The Nimdzi Language Technology Atlas 2019 is out. The number of individual products mapped has increased from over 400 to over 500. Let’s have a look at what’s new and what has changed in language technology over the last year.
With the introduction of AutoML in July 2018, Google Translate became trainable. Microsoft upgraded its Custom Translator as well. This removed a key differentiator of the dedicated machine translation (MT) vendors, such as Kantan MT, Omniscien, and Systran. Now, popular MT is trainable, and customization is almost as easy as uploading a translation memory into the cloud. Therefore the previous classification of “Generic” and “Trainable” MT became obsolete. We retired it in favor of a softer distinction, depending on how you approach the piece of MT technology – “ready-to-use” or the “build” approach. We also removed the MT toolkits category. MT has progressed too far to build it from scratch anymore.
Next, MT is to add predictive quality analytics and make them stick with the professional community. Moreover, a few key MT vendors still need to add customization ability to their engines, chief among them being DeepL and Yandex Translate.
New terminology solutions have hit the market in 2018-2019, developed by language service providers (LSPs) such as Semantix, Bureau Works, EGO Translating, Institut für technische Literatur, and Toptranslation in Germany. The new terminology tools typically come as a supplement to a translation management system. They add a workflow for the terminologist to approve terms, the ability to link terms with images, group terms into clouds or “trees,” and a log-in integration with Active Directory that allows all Windows users within an organization to access terminology without buying additional TMS licenses. The next goal for terminology tech is to integrate with more content creating and authoring tools in documentation, marketing, and support departments.
New media localization tools and functionalities are a response to the content boom of TV series and films. TransPerfect launched Media.Next suite, Omniscien announced Media Studio, dotsub rolled out workflow management in videotms.com, while CAT tools such as memoQ, SDL Trados, and Wordbee added video players to facilitate subtitling. At the same time, established media localization systems such as Zoo Dubs and Subs, Subtitle NXT and Ooona continued to grow and gain traction.
New technology in media localization is steering towards text-to-speech and machine translation technology to shift to human post-editing environments. It’s also catching up with translation management systems on workflow management, collaboration, and integration functionalities.
New launches in remote interpreting petered out once the market has reached platform abundance. It became apparent that remote interpreting is going to be more difficult than taking video chat technology from Google WebRTC and adding a custom interface with a logo to it.
Instead, new development centered on management functionalities, to help small and medium-sized LSPs automate the hundreds and thousands of bookings they manage every month, primarily those in the United States. Companies like Lango, uSked, and Total Language make their appearance on our map for the first time this year, as they make their presence in the market known.
In the future, it is likely that technology developers will make remote interpreting call centers easy to build and accessible to small companies around the world while integrating all of them into networks via APIs. Interpreting has yet to make heavy use of AI, speech recognition, and machine translation.
To highlight the fact that the top LSPs already make millions in data services and AI training, and have started to automate these service lines, we’ve added the Machine Intelligence category to the landscape. “Production platforms” comprise software that allows miscellaneous tasks such as tagging, annotation, data labeling, and voice transcription to run at scale with hundreds of data workers taking part. So far, Nimdzi has learned about only a handful of products in this class. All of them are proprietary products developed by large LSPs. So far, we haven’t identified a commercially available technology in this niche. “Natural language understanding” includes machine learning systems with APIs that LSPs use to build chatbots.
Language AI is a field that comprises hundreds of technologies and startups. However, LSPs only use a small portion of them to deliver language services, which we wanted to reflect upon in this year’s Technology Atlas.
Technology vendors continue to pop up and grow, but the whole language services tech field remains relatively small in terms of total revenue. Our estimate for the market size is just shy of USD 800 million, or about 1.6% of the services market.
There are no technology unicorns in the language services industry. In business core functions such as dev ops, marketing, and support, the likes of Atlassian, Hubspot or Zendesk continue to rake in billions. But no one has been able to do so with language tools — at least not yet.
The infographic above is publicly available and can be reused with the source quoted. It will be updated on an annual basis (or more frequently depending on the pace of innovation).
The spreadsheet is only accessible to Nimdzi clients. It is updated continuously as information about new tools becomes available.
The language services industry is undergoing a profound transformation with the emergence of cutting-edge technologies such as ChatGPT and large language models (LLMs). These powerful language generation models have captivated the attention of businesses and language professionals alike, offering exciting possibilities for translation, localization, and content creation.
It’s already been six years now since Google revealed that Google Translate processes 146 billion words a day — three times more than what all the professional translators in the world combined can do in a month. That was 2016 and things haven’t really slowed down in the machine translation (MT) universe since.
We recently introduced you to the two- (or five-) second rule, which is essentially the reaction or decision-making time a linguist should spend judging whether to post-edit a segment of machine translation (MT) output or to retranslate it. This rule of thumb aims to help increase the linguist’s productivity when working with MT.
If you’re a driver, you’ve probably heard of the two-second rule. Staying at least two seconds behind any vehicle is considered a rule of thumb for drivers wanting to maintain a safe following distance at any speed. The two seconds don’t represent safe stopping distance but rather safe reaction time.