Cover lesson · MFL · Year 9 · Advanced

Reviewing online translations

DCF: Producing → Evaluating and improving digital content

Name: Class: Date:

Cover teacher — you need no French

Print: this sheet, one per pupil, and the stimulus sheet "Élise's email, and what the machine made of it", one per pupil. The stimulus carries the French, the machine's English and a glossary; the worked example carries the full answer key. No computers needed — and this lesson is better without them, because the point is to judge a translation rather than generate one.

Run: 5 min read both texts · 5 min learn the four error types in the table below · 20 min complete the error log — nine numbered passages, but only eight are wrong · 12 min write the clean translation · 5 min the judgement question · 3 min swap and check.

Collect: this sheet from every pupil. The error log and the clean translation are what the class teacher marks.

What you are doing

Free translation tools are good enough that the mistakes they make are the hard ones to spot: real English words, in real sentences, meaning the wrong thing. You have a correct French email and a machine's English version of it. You will name and fix every error, produce a clean translation, and then say when you would and would not trust a machine with this job.

The four error types

Type	What it looks like
False friend	A word that looks like an English word but means something else. Librairie is a bookshop, not a library.
Literal idiom	An expression translated word by word so the meaning falls out of it. Il pleut des cordes is not "it rains ropes"; it is "it is pouring down".
Wrong register	The right meaning at the wrong level of formality — a friendly text turned into a business letter, or the other way round.
Grammar carried over	French structure left standing in the English: an article that should go, a tense formed the French way, a name never translated.

Step by step

5 min — Read Text A, then Text B. Do not mark anything yet.
5 min — Read the four error types above and the glossary at the bottom of the stimulus sheet.
20 min — Complete the error log. One row per numbered passage. One of the nine is correct — say so and say why, rather than inventing a fault for it.
12 min — Write out the whole email in clean, natural English in the space provided.
5 min — Answer the judgement question.
3 min — Swap. Check your partner spotted the correct passage and did not "fix" it.

Error log

No.	The machine's English	Error type	What it should say	The French that caused it
1
2
3
4
5
6
7
8
9

Your clean translation

The whole email, in English a friend would actually write. Keep Élise's tone: she is fourteen, she is writing to someone she likes, and she is slightly stressed about a deadline.

Judgement question

A classmate says: "There is no point learning this, the translator got most of it right." Answer them in three or four sentences. Say what a machine translation is genuinely useful for, what it must not be used for, and how you can tell the difference without already knowing the answer.

Success criteria

I find all eight errors and I leave the correct passage alone, saying why it is right.
I name the type of each error — false friend, literal idiom, wrong register, grammar carried over — rather than just marking it wrong.
For every error I quote the French word or phrase that caused it, so my correction is evidenced and not guessed.
My clean translation reads like natural English at Élise's age and keeps her friendly tone throughout.
My judgement gives a test somebody could apply to a translation of a language they do not speak.

If you finish early

Two mistakes in Text B are not numbered. Find them. One is a false friend and one is an error that appears twice.
Translate your clean English version back into French without looking at Text A, then compare. What did you lose, and is it the same sort of thing the machine lost?
Write three sentences in English that you think would defeat a translator — an idiom, a joke and something that depends on knowing who is speaking — and say why each one is hard.