2.3.2 Interlingual Machine Translation
The next generation of machine translation seemed to be able to overcome these barriers by introducing one more step. Interlingual Machine Translation Systems also analyse a text first. Now all the words and constructions are converted into a neutral language, the so-called inter-lingua. This language is designed for saving every information the system can find out analysing the text. For example, the interlingual system tries to extract the main idea or subject of a text by comparing words used in the source text with special dictionaries of different topics. The interlingua itself can be compared to the easy-to-learn and uniform language of Esperanto. The Interlingual System strategy is similar to the approach by which Latin texts have to be translated: Step by step questions like "What happens?", "Who is acting?", "When does it happen?" and much more. Knowing a lot about the text, the results and information are cached in the interlingua. The interlingua consists of structure, word and background information that can be extracted and detected by the source text. The translation system transfers these results in the interlingua into a new text in the target language. The essential advantage of this system lies in its ability to analyse even complex structures of source texts and its verb pattern recognition. Compared to Direct Machine Translation it does not need such an enormous dictionary because, for example, not every verb or adjective form has to be saved. This is needless because the system knows how verbs are conjugated and knows comparison. The interlingual system is the most popular one at the moment because it seems to be able to generate useful translations to understand a text of a foreign language or at least pre-translations which require only a few corrections.