One Simplified Character in the Middle of a Traditional Post
Last week, while reading a Traditional Chinese caption for a card news post, I found 「之后」 sitting where 「之後」 should have been. Korean readers barely notice, but Taiwanese readers spot it instantly. The cause was simple: the AI was writing in Traditional, and a Simplified habit leaked out mid-sentence.
The Fix: One Page of Slip Characters
1. Create one memo file. Something like zh_check.md is enough.
2. Write down ten characters that exist only in Simplified, with their Traditional counterparts.
3. Before publishing, have the review bot scan the text against this file.
后→後 里→裡 发→發 么→麼 这→這
们→們 说→說 还→還 过→過 时→時
The Instruction for the Review Bot
In the Traditional (zh) body, if you see any character from the left side of zh_check.md, tell me with the line number. Do not fix anything, just list them.
There is a reason for 「do not fix anything」. If you leave it to auto-correction, it may touch words that are already correct in Traditional, such as 「皇后」. The bot finds, and a human decides what to change.
Results
We applied it to 40 Traditional Chinese posts over two weeks. Slips found after publishing dropped from 7 a week to 1. The remaining one was a character not on the list, and once we added it the list grew to eleven.
Tip of the Day
A short list of only the characters you often get wrong works better than a long set of rules. Add one line each time a new slip appears. A non-developer can start with nothing but Notepad.