Services publics et Approvisionnement Canada
Symbole du gouvernement du Canada

Liens institutionnels

 

Avis important

La présente version de l'outil Favourite Articles a été archivée et ne sera plus mise à jour jusqu'à son retrait définitif.

Veuillez consulter la version remaniée de l'outil Favourite Articles pour obtenir notre contenu le plus à jour, et n'oubliez pas de modifier vos favoris!

Rechercher dans Canada.ca

The ups and downs of online collaborative translation

André Guyon
(Language Update, Volume 7, Number 1, 2010, page 33)

Let’s begin with a short definition: collaborative translation is done by a group of individuals working together. In other words, coordination and discussions between participants are part and parcel of the translation process.

Despite what some technology gurus believe, collaborative translation and collaboration between translators are nothing new. For example, in the early 1990s, Translation Bureau professionals collaborated to translate the North American Free Trade Agreement in record time.

Also, translation suites, which allow translators to store sentences as soon as they are translated so colleagues can use them almost instantly, have been in use for many years.

Sometimes, suites have made it possible for translations to be delivered faster, with less effort, and at a lower cost, particularly for documents that have undergone multiple revisions prior to final delivery.

More recently, for the translation of very large projects, the lines have blurred between teamwork and community work. Mozilla’s Firefox browser and the online free encyclopedia, Wikipedia, for example, were translated by volunteers. Before the Internet age, collaborations and community involvement were local rather than global.

Huge community projects provide volunteers with a tolerable minimalist interface to translate or revise one or two paragraphs on occasion. The work always gets done in spite of the rudimentary technology that is used, which bears no resemblance to the tools normally used by translators.

Paradoxically, translating a Wikipedia article can take you on a voyage from seventh heaven to the fires of hell. Quite often, the process is better and more rewarding than are many paid translations. However, the work interfaces are a huge step backwards, with each enhancement to an article (tables, hyperlinks and complex formatting) adding time to the translation process.

The commitment required of Wikipedia volunteers is quite similar to the requirements of codes of ethics generally adhered to by associations of translators. The process even includes the expression of doubt and the possibility of having experts validate passages where the translator thinks that such an opinion is warranted. Salaried employees and freelance translators can only marvel at such a wise approach. Volunteer translators are asked to fill out a translation tracking file, etc.

However, I have saved the best for last! The instructions indicate clearly that the object is not just to produce an accurate translation, but also to deliver a good article that will appeal to the target audience. In other words, the translator is urged not to shy away from adapting, lengthening or shortening the text as required. How many translators can only dream of such freedom when their contract requires strict accuracy?

In short, in the non-hierarchized collaborative environment that is Wikipedia, there are almost as many control mechanisms as there can be in the workplace, but with far more freedom of expression.

Alas, after the exhilarating joy of reading these rules and suggestions, the translator must return to a reality that is not as pretty. The working interface was never really designed for translation. It’s a little like a flashback to the old days when we had to edit texts in HTML by hand. Try inserting a hard space in a wiki article, for example, and you will see how much work is involved.

In addition, these sites offer almost none of the tools translators are used to working with, such as spell checkers, grammar checkers, translation memories and definition and terminology management tools.

However, some recent environments, such as Google, offer the opportunity to use glossaries, translation memories and machine translation, all shared in real time. The Google environment supports any text, but the company reserves the right to use anything put into its interface.

The tools provided by Google can now be used to translate material such as Wikipedia articles, which should lead to substantial advances in the amount of content available in different languages.

The other side of the coin

Collaborative and community translation also has its drawbacks.

For example, for-profit corporations do not hesitate to cloak themselves in humanitarianism to convince the masses to translate their Internet products for free. Their “line” goes something like this: “We’d love to be able to deliver our product in your language, but unfortunately we don’t have the means to do so. However, we can help you do the work for your linguistic community if it’s really important to you.” Sometimes it works for them.

Today, associations of software or e-commerce merchants are trying to figure out how to convince the general public to translate for free or for next to nothing. They sometimes claim that their goal is not to save money, and that translation done by users is of higher quality and better suited to their needs. The argument is clever, but when they talk amongst themselves, the focus soon changes to all the money they are saving.

Fortunately, we can see that not-for-profit projects, like Wikipedia or Firefox, have much higher‑quality translations than many of the products of multinational corporations masquerading as subsidiaries of the Red Cross.

For better or for worse, tools that facilitate collaborative translation have emerged. Nothing prevents translators themselves from taking an interest in and benefiting from these tools, even though they are far from being on the cutting edge.

All indications are that the line will blur gradually between collaborative tools designed to translate the occasional paragraph and professional tools designed for ongoing work and for processing large numbers of texts.

For example, there is nothing stopping anyone from integrating open-source translation memory management software in whole or in part into a wiki or blog or modifying the text editor to provide an environment in which it is possible to translate for more than an hour without becoming woozy.

There is also no reason not to include an import-export format in a wiki or other collaborative platform so translators can work in their accustomed comfort and submit pages or paragraphs via the collaborative tool.

This solution is without question what most translators would favour. Developers, on the other hand, tend to prefer integrating translation functions into the interface. This is normal; everyone prefers to have a home field advantage.

Collaborative translation tools are always web-based (nothing to install, simply open your browser and go to a URL such as www.somethingorother.org). Several people can share the task of translating a text and thus deliver the product in less time than if they worked alone, all without having to leave home, even if they do not live on the same continent.

Though far from perfect, wiki-type interfaces do have some definite pluses, such as real-time management of the volume of text translated and automatic division of work.

In essence, rather complex mechanisms have been created to allow a very large number of people to translate large volumes of text. Nothing prevents language professionals from taking advantage of this. Access to some platforms is free, and on the horizon there are platforms that will not only be free of charge, but free of ties to corporations.

If we were to use all the benefits that today’s web-based tools, with their collaborative platforms, have to offer, we would have a system that would enable us to share a text of unlimited length among an unlimited number of translators and revisers. After a job was completed, the system could provide statistics such as how many words each translator translated or how much time each translator spent doing the work.

We would have revisers starting work minutes after translators instead of hours later. Peer review would be almost instantaneous, and validation of questionable parts by peers would be done by experts in the field.

There would be a free translation memory updated in real time, with a tool to record translation difficulties on the fly and create minimalist terminology records. Clients would even be able to add notes during the translation process, indicating any concerns or reservations they have about certain choices made by translators.

You never know, the future may be nearer than you think. Many initiatives are already underway, some of which include a great deal of openness and sharing.