The goal of full post-editing is to make the most of the usable parts of the MT text, and at the same time, to make the translation linguistically correct, stylistically appealing terminologically accurate, and consistent.
This machine translation output comes from an engine that was trained by Linearis Translations. The engine has been trained for this specific language pair and domain of expertise using similar content.
MT suggestions in CAT tools
Any matches above 70-75% come from the TM(s) and are handled as in any project. Anything below a 70-75% match is machine-translated content and requires post-editing. If there is no existing TM leverage, the whole text may be machine-translated.
Post-editors can choose whether to opt for the MT match, a lower TM fuzzy match, or come up with their own translation. But the goal of post-editing is to identify usable parts of the MT text and build around these rather than ignore MT suggestions completely and start translating from scratch.
Recommendations for Full MTPE
— Make the translation linguistically correct, stylistically appealing, terminological accurate, and consistent.
— Beware of mistranslations which are common in engine outputs. Ensure that no information has been accidentally added or omitted.
— Edit any offensive, inappropriate, or culturally unacceptable content.
— Retain as much as possible from the MT. If it is correct and sounds natural in the target language, donʼt change it.
— Make quick decisions (3 to 5 seconds max per segment). If you decide that you cannot easily correct the MT output, please retranslate the text from scratch (or from the translation memory output).
— Make the target text fluent. If it sounds fluent and natural, itʼs enough.
— Meaning is a priority. Target text should be free from errors and inaccuracies in meaning.
— Pay extra attention to specialised terminology or cultural references.
— A quick glance through the edited segment will ensure no technical errors are present (things like merged words, double letters, articles, spaces etc).
— Ensure that the formatting is correct.
— Pay special attention to how acronyms and abbreviations are translated and use the TM search function to your advantage.
Suggested MTPE method
1. Global overview
Skim the text first in order to have a general idea of the content and detect recurring errors. Use the “Find and Replace” tool to change frequent errors automatically.
Read the target segment first, then the source, and check that everything is translated with no mistranslations, additions or omissions. Also carefully check the terminology.
3. Quality Control
Run an automatic or manual QA to check tags, consistency issues, punctuation, numbers, double spaces, spacing rules, dates, measurement units and glossary terminology. Also run a spelling and grammar check.
Please signal any repetitive MT issues or terminology issues to the PM (or in the feedback form that you might have received from your PM). This will help to fine-tune and improve the MT engine output for future projects.
✓ Correct grammar, syntax, semantics, and punctuation.
✓ Ensure that no information has been accidentally added or omitted.
✓ Ensure all tags are correctly placed.
✓ Use as much of the raw MT output as possible.
✓ Ensure that key terminology is correctly translated and that untranslated terms belong to
the clientʼs list of “Do Not Translate” terms.
❌ Hesitate too long over a problem.
❌ Worry if style is repetitive.
❌ Make preferential changes.
USE MT SUGGESTIONS IF:
✓ A large portion of the sentence is correct.
✓ The raw MT quality is very good with only minor corrections needed.
✓ The raw MT quality is not so good but still faster to correct it.
✓ The MT has the correct meaning.
DONʼT USE MT
❌ The raw MT doesnʼt make any sense.
❌ You need more than a few seconds to understand it.
❌ There are errors requiring rearrangements of most of the text.