Home » Articole » EN » Business » Translation » Computer-assisted translation » Machine translation » Approaches and evolution in machine translation

Approaches and evolution in machine translation

machine translation


Direct approach:

  • word for word translation of the source text to the target text
  • reformulation of the word order in the target text translated

Interlanguage approach:

  • construction of the interlingua representation of the source text
  • construction of the target text representation

Transfer approach:

  • lexical and syntactic analysis of the source text
  • transfer of translated lemmas and target language structures

Proposal: Sequential (or progressive) approach:

  1. First: Automatic correction of the source text (1 → 2)
  2. Then: Detect idiomatic expressions in Source Language in Source Text > interpret them in Target Language > retranslate the word-for-word in Source Language > replace in the Source Target (2 → 3)
  3. Then Detection for Interpretation (Source Language → Target Language + Translation Word for Word → Source Language) and replacing only proper names in the Source Text (3 → 4)
  4. Then: syntax reorganization in the Source Text (4 → 5)
  5. Finally: Translation of Source Text word for word (5 → Target Text)


In late October 2005, the press announced significant improvements. The International Center for Advanced Communication Technologies, run jointly by Carnegie Mellon University of Pittsburgh and the University of Karlsruhe in Germany, were unveiling a computer system for instant translation. A Chinese student, wearing eleven electrodes on the face and throat, gave a speech in his language which was simultaneously translated into English and Spanish. Researchers report concluded that “The results are not perfect” and that “it may be difficult sometimes.” In fact no item indicates that the German and American journalists were able to discuss with the Chinese. Journalists have also generally failed to mention that, when Dr. Waibel announced it would take the issues of German and American journalists, the computer has heard something like: “So there we glycogen alternating between questions Germany and America. “. This type of translation software will also require improvements in speech recognition, including blurred by ambient noise and poor pronunciation.

The current translation devices are useful for limited situations, such as to reserve a room in a hotel. “If I go to Beijing, I can go down to the Hilton without any problem,” said Stephan Vogel, a Carnegie Mellon researcher.

Existing products and services

Free online translators: They allow to translate words or texts of limited size.

Languages generally available to the translation: German, English, Spanish, French, Modern Greek, Italian, Dutch, Portuguese.

It is also found in translations in English: Arabic, Traditional Chinese, Simplified Chinese, Korean, Esperanto, Japanese, Russian, Swedish

Free software:

  • GPLTrans
  • Linguaphile
  • Traduki
  • Apertium
  • GoldenDict
  • WEATHER System
  • myMT


Various projects seek to improve the performance and speed of translation engines (including Google), hoping one day to overcome the “language barrier” and translate in real time (as was imagined by many writers of science -fiction) a foreign language, or even be able to “communicate” with some animals via a voice processing software.

  • An ongoing project in Japan is to equip a mobile phone with an automatic multilingual translator. This project aims initially to be displayed on the phone screen translation of simple sentences and words spoken in Japanese or other languages in seconds and autonomously, that is to say without depending on a server.
  • Google announced in early 2010 almost instant voice translation application (speech-to-speech translation)
  • Advances in machine translation will likely be in conjunction with those of the Foreign Language Teaching computer.

Leave a Reply

Your email address will not be published. Required fields are marked *