I’ve been teaching English language learners in Japan for more than a decade, and in that time I’ve developed a keen sense of their strengths and weaknesses in the language. Not all learners are the same, of course, but there are patterns that I know will reliably surface with each new batch of learners. For a long time, however, there was a particular problem that I was reluctant to take on in my classes, pronunciation.
Not to make excuses for myself, but because I’ve been teaching homogeneous groups for so long, I’ve grown accustomed to the first language interference that affects Japanese speakers’ pronunciation of English. In other words, I’ve stopped noticing the systematic flaws in the pronunciation of my students.
A lot of those I teach end up studying overseas in America or the United Kingdom. Students returning from these experiences often reported some kind of failure to communicate that stemmed from poor pronunciation. Most recently, a student told me of how his attempt to order a soft drink, “Coke, please”, resulted in him being handed a cookie by the staff at the counter of a fast food restaurant.
Over the years, I’ve occasionally done one-on-one pronunciation work with individual students. I’ve also worked with entire groups on phonemes that are particularly challenging for Japanese learners. However, I’ve always struggled to incorporate regular, personalized feedback on pronunciation. The main reason for this is time. It’s just not possible to provide every learner with feedback on pronunciation, or so I thought…
A Set of Artificial Ears
Speech transcription first became a feature of Poodll when we were working on creating captions for video resources uploaded by teachers. We thought it would be of benefit to teachers and students if we could automatically generate transcripts of audio and video files. It wasn’t long, however, before we started to notice other uses. (Speech transcription helped us to create our auto-graded oral reading application, ReadAloud).
The first example of using speech transcription for pronunciation practice came from a group of students preparing for an English speech contest. Because I would be serving as a judge, I thought it best to avoid coaching the participants personally. However, I still wanted to help out with pronunciation, so I set up Poodll recording assignments for them with automatic transcription. The participants practiced recording their speeches and then compared the transcripts generated by Poodll to their original speeches. It was a great experience for them, as the transcripts revealed systematic errors in the way that they spoke certain phonemes. With the feedback provided by the speech transcription, they were able to rehearse their pronunciation to the point that they felt confident they could deliver their speeches in a way that everyone listening would be able to understand.
Since that first experiment with the speech contest participants, I’ve added transcription to a wide range of recorded speaking tasks in my courses. One of the expectations is that students listen to their submitted recordings and follow along with the interactive transcript to see where their pronunciation needs additional work. It’s created a nice feedback loop where none existed previously. Students now understand that pronunciation needs to be a focus, and their efforts to improve are independent and self-directed. It’s a win-win for time-strapped teachers and students alike.