Translating lists with forbidden words – a special challenge

Operators of discussion platforms are facing the challenge that certain posts and the language used in them are not wanted. For this reason, such posts should quickly be recognised, filtered out and deleted. Therefore, lists with unwanted words are created, which set off a kind of “alarm” and cause examination of the post or even automatic deletion if they occur in posts. If you have already tried to compose such a list yourself, you will quickly realise that it is difficult, if not even impossible, to create an extensive list without also putting words on the list which can be classified both as a “normal” word and as an “unwanted” word depending on the context and usage.

Operators of multilingual websites are now increasingly approaching us with the request to translate these created lists of words into another language. This does not sound that difficult at first; however, at second glance, it is often more difficult than at first assumed. Only looking up in the dictionary is anything but sufficient.

On the one hand, the translator suddenly has to know a “technical language”, which is not taught or systematically recorded anywhere and also always has to be up-to-date and possibly varies according to region (I’ll only say “swearwords” – who is able to name 1,000 swearwords on the spur of the moment?).

Any questions?
Just call us:

+49 89 67989548

On the other hand, there are also cases in which a platform operator would like to exclude more words. For example, a crowdfunding platform that considers some causes to be “worthy of support through donations”, but not others, and has therefore created a blacklist for various causes. On such a list of keywords, our team stumbled upon the following two English words, among others: “heart attack” and “horse”. We wondered what was reprehensible about a “heart attack” and a “horse”. On examining the list more closely, we came to the conclusion that “heart attack” was put on the list because of “attack” (and is to be rated as a “permitted word” in the combination “heart attack” by the system, unlike the word “attack” on its own), whereas “horse” is a synonym for “heroine” in a certain milieu. The term “heart attack” is therefore irrelevant in the German list and is to be deleted. The same applies to “horse”, but terms such as “bath salt” and “herb mixture” are possibly to be added instead (for those who cannot follow me: these are also code words for other products).

Thus, a lot of tact and thinking is necessary – often also a dialogue with the client since such a list can possibly not be simply translated, but partly also has to be developed directly in the target language.

And: Machine translation tools would definitely have failed with the list – they would just have chosen the first meaning from the dictionary.

Any questions?
Just call us:

+49 89 67989548