Machine translations (MT) are readily available today. The systems are fast and inexpensive, and the results are often surprisingly helpful. However, it is also extremely important to consider certain aspects, such as: Does the MT system protect personal data sufficiently? What mistakes do such MT systems often make? And under which circumstances should I not use a MT at all?
Important:
Always label a machine translation so that readers can adjust their expectations accordingly: "This text has been machine translated."
When can I use machine translations?
For some situations, machine translations are a very good option:
- To get a basic understanding of a text written in a foreign language (e.g. paper, article)
- For a quick translation where quality is not important
- If you have a good knowledge of the target language and can correct any serious mistakes
Examples:
- Everyday communication and information (emails, invitations, newsletters)
- Content that is quickly outdated (e.g. website announcing an event)
- Social media posts (e.g. Instagram, X)
When do I need a human translator?
Do you need a creative text that is tailored to a specific target audience or situation? Should corporate language be used in your text? Is data protection an issue? In these and other cases, please contact our translation team for assistance.
Examples:
- Prominent, eye-catching texts such as flyers, brochures, websites, speeches
- Texts with personal data (free online tools save everything!)
- Contracts, legal texts
- Texts in bureaucratic language (e.g. travel expense reports)
- Texts using university-specific wording
- Texts with recurring terminology, as machine translation tools do not always select the same translation for the same term (lack of consistency)
- Texts that require a precise, high-quality translation
- Texts that leave room for interpretation and include idiomatic expressions and nuances
Frequent errors in machine translation (MT)
Machine translation (MT) saves us time and often produces good results. However, it is essential to keep the following in mind:
Machine translations always need humans to double check them
- At first glance, a translated text may sound great. As a result, you may miss mistakes if you don't check it thoroughly
- MT does not notice errors or inconsistencies in the original text
- MT does not have intercultural communication skills
- MT does not consider that additional information may be necessary for international readers
Unclear choice of terms
- MT sometimes chooses a translation that is unsuitable for the context (MT selects the translation option that is statistically the most likely – basically, the system guesses)
- MT translates the same terms differently in different sentences (confusing for readers)
- MT does not use corporate language (internal style guidelines and glossaries)
Technical weak points
- Incorrect sentence structure
- Incorrect or missing gender
- MT omits terms it is unfamiliar with