Machine translation is a field of computer science that focuses on converting written text or verbal speech from one natural human language to another. Research in this field began in the 1950s and has since advanced to the point that, as of 2011, several machine translation systems were available for public use. Completely accurate translation is very difficult to attain using machines, however, and many challenges in this field had yet to be solved.
The general technique used in machine translation is to convert words from one language into their equivalents in another tongue, using a computerized dictionary. Human languages are complex, however, and phrases can often have multiple meanings, so effective translation programs must take entire sentences into account. These programs must compute the meaning of each word. For instance, the word "book" can be a noun as in "he read a book" or a verb such as "he will book a flight." By analyzing context, programs can determine the best words for the translation.
Machine translation algorithms must also consider the grammar structure of the target language. As an example, for the English words "red shirt," the direct translation into Spanish is "roja camisa." Correct Spanish grammar places adjectives after nouns, however, so proper translation must follow this grammar rule and rearrange the Spanish result as "camisa roja." Grammar rules might be simple to apply when only two words are involved but can become complex and difficult for machine translation to process when entire sentences and paragraphs must be converted.
Despite advances in machine translation, computer programs in the early 21st century still were prone to mistakes. A technique to solve this problem is machine-aided human translation, in which professional human translators use the program. These professionals are aware of the limitations of machine translation, so they are able to find and correct common mistakes. Combining human and machine translation services allows large amounts of text to be converted both quickly and accurately.
