Meeting the Modern Challenges of Language Technology

The language services market at large is undergoing waves of change often founded upon advancing technology.

Modern localization challenges require increasingly advanced translation and localization solutions optimized for multiple content types, formats, and stakeholders. Not to mention each industry vertical has its own considerations, and each new tool or platform needs to integrate to dozens of others to meet the needs of varied users.

This also means today, more than ever, a few key trends—and how to respond to them—can make or break a company.

XTM International CEO Bob Willans talked about a few key trends that are driving change for his company, and how they intend to meet the challenge head on.

Integration and Interoperability

The language technology landscape is increasingly becoming more complex as startups try to find their niche, more established players improve their functionality and features, and leaders try to maintain their market share through innovation.

For users of language technology, all this complexity is driving them to require more interoperability and integration.

“More and more of our customers want integrations with systems,” Willans said. “You gotta be interoperable. You gotta be able to connect to web content management systems, different machine translation engines—everybody wants to be able to connect, and there are different levels of connectors.” At XTM, for example, Willans said the need for robust terminology and dedicated desktop quality assurance systems meant integrations with TermWeb and Xbench were very important for clients.

Data, Data, Data

Before machine learning brought forth an insatiable hunger for training data, the Big Data boom years earlier that continues to this day opened new possibilities in terms of how to leverage information for business analytics and intelligence. Today, language service buyers continue to want more data capacity, analysis capability, and visualization.

“We’re finding a lot of customers are requesting a lot of cost information, business intelligence information that we’re exposing via our API and reports and graphs within XTM,” Willans said. “People just want data—they want to access their data, they want to be able to manipulate it in a way that suits them.” Furthermore, the rich information output from highly automated processes like continuous localization is complex: clients want to access actionable insights and not just raw data.

And these trends do not exist in isolation. The need for more data and control over data spills over to the need for more integration and interoperability. Willans pointed out that XTM has integrations with business intelligence tools like Domo.

Scalability with Functionality

Speaking of trends not existing in isolation, Willans shared that scalability and functionality are demands that naturally flow from clients that regularly crunch big data.

“Scalability is a big thing. Our target market are large enterprises and many are processing very large volumes of data, so they need to be assured that any platform that they invest in can manage that data seamlessly,” he said.

And as customers scale upwards, they want to make sure their provider can likewise scale their service to their needs. “There’s a high level of functionality there that we find our target market of large enterprises are really appreciative of,” Willans said, noting that XTM’s feature list has grown so much that they turn off many of them by default. Users can choose to switch them on when they require, otherwise for users who need limited functionality, there is the risk of “the product becoming so laden with options that it could become complex to use.”

“Customers always want more and more features but at the same time we have to try—and it’s a continuous struggle—to keep the interface clean and simple and as uncomplicated as possible,” Willans said.

Automation

The unstoppable march of neural machine translation (NMT) falls under the macro trend of automation, and this too, is closely associated with scalability and functionality. Willans explains that clients want to be able to do as much as possible on a single platform even as they grow in scale—from continuous projects to transcreation to relay translations and beyond.

Therein lies the promise of localization process automation features, such as templates, group management of translators, and of course, NMT.

“We’re finding that more and more large enterprises are definitely using NMT. They’re seeing the benefits of it. They’re seeing the cost reductions of using it,” Willans said. An important caveat, however, is that it ultimately still boils down to the content. “If you’re talking about a customer-facing or legal document, then, clearly, that’s very crucial that it’s 100% right,” and NMT, even at its current level, is not enough for these circumstances.

Still, NMT adoption is definitely on the rise. “We’re often asked how many different engines we have,” Willans said. “We’ve recently added DeepL, Yandex, Amazon—all of these engines are leaders in their particular language pairs.”

Of course, as probably one of the most exciting ongoing developments in language technology, Willans said they are exploring further possibilities with NMT. He said with the help of their partner Crosslang, they have been working a white label version of XTM specifically suited for a vertical industry.

“So we train MT engines, specifically for legal, built around a white label version of XTM that has been customized for the legal industry,” he said. “We’re looking at other vertical markets like oil and gas, where there is a huge amount of scope for machine translating documents.”

“Absolutely Optimal for the Translator”

It is plain to see it is no simple task to not only meet the demands for integration, data, scalability and functionality, and automation, but also at the same time guarantee that the solution for each works well with one another.

So how does Willans tackle these ongoing trends and offer value to XTM’s users? During the upcoming XTM Live London 2019, XTM will be releasing XTM v12, their answer to all these challenges and trends.

“The key thing is XTM workbench,” Willans said, referring to the redesigned XTM editor-slash-dashboard. He explained that the original editor was developed in 2004, and now that technology and design principles have moved on, seeing all these challenges laid out before their users, XTM found an opportunity to redevelop from the ground up for a holistic XTM workbench that is more intuitive and flexible.

“We’ve taken the opportunity to just stand back and say, okay, let’s redesign this from scratch. Let’s use the latest technology to get better performance to redo the layouts so that ergonomically, it’s absolutely optimal for the translator,” he said. “We’ve been working very, very closely with customers, with trained linguists to ensure that our designs meet their requirements”

“It’s gonna be a benefit to the translators, who are, in effect, the majority of the users in XTM,” Willans explained. “Even though we talk about project management functionality, majority of the users are translators—they’re working on XTM all day long so any enhancements we can make to their productivity are very significant.”

Join Willans with the rest of the XTM team and speakers from Adidas, Nestlé GLOBE, and Swiss Post to learn about these very trends that XTM v12 resolves, and see for yourself how the new XTM workbench works.