Why is 'You're taking my hair' a mistranslation? — Translation
Swap the words, and you get a sentence. Carry the meaning, and you get understood.
1 · Think first
Translating isn't swapping words one at a time. It's grabbing the meaning of the whole sentence first, then building that meaning again in another language.
Old translators swapped in words and short chunks like a dictionary. Today's AI, a lot like yesterday's summarizing, lets the words in a sentence look at each other to grab the meaning first.
In 2016, Google first put this method into a real translation service (Chinese to English came first). That's when clunky word-for-word translations dropped sharply.
2 · Experiment · Before & after
Try it — Word-swap translation vs. meaning translation
We translated the same Spanish sentences into English two ways. Switch back and forth between the two, and you'll see where it falls apart.
Figure
Three steps of translating the meaning
Notice there's no box for swapping words at all.
3 · Quiz
Quick check
Three questions. Pick an answer and it's marked right away, with the reason why.
Today in one line
Today in one line
Swap the words, and you get a sentence.
Carry the meaning, and you get understood.
Learn more
- Google's official post saying it had put neural translation into a real service — A Neural Network for Machine Translation, at Production Scale (Google, 2016)
- The technical paper explaining that translation system — Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation (2016)
- The paper that, instead of squeezing a sentence into one lump, let the model pick where to look — where attention got its start — Neural Machine Translation by Jointly Learning to Align and Translate (2014)