And how Google Translate works.
Users of Google Translate know that it isn’t perfect. Sometimes word meanings are mixed up, sentence structure isn’t quite right, or it just doesn’t quite sound like human translation. On the other hand, nobody can speak all of the world’s languages, and paying for a real-live person to do a translation job can be expensive and time consuming.
Earlier this week, many users trying to translate from Ukrainian to Russian were in for an unfortunate surprise. The “Russian Federation” was translated as “Mordor” (an evil country in the Lord of the Rings), Russia’s Foreign Minister’s name was translated as a “sad little horse,” and “Russians” was translate as “Occupiers.”
The issue has been fixed since Tuesday afternoon, but many screenshots have been circulating around the Internet.
SEE ALSO: Galápagos Tortoises Are Now on Google Street View!
“[N]ot all translations are perfect, and there will sometimes be mistakes or mistranslations. We always work to correct these as quickly as possible when they are brought to our attention” an unnamed spokesman from Google told BBC while explaining that the algorithms used in computer translation are very complex and depend on context.
As this video from Google explains, their translation software doesn’t learn like humans do. When people learn a language, they study vocabulary and grammatical rules to link words together into sentences. Computers, on the other hand, study millions of documents translated by human translators to find statistically significant patterns and learn rules to translate other documents.
Political conflicts stemming from Russia’s annexation of Ukraine’s Crimean Peninsula mean that Ukrainians don’t always speak fondly of Russians. Sometimes they do use terms like “Mordor” to reference Russia, and “Occupiers” when talking about Russians, so it is possible that the errors were simply reflecting derogatory terms actually used.
Regardless of how it happened, remember to be careful when using automatic translation services and have someone who knows the language read over the output whenever possible.
