Criticism to Specification
School correction culture teaches you to spot every mistake and demonstrate the right form. In ordinary conversation not every correction improves the meaning; some only redistribute status.
If the meaning is clear and the conversation is not about language, constant corrections rarely help. They break the thought and shift attention from the content to the standing of the speakers.
It is particularly out of place to correct someone:
- in public;
- during an emotional account;
- in a language that is not their first;
- to display your own knowledge;
- when the mistake does not affect the meaning.
But a blanket ban on corrections is odd too. A language learner needs feedback. A mistake in a technical term can lead to the wrong action. A typo in an official document sometimes changes its legal meaning.
It is worth taking into account the task and the permission:
Would you like me to correct your pronunciation?
The exact wording matters here, may I check it?
There is an error in the document that changes the meaning.
Form matters. “I think a word is missing here” sounds different from “you’ve written it illiterately again”.
Language exists to convey meaning, but form carries information too. A mistake can be immaterial in conversation and critical in an operating manual.
What you correct is not the person but a specific text or skill.
The same principle works well beyond language: good feedback corrects a specific action, names the consequences, and does not turn an assessment of a result into an assessment of a person.
You explained a difficult thing clearly.
You prepared the meeting well.
I liked the way you resolved that conflict.
This is more useful than a general “you’re a genius”, because the person can tell which action is worth repeating.
Sometimes, though, what is needed is not praise but thanks.
“Well done” places the speaker above and issues a verdict. “Thank you, that saved me two hours of work” shows the effect of the help and treats the relationship as one between equals.
Public praise is appropriate if it does not embarrass the person and does not claim for the speaker the right to grade everything around him. A manager really can note the team’s work. A colleague can thank another colleague. A friend can be proud of a friend. Formal hierarchy is not required here.
Criticism is better delivered privately, unless the matter requires a public correction. A useful structure:
- the observable action;
- its consequences;
- the expected standard;
- a possible solution.
The obligatory sandwich of praise, criticism and more praise soon becomes transparent. After the words “you’re doing great work” the listener is already waiting for the knife.
Better to be direct and respectful:
The units are mixed up in the report. That makes the calculation unusable. Let’s check the table and add an automatic control.
Thanks get stronger with specifics:
Thanks to you we made the deadline.
You took off my hands a conversation I was dreading.
Your help let us verify the solution before launch.
Help does not always create a debt. In healthy relationships people sometimes do good without keeping score. But large favours are better discussed in advance, if one side may be expecting a particular favour in return.
Permission changes the status of a comment. The question “do you want a review, or have you already sent it?” takes a second and splits the situation in two. In the first case the person gets help. In the second he learns that his work has been graded and can no longer be fixed, and takes away only the second part. The same question saves the corrector time too: about half of all comments on an already-sent text are of no use to anyone, including their author. What helps a text that has been sent is not a review but a list of things to bear in mind next time.
A comment on a text that cannot be changed is not feedback but a report on one’s own powers of observation. The difference is usually obvious to the recipient and almost never to the sender.