Start a new topic

Basic principles for Post-editing Machine Translations

Can someone give me some basic principles for post-editing machine translations?

Where to stop?

Customers are increasingly asking to offer PEMT, where Interactive MT or On-The-Fly MT is a much more sensible choice:

(1) With IMT you can use your own glossary of preferred terms. This ensures consistency and the use of the correct technical terms.

(2) IMT allows automatic conversion of punctuation marks (quotes, dashes, etc.) to your favorite punctuation marks.

(3) IMT does not suffer from interfering markup (tags) in the middle of words, translating words into parts and glueing the results together–resulting in nonsensical translations.

(4) With IMT, you benefit from previously translated segments: in subsequent segments you do not have to 'set things straight' for the umptieth time. Towards the end of the project, your amount of work decreases, especially if you feed your glossary along the way.

(5) With IMT you can have multiple MT systems make suggestions in parallel and choose the best.

(6) ...

I think that CafeTran Espresso is very well suited for IMT :).

Seems like this important topic doesn't interest other users of CafeTran Espresso, or at least, nobody bothers to discuss it.

Not a problem.


De roepende in de woestijn.

BTW, not much of interaction about this topic on Proz, either. Whereas in the Dutch Translators' Coffee Corner a vivid discussion followed. 

Login to post a comment