Google recently launched a Translate Community, inviting language-lover volunteers to help improve the accuracy of Google Translate services.

At first sight, this seems like a win-win situation, as long-suffering Google Translate users have often voiced their frustration with the quality of the translations rendered, while Google gets free knowledge (and labor) from a virtual community of volunteers who speak more than one language. (This is another issue: Google is a multinational corporation operating with billions of dollars annually; one would think that it could afford to pay professionals to do the job right!).

Is All that Glitters Gold?

Nevertheless, it might make sense to heed the old adage, “If it seems too good to be true, it probably is” because – as anyone who works in translation knows all too well – speaking is not translating, any more than walking is dancing.

Together with all the usual issues associated with crowdsourcing (susceptibility to malicious attacks, quality of work in general, reliability of contributors, problems with communication, and so on), there are drawbacks that are specific to collaborative translation/editing on a massive scale. This can be further complicated when editing a machine-generated translation, i.e., one that from the beginning lacks the human translation of its meaning. It seems rather like weekend DIYers building a skyscraper on a foundation of sand…

The Importance of Human Expertise

No matter how sophisticated, to date no machine translation system has been able to approach the professional translator’s sensitivity to language variant, context and register, essential elements in any good translation…and something that language-lovers volunteering their time and effort cannot be expected to contribute with their efforts, no matter how well-meaning or dedicated they may be.

The Pareto principle (aka the 80-20 rule) tells us that, for many events, about 80% of the effects are generated by about 20% of the causes. It remains to be seen if the Google Translate community will have enough knowledgeable volunteers who are active and expert enough to provide the accurate information needed to guarantee translation accuracy in sufficient volume, and how this will balance out with the contributions of those who are not.

A Bad Translation is Worse than No Translation at All

One of the most critical issues, however, might actually be one that is not commonly addressed: the illusion of reliability. Most translators have dealt with clients who want them to compete pricewise with machine translation (which is, of course, impossible) and have found themselves having to explain the pitfalls of Google Translate and other machine translation services in order to justify their rates. Some clients have finally “discovered” (perhaps after a negative experience!) that machine translation is not suitable for use in any but the most limited of circumstances, but it remains to be seen whether this new project to improve Google Translate services will lend a false illusion of reliability to what will nevertheless remain nothing more than machine translation.

Will it be possible to rely on the accuracy of a machine translated document by the year 2029? Will human levels of translation quality be reached by machines and software programs?

Translation and interpreting services have been traditionally considered human activities with little to almost no space for technical interventions. However, technical developments in language and translation software have led translators and interpreters to assume that their job will be highly influenced by machines by the year 2029.

Language is a living entity. It’s much more than a collection of isolated words and expressions. Each language embodies a cultural background, cultural concepts and a certain level of subtlety that even the most accurate and highly trained translator cannot translate perfectly.

In the video below, Mr. Ray Kurzweil –a well-known inventor, author and futurist- points out that translators and interpreters should embrace language-related technology advancements as a means of expanding their translating and interpreting abilities. No machine will ever be able to capture the subtlety in all languages as many expressions simply cannot be translated isolated and without context. He adds that even though there’s a natural resistance against translation machines, the truth is that the translated documents, both verbal and written, that they produce tend to get better and sound more natural over time. He accepts that machine translators may not be useful to translate romantic sentiments or more poetic forms but they are actually adequate for translating business discussions and everyday conversations.

According to Ray Kurzweil, when these technologies are first introduced, they tend not to work very well and people tend to dismiss them. They are perfected; they improve their performance and sneak up on us, and even though they seem revolutionary, they’ve been around for years already.

However, Kurzweil is cautious and emphasizes that translation technologies will not replace human translators and interpreters. By the year 2029, machines will be able to provide human levels of translation quality in certain type of translations and in certain translation fields but people’s need to learn foreign languages in order to enjoy and understand a literary piece in its source language or the need to rely on a professional translator to understand the meaning of a poetic writing will not be altered.

In a globalized world, accurate translation services are in great demand even if the economic context is not the best one. Translation companies can and should take advantage of translation technologies as they become available as they are useful tools that help them be ready for globalization and provide their clients with expanded language services.

Ray Kurzweil on Translation Technology from Nataly Kelly on Vimeo.

Amara is a large source of non-profit, crowdsourcing translations. The platform was launched by the creators of YouTube to translate as much video content on the web as possible, into as many languages as possible, as quickly as possible and as cheaply as possible.

What are the positive aspects of crowdsourcing translations?

More and more people worldwide choose to watch documentaries, films, TV series and other kinds of video content via the Internet instead of on television or via cable. The difficulty lies in not always being able to understand the languages spoken in online videos or in not being able to hear the content of the videos if you happen to be deaf or hard of hearing. Amara is a platform which hopes to eradicate these issues through crowdsourcing translation and so far it has proven itself to be very, very successful.

On average, Amara can translate and upload captions onto any film in 22 different languages within 24 hours. This effectively means that, in an incredibly short space of time, the content of that film can be watched across the globe and be made accessible to the deaf community or the hard of hearing. The most astonishing factor of Amara’s success is that the people who translate are volunteers – they’re people from the online community who simply sign-up to the platform and start translating in languages that they speak.

What are the possible problems faced by crowdsourcing translation platforms?

One of the biggest feats involved in the management of video content online, however, is the sheer volume involved. Translating online videos into 22 languages in the space of 24 hours is an impossible task for any ordinary translation company to take on, particularly when taking into consideration that all translators in translating companies are paid for their work. Amara recruits translators from all over the world for free, simply by reaching out to a community of online video-content enthusiasts who are only too happy to help when it comes to making internet content available to all.

One of the main issues with platforms like Amara is whether or not they are sustainable. If the Amara community begins to dwindle and volunteer translators stop translating at any point, the system will fail. In addition, there are lots of measures which have to be put in place in order to check the quality of the community translations and these measures require time, manpower and monetary investment. Amara’s philosophy is, without a doubt, a positive step forward for globalization and for the deaf/hard of hearing community, but the maintenance of the platform might prove to be unmanageable in the years to come.

What do qualified translators have to say?

Most translation companies and freelance translators who hold high-quality translation certificates show little support for platforms like Amara for obvious reasons. The idea that “anyone” can produce trusted, quality translations through crowdsourcing undermines the skills, qualifications and experience of professional, paid translators. It would appear that few people would argue with this point.

Crowdsourcing translations are no real match for the quality of paid translations by translation experts, but the translation of online video-content is such an overwhelming huge task that in some cases online users would argue that some kind of translation is better than no translation at all. Amara gives the deaf and hard of hearing community access to more online video-content from all over the world than ever before. The positives of this fact are indisputable.