English Speaking Assessment: How to Give Transcribed Conversation Tests

In my previous article, Why Koreans Struggle to Speak English, I mentioned that correct grammar comes after a lot of speaking, not before. It is a result, not a prerequisite. The solution is less instruction and more production. More production – a lot of speaking – requires a way to test a lot of speaking.

Such testing must be practical: easy to give and grade, accurate, and available to every English teacher, both Korean and native speaker, regardless of their teaching experience and regardless of their confidence to grade speaking ability.


First, the test had to be communicative. My classes were two- and three-person conversations and so should be the test. Second, it had to be practical, easy to give and grade. Third, it should have impact, it should improve behavior. To provide feedback and error correction, transcribing was added. Transcribing provided personal feedback, and data, which gave the test great accuracy. Serendipitously, the data also proved improvement. Eureka! Here then is a practical, authentic, communicative test which both measured and improved speaking ability in one fell swoop.


1. Three students of similar ability have a 20-minute conversation.
2. The conversation is recorded on digital recorder.
3. The audio files are emailed to the students and they transcribe the whole conversation on Word.
4. Students compile their own data (total words, average-words-per-utterance).
5. Voilà! Students get extensive personal feedback and teachers get accurate grading data.


Six years of transcript data proves a strong correlation between the total number of words spoken and speaking ability. The average words per utterance does not have a strong correlation to speaking ability (because often low level students memorize a long passage and then clam up). However, average words per utterance is a very strong indicator of improvement.

It's possible to see a great level of improvement in only seven weeks. For 10 years, students have had non-communicative, grammar-based, teacher-centered, lecture courses. They are ready to speak, and when they do speak, most show great improvement. This is visible, measurable improvement for the previously invisible skill of speaking. The test has impact – it produces and proves improvement.

Here then is a way to directly grade the product of communicative speaking classes.

1. First and foremost, an accurate speaking ability grade can be rendered without recourse to the standard grammar, pronunciation, accuracy, fluency, comprehension and vocabulary scoring. There is no need and little benefit to tediously compile those indirect and impactless indicators.
2. Second, it is transparently fair. Students can clearly see all facets of their – and their classmates – test performance.
3. Third, it has great discriminality. It spreads the grades out by performance, much more so, and more precisely, than grammar, et al.
4. Fourth, it measures how much students do, not how much they do wrong. By rewarding how much students do, it impacts them do to more.

Transcribed communicative tests are the solution. Students do all the work and get extensive personal feedback. Teachers get accurate grading data which also proves improvement.


1. For the same reason that every ballet studio and health club has mirrors, every speaking test should be transcribed: Self-monitoring, self-correcting feedback.
2. Each transcript contains an overabundant gold mine of self-correctable errors. (My brother, she is in the army, requires no teacher input.)
3. With the word-count, students are graded by how much they do, not how much they do wrong.
4. Average-words-per-utterance provides a visible measure of improvement.


The fixation on grammar has handicapped Korean students for way too long. This is Einstein’s definition of insanity: Repeating the same action but always expecting a different result. The article in the Korea Herald shows again that the results are not improvement.
Most university freshmen in Korea are advanced students but beginning speakers. Beginners in any skill who do a lot improve a lot. So, step aside and let them speak. Here is a way to grade that speaking.

Communicative ability will only flow from communicative tests. Here is that test. For teachers and countries with a communicative will, here is a way.

Click here to see Gunther's test as well as download a free test transcript and data template.  

Gunther Breaux

Gunther Breaux came to Korea in 1996. He's an associate professor at Hankuk University of Foreign Studies, teaches English conversation, has an MA TESOL and is the author of several popular ELT books. Email: plangbro@gmail.com Websites: jazzenglish.com & talk-or-walk.com

One thought on “English Speaking Assessment: How to Give Transcribed Conversation Tests

  1. Pingback: Why Koreans Struggle to Speak English | Profs Abroad

Comments are closed.