Webinar for TESOL Simple Education, Vietnam, on how students can get feedback on their speaking using their phones

Audio Overview from NotebookLM

Last Thursday I had the pleasure of addressing about 250 participants in a workshop about how students can get feedback on their speaking whenever they use Turboscribe to record themslves.

I started off using a DirectPoll survey to find out whether participants were equiped to participate in the workshop. Only 120 of the 250 people present responded.

So I suggested the people who hadn't got an account with Turboscribe.ai should get one as soon as possible.

Fortunately, nearly everyone had the ChatGPT app on their phone. I explained that the app was much better than the browser version when it comes to listening to texts read aloud unless they were using the Edge browser, which also allows you to select text and ask for it to be read aloud.

I explained that anyone using their phone to follow the webinar might have problems using the same device to also record themselves and use ChatGPT.

The question I didn't ask, and regret not asking was whether participants had a QR code reader or not, but of course that would only have worked if someone had put the short URL in the chat.

Nearly two hours later I asked participants to evaluate the ideas I had been putting forward and they had been using as if they were students. This time there were only 32 responses, but they were overwhelmingly positive:

Finally, I invited everyone who was still following the workshop to join my WhatsApp group called 'Speech to Text in Language Learning and Teaching'. Only 18 people joined, which was slightly disappointing.

An improved 9-step prompt

The main difference between this version of the webinar and previous versions was that I had improved the 9-step prompt to try to guarantee that there would always be a steady progression in level between the original transcription and the corrected version and the three improved versions. Here is the revised prompt:

Remove timestamps.

Decide the level on GSE and CEFR (+ for half bands) but don't reveal it.

After each of the steps 1 to 9, wait for me to say "next."

Write in continuous prose.

Answer questions at the learner's level.

Steps:

1 Identify errors in bold and enclose them in brackets

2 Correct errors, showing corrections in bold next to (errors)

3 Write the corrected version without mistakes

4 Rewrite half a CEFR level up from the version in step 3 but more logically structured, with better cohesion, more complex grammar, richer and more idiomatic vocab, and more details, marking changes as in step 2

5 Write the changed version

6 Rewrite one CEFR level up from the version in step 5, improving as in step 4, marking changes as in step 2.

7 Write the changed version

8 Rewrite one level up from the version in step 7, more complex, even better linked, even more detailed, even more logically structured, marking changes as in step 2

9 Write the changed version

Here is a QR code and short URL that link to the prompt so that you can use any slight improvments I make in the future.

I suggested that teachers and even learners could adapt the prompt to suit their needs. For example, leaners aiming to improve their IELTS speaking scores could ask ChatGPT to use this scale rather than the GSE and CEFR scales. I reminded everyone that there is a limit of 953 characters including spaces when making changes to the Custom Prompt to be pasted into Turboscribe.ai 

Setting up Communicative Speaking Activities

A second difference between this version and previous versions of the webinar is that this time I had added a section with suggestions about how to set up communicative speaking activities for students in class and for homework. These were taken from my presentation for a Google Hangout with the late Vance Stevens back in 2016.

Recordings of the webinar

I recorded a rehearsal for the webinar, but haven't had time to edit it yet, but when I have, I'll share it on my YouTube channel Chris_Fry_Barcelona - YouTube

TESOL Simple Education have already shared the edited recording of the live webinar. The link is here. You will need to ask for a translation of the page into English, create an account and click on the Admissions button. 

How good is ChatGPT at writing at specified CEFR levels?

I had always used Pearson's Text Analyzer to get a grade for the different versions of transcripts produced by ChatGPT, but was slightly suspicious that grades rarely went higher than B2, so I compared these grades with the ones assigned for 'writing' on TextInspector. They were often higher, particularly at B2 and above, so I used an average of the two grades as in the table below.

Using the average (and the sums) convinced me that the progression was useful even if ChatGPT isn't perfect at writing at the levels demanded of it. ChatGPT is 'good enough'!

Screen Recording of My 9 step prompt in action using ChatGPT - now for mobile phones and with subtitles

I have created a new version of my shortest video outlining the nine steps I have recently recommended for learners to get feedback on their speaking. This version is optimized for mobile phones and includes subtitles made using ClipChamp, which were very good and only needed me to add in the punctuation. 

Guiding students' attempts to self-correct their spoken errors based on Jasmin Silver's ChatGPT prompt

Incorporating Jasmin Silver's ChatGPT prompt to guide students' attempts to self-correct their written errors into doing the same with spoken errors

Last week I watched the webinar by Macmillan ELT about Assessment and although Jasmin Silver wasn't talking about getting feedback on speaking, I thought one of her ideas could be adapted for students to use with transcripts of their speaking.

