From fairly on in the story of computer science, researchers have been trying to build computer programs that could translate one language into another. While many other really difficult computing problems have been solved- they can now communicate at stunning speeds, distinguish one human face from a group of thousands of others, play chess as well as the best humans, and a do more than the early pioneers ever dreamed of- machine translation remains an open problem.
Even translating French into English is something that a computer alone cannot do. The two languages share common roots and both have a grammar and vocabulary which is perfectly well understood by everyday speakers and scholars alike, so why can’t computers?
The answer lies in the complexity of language. With the exception of artificial creations like Esperanto, languages grow up over many generations. They may follow a rough set of rules in most cases, but there are always special cases like irregular verbs, slang, and double meanings. Some words have a particular meaning within a certain context and another meaning otherwise. For computers, the lack of hard and fast rules causes all kinds of problems, and because computers cannot yet understand human language, they can’t make the logical and intuitive leaps that help human beings ensure that their spoken and written words make sense.
Although computers cannot perform translations like real people can, they are helpful. A translations service might make use of multi-language spell checking software, for example. One day a computer program might be able to translate one language into another without losing meaning or flow, but until then they’ll simply be used to help the process along.
0 Responses to “Computers and language”