Luiz Octávio Barros was asking about scripts we feed ChatGPT the other day. I mentioned this:
Which is a very boring example of an (over the word count) type IGCSE ESL text I generate for correction practice. I’m not *massively* into correction tasks like this, but my learners do need practice and reminding of common errors to look out for when self-correcting.
Anyway, I’m still not a massive ChatGPTer. I’m trying to work out how to best use (or get learners to use) the tool productively/efficiently/purposefully/etc. Thought I’d start documenting some of my ChatGPT uses, however off-the-cuff/unpolished, and reflect on their value (if any).
Today’s lesson was a springboard to our new unit on food (Cambridge IGCSE ESL, Part 1, Unit 3…). My hour-long primer / vocab review / diagnostic type thing involved activities such as:
- One minute challenge! How many words do you know related to: cooking verbs, adjectives related to food prep, tastes/flavours, cuisines, types of diet, etc.
- List three dishes from your country. Make notes on the dish name, when it’s traditionally/typically eaten, ingredients, how it’s prepared, what it tastes like. Describe each dish to classmates – which dish sounds most weird/delicious/like own cuisine/etc
- Student A: you’re bringing your mate from another country to a local restaurant. The whole menu is in your language and they can’t read it! Listen to their questions about dietary requirements, preferences, etc. Explain and recommend dishes. Student B: you can’t read the menu, you’re fussy. Ask for recommendations. Etc. (This is like my lesson for Onestop English here)
- Discussion topics on food, like ‘my food heaven/my food hell’
So, how did I add ChatGPT into proceedings? Was it valuable? Did it get in the way?
One minute challenge:
After learners reviewed existing vocab and played a little word tennis, we asked ChatGPT to list 10 common items for each topic. Learners (just for fun) saw how many items matched their own guesses. Then they reviewed the other words to see what they already knew. They fed ChatGPT instructions to provide a translation in brackets after each item, then I asked them to relate new words from ChatGPT output to their own cuisine (which food from your country is [taste/preparation method/etc]?)
Verdict: It was a bit of a vocab upgrade. ‘marinated’ and ‘fermented’ came up, both being useful vocab for Koreans describing their cuisine. It didn’t offer much, but it was quick, generated some chat, 5 minutes well spent. Prone to less useful vocab off-the-cuff though so not that principled.
Describing dishes from your culture:
I asked learners to note down info about dishes from their cuisine in a table under these headings: name, when is it traditionally eaten, ingredients, preparation, taste. They struggled to generate ideas and there were vocab gaps. So, we asked ChatGPT to provide examples, and do so in note/bullet point form. That was good support and helped with activity flow, but it took out a bit of independent research from learners. Still, it gave learners some of the key vocab to work with and to store for future practice:

Verdict: This is an IGCSE ESL speaking-type question (‘Can you describe a typical dish from your country?’), The model from ChatGPT did help with some specific vocabulary. Learners could use ChatGPT to help with planning for speaking tasks, but you’ve got to tell it to be concise, and maybe only get *some* support (like two examples per box) or support for certain info only.
I think this would be better used as the basis for a table completion task first. I’ll adapt next time. It did give me an idea to quickly make barrier games though, which was cool and seems obvious now.

Mock menu for a roleplay:
So, whenever I’ve done this activity before I just get learners to search for an authentic menu in their own language online. This proved a bit faffy, so I just asked them to get ChatGPT to create one. It didn’t look the best, but it served a purpose.
Verdict: meh, nothing much really, just a quick prompt.
Speaking/vocab tasks
We used 6×6 dice-roll grids of tasks/questions in our Pearson IGCSE ESL book to review prior knowledge. You roll twice to get coordinates for your question or challenge. You could make them quite easily using ChatGPT:

This was pretty useful actually, as chats using this brought up some good vocab and discussion. And with some you can end up springboarding to off-the-cuff tasks, like how ‘describe a time you tried a new food’ becoming a mini stage on telling better anecdotes.
On an aside from the rather generic ‘food’ topic, I can see how I can use that type of grid for a knowledge review in a weeks’ time of Boy Overboard with the Year 7s (although based on the examples below I’ll have to check it’s all relevant to Chapters 1-5!):

Overall:
A few good experiments. Probably the support for notetaking was my favourite. I can see how I can use this type of support earlier in the year then reduce it, get learners to add more ideas, etc. Very quick to create. As was the info gap based on it, so I’ll bear that in mind.
Nothing ground-breaking and mostly in the ‘time-saver’ category I guess. And perfect for these start-of-a-topic get them chatting type activities.
Let’s see what the next experiments bring.
Categories: General, reflections

Leave a comment