for lower secondary school students in Sarawak / Kho Chung Wei
English language education in Malaysia is undergoing a reform, where students are expected to graduate secondary schools with the ability to use English in daily situations, including the workplace. However, some students enter secondary schools without even acquiring the basic functional literacy i...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/15210/1/Kho_Chung_Wei.pdf http://studentsrepo.um.edu.my/15210/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | English language education in Malaysia is undergoing a reform, where students are expected to graduate secondary schools with the ability to use English in daily situations, including the workplace. However, some students enter secondary schools without even acquiring the basic functional literacy in English. The role of English language teachers especially in the lower secondary is to provide remedial instructions to narrow the language gaps of these students. A useful tool to support the work of these teachers is a user-friendly and efficient diagnostic test that can identify the students’ language gaps as accurately as possible. This study was an attempt to develop such a test. Using cross-sectional survey design, the test was administered to a representative sample (N=3,086) of lower secondary school students in the southern zone of Sarawak. Data analyses indicate that the item response data best fit the between-item multidimensionality Rasch model, suggesting that the diagnostic English language test measures six related unidimensional latent variables. These variables were positively correlated to each other as expected from a multidimensional test of the same construct. However, not all the items and cases fit the model. Out of 90 items and 3,086 cases, the responses of three items did not fit the Rasch model when the test-takers’ abilities were not targeted by the items while 5.83% of the cases underfit the model. The misfits occur most probably due to guessing. There were also 21 problematic items that could not discriminate test-takers with low ability from those with high ability. Moreover, it was found that there were gaps in the item distribution across the range of test-takers’ abilities for five of the six dimensions although the overall item difficulties were normally distributed. In terms of differential item functioning, one of the items was found to have moderate to large DIF across grade levels and age cohorts while another item had slight to moderate gender DIF. The DIF that exist across ethnic groups, native language clusters and geographical areas were considered negligible. The study has demonstrated that the diagnostic test has promising potential, but much work still needs to be done.
|
|---|