AI and Language Automation: Opportunity or Calamity for Localization Services Providers?
By Stephan Lins, CEO, MediaLocate
Localization is a complex process—some of it is automated by tools, but much of it is still a human-driven, manual undertaking. So it’s no wonder that recent AI advances in Machine Translation (MT), as well as the allure of automated one-click translation platforms have caused a stir in the translation and localization industry and some fear that this development might spell doom for language professionals and perhaps even be the end of language service providers (LSPs) altogether.
So, is complete push-button localization imminent or hyped?
In short, it’s neither. Sure, MT and off-the-shelf process automation tools have come a long way, but they’re not nearly good enough to put most companies’ localization programs on autopilot. What large buyers of language services still need the most, is customized engineering solutions. Rather than a tech promise of “one-size-fits-some-sometimes” by a growing number of MT engines and l10n tools providers, many Fortune 5000 translation buyers simply need their translation providers to be able to develop customized automation solutions that specifically hit the mark for them and for all of their splintered language needs and authoring systems.
Contrary to common belief, many of the world’s most visible companies still rely on dated or non-standardized content authoring and management systems, disjointed resources, and habitual processes or deeply integrated tools—all with marginal efficiency.
So what should globally expanding companies look for and on which type of language service provider should they place their bets?
- Traditional translation agencies are really good at finding, placing, and managing translators, but technology and language automation are not their forte.
- Localization tool developers do offer translation and management tools, and they can make a big difference, but they are expensive and without significant customization and ongoing support and resource commitments, such tools often don’t reach their full potential.
- Most digital or instant translation platforms are simply crowdsourcing marketplaces or MT engines, and regardless of their attractive UIs, they are by their very nature designed to channel a client into a templated, limited-option offering with marginal quality.
On the other hand, language-and tech-savvy, yet tools-agnostic LSPs, who can develop specific and sometimes downright unorthodox solutions to complex localization dilemmas, offer a distinct competitive advantage. They can provide the language know-how and technology, but won’t hold buyers hostage to any one captive solution. Instead they offer agility to switch or adapt to ever-changing automation tools and market developments.
Here’s a short list of use cases where customized localization solutions can add significant value, speed and quality to your global language expansion:
APIs: Is your content hosted in a content management system (CMS)? Consider connecting it to your LSP’s translation systems (TMS) via API. This will simplify the handoff and delivery of translations, making your processes truly agile.
Services, tools, and ways of translating will undoubtedly be different tomorrow than today. But the need for customized language solutions and LSPs will not disappear any time soon
Top LSPs offer APIs and connectors for leading content management systems such as Adobe Experience Manager, WordPress or Drupal. If a connector for your system is not currently available, translation providers with a savvy localization engineering team can develop custom connectors for you. And since they provide tools and language services, they can do it for less than strict tools companies who must turn a profit on their product alone.
Machine Translation (MT): An MT assessment for your content type(s) may determine whether MT is a viable option for your company, but you need trained engineers who can run pilots, determine quality scores and identify best MT engines for certain languages and content types. There are many good MT engines currently on the market, and Google, Microsoft, Amazon, eBay, Salesforce and others are developing or acquiring their own. Therefore, MT is here to stay, but not without significant customization and various levels of human post-editing.
Large projects. Multiple formats and stakeholders. Impossible turnaround: The early bird gets the worm and time-to-market is more critical than ever. But global marketing campaigns are now also more complex than ever. Our company recently completed a project for a large software client, delivering nearly 5,000 localized Photoshop files for 72 target countries with individualized content in a matter of days. This process, which used to take nearly a month, was streamlined and automated by our localization engineers to just a few days. If your project is large-scale, has multiple stakeholders, and many regional specifications, consider having your provider write custom scripts to automate key points of the process. The result will not only be much faster, but free of error-prone human touch points.
Mark-Up Management and File Parsing: Most content authoring systems create tags in and around source texts, with unique instructions and descriptions that call for specific actions or don’t require translation. Different string lengths and sizing limits (the available space for translated text strings to appear on a website, software or app) may need to be controlled based on dynamic parameters. Whether you choose MT or human translations, these tags need to be identified and isolated. This often requires custom-scripts and a keen knowledge of content types and language nuances.
Testing and QA: While there are a number of automated testing apps, linguistic and functional software testing still requires an engineer familiar with the website or app itself, as well as knowledge of language specific localization bugs and pitfalls. An experienced LSP will help develop a testing plan with clearly defined test cases and can lead a team of testers who then run through each of the cases in their native languages.
Multimedia: Although there are tools to automate text entry and placement in various multimedia formats, due to text expansion or compression in different languages adapting animations to fit is still very much a human-driven creative process that ensures the same message and quality in translated text. Tools are also being developed to automate the process of synchronizing video to voice, but both run into issues of unnatural pacing, as well as inconsistency in flow and intonation. If stretching/shrinking a video to fit a voice track, there are usually speed ups and slowdowns in certain animations that do not fit very well. Matching a new voice track to the existing timing of a video usually requires planning in the translation stage to match the length of the original script, and sometimes even requires retranslation and rewording on the spot to come close to the original timing in the target language. The best approach is taking creative liberty with the video and implementing new animations as needed, both in conveying the message on screen as well as synchronizing the voice.
Localization Intelligence Portals: Still tracking your projects on an excel spreadsheet? Agile and forward thinking LSPs can build tailor-made portals that track project status, TM metrics and lookups, localization spend and savings, historical reports, or any data point of importance to you. Custom portals not only provide instant 24/7 information access, but also the ability to centrally approve and launch projects and retrieve translated files, all in one convenient centralized repository.
In summary, there is no question that technology has and will continue to drastically change the translation and localization industry. The types of services, tools, and ways of translating will undoubtedly be different tomorrow than they are today. But the need for customized language solutions and for LSPs who can deliver them will not disappear any time soon.