What is the new online translation tool?

The European Union (EU) has backed the development of a new online translation tool. The aim of this is that the building machine translation technology that produces reliable results. The organisation hopes the technique will also enable the creation of multiple accurate web documents.

Many machine translations rely on ‘learning’ from the information they receive. This means the translations they produce improve the more content they collect in a particular language. Unlike many of these tools, EU-backed initiative Multilingual Online Translation (MOLTO) has domain-specific grammar built into it. While it sounds complicated, the European Commission claims no training is required to learn how to use it.

It’s little surprise the EU is backing the project, as the institution currently has 23 official languages. As a result, translation services are integral to the organisation. While efficiency is obviously a key criterion for any machine translation being rolled out on this massive scale, this should not come at the sacrifice of accuracy.

Quality and efficiency

The goals the project has set for itself are certainly steep; to enable users to translate text from and into numerous languages, with high-quality results, in real time. It is built on the concept of semantic systems of grammar that are domain specific, and interlingua-based translation.

In the field of automatic machine translation at the moment, users have two choices. They can use open-domain tools such as those provided by search engines, but the results can be inaccurate. This means the information is unreliable. Alternatively, they can use a domain-specific high-quality translator, but this means spending money and the system can also be less than efficient. However, the European Commission hopes MOLTO will offer the best aspects of both these options.

Using MOLTO, web documents can be created in multiple languages automatically. Therefore, the results will be domain-specific and so better quality. Documents that are suitable for this form of translation and duplication include patent descriptions, encyclopaedia pages, e-commerce information, interfaces, manuals, and contracts are all documents that are suitable for this form of translation and duplication.

According to the Commission, MOLTO users do not have to be able to speak a language fluently to use the system.  They also don’t have to be a technology expert. As a result, it hopes that one day in the future people will download the tool for free and get the web documents translated into several different languages at the same time. “Ultimately, it could revolutionise automatic online translation, to the benefit of citizens everywhere,” the Commission claims.

MOLTO is just one example of the technological breakthroughs that are being made in the field of machine translation, particularly in improving accuracy and efficiency. However, Language Insight continues to urge caution on relying too heavily on these tools.

Choosing your translation

We believe that translation grades can be broken down into four categories: machine, gist, for-information and for publication. Machine translation can be very efficient but the results tend to be the least reliable of all the grades. Gist, for-information and for-publication translation should be produced by a human translator – and the for-publication grade is the crème-de-la-crème of these.

If you simply require a quick translation of a few words then the online translation is a useful resource. The results may not be completely accurate but you can use your own judgement to make sense of them.

Market researchers wishing to translate questionnaires they have had back from international fieldwork, meanwhile, may feel that a machine translation is suitable if they are just after the gist of the text they have in front of them. This may be the case, but similarly, they might interpret the translated results incorrectly and miss valuable information as a result. Similarly, a document you have no intention of publishing but need to distribute to colleagues working in your international offices could be roughly translated using the computer software, but you should be aware the reader may find the information confusing. If it is a patent description or a technical manual, for example, this could lead to errors.

Language Insight continues to watch the improvements being made in the field of machine translation. We believe in the future that technology will not only benefit businesses but translators too. Until then, it reminds users that they should carefully consider what type of translation they need. They should consider what it is being used for before making the choice between man and machine.