UTL

Universal Translation Language

The evolution in the field of machine translation in the last few years has proved the collaboration between the man and the machine (particularly in form of pre-edition) to be one of the most promising ways for improving the efficiency and accuracy of the translation process. The UTL project intends to be the ultimate step on this direction.

The need for translation of documents, web sites, catalogs, etc, into several foreign languages is very high nowadays. If we want to translate any text into 10 or 15 languages, there are basically two possible methods:

1. Hiring 10 or 15 human translators to do the work (with or without computer assistance). Results are:
         - Very high quality
         - Very high costs

2. Using machine translation. Results are:
         - Poor quality
         - Very low costs

With UTL there is a third method available:

3. Using UTL translation (one human translator + machine translation). Results are:
         - High quality
         - Low costs

What is UTL translation?

UTL is a new method for translation which is under development, and involves the participation of one human translator and a machine translation software.

Current machine translation programs translate between natural languages (e.g. from English into Spanish). Results are poor because machines are not smart enough to understand natural languages. Machines make frequent mistakes when analizing the source text, and these mistakes are transferred into the target text.

UTL is an artificial language easy to understand for machines. If a human translator translates the original text into UTL, and gives this UTL version to the computer, the computer will be able to understand it and translate it correctly into other languages.

As a result of this method we can get 10 or 15 translations of top quality machine translation following just one hand-made translation.

How is the UTL language?

UTL (which stands for Universal Translation Language) is a human constructed language mainly intended for professional usage. It is based on Esperanto, an artificial human language whose suitability for machine translation (MT) has already been observed by other developers in the past. Any document written in this language can be machine-translated into natural languages with a high level of accuracy.

In this regard, UTL works similarly to so-called controlled languages (cf. controlled English, in any of its versions), which are a subset of a natural language intended for high quality MT. The advantages of UTL over traditional controlled languages are a wider scope of applicability and higher expressive capabilities. Traditional controlled languages are based on natural languages that have been deprived of many of their grammatical or lexical resources in order to avoid ambiguity. UTL, however, is a complete, non-clipped language capable of expressing virtually any concept, syntactic structure, verbal tense, etc., without limitations. UTL is prepared to make use of its own words and words from other languages as well, which, combined with UTL's powerful grammar, makes it a perfect tool to express any kind of content in any domain, in a machine-tractable form.

-Other pages on this web site:

 A more extensive description of UTL project (in .pdf format).

The UTL language.

 Some notes on UTL's syntax.

 Our links page.

NOTES:

   A parsing tool has been prepared which analizes correctly and unambiguosly a fair variety of UTL sentences, including complex ones. 

   The UTL project was presented in the MT Summit VIII, held in Santiago de Compostela, September 2001. The contribution ("Multilingual Authoring through an Artificial Language") is available at their site.

   (Later note: As of 2004, the author's work on UTL has been discontinued. However, the project remains open to any individual or institution who wishes to take over and further develop this approach.)

---·---