I Don't Speak English. I Stopped Trying to Fix That.

What turned out to be solvable, what didn't, and where the line actually falls.
The situation
I do not speak English. I cannot read it well enough to check my own work.
My wife and I run a cooking class out of our home in Arashiyama, Kyoto. Every guest is a visitor from abroad. I built the website, I answer the booking emails, I handle enquiries, I run the social accounts. All of it in English.
This is not new for me. I spent twenty years in restaurants in Kyoto, and plenty of our customers were foreign. Some nights the whole room was. I could not hold a conversation with any of them, and the business ran fine.
It did not always run smoothly. I remember one dinner service where guests wanted to split the bill individually. Our dishes were shared plates, ordered for the table, so separate cheques were not really possible. They were not native English speakers either, so the translation app was not much help. I explained, they did not accept it, and in the end we went through the whole bill item by item together.
That is the kind of thing that does not get solved by knowing more vocabulary.
I tried to solve it, and found out I couldn't
In 2025 I tried real-time translation. Earbuds, live translation, the whole idea being that I could finally join the conversation in my own kitchen.
It failed in three specific ways.
When two guests talked at once, it collapsed entirely. We play Western music during class, and it kept picking up the lyrics. Even with one person speaking, the timing lagged enough that the moment had passed by the time I understood it.
For listening only, it was borderline usable. But it was exhausting, so I stopped.
The tools may have improved since. I do not know, because that experiment changed my mind about something bigger.
I stopped studying English. Almost entirely.
Not out of frustration. It was a decision about where my hours go. The conversation problem already has a solution in this house, and it is not me. Meanwhile there was a job nobody else was doing.
Getting people to come here in the first place. That is my work.
I would not call cooking my strongest skill. I have no idea what my strongest skill is, honestly, and I am not sure I know what I most enjoy either. But I know what I find interesting, and it is this: thinking about why someone would choose us, doing something about it, and then watching whether anyone turns up.
That started a long time before this business did.
Somewhere in my restaurant years I picked up a book about marketing — the first book I had read properly in years. The argument was that nobody wants the drill, they want the hole. What people buy is never the object in front of them. I was genuinely startled by it. I read it several times and started recommending it to the younger staff.
I stopped recommending books eventually. I will come back to why.
In the restaurants, the more senior I got, the more of this work landed on me. That is just how it goes. It was interesting and it was exhausting.
The difference now is the distance between me and the consequences. Back then I was doing the same kind of work, and somebody else carried the cost of getting it wrong. Now I carry it.
That turns out to be the part I like most.
Everything else turned out to be solvable
The website. Booking replies. Enquiries. Social posts. The confirmation email that goes to every guest.
All of it in English, all of it done by me, all of it through an AI assistant.
The workflow is duller than people expect. I write in Japanese first, always. Not rough English to be corrected — I write exactly what I mean, in my own language, with all the nuance I would use with a Japanese customer. Then it gets converted. The thinking is mine. The last step is not.
So the line is not where I thought it was
I had assumed the dividing line was English and not-English.
It is not. It is live and not live.
Anything written, where nobody is waiting on the other end, I can do. I can take as long as I need. Anything happening in front of me, in real time, I cannot — and that is not a vocabulary problem, which is exactly why studying harder was not the answer.
That distinction took me an embarrassingly long time to see.
How I check work I can't read
I do not ask my wife to check the written material. It would take too long, and there is too much of it.
What I do instead: I take the English that comes back, have it translated into Japanese again, and read that. Sometimes I put the same text through a different AI and compare. If the round trip says something I did not intend, I know something went wrong.
Is that the right method? I genuinely do not know. But bookings have gone up and guests seem happy, so I do not think it is producing anything badly broken.
I will also say that Japanese translation has improved noticeably between 2024 and now. The version of this that I am describing would not have worked three years ago.
Where it nearly went wrong
A few weeks ago I had to write the hardest message of the two years.
Our website takes instant bookings. A guest booked a date that should have been closed — one menu had been left open when the rest were blocked. Our error. Every nearby date was full. I had to write to someone whose travel plans I had just damaged.
I wrote it in Japanese first, as always. Then I had it converted. And when I compared the two versions, they said different things about the single most important point.
My Japanese said I needed to ask the guest to cancel from their side. The English said we had already cancelled and they needed to do nothing.
The English was not wrong as English. It was fluent and it was polite. It was wrong because I had not been precise about my own situation, and it filled the gap with the most natural-sounding version.
The bottleneck was never the translation. It was whether I had stated my own situation clearly enough to be translated.
When this goes badly, it will not be because of vocabulary. It will be because I was vague in Japanese, and the vagueness came out the other side sounding confident.
The part I cannot do
My wife interprets every class. Every session, in the room.
I do not think I could run this without her, and I have stopped pretending otherwise. The real-time half is hers.
What I did not expect is that this turned out well rather than badly. It is not a gap being covered. It is her part of the business, and she is doing something she likes doing.
Put simply: I get people to the door. She is the one who meets them. Neither half works without the other, and I am no longer interested in trying to do both.
What I would tell someone in the same position
Do not assume the language is the obstacle. Work out which parts of your business are live and which are not. The second group is probably larger than you think, and it is the group that has a solution.
Write in your own language first, every time. Produce meaning, then convert it. Do not try to produce English.
Be more precise than you would be with a compatriot. With a Japanese customer I can be vague and they will read the situation. An AI will not read the situation. Vagueness in, confidence out.
Decide where your hours go, and be honest about it. I am not studying English. That is a choice, not a failure, and I would make it again.
Work out what your actual job is. Mine is getting people to the door. It took running this alone to see it clearly, and it was a good thing to find out, not a disappointing one.
Why I stopped recommending books
I said I would come back to this.
After that first book I started reading again, properly, for the first time since I was a child. And I kept pressing things on people. In a restaurant, in a position like mine, that is easy to do.
Then I stopped.
What changed my mind was standing in a shop with a book someone had recommended to me, opening it, and feeling absolutely no desire to read it. It was not a bad book. It was just that the person recommending it and I were standing in completely different places, and nothing about their enthusiasm could move me to where they were.
I had been doing exactly that to other people for years.
So now I answer if I am asked, and I do not push. That applies to this as well. I am writing down what happened to me and what the numbers were. I am not telling anyone to do it.
One thing I am clear-eyed about
If the tools went away, or I lost access to them, this business would end. Not shrink. End.
I know that. I have thought about it. I do not have a second plan.
That is worth stating plainly, because people who write about AI tend to write about it as an improvement to something that already worked. For me it is not an improvement. It is the floor.
Satoru Nakajima runs Kyoto Green Cuisine, a home cooking class in Arashiyama, Kyoto, with his wife Akiko.
This post is part of a series about the business side of running it.
