Can Translation Technology Offer All The Answers?

The conference of the UK’s Association of Translation Companies, an organisation of which Alexika has been a full member for over a decade, was held in the seaside town of Brighton this year. The event was extremely well organised and well attended, and many thanks are due to the organisers.

The question of whether technology can offer all of the answers for those buying translations was debated at length on the Friday afternoon. More provocatively – is the human translator heading for extinction? It has been a couple of years since we wrote on this subject so it is time to revisit. The ATC debate was excellent – contributors came from all over Europe and had vast experience of the language industry. Some had particular technical knowledge of the workings of machine translation technology such as the ubiquitous ‘Google Translate.’ To be blunt though, it was not a debate that changed minds – there was a consensus in the room.

Our views on translation technology remain broadly the following:

Translation is generally a post-graduate profession.

A post-graduate education and level of language skill are generally required to create a translation that reads as though it was written in the target language and uses the correct terminology for the subject area. A professional translator is then in a position to take advantage of the latest language technology.

The technology that is used by most translation professionals is generally referred to as ‘translation memory’ and includes ‘terminology management.’

This is now reasonably mature – the market leader, SDL Trados Studio (which we use ourselves – and train others to use – at Alexika), is currently celebrating a 30 year anniversary. At the core of this technology is software that builds a database of words as the translator works, so anything that has been translated previously will automatically pop up on the screen – this can save money on repetitive texts (so no need to translate the same thing twice) and improve quality through improved consistency. This technology gets better as the years go by, and also helps the management of the translation project process.

Machine translation tools such as Google Translate are a completely different technology, which is great for getting a quick and free gist of something in a unknown language.

Google Translate is a statistical based engine that is processing vast amounts of translated texts, and works out the statistical probability that a new translation will be the same as an old one. A more sophisticated branch of this technology is where a basic machine translation engine can be ‘trained’ in a particular field, again by processing lots of data. In some cases – such as large and repetitive texts – an acceptable translation can be produced cost-effectively by having a human linguist ‘post-edit’ the machine output. It is expected that this will be a competitive means of offering an acceptable level of translation for certain types of requirement in the years to come.

In a room full of professional linguists and translation professionals, no-one was of the view that the human linguist will not be required in future. But there was a consensus that technology continues to be a great help to language providers and clients alike.