Writing Feedback: Have AI Label Errors by Type Instead of Rewriting
Lin writes a short email to the host; AI labels errors by five categories without rewriting, Lin makes the corrections, and an over-correction that treats style as an error is caught.
Today's Goals
- Have AI mark only the location and category of each error, and do the rewriting yourself.
- Organise feedback by five error categories to find your recurring errors.
- Spot over-corrections that treat style preferences as errors, and keep your own voice.
On Wednesday 28 October, Lin practises scenario C5: a short email to Marta, the host in Valencia, confirming arrival and asking two practical questions. Writing is where AI help is most tempting and most likely to backfire. Paste a draft, ask "can you improve this?", and a polished email comes back in seconds. It is correct, fluent, and not really Lin's: a different tone, a different length, sometimes even different facts. Lin learns almost nothing from it, except that the assistant writes Spanish well. Today's approach is slower and works better. The assistant points to errors by category, Lin fixes them, and the assistant checks the fixes. Then Lin looks at what happens when the assistant is asked to "improve" the email anyway.
Plain-Language Walkthrough
Why a rewrite teaches less than a label
When an assistant rewrites your text, it does the most valuable part of the work for you: noticing what is wrong and producing the right form. You read the result and agree with it, and agreeing feels like learning. But next week, writing a new email without help, you will make the same errors, because you never had to produce the correct forms yourself.
A label works differently. If the assistant says "error in 'mi maletas': category E2, agreement," you still have to work out the fix. That small effort of retrieval and correction is what changes how you write next time. It is the same idea as answering before looking at the key: the work you do is the learning you keep. A good writing tutor rarely rewrites a student's sentence. They underline it, name the problem, and hand it back. Lin asks the assistant to behave the same way.
Sign in, then link your WeChat mini program account on the account page. Linked accounts share one free article per day. Unlock additional articles with rewarded videos in the mini program.
Scan in WeChat to open Dayward AI. Open the same course and chapter, choose the rewarded video, then refresh this page after completion.
Interview questions
Why have AI label your errors instead of giving you a corrected version?为什么让 AI 标注错误，而不是直接给出改好的版本？
Common in ChinaCommon overseasBasic#language#writing🔒 Unlock this chapter to see the answer keys.
What is the use of organising errors by category?把错误按类别整理有什么用？
Common in ChinaCommon overseasBasic#language#error-analysis🔒 Unlock this chapter to see the answer keys.
If AI rewrites a colleague email in a very formal tone, why not simply accept it?AI 把同事间的邮件改得很正式，为什么不一定要接受？
Common in ChinaCommon overseasIntermediate#language#over-correction🔒 Unlock this chapter to see the answer keys.