What is Machine Translation?

Machine Translation (MT) is the technology that allows you to input text in one language and have a computer translate that text (from its own dictionary resources) into a different language. MT's goal is not a perfect translation, which no computer can give you; rather, its purpose is to give the user as good an idea as possible of the meaning of what he is reading.

What Machine Translation ISN'T
The goal of MT is not to translate text as well as a human translator could, but to offer the non-English-speaking computer user a solution to the lack of software translated into their own language. Although it can't produce a perfect translation, MT will at least give the non-English speaker the gist of what is written, which is far better than no translation at all.

Background

Research and development in the field of MT began nearly 50 years ago and has encountered many obstacles, including many years when all research stopped due to lack of interest or funding by any government. With time, increasingly sophisticated computers have been programmed to mathematically decode text from one language to another. However, the problems of word ambiguity, irregular syntax, and context-dependent multiple meanings of many words still make high-quality machine translation impossible.

Reliability

How reliable is machine translation? That depends. If the text to be translated is very complex, it is certainly much more difficult for a computer to translate it clearly. But even simple sentences containing words with double meanings can lead to incomprehensible translations. Unfortunately, we can't realistically expect machine translations to be very accurate, but rather to be an acceptable alternative to text that is otherwise impossible to understand.

Where MT Works Best

MT works best within narrow contexts. Computer application menu items and commands, for example, are fairly standard and can be translated with a high degree of accuracy. Dividing the whole dictionary of a given language into specific areas that use specialized vocabulary can go a long way toward making MT more reliable (see "Dictionaries" below).

Dictionaries

The most reliable MT programs offer several restricted dictionaries, or sets of vocabulary -- for example, medical, literary, or electronic contexts -- which the user selects according to the nature of the text to be translated. Thus, certain words used in a scientific context, for example, would always be translated in a certain way, while in a different context, the same word might well be translated differently. This system raises the level of translation accuracy significantly, so that if an MT system knows that the text to be translated involves the travel agency business, a "card" is probably a credit card, not an ace of spades.

Conclusion

Presented with text in a language we can't understand, we'd all be happy to have instant accurate translation available on demand. Unfortunately, this is still a dream. Instant translation must be done by imperfect machines, while human translation both takes longer and costs more. LingoWare offers users a compromise that works well in a difficult situation: fast, free machine translation that breaks through the language barrier and makes the world's software accessible to everyone.