This determines the results
Measure the right things, and you incentivize the right things.
Japanese students are notorious for having poor English. This can even be in spite of getting the highest marks in a test (I’ve seen it before). Is this because the students don’t know how to convert their knowledge of grammar and vocabulary into communication? Or is it because they have zero incentive to even do that if it’s not on the test?
Pretty simple, in theory at least. Make the test a test of communication ability. You’ll get students able to communicate. Imagine that!