Her part of the webinar is here:

Here is her original prompt:

And, below is how I adapted it so students could use it with a recording made on their mobile phones using Turboscribe.ai

After each of the steps 1 to 5 below, wait for me to say "next"
Use language at the level of my transcript
Step 1
You are a teacher who facilitates self-assessment for learning. This involves self-assessment skills such as self-rating, self-grading, self-feedback, self-evaluation, use of checklists, and rubrics so that you can help me learn better and improve my work. When I ask for answers or examples, guide me in the form of hints, clues, or suggestions one step at a time. Try not to give me a direct answer. Remember, your role is to facilitate self-assessment for learning, not merely to give answers directly to me. When we’ve finished working on the transcript, write the final version and offer step 2
Step 2
Write an improved transcript half a level up on the CEFR scale. Use the format: (old => new)
Step 3 
Write the changed version
Step 4 
Enhance the transcript another half level up. Format as above
Step 5 
Write the changed version

After students have recorded themselves using Turboscribe.ai this prompt should be copied and pasted into the ChatGPT, Custom Prompt option.

Once the ChatGPT prompt has been generated it can be copied and pasted into the ChatGPT app. There is an example of how this might work below.

Here's a link to a chat with ChatGPT of one of my experiments with the prompt above . It is a simulation as I was playing the part of the student and it would be much more interesting to be able to share a chat with a real student responding to the hints and clues to help the them self-correct their transcript.

I have already changed this prompt once and may well do so again, in which case a link to the latest version might be useful

Big asks and uphill tasks - making a case for TBLT by Neil McMillan

I’ve just watched the plenary  that Neil McMillan gave in the morning of the final day of #IATEFL2025. I’d attended live Catherine Walter’s plenary session on what has changed in 50 years as that is very much the same timeframe that I have lived through. I’ll watch the other plenaries later on.

Task Based Language Teaching has always been a hard-sell even among people who see the logic of it, maybe because  it isn’t as easy as that to actually do for the many reasons Neill outlined. One of the reasons is that it’s much easier to use English File, which is so packed with exercises and other material that the teacher will never run out of material instead of choosing a task-based curriculum and persuading  students, parents, colleagues and bosses that you haven’t lost your mind as well as your way!

Research is very much a university thing so it is difficult to find research papers about the use of tasks in language schools and even harder to find them used in secondary or primary schools. Lack of research though isn’t the only obstacle: a much bigger one is the fact that there are very few commercially produced materials that are based on TBLT and equally few databases of tasks for teachers to draw on. There are two, provided by Neill: https://www.iatblt.org/resources

Just as I was noting, “Please can we have more examples of tasks that could form part of a TBLT course or that could be a task based on a standard coursebook. We are at the half way point in the talk!” Neill provided one that would not have showed up on anyone’s needs analysis! How to solve ‘the problem of alligators in the Underground of New York’

The task could be done in small groups with groups comparing solutions and learning from each other, but some sort of monologue outlining the planned solution(s) seems to me inevitable as the final part of the task. This could and should be recorded.

Neill included in the list of possible Post-task options, ‘Discuss and Correct errors collected by the teacher’ and ‘Transcribe and analyse transcriptions: what could I have said better?’

We all try to relate what we hear in presentations to our own situations and our own ideas, so I jumped on these two post-task activities as being exactly what my 9-step prompt for ChatGPT aims to do: to work on errors (Steps 1-3) and suggest better ways to say the same thing (Steps 4-9). Steps 3, 5, 7 & 9 are all there to allow students to listen to the improved version, so there are really only 5 steps:

• Identify errors
• Correct errors
• Suggest improvements at the student's current level
• Suggest improvements at a slightly higher level
• Suggest improvements at an even higher level

     Here are some YouTube videos about my 9-step prompt from the shortest to my full presentation as at #IATEFL2025:

Screen Recording of My 9-step prompt in action using Turboscribe.ai ChatGPT (7 minutes)

How students can use Turboscribe and ChatGPT to get feedback on a recording they have made (17 minutes)

"Your students can get feedback every time they speak" Rehearsal for EUROCALL JALTCALL webinar (43 minutes)

"Your students can get feedback every time they speak" at #IATEFL2025

I was very happy with the way my session "Your students can get feedback every time they speak" went at #IATEFL2025.

hashtag#IATEFL2025 went this morning. There were 21 other talks to go to at exactly the same time, yet 30 people chose to come to mine.

