How to Test Speaking Skills in Japan

76 Part 2: How can we mark & administer speaking tests? ever, if you go with pairs or threes in each group, then you’ll need to pay attention to the pairings in order to better match ability levels and avoid problems People influence each other when they speak in various ways, so if there is a big proficiency level difference between two stu dents, this could have a negative influence on test validity (Na tion, 2013) What can be done about this situation? Well, if the differences are not so pronounced, then you could go with a random-partner approach This is often best because students, not knowing with whom they will partner with, will not be able to prepare too much in advance In other words, you’re more likely to get an accurate accounting of their spontaneous ability to have coherent conversations Random pairing is also the easiest and most practical way to go—just pull their names out of a hat, draw chopsticks, or shuffle up their name cards when you want to choose partners This is a good way to go if you feel everyone is more or less at the same ability level Another idea for sorting students is a semi-random approach Here you identify a few students who are either significantly above or below the general class average for ability Then, stack the deck in some manner so that this student will not end up partnering with that student This approach is also easy to do—just plan ahead of time who should not partner together, then make sure those stu dents don’t get together when you pull names out of your bag Finally, if the level differences are pronounced, you’ll need to assign the pairs yourself ahead of time This can be tricky at the beginning of the year when you don’t know your students well You’ll need to pay careful attention during class practice activities and keep careful notes about how everyone is doing or consider giving some sort of placement exam (Breaux, 2016) Another idea