With machine translation becoming more and more popular, many people are beginning to think that the computer will take human translators out of the picture. After all, who wants to pay for something that can be found for free online with just a few clicks? Computer translation would have been the perfect invention if not for linguistic and grammatical differences between languages.
Whenever I’m feeling tired or bored while working, the first thing I do is run some text in English through a translation engine, to see what it comes up with in Bulgarian. It’s mission impossible, and always good for a laugh. If you need to understand the gist of what a text says, machine translation might do. But it is unadvisable to show this text to anybody else or refer to it as a “translation.”
What makes it impossible for a computer to translate a text from English to Bulgarian?
- Bulgarian is more descriptive. For example, verbs like matter, backslide or mislay may require three, four or even five words to translate them into Bulgarian. It is necessary to use two words to translate reflexive verbs. Translation engines never seem to get this right. They usually just use the first word and the meaning is completely lost.
- Bulgarian is gender-sensitive. This means that we use different words if the subject is feminine or masculine. For example, if you say “Welcome” to a man, it will sound different from the “Welcome” you say to a woman.
- To make it even more complicated, objects have different genders. For example, a car is feminine, while a motorbike is masculine. In English, the use of “he” or “she” for inanimate objects is not very common, with a few exceptions like “ship”, which is feminine. The Bulgarian word for “ship” is masculine. So, translation engines go crazy and the translation gets mixed up.
- Bulgarian is number-sensitive. It is the same as with gender. An entire sentence can change if there are one or more persons involved. For example, if I replace “He went to the shop” with “He and his brother went to the shop”, I would also need to change “went” and use a different word. Computers often fail to do that.
- One English word can have different meanings in Bulgarian. For example, “go” has more than thirty meanings. This is quite confusing for translation engines, because they do not consider the context in which a word is used. A computer translates “It’s all gone” in English into “Everybody went home” in Bulgarian.
- A machine simply cannot translate idioms. This is my favorite point. There is no way to explain to a computer that there are no bananas in the translation of “went bananas” and no cake in “piece of cake”.
- There are deliberate mistakes. People who supply the data for the translation machines can be quite the jokers and drop translation bombs into the user’s hands. Translating “I am just asking” as “I am just @ss king” can be very funny, unless you’ve used a machine translation for an official text.
Computer translations can be useful and can save us time when we’re in a hurry, but we have yet to invent a machine that can translate as well as a human can. Until then, we can always translate the lyrics of that song we like so much with the help of a computer. For marriage certificates, though, we’d do better to rely on human translation.