English conversation after 18 years. Let's try studying every day by building a custom learning app with Codex.
I recently accompanied my daughter on her short-term study abroad trip, which gave me the chance to use English for the first time in a while.
The last time I spoke English on a daily basis was when I was 20 years old. I studied abroad in the United States and used English almost every day.
Since then, for about 18 years, I have had almost no opportunities to speak English.
Spending time in Cebu this time, I found that the words came out more easily than I had expected. I didn't have much trouble interacting at hotels or restaurants, and I was able to have normal conversations with the teachers at the language school. It seems I hadn't completely forgotten it.
However, my own feeling was that my English ability was about at the level of my first month in the U.S.
I can speak the bare minimum, but nuanced expressions don't come out easily. It feels like I'm managing to get my point across by replacing the content in my head with simple English.
Things like jokes or slang didn't come out at all.
I suppose that's to be expected if I haven't used it daily for 18 years.
It's a waste, isn't it?
I should have been able to speak better before, but I haven't used that skill for a long time. It doesn't seem like I need to relearn from scratch, but I feel like I could regain some of my intuition with a little practice, so I've decided to try speaking a little bit of English every day from now on.
Studying intensely doesn't last
While I want to regain my intuition for English, I don't intend to study for an hour every day or memorize a huge number of words. I'm not very good at pushing myself hard over a short period. I know that if I get too enthusiastic at the start, it eventually becomes a burden and I end up quitting.
On the other hand, I am good at continuing things that have a low burden every day.
The same goes for my notes, muscle training, and abstaining from alcohol; I have kept them up by incorporating them into my life without increasing the daily load too much. I believe my special skill is continuing things lightly every day.
For this English project, I'm not aiming to significantly improve my English ability in a short time. By speaking English for about 5 minutes every day and continuing that for a long time, I will gradually bring back the sense I haven't used for 18 years.
The problem is what to do in those 5 minutes. I thought about that this time through the implementation of an app.
Practicing 3 types of phrases every day
The first thing I thought of was a way to practice short English phrases every day.
The content will consist of the following three types.
Business English that can be used at work
Everyday English that can be used for travel and daily life
Reactions to make conversations a little softer
Initially, I limited the third one to jokes, but I'm not the type of person to crack witty jokes, so I stopped that.
Since I'll be dealing with expressions that can be used lightly in conversation, such as "I wish!" or "Story of my life," a name like "conversational reactions" seems appropriate.
For example, for business English,
Let me check and get back to you.
—that's all there is to an expression like that.
Listen to the audio, speak it a few times while looking at the screen, and then try saying it with the English text hidden. Finally, swap out some of the words to create a sentence you would actually use.
For instance,
Let me check the data and get back to you.
—if you do that, you can use it as-is in your own work.
The idea is not to memorize a single sentence as it is, but to learn it as a reusable pattern.
Don't just memorize the English sentences created by AI
You can start right away if you just have the AI come up with daily phrases for you.
However, you won't know if those English sentences are actually used often, if they are natural expressions, or if they are worth the time you are spending to memorize them right now.
Since you are putting in the effort to build this up every day, you want to avoid expressions that are rarely used or that are unnatural in conversation even if they appear in textbooks.
Therefore, I check learner materials from Oxford or Cambridge, English corpora that collect actual conversations, and select expressions that have a solid basis.
On top of that,
Is it likely to be used in my work or daily life?
Is it short and easy to say out loud?
Can it be applied by swapping out parts?
Is it not outdated?
Is it not a rude expression?
I narrow them down based on criteria like these.
Instead of letting the AI freely create phrases, I will have it take on the role of customizing selected phrases for my own use.
For example, I can use it to check the grammar of English sentences I've written, or to suggest example sentences that fit my work in management, data analysis, or marketing.
Rather than leaving the selection of phrases entirely up to the AI, I will use the AI to adapt material chosen based on reliable information to suit my needs.
The app for my personal use will be developed using sprint development.
Having thought this far, I passed the requirements to Codex and began consulting with it.
At the moment, I plan to build it in a format called PWA, which can be opened like an app from a smartphone's home screen.
Instead of building a large system on the cloud from the start, learning history and the example sentences I create will be saved on the smartphone itself. No login is required.
I will also keep the screens to a minimum, limiting them to the following four.
Today
Review
Collection
Settings
When you open the app, three phrases for the day will be displayed.
When you start practicing, you first listen to the audio and imitate it while looking at the English text. Next, you hide the English text and speak it, and finally, you choose from 'Could say it,' 'Could say it with a hint,' or 'Could not say it.'
Depending on the result, phrases you couldn't say will appear for review sooner, while those you said easily will appear after a longer interval.
I plan to have it show two new phrases and one phrase learned in the past, rather than making all three phrases new every day.
This way, I can increase my new expressions while also reviewing expressions I've started to forget little by little.
Initial test run with just 60 phrases
Even if I prepare hundreds or thousands of phrases from the start, I'll end up rewriting them anyway, so I'll start with the following 60.
20 Business English phrases
20 daily English phrases
20 conversational reaction phrases
Learning two new phrases per day covers about six weeks. I will actually use these 60 phrases to check if they are too difficult, if three a day is too many, and if the review intervals are appropriate.
I will ignore the days I missed.
What I want to emphasize most with this app is not letting learning tasks pile up.
In English learning apps, the number of consecutive days studied is often displayed prominently. It's fun while it continues, but once it breaks, motivation can suddenly drop.
This time, I won't make consecutive days the main metric. Instead, I will display how many days I practiced in the last 7 days and how many days I managed this month.
Phrases from days I couldn't study will not be carried over to the next day. I won't create a situation where 9 tasks are waiting for me when I open the app after a 3-day break. I will simply resume from that day's tasks.
I will also record separately the days I only listened to the audio once, the days I only spoke one phrase aloud, and the days I practiced all three.
I will prioritize being able to come back without quitting over doing it perfectly every day.
Do not include AI features from the start
Although it is an app built using AI, I have decided not to include AI features in the first version.
At first, phrase display, audio playback, hiding English text, recording and playback, review, and learning history are enough. Automatic generation of personal example sentences, English correction, audio transcription, pronunciation evaluation, notifications, and synchronization across multiple devices are postponed for now.
If I try to include all of these, building the app itself becomes a huge task. That would be putting the cart before the horse. I will use it for a few days with only the minimum features first, and then add features based on the problems I actually encounter.
I will adopt a style of adding only what becomes necessary later, rather than building features I'm not sure I'll use first.
Reinventing the learning method itself
What I am trying to do this time is not to complete one English learning app.
It is to think about the learning content, time, and management method that suit me, actually try them out, and remake them if they don't fit. If three a day is too many, I'll make it two. If five minutes is too long, I'll make it just one phrase. If there is too much review, I will reduce it.
Conversely, if I want to speak more, I will add 30 seconds of speaking at the end. Existing teaching materials and apps provide learning methods that are easy for many people to use. However, that does not necessarily mean they fit my life or personality.
This time, instead of adapting myself to the teaching materials, I will adapt the teaching materials and the system to myself.
I don't think I can bridge an 18-year gap, but I hope this will at least be the catalyst that helps me get to the point where my plans for traveling abroad start to take concrete shape.