When translating files, you have an option to enable the feature Translation quality evaluation. Once enabled, the feature will run a quality check of your translation, segment by segment.
It will identify any issues based on different category and severity (see definitions below) and generate a report for the translated file.
The translation quality evaluation feature is not available by default as part of the Translation Flow, but it can be added as a beta feature upon request.
If you are a beta customer, you'll see an option to turn on translation quality evaluation for file translations. If you are interested in testing this feature, reach out to your Account Executive or Customer Success Manager.
It is also possible to test the translation quality evaluation tool independently from the Translation Flow.
Learn more about how to enable the translation quality evaluation feature in this article.
The feature won't automatically fix the identified issues, but flag them in a report including a short explanation. Based on the report, you can decide whether you would like to request a review of the translated file by another team member.
Supported languages
Translation quality evaluations are available for the following language combinations:
- English ⇄ German
- English ⇄ French
- English ⇄ Italian
- English ⇄ Spanish
- English ⇄ Japanese
- English ⇄ Korean
- Japanese ⇄ Korean
- Japanese ⇄ Chinese
Combinations including English as target language support both English (American) and English (British).
Combinations including Spanish as target language support both Spanish (Spain) and Spanish (Latin American).
Combinations including Chinese as target language support Chinese (Simplified) and Chinese (Traditional).
Supported platforms
The translation quality evaluation feature is supported on the DeepL website for file translations. To run a translation quality evaluation, the file needs to be stored via the Storage option via the toolbar on the file translations page.
Learn more about storing file translation in this article.
This feature is only available to customers whose contracts include Amazon Web Services (AWS) as a subprocessor. More information about DeepL's data infrastructure and sub-processors can be found here.
Translation error categories
Accuracy
- Addition: The translated text includes text that was not part of the original text.
- Omission: Content from the original text is missing from the translation.
- Mistranslation: The content from the translation doesn't match the content of the original text.
- Overtranslation: The translated text is more specific than the original text.
- Undertranslation: The translated text is less specific than the original text.
- Untranslated text: Content that should have been translated has been left untranslated.
Fluency
- Punctuation: Punctuation (e.g., periods, commas) is used incorrectly.
- Spelling: Incorrect spelling in the translated text.
- Grammar: Issues related to the grammar or syntax of the text, other than spelling and orthography.
- Inconsistency: The text within itself is not consistent.
- Character encoding: Characters are garbled due to incorrect application of an encoding.
Style
- Awkward: A text is written with an awkward style.
- Inconsistent style: The text uses inconsistent stylistic conventions across sections.
Miscellaneous
- Miscellaneous: Errors that do not fall into the other categories, but are nonetheless clearly identifiable as mistakes.
Translation error severities
Critical
Errors that make the content unusable or change its meaning a lot. For example when the meaning is completely reversed, when important information is missing and the main message is destroyed, or when the content is clearly not right for the intended audience.
Major
Errors that might mislead the reader or change the meaning a lot. The general idea is clear, but specific details are wrong, the terminology is not correct, or there are too many grammatical issues with the sentences that make it hard to understand.
Minor
Errors that don't make the text unreadable, but they can make it harder to understand and less appealing.
Ambiguous
A potential error was found but its severity cannot be determined due to missing context, such as an ambiguous original text, or unfamiliar internal terminology. We recommend a check by a subject matter expert in this case.