How can translation agencies make the most of online localization platforms? Jan Hinrichs from Beluga Linguistics sheds some light on the subject.

By Estelle on January 16, 2018

Jan Hinrichs, Founder and CEO at Beluga Linguistics

WTI: What gave you interest in the translation business and for how long have you been running a translation agency?
J.H.: I came to translation through my work at XING, a company my brother founded back in 2004. I helped him to translate the website into Spanish first and in the two following years I was actively involved in the management of the translation process into another 15 languages and as a country manager for Spain. In 2006, in agreement with XING’s management, my wife and I founded Beluga after realising that we could run the translation part better as a standalone company.

WTI: Can you tell us what kind of companies your are mainly working with?
J.H.: With XING as our first client we moved ahead and were lucky enough to get in touch over the years with companies like Last.fm, Moo, Bebo, Lookout, Swatch, MyTaxi, Tinder, Adroll and other high profile companies from tech and global brands. Our experience in setting up and running translation teams for fast moving companies has been a perfect match for businesses with ongoing translation needs for their digital content.

WTI: You are specialized in software translation. Which kind of software are you mainly working on?
J.H.: Apps, websites, blogs, dashboards, help center, emails, support content, etc.

WTI: You are our oldest customer, when did you start thinking about using an online localization tool?
J.H.: Online translation tools have been key for companies with ongoing translation workflows ever since SaaS was invented. Social networks spearheaded this development.
At XING we had built up a homegrown editor which helped us to scale easily and run daily updates. When we on-boarded new clients later on, we found ourselves building up editors with our clients internal staff again and again. It was very time consuming and the success depended heavily on the resources our clients were able/willing to dedicate to the process. It was time to get an independent third party tool in the middle.
We briefly launched an open source editor called FIT but this project unfortunately died because of lack of volunteers. 
Through Last.fm - which was our third client back in 2006 and who’ve been trusting us for more than a decade with their localization process - we got to know Edouard, who at the time worked at Last.fm and helped us get the editor working there. When we pitched him to join the project he came up with a better idea: WebTranslateIt! We were lucky enough to be the first ones to benefit from his unique skills and could roll out many projects through this platform.

WTI: How did WTI improve the translation process for you?
J.H.: The support and responsiveness of the WTI team has been just great and has allowed us to solve obstacles in our projects within no time. We can easily set up projects for our clients without any technical personnel involved. When technical knowhow is needed and we can’t help any more the WTI team is always there to solve potential issues.

WTI: How exactly do you use WTI? Do you centralize all of the projects of your customers? Or do you have them open their own account and then handle the translation process for them through their account?
J.H.: We usually help our clients to open their own accounts and onboard ourselves as managers within their account to help them with the setting up. While they connect via api with their repositories we manage the human part of the process.

WTI: What kind of feedback do your translators give you about WTI?
J.H.: WTI is one of those editors that is easy to use, stable and that gives translators most of the things they need. Something what we do miss a bit at WTI is a segmentation on a sentence level to process fuzzy matches better. Currently there are only suggestions but the stats do not bear them in mind.

WTI: And we will certainly be working on that. Because of the way we communicate on diverse media and platforms, the content that must be localized is always evolving and needs to be turned around rapidly, do you have a lot of customers using WTI to provide continuous delivery in localization and what do they think of the process?
J.H.: 90% of our projects are ongoing projects that require a platform like WTI and thanks to the synchronisation tool the process is pretty smooth.

WTI: A growing trend to meet the challenges of the localization of constantly evolving content is machine translation post-editing (MTPE), our platform allows its use as well. Do you rely a lot on MTPE or do you prefer to have translators issue a first translation before proofreading?
J.H.: Machine translation has made a major step forward a year ago when Google launched their Neural Machine Translation (NMT) engine. We have seen a big jump in quality. We can’t use MT for everything but it is already a great help to speed up translation work. We usually enable MT results to be shown in the suggestions from WTI. The translators can then decide if they want to use them or not. Paired with adaptive NMT translators get more productive and can do more in less time.
It is important to understand as well that MT can help translate content that couldn’t be translated by human translators because of time or cost constraints (Microsoft or the EU have been working with MT for years because of the vast amounts of content they need to publish). Today Neural Machine Translation engines do offer in certain contexts very good results a human only need to edit slightly. I believe that in near future many initial translations will mostly be done by NMT and human translators will concentrate on post editing and higher level translations with more impact.

Do you need professional help translating your website, software or app? Or simply want to stay tuned to Jan’s outlook on translation and localization? Follow him on: Twitter, Facebook, Instagram and Medium.


And don’t forget to follow us on Facebook, Twitter and LinkedIn.