As part of Lingoport’s webinar, “How G11N Technologies Adapt to Agile Development Accelerations” we interviewed featured guest speaker, Ben Sargent, Founder and Solutions Architect of Group-Q. Ben provides a sneak peek into his thoughts and predictions around the future of G11n and supporting technology.
1. How has the localization industry evolved since your career began?
Ben Sargent: In the 1990s, LSPs would replicate the build environment and conduct bug tests and fix iterations, then deliver a “golden master” CD-ROM ready for duplication in the kitting factory — the final product would be in a shrink-wrapped box for retail. We’ve come a long way in 20 yrs! When the software publishers adopted the principle of simultaneous ship, with localized versions released the same week as the English, there were no tools for managing software updates. We had thousands of files in hundreds of directories and received updates after beginning the translation. Localization engineers griped they weren’t being engineers any more, instead they had to act as file librarians.
That’s why tools like Lingoport Resource Manager (LRM) are so important – the industry is ridding itself of librarian tasks. LRM seamlessly manages the updates so developers can keep working on code and changing the user experience even after starting the translation process. You can track which strings have changed and immediately push them into the translation workflow. Or you can hold them back and only send updates when you want, if that works better in your process. Automated dashboarding gives you the current status of each job, string, and language. Hence the manual, mundane tasks are no longer part of the workflow. And that’s a good thing for everybody!
2. What would you describe as the leading challenge localization experts face today and in the next 5 years?
Ben Sargent: For the big translation buyers, capacity and throughput is the perennial issue and that has not changed, nor is likely to change. Automation is great, but you still need qualified experts on the receiving end of those workflows. Translation providers have had to absorb massive growth in volume over the years, and machine translation does not change that either. Right now we have adaptive neural machine translation that performs as well as human professionals for first pass translation in some environments. Of course, experienced linguists still need to review and correct the MT output, the same as with human translation. And that’s how we know MT can outperform humans, when the edit distance in the review step is lower and the throughput is faster.
So the biggest challenge still remains, where do you find those linguist resources? How do you train them and organize them and keep them motivated to deliver consistently excellent work under very tight turnaround times to meet the demands of agile and continuous localization? In that sense, technology is the easier part compared to the humans. Even with all the automation, the industry must constantly find, recruit, train and manage linguists to work in an increasingly automated world, where they are expected to add-value to a greater and greater number of words every day. So capacity and throughput are still the fundamentals that both the technologists and the humans strive to overcome.
3. What would you describe as the most impactful L10n technology innovation over the last 5 years and why?
Ben Sargent: Neural MT would have to be the most recent game-changer. But other forms of machine learning or AI are having an impact too. Several Group-Q partners use AI to optimize their production workflow. One example is an algorithm that looks at 30 different data points to screen linguists for job assignments. For some accounts, this is followed up by human vetting and training. In other workflows, it fully automates job assignments.
So-called “lights-out” project management is where the translation management system or TMS parses and preps the content, assigns the best available resources, routes the job through a multi-step workflow, and then post-processes and delivers the content, with zero intervention from a project manager. Lights-out workflows can be rules-driven to start, but eventually all that job data becomes fodder for machine learning, and you end up with AI-driven process management. Last year, CSA Research published data showing that 10% of LSP respondents on a survey were already using some form of machine-learning based AI. Lights-out workflow is used for less than 10% of jobs by most of the companies that have it, but that number is growing.
The other big game-changer is service-oriented architectures, enabling responsive workflows generated in real-time, where each step can be sent to a separate machine actor or human worker, based on the job requirements as interpreted by the machine brain. These systems will determine job requirements from meta data and by reading the content. Lingoport components including Globalyzer and LRM are good examples of services that can be called by such a system. So what we can do with technology now is super fun, super cool, and moreover impactful!
4. If you could only provide one piece of advice to L10n teams, what would it be?
Ben Sargent: Too often client managers are just trying to solve a tactical problem, their hair is on fire, they are overworked and under supported by their own management. Thus, they don’t share key information with their vendors, such as corporate strategic initiatives, organizational transformation, and long-term goals. KPIs are not connected to the larger vision. Budgets and timelines are hidden. Vendors are kept in the dark and get the impression that the client-side managers are not being fully transparent with their challenges. Balancing tactical needs and strategic initiatives begs for closer collaboration between localization buyers, suppliers, and language technology vendors.
At Group-Q, we offer the expertise to solve complex issues by developing bespoke programs, and we work hard to get beyond the tactical needs to also address strategic goals. Good vendors are strategic partners, helping client-side managers shoulder the demands of their markets, the constant pressure for growth, cost reduction, and delivery schedules. The culture that “L10n is a cost center” has not gone away, but the right service provider can help you make the business case for why localization is a revenue driver and customer retention tool. Clients that provide transparency to vendors about corporate initiatives, goals, objectives, and timelines enhance the strategic nature of the localization process. Think back to the days of “waterfall” product cycles — indeed slower development taking products to market, but the teams from all sides worked more cooperatively and planned their goals to meet the market demands. The faster pace makes that harder today, but suppliers want to be engaged and help. With the right inputs, that cooperation adds more value.
One problem is that we still don’t have a career path through-line from localization management to executive management in the enterprise, with the inevitable result that strategic planning is absent or inadequate. At Group-Q, our advice to international product managers, localization directors, and global procurement specialists is to become globalization champions, manage up, and advocate for strategic planning for localization as a driver for global market share and brand equity. We help them develop and present their business case. That’s how we make progress and how we help our clients succeed!
Date/Time
- Date: February 27th, 2020
- Time: 9AM Pacific, Noon Eastern, 18:00 CEST
- Duration: 45 minutes, plus audience Q&A
Group-Q assists companies to solve tactical needs and achieve strategic objectives, tapping multiple eco-systems of localization skills and technologies through our portfolio of partners and preferred suppliers.
As part of Lingoport’s January webinar, “