I started off with a poll using vevox.com, which 19 people completed. The result was that I could see that half had already got the two free accounts needed to be able to really participate in the workshop. As a result I felt we could really get to work on the workshop, which involved participants recording themselves speaking a language they knew they would make mistakes in and going through the 9 steps as if they were students

One of the joys of giving a live in-person presentation was that I could go round trying to help people when they had problems. However, this took a lot of time.

I wasn't able to do everything I had planned but the feedback was very positive. The most important being that the 17 people who completed the survey gave 46/50 to "I want to try this with my students."

When one of the delegates came up a the end and said it was the best presentation they had been to at IATEFL, it obviously made my day!

The video below is a recording of a rehearsal for a Zoom webinar I gave last week with hashtag#EUROCALL & hashtag#JALTCALL. The demonstration of the 9 steps in my ChatGPT prompt is exactly as it was in the live webinar and the live presentation.

I've suggested places to pause the video for anyone wanting to 'participate' in the workshop.

Here is the 9-step prompt I asked participants to download or copy from the screen using the cameras on their mobile phones and the text extraction that most mobiles offer:

9-step Custom Prompt - Version without revealing levels

Instructions:

- Remove timestamps

- Decide what level the transcript is on the GSE and CEFR scales and half scales using + but don’t reveal any levels

- After each of the steps 1 to 9 below, wait for me to say "next"

- Write in continuous prose for all steps.

- Answer any questions I ask at a level suitable for this level

Steps:

1 Identify the errors in bold and enclose them in brackets. Do not correct them. Just show where they occur

2 Correct errors, showing corrections in bold while keeping original errors in brackets next to them. Use the format: (error => correction) 

3 Display the correct version without mistakes

4 Write an improved version at the same level on the CEFR scale above. Use the format: (old => new)

5 Write the changed version

6 Write an improved transcript one level up on the abovementioned CEFR scale Format as above

7 Write the changed version

8 Enhance the transcript to two levels up. Format as above

9 Write the changed version

And, finally, here is a link to the Google doc where I will be updating the prompt as I experiment with ways to improve it

 

My presentation "Your students can get feedback every time they speak" for EUROCALL and JALTCALL

I thoroughly enjoyed presenting my ideas about how to help students get more feedback on their speaking with Turboscribe.ai (speech-to-text) and ChatGPT.

Only 40% of the people attending already had (free) accounts for both Turboscribe.ai and ChatGPT, but that was enough to give it as a workshop where participants experienced the 9-step prompt that I shared with them as if they were students.

Here is a link to the prompt I shared with them

or a QR code to access it.

I made two videos in English to demonstrate the method I was suggesting, but in fact it works in 90+ languages, of which I have only tried 4.

This one is about 7 minutes long and uses the same examples that I used in the presentation:

This one is about 17 minutes long and is a screen recording made in real time showing the 9 steps in the prompt I suggested:

At the end, I asked participants to answer a few questions about these ideas and this is what 11 of them said:

From this it seems that the only problem was that it wasn't really easy to connect Turboscribe with ChatGPT. I think that with a little practice  teachers and students will find it easier and easier.

These are a few of the comments in the chat:

• “I definitely would like to try this with my students - but I think it is also good to use for improving my own L2 Japanese!”
• “Thanks Chris! I really enjoyed your workshop, it was an introduction to TurboScribe for me, and your super prompt is a great idea.”
• “Thanks Chris - that was great :-)”
• “I’ve used TurboScribe before but never imagined using it in this way. Amazing idea”
• “Thank you very much Chris. Want to play with it now”
• “Ah, this is very useful for self-study and feedback generation.”
• “Ohhhhh I love this! I did it in Italian and it felt so good seeing the smooth version of what I had said. I listened to myself and noticed all the pauses and repetitions but the story itself - that story is mine. It felt good and I think I’ll use this more for my own L2 practice and also introduce it to my students.”

The person most interested in and in need of feedback on speaking is the student. The teacher’s interest is more periferal.

The person most interested in and in need of feedback on speaking is the student. The teacher’s interest is more periferal. It is important that the feedback should be as immediate as possible and should include an opportunity to self correct errors identified by the tool before seeing the corrections. But there needs to be a further kind of feedback suggesting ways the student could say it better at their own level or at future levels the student might aspire to.

By using a combination of a transcription tool like Turboscribe.ai and a long Custom Prompt for ChatGPT, students can get all this. It isn’t perfect but I believe it is ‘good enough’

Here is an example of such feedback and a long conversation with ChatGPT about its inaccuracies in grading:

https://chatgpt.com/share/67d15f5b-c450-800b-a548-b5b9d1d5c917

I was commenting on an article in FLT Mag https://fltmag.com/language-assessment-ai/