Machine Translation (MT) refers to automated language translation. The concept has been around since the 1600’s but has come into its own beginning in the twentieth century. Along with the invention of electronic calculators came the development of ways to adapt computer technology to language translation of documents. Research became prevalent at universities in the mid 1950’s to develop and test machines to perform tasks previously only possible by human translators. A lot of research and funding was devoted to translating Russian into English by the United States government and Universities, related to the economic and political issues of the time. When progress slowed down, focus shifted to developing translation tools that could help people doing language translation by hand, such as electronic dictionaries and text processing devices.
In the 1980’s, the focus was on translating additional languages and developing different types of systems. Countries around the world made contributions and developed MT programs through research and collaborations. The increasing prevalence of computers and data processors helped to support the expansion of research and breakthroughs in the field.
In the 1990’s different methods were developed and used. There was a shift in the research from purely academic to applications used in business settings. An important development in the field of MT was growing success in speech translation. The introduction of the internet and subsequent formation of speech recognition software meant MT was becoming more widely developed and used around the world.
There are two types of Machine Translation: Rule-Based Machine Translation (RBMT), and Corpus-Based Machine Translation (CBMT):
1- Rule-Based Machine Translation is based on dictionaries and grammatical structure of both the source and target languages. The rule based approach began in the 1970’s. Examples include Systran, Eurotra and newer systems such as Apertium. RBMT includes several methods: direct (based on dictionary definitions of the words), transfer (uses language analysis), and interlingual (source data is converted to an abstract language) machine translation. The grammar rules and words in each language are matched in order to translate words and from the source language to the target language.
2- Corpus-Based Machine Translation relies on data analysis of a collection of electronic media in source and target language. As the database grows, the quality of the translation improves. This method uses probability and calculations to translate based on the bilingual text corpus of languages. Examples of CBMT are Statistical (uses statistical models based on bilingual text corpora) and Example Based Machine Translation (based on a data base of sample sentences from both languages and substituting the particular variables to translate additional text).
Anyone who has studied a second language understands the complexity of language translation. Unlike using computers and other machines for more straightforward tasks with one possible outcome, language translation requires expertise to convey the meanings of the individual words, but much more importantly, how the meanings can change in context or depending on the emotion of the speaker or writer. There has been vast improvement in the last 20 years, and now most computer users have tried tools such as Google Translate and other automated web-based translators. There are still many challenges to the field of MT and a lot of additional research and development opportunities for future breakthroughs that keep improving existing software, and developing new advances.
Hutchins, J. (2005, November 1). The history of machine translation in a nutshell. Retrieved July 15, 2014, from http://www.hutchinsweb.me.uk/CUHK-2006.pdf
Van der Meer, J. (2005, April 1). Different Approaches to Machine Translation. Retrieved July 15, 2014 from http://www.multilingual.com/articleDetail.php?id=1082