Translating with DeepL and other tools

Machine translation tools can help you communicate with people who do not speak your language well.
However, there are a few things you need to watch out for.
 

Important information

  • Even with the paid version, personal data may not be put into DeepL or similar tools for translation!
  • It is often helpful to note that the text was translated using a machine translation tool. In English, for example, “This text was translated using a machine translation tool.”

Typical examples of using machine translation | What is it good for? And what should be avoided?     | Correcting texts translated by machines | Contact
 

Example 1

You want to send an email to a student in a foreign language.
You have good knowledge (at least B1) of the foreign language.

Please note:
Emails first need to be anonymized (this includes names and birthdates but also many other kinds of information).

  • Good way to use machine translation tools

Example 2

You want to understand a text written in a foreign language for your own information. You use the tool to translate the text into your native language.

Please note:
Even if the translation is fluent, it can include content errors. You should take the time to review the content of the translation.

  • Good way to use machine translation tools

Example 3

You want to send an email in a foreign language.
You do not know the foreign language or your skills in it are weak (A1/A2).

Please note:
Machine translation tools can produce errors in content. If you cannot understand the translated text, then you cannot check to make sure there are no errors. You therefore cannot guarantee that the content of the email will be correct.

  • Caution is advised.
    For university purposes, only use machine translation tools in these cases if correct content is not important.
    If it is important that the content is correct, then you should either translate the text into a language that you and the other person both understand or ask the university translator for assistance.

Example 4

You want to translate / edit your academic article before submitting it to a journal.

Please note:
The free versions of translation tools use the data entered to improve the system. Please carefully consider whether you really want to provide your unpublished research data for this purpose.
In both the free and fee-based versions, the tools produce content errors fairly regularly, and these can often be quite serious.

  • Caution is advised.
    The university translator coordinates translating and editing jobs with a pool of external expert translators and editors who can guarantee that the content and language of your article is correct.
     

What is it good for? And when should it be avoided?

Without any additional editing
  • Texts for your own use if you only need to understand the basics of the content
    Example: Journal article in a language you do not know and you’d like to know whether it’s relevant for your work.
With additional editing
  • Informational texts if errors in content would not have any serious consequences
    Example: Often information leaflets, informational emails
  • For texts with a very tight deadline
    Example: Informational emails
  • Help with rewording for many types of texts
    Example: Individual sentences in your (research) work if you cannot think of a good way to write it in the foreign language
Not suitable
  • Texts that contain personal data  
    Example: Email about a student’s personal situation
  • Texts that need to be stylistically appealing
    Example: Marketing texts
  • Informational texts if errors in content could have serious consequences
    Example: Occupational health and safety texts

Back to the top

Correcting texts translated by machines

When a machine translates a text and a person edits the output, this is called “post-editing.” When post-editing, you need to pay attention to different errors than when you edit a text written (or translated) by a person.
 

What to pay attention to – real-life examples

Content errors

Machine-translated texts need to be reviewed very carefully for important errors in content. 

A text from early in the Covid-19 pandemic was written to describe what staff members who were in hotspots during spring break should do. They were advised to work from home.
 

OriginalMachine translation in English
"…wenn Sie nicht in Home Office arbeiten können, muss Ihr Vorgesetzter eine Freistellung in Erwägung ziehen.""…if you can’t work from home, your supervisor will need to consider taking time off."
[Back translation in German: Ihre Führungskraft muss überlegen, ob er/sie sich ggf. freinimmt.]

Content disappears

In a journal article, a table was translated. At first glance everything looked fine, but a closer review showed that an entire line was simply missing. This phenomenon has also been observed with relative phrases or even negations within sentences.
 

Incorrect references

Currently, machine translation tools still process each sentence separately. This means that references across or even within sentences can be incorrectly translated.
 

OriginalMachine translation in English
"Die Universität Hohenheim ist eine hervorragende Universität. Sie ist auch sehr schön.""The University of Hohenheim is an outstanding university. She is also very beautiful."

Inconsistent use of terminology

Today’s machine translation tools are not able to clearly separate terms from one another and translate them consistently without a glossary. This means that terms or names of institutions are translated in different ways within a text or, as in the example below, one English term is sometimes used for different German terms.
 

OriginalMachine translation in English

"Für die verschiedenen Bereiche bzw. Laboratorien werden vom Strahlenschutzverantwortlichen Strahlenschutzbeauftragte bestellt.

Für die Isotopenlaboratorien liegen Umgangsgenehmigungen vor, die beim Strahlenschutzbevollmächtigten, bzw. bei den Strahlenschutzbeauftragten eingesehen werden können."

"Radiation protection officers are appointed by the radiation protection officer for the various areas and laboratories.

Handling permits are available for the isotope laboratories, which can be inspected by the Radiation Protection Officer or the Radiation Protection Officers."

Contacts

Dr. Kelly Neudorfer

Web Editing and Development
Translator
+49 711 459 24498
E-Mail

Further information

  • Purchasing DeepL licenses from KIM: Email
  • Training sessions on using DeepL at the university (for administrative tasks or social media) are offered regularly and advertised on the F.I.T platform.