What should localization buyers know about linguistic AI?
05 Oct 2022
6 mins
Procurement is the act of obtaining goods or services for business purposes. Localization procurement is the act of deciding which Language Service Provider (LSP) to trust with your content. This is a critical function, even more so in the current uncertain economic climate where it would be all too easy to bow to price pressure at the expense of a sustainable localization strategy with a responsible partner.
How can LSPs help procurement balance cost and quality? The answer is through innovation which enables enterprises to emerge stronger from these economic challenges. Linguistic AI innovation can inform a strategy that increases efficiency and rationalizes cost without jeopardizing quality.
Traditionally, localization buyers are reliant on complex RFP documents to question LSPs. However, the localization landscape is now more complex than the traditional speed, cost and quality paradigm. We see the emergence of sophisticated technologies which are often not well understood.
So perhaps we should consider a novel approach to localization procurement, one which centers on a partnership between procurement, enterprise localization managers and innovation-driven LSPs to expertly guide and advise through this new and evolving landscape.
The foundation of this new approach is an Machine Translation (MT) content Strategy – mapped and prioritized with the end user in mind. It’s an LSP’s job to make buying decisions easier by providing guidance, best-practice recommendations and increasingly powerful content insight tools that quickly pinpoint the value of content and help decide where to focus localization budgets.
The onus is on LSPs to expand the linguistic AI capabilities that go beyond text-in-text-out machine translation, building further on the same Natural Language Processing principles that made translators more productive in the form of neural machine translation (NMT). Linguistic AI features will enable buyers to make highly data-driven decisions and grow and nurture trusted partnerships with technology-mature LSPs.
What are these new linguistic AI features?
A good localization strategy is focused on content. Not all content is equal and different contents require various levels of technology and human touch. Linguistic AI technologies can provide insights on content complexity and potential upstream localization costs, helping to drive more accurate business decisions.
Today, Linguistic AI technology offers AI-driven content analyzers, providing users with insights such as domain recognition. This is a straightforward way to categorize content and gauge future complexity, leading to more accurate pricing, appropriate quality processes and a framework assuring relevance for the consumer.
Other features such as MT quality estimation can help predict the validity of using MT. The technology attributes a quality level to a machine-translated segment, allowing the user to make an informed decision on what to do next, for example publish directly or send to post-editing, saving time and money in the process.
Deciding which content is valuable prior to localization can be a time-consuming and labour-intensive task. This is where multilingual content insights and dynamic summarization provide new and innovative ways of solving the content challenge. The technology makes it easy to understand large documents quickly and decide whether they should be localized.
Needless localization or inefficient use of human expertise negatively impacts already hard-pressed localization budgets, with the risk of relevant content not reaching target audiences.
Where does the agency come in?
Having more agency means taking responsibility and shaping those aspects of life or work that directly affect us. In the constantly changing localization industry, this can be a challenge for enterprise localization managers. It is key to engage and align with these AI-driven changes to ensure sustainable cost efficiencies in partnership with a trusted and supportive LSP partner.
Enterprise localization managers can work with procurement towards a roadmap based on Linguistic AI advancement with a clear MT content strategy in mind. Based on a MT technology first approach, it is possible to offer the right custom MT solution with a targeted level of human effort, ranging from full post-editing plus review to light post-editing. This is made possible through up-stream auditing of enterprise content needs.
Localization managers are arbiters of quality but will see their status and influence grow through becoming content enablers – understanding and serving the needs of all enterprise content stakeholders and lending linguistic expertise to critical business decisions.
Recognizing and reaching out to the wider enterprise content constituents helps break the silos in which content often exists and helps to better articulate the priority, the urgency and the audience requirements that are key for an informed localization strategy. Crucially, given that Linguistic AI progress is predicated on plentiful supplies of data, a virtuous circle is likely to emerge that understands the existing challenges but also provides the only sustainable response which is data for building more and better tools.
Innovation is key
Insight-based tools are at the heart of a true innovation driven-approach and benefit the whole localization ecosystem. A linguistic AI program for an enterprise can only evolve when the LSP starts to move beyond test-in-text-out standard MT solutions. Such programs unlock key information transparently and much earlier in the localization life cycle, both for enterprise stakeholders concerned with costs or the LSP charged with content workflow decisions and quality deliverables.
Procurement should not focus exclusively on obtaining more and more discounts but on creating a sustainable and long-term value proposition with a trusted LSP. Likewise, localization managers need to look beyond quality and pave the way for a fluid content delivery to customers.
Now is the time to grow from the proven AI and machine learning focused machine translation post-editing use case to encompass Language Weaver’s complete Linguistic AI technology, bridging the gap between MT technology and MT services to address real-life enterprise content challenges.