“Time is our most valuable asset, yet we tend to waste it, kill it and spend it rather than invest it.” Jim Rohn.
This is particularly true in the localization world, where time – and by extension time savings – is a highly valued commodity. Translators and project managers working to deliver high-quality projects to strict deadlines know this all too well.
The solution is more automation while maintaining the quality our customers are expecting. The work of translators is already highly automated through CAT tools and translation management systems. Further automation needs to rely on innovative AI-driven tools which provide additional, upstream insight to activate greater time savings. These tools, coupled with performance data that is collected post-production, combine to create a virtuous cycle for future performance improvements in content localization.
The goal is not to have all content translated by human translators but to align our approach with the content requirements. Human intervention needs to be targeted and meaningful. Evolving linguistic AI innovation helps to pinpoint where the human touch is needed and which content can be delivered with less or without human intervention. For example:
- Content Insights allow the end user to quickly determine what a document is about and to action the translation of that content according to their business needs.
- MT Quality Estimation can indicate the quality of machine-translated output. Understanding the quality at a glance allows the end user to make a business decision on whether the quality is good enough for their end goals or whether the content should be sent to a human for post-editing. Post-editors can focus their efforts on segments flagged through Quality Estimation that require more editing, which will allow them to be more productive in their work.
- Adaptive MT allows translators to train their own MT models with data sets that they have collected over time, making the MT output more relevant to their projects.
And to think, these are all considerations before the human translation process has even begun!
The main threat to future time savings is a lack of integration into existing workflows. If these new linguistic AI tools and capabilities are not seamlessly integrated into future workstreams, translators are at risk of losing time by having to switch between different environments with serious implications for their productivity.
Of course, technology automation is nothing new to a translator. Translators are regularly required to work with platforms, CAT tools, and terminology tools from different providers. Both established and new technology providers need to be conscious of protecting the user experience and reducing learning curves when implementing new features.
How can a linguistic AI enablement strategy help?
The new linguistic AI features need to be properly understood to generate the expected benefits. An informed enablement strategy helps to eliminate uncertainties and reservations for both customer and LSP.
These AI tools can be used to provide data that ultimately helps customers to make better decisions on how to process their content and improve their ROI, for example through investing in custom MT technology. On the other side of the fence, AI tools offer LSP project managers a much higher degree of predictability, improving time and delivery management. We expect AI-driven data points and touchpoints to become the foundation for a new collaboration between key stakeholders where data is in effect the “currency of trust.”
A seamless experience built around AI tools offers so much more than just speed and efficiency benefits. Monitoring the trajectory of data and feeding insights back into the development process will lead to better outcomes overall. The performance of content insight tools, quality estimators, adapted NMT models, and post-assignment data such as edit distance and quality scoring will combine to provide a holistic view of a localization project within one consolidated workflow. The industry as a whole will benefit from this new linguistic AI paradigm, from technology providers to LSPs, translators, and of course buyers.
Translator agency comes of age!
Having looked at how linguistic AI tools can benefit LSPs and customers, what about the freelance translators without whom our industry couldn’t function? Freelance translators experience the same pressure as staff translators to deliver high-quality translations in ever-decreasing timeframes.
However, they don’t have the support system of working with an in-house team of project managers, fellow linguists, and technical teams. For a freelance translator, working with tools that increase efficiency and maintain quality is vital. Many freelancers have amassed a treasure trove of linguistic assets such as translation memories (TMs) and glossaries during their careers. Language Weaver's adaptive machine translation allows individual translators to use their human-quality bilingual resources to train MT models, leading to more accurate and tailored MT output for specific domains or projects. Adopting this new paradigm is an important step in helping translator agency come of age.
Of course, this is not the first time that we are on the cusp of a seismic shift in the role of translator and project manager. Early discussions with translators around post-editing – the first linguistic AI use case – were characterized by stark and opposing positions on the effectiveness of MT.
Translators had understandable misgivings about a technology that would change their way of working and livelihoods. The answer then, as today, is an open and transparent discussion and a move towards structured learning. Initiatives such as the RWS post-editing certification program or the RWS Campus community are creating a foundational learning pathway and establishing best practices for the industry. Programs like these lead to a greater understanding of machine translation, post-editing, and linguistic AI technology, paving the way for a smooth introduction of linguistic AI innovation. This is crucial if we want to see universal adoption, an achievable technology ROI, and greater enterprise content reach.
Conclusion: A better use of linguistic expertise?
In the 90s the term “information superhighway” was coined to describe the rapid access to information on the internet. Today we are talking about content. Content is information presented in a digestible and nuanced form and localized in your native language to reach global audiences. If we want to achieve our vision of a “content superhighway”, a structured rollout of new linguistic AI technologies and processes is a must. Only then can we meet the content explosion head-on and achieve the scale of targeted and personalized content needed to meet demand.
Language Weaver technology deployments facilitate key business decisions and integrated processes. A forward-thinking strategy of combining technology with linguist expertise ensures that local language content is available when and where it is needed.