Read for meaning, not flow

Module 1 : What AI already does in translation

Plan du cours

Cette leçon n'est pas encore dans votre langue, la voici donc en anglais.

Petit test

Guess first, no penalty. A machine translation reads smoothly, with perfect grammar. What's the biggest risk?

MT (machine translation) tools are built to produce natural sentences. That's why their errors survive a quick read: a wrong sentence that reads well doesn't look wrong.

So read for meaning, against the source, line by line. Watch hardest for negations ("not", "never"), numbers, units and conditions ("only if", "unless").

Spot the changes

The English source of a patient leaflet: "Take one tablet in the morning. Do not take this medicine more than twice a day. Keep out of the reach of children. Stop taking it and call your doctor if you get a rash."

Trouvez les erreurs

Tap the two Spanish lines that don't match the source. Tap a line that's fine and you get a hint.

MT draft (Spanish)

0 sur 2 trouvées

A check prompt for the risky words

A prompt is the message you type into the tool.

Here is a source text and its translation: [paste both].
List every negation, number, unit and condition in the source.
For each one, quote the matching words in the translation, or write "missing".
Don't rewrite anything.

Complétez les trous

Fill the gaps so the check catches a reversed meaning.

Prompt

List every in the source. For each one, . .

Try it

Vos étapes, environ 2 minutes

0 sur 1
  1. 1.Run the check prompt on one translated text from your work, and fix anything marked "missing". Add the prompt to your post-editing kit.2 min

Exercices

Question 1 sur 2

You review a machine-translated paragraph that reads smoothly and has no numbers. Which check is most likely to catch a sentence that reverses the source?

Tuteur IA

Posez vos questions sur cette leçon et faites-vous guider dans l'exercice « Essayez ». Le tuteur fait partie de Learn Pro.

Voir Learn Pro