The New Scientist has recently published an article about borrowing words to allow machine translation into lesser-known tongues. Computer aided translation can make French, Spanish or even Japanese accessible to English speakers, but it requires similar documents in each language to learn how to translate. For rare languages, finding the reference material is a hard task. Therefore, a software that borrows words from existing tongues could be a great help to translate into little-used languages.
Luis Leiva and Vicent Alabau of the Polytechnic University of Valencia created a system called Culturally Influenced Interlanguage (CI2). CI2 exploits the similarity in words and grammar found in language families like the Romance languages. This system translates items into minority languages by borrowing words from languages in the same family. The result is not grammatically correct, and it may contain spelling mistakes, but to a minority language speaker the meaning should be understandable.
CI2 has been tested by asking 17 native Spanish speakers to read a selection of Swedish sentences translated into CI2 Spanish via Italian, French and Portuguese, along with direct translations in all three languages. The volunteers said that the CI2 text was easier to understand than the Italian and French versions, but the Portuguese translation was clearer still. In May, Culturally Influenced Interlanguage will be presented at the Conference on Human Factors in Computing Systems in Austin, Texas.