Can language students be trusted with commercial translation tasks? You probably heard of the $15M Series B financing to fund a web startup that is promising to use language students along with an online crowdsourcing methodology to disrupt the translation industry.
The new startup develops a language training platform for users to learn new languages for free. Part of their learning involves translating sentences from and into the language they are learning. Translations that correlate among students will then be sold to translation buyers.
To paraphrase a colleague, this is like taking differential equations problems and outsourcing them to a class of fifth graders hoping that if enough fifth graders derive the same answer, the answer can be assumed correct. I would like to add however that in this case, they are dealing with language and its varying nuances and not an analytic subject like mathematics where only one answer is correct.
A reputable translator/blogger brilliantly referred to this process as “Clown-Sourcing”! Another professional translator commented online “if crowdsourced translation already seems doubtful enough, crowdsourced translation by students sounds like a horror movie.” An industry professional added, “They are trying to create the world’s slowest and poorest-quality MT (meat translation) system.”
I am assuming or at least hope that this startup will use its students’ translations from the language they are learning into their native (another assumption) language, not the other way around. But there is no indication of that.
My guess would be that for very simple sentences, and if the translations are done into the native language, and not into the language being learned, the translated text may be of reasonable quality if enough surrounding context is provided. But those who have been in the commercial translation – localization business long enough know that there are seldom easy translations to be had. Each industry and even company has its own terminology, unique style, subject expertise requirements, and inadvertent complexities.
Perhaps there is a market for simple translations performed for free by language learners. But is it large enough to justify the massive investments being made in building the necessary learning platform and technology infrastructure? Time will tell for sure!
Meanwhile, at GlobalVision and CloudLingual, we only use native translators with a proven record of translation abilities and mastery of subject expertise, not a statistically likely correct translation of a sentence outside its context. We use innovation to streamline and improve the result not to take unnecessary risks to lower cost. If you want your translations done correctly, don’t rely on unproven disruptive innovation, or you are likely to end up with is a severely disrupted client base!
Update on October 2015
Exactly 3 years later, an article was published about this company that has already received $45M in funding: “Initially, the company made money by crowdsourcing translations and selling them to private companies. But [the founder] said it wasn’t the best model because they needed a large sales force and “it’s a race to the bottom as far as price.” Obviously, when you sell low quality translation, your price will erode severely! The company now sells English proficiency tests, which fits well into their core offering. This seems a much more reasonable business model to us and confirms our prediction that quality translations matter when one is selling a professional service or product!
This whitepaper presents applicable ten tips that will help you navigate through the requirements of your translation projects and deliver a quality product that your international end-users will thank you for. Download it now!