Skip to content

GTS Blog

by Dave Grunwald, CEO of gts-translation.com

Archive

Category: crowdsourcing

The translation startup space is expanding. Last month, Japanese translation company MyGengo announced that it had raised an additional $5.25 Million in funding to develop their online translation platform. This brings the total capital raised by MyGengo to $7 Million. No small amount. Website translation company Smartling recently raised an additional $10 Million. With war chests that large, these companies are likely to challenge the localization industry in many ways.

It now appears that MyGengo may be feeling the heat from another Asian translation startup called OneSky. And while the implementation each company offers may differ somewhat, these companies share a common vision: to enable software companies to reach new markets-especially in Asian countries-by offering cost-competitive software localization services bundled in online API tools that integrate with the client’s development process.

Based in Hong Kong, Onesky is being advised (and possibly funded) by Greg Sung, founder of aNobii, and Garry Tan, Venture Partner of Y-Combinator and Co-founder of Posterous.

OneSky Online Translation Quote

OneSky Translation Quote Interface

OneSky’s website provides online quotes for professional translation on their Home page. This can be done either by pasting in text or by uploading software string files. The quotes are provided by translation companies that partner with OneSky. In providing this service, OneSky functions similarly to Cloudwords: another startup I wrote about last September.

For software companies interested in ongoing localization or crowdsourcing, OpenSky offers a localization management system with several pricing tiers and integration options: OneSky offers an API which connects the software app to OneSky’s management server. The strings are translated on OneSky’s server and then populated into the software app. Since this process is dynamic, new or modified strings in the source code are detected and automatically sent to the OneSky platform. All translation done on the OneSky platform are sent to the software application and automatically updated. The OneSky platform also offers crowdsourcing options which enable the users of the software to vote on translations; the voting mechanism can be setup to control the localization process and update translations based on user feedback.

OneSky Development IntergrationOneSky offers a wide range of libraries for the various development environments, web and mobile platforms. The most popular ones are covered: this includes the iPhone and Android mobile platforms. The OneSky blog brings several case studies and testimonials from software companies that have used OneSky’s platform for localization management and crowdsourcing. These success stories indicate that OneSky is a company we may hear of more in the future.

Enhanced by Zemanta

Last February, I wrote a post entitled Is the translation crowdsourcing market getting too crowded? That post, which discussed some of the new online translation companies, asked some basic questions: is there enough business to support all of these companies? and if this is such a good business, how come more translation companies are not getting into it?

But I neglected to address the most important issues: what is the quality of these translations and how cost-effective are they?

Ever since I wrote that post, I wanted to write a followup post by trying out some of these services and reporting on the results. So I prepared a small file which contained about 400 English words of technical text and tried to order a Spanish translation of that file from a few of the online companies. The results are presented in this post. Surprisingly (to me at least), the results indicate that buyers should stay away from these services and stick to the old fashion translation agencies. At least if they want good results.

One general comment: at GTS we charge about $0.15 per English word for English into Spanish technical/medical texts, or $60 for our test file. This includes translation which is reviewed by a second translator, to yield high quality results. For higher volumes of work, we can discount this price. So as a rule, I decided not to order from any online service that charges more than $0.15 per word.

Ready? Here are the results: continue reading…

The Microsoft Logo

Image via Wikipedia

In the wake of the API translation wars, Microsoft made an announcement on their blog today that they are releasing a new version of their Microsoft Translator. The new release brings collaborative translation to the forefront, allowing users to custom create their own translation community to translate their website.

Microsoft has also reiterated their commitment to maintaining their free API:

I wanted also to reiterate that our API is not going away, and we continue to be focused on adding value to our partners through continued improvements.

This is reassuring news to the developer community, after the stunning news of Google’s shutdown of their Translate API; this was followed by a recent annoucement that they will keep the API but will start to charge for it.

I will bring more details about the new MS Translator release as I learn more about it.

Enhanced by Zemanta

Microsoft Research will be holding its Microsoft Research Latin American Faculty Summit 2011 in Cartagena, Colombia on May 18–20, 2011. One of the talks on the agenda will be delivered by Chris Wendt on Sharing Research Worldwide: Advances in Automatic Translation. Chris is the principal manager of the Microsoft Translation team. If Chris gives me a link to his presentation (if it will be streamed live), I will post it so we can all watch. Or perhaps a link to the slides.

The summit will also feature a DemoFest with two booths on the topic of language translation:

  • Involving the Community in Creating High Quality Automatic Translations. This booth will demo mechanisms for community involvement that are able to push the quality of an automatic translation to a level that satisfies even the most demanding users, and how to make this a fun and compelling exercise.
  • WikiBhasha—A Multilingual Content Creation Tool for Wikipedia.

I saw an interesting post on Blogos (Multilingual Magazines blog) about a new language learning website which will help translate the web using the power of the crowd. This concept is the brainchild of CMU Professor Luis von Ahn who is the creator of reCaptcha, a company which was acquired by Google in 2009. The new startup which is called Duolingo follows a similar concept to that of reCaptcha. People want to learn new languages. So why not get them to translate web content while learning and use the translations for real work. An article in Techcrunch from a few weeks ago predicted that Google will acquire this new company within 6 months.

The idea is great. But will it be the greatest idea since sliced bread? I am not sure but here are some interesting facts to consider.

  • Unlike reCaptcha, this service does not fulfill an essential need. Language learning is cool, but most people can live without it. U.S. customers currently feel little urgency to learn foreign languages despite the increasingly global economy (this quote was taken from a Motley Fool article about the language learning company Rosetta Stone). Captchas, on the other hand, fulfill a critical need. Without Captchas, spammers and hackers would own the Internet.
  • reCaptcha asks people to perform a very menial task (reading words). Duolingo is asking people to learn a language and to translate between two languages. This will take a much more complex and broader delivery platform than reCapcha.
  • The US stock market does not believe in the language learning market (see following stock chart and the Motley Fool article). There are many free online language learning systems such as Berlitz, Disney’s Publishing Worldwide, CBS’s Pimsleur, LiveMocha and McGraw Hill.
Rosetta Stone 2Y Stock Chart

Rosetta Stone 2 Year Stock Chart

Will Google want to get into the language learning market? This would to be a new area for them. But it may be attractive to Google (or another company with an ambitious MT development program) as a source of getting large training corpuses. It’s worth keeping an eye on this company but they have a long way to go. Let’s see how the launch (scheduled to take place in a few weeks) unfolds. Let’s see how many people use this as a language learning platform.

It really is a great idea. But I had another thought: instead of teaching people new languages through translation of text, let’s train people as translators. There is a shortage of translators for low-cost website translation. Getting new people into the industry as qualified translators while translating a whole bunch of content would really be killing two birds with one Rosetta Stone.