A Study of variables in the online Korean speaking tests for academic purpose

Foreign students learning Korean took part in an online speaking test to evaluate their ability to speak Korean. The major research questions covered the effects the task types have on L2 test-takers' proficiency, test types, and raters. Furthermore, the authors investigated what the results imply regarding the design and implementation of Korean-language tests. The authors divided the 394 participants into four proficiency levels: mid-low, mid-high, high-low, high-high. Six native speakers were invited to assess the ratings. The findings indicated that the effects of Korean proficiency levels on ratings differed depending on test type. The subjects of the mid-low level showed a wide range of logic across the tasks, while those of high-high level a narrow range. Moreover, the degree of difficulty on both levels could be ordered as follows: 'story (experience) > explain (table), persuasion (dialog) > persuasion (read) > persuation (lecture) > explain (picture)' for the mid-low group, and 'persuasion (lecture) > persuasion (dialogue)> explain (picture) > explain (table) > persuasion (read) > story (experience)' for the high-high group. The most difficult task, however, was the different test types: the persuasion task (lecture) on the B type and the explanation task (picture) on the A type. The study revealed that rater training should be strengthened in the case of an online, non-interview speaking test.