The ins and outs of web localization management

In web localization, some things remain the same and others have significantly changed in the past five years. Previously, initial project stages centered around export/import workflows and building appropriate filters for exported files in a CAT environment. Teams would also evaluate the feature set and global readiness of the host content management system (CMS). Oftentimes, CMSs lacked native internationalization options and therefore required extending their feature set through plugins to make the site “localization ready” before externalizing content. This process of configuring plugins for enterprise sites can add unforeseen tasks to sprint cycles and delay the timeline for project start.

Today, newer headless CMSs natively support some of these features. Contentful, for example, now has out-of-box locale settings. A content manager can select and configure these settings without the need for a development team. Validation settings at the field level are also available to help prepare for exporting the right fields for localization. CMSs have also taken proactive steps to minimize the amount of work a plugin needs to do, which allows businesses to launch their websites faster and reduces the time-to-market for global audiences. However, there is still room for improvement in terms of adding localized SEO terms, and key term search functionalities within the CMS.

Batch processing and multiservice optimization

Most websites being localized into multiple languages have massive volumes of content, and batch processing helps in tackling them. Batch processing is a vital element in processing large volumes efficiently for desired site launch dates. A project manager works closely with clients to identify content priority and the best approach for the various elements involved in the process (translation, transcreation, MT, etc.). The organization of batches in turn is influenced by several client-driven factors that lead to workflows tailored to purpose:

• Availability of source content
• Level of priority (e.g., driven by content visibility, traffic to page, SEO implications)
• Type of content (static vs dynamic)
• Translation approach for the content (e.g., translation, transcreation, MT)
• Need for client review and availability of reviewers

Today’s websites are more media-rich than ever before, and some content cannot be batch-processed. Many web localization projects in fact require multiple services, including text translation, audio-visual localization, and transcreation.

As an example, Acclaro provides localization services to an online marketplace that sells high-end luxury goods. Their content includes product home pages, product descriptions, static site content, marketing assets, and more. To tackle these types of projects, Acclaro works with linguists to provide a range of services for each content type, combining machine translation, transcreation, search optimization, and testing. Linguists need to be able to learn the “voice” that the customer wants to project to the marketplace across content types.

We are also seeing an increasing number of customers with widespread content types. Therefore, skills like SEO optimization and linguistic testing, and mastering the latest translation technology are all increasingly important for linguists to have.

Linguists have to be ready to shift mental gears when working on different types of content from customers. It used to be that website content looked more or less the same across pages, written in a tone suited to the domain or vertical. Now, even though the content comes from the same company, the style and tone requirements vary greatly. Adding to the mix, social media has raised the bar on user expectations, and potential customers expect to be wowed by swiping or scrolling to rich content on any device. This has impacted websites dramatically, as companies look to provide meaningful user experiences for their audiences in the same vein.

AI and automation of processes

Transcreation or outright copy origination in-language for the marketing parts of a website is an essential part of website localization. Experimentation with AI-enabled translation and transcreation is on the rise, and it’s a very exciting time for AI and a lot of potential for our industry.

There is a lot of buzz around AI, some warranted, some not. NMT itself has not fundamentally changed. Its successful deployment depends on well-established processes, i.e. how clean and well written the source content is, the vertical, the baseline performance of the given NMT engine, the performance of the platform depending on the volume of data, its trainability, and other factors.

AI-enabled translations up to this point involve algorithms that automate a data processing step, such as automating a QA check at scale, or identifying non-translatable elements in text. Generative AI is opening the door to interesting use cases with respect to marketing content. Perhaps soon prompts can be given to create marketing material in the target language itself, bypassing the translation step entirely. As of now, we are still witnessing AI “hallucinations.” This is vexing because the AI produces very fluent looking results but then requires a very careful eye to discern between correct and incorrect statements. Translators might need to be relied upon to verify source data.

Localization project management tasks are traditionally manual and time consuming, pulling information together across many systems in the end-to-end process. As such, they are ripe for improvement via AI. But so far, I have experienced more of an impact from AI on language production than on project management tasks.

Increased connections between translation management systems (TMSs) and secondary business management tools sharing all the metadata used to manage translation projects between systems and trigger actions will bring great efficiencies to the role of project manager. The potential from AI solutions in place is something exciting to see and which could help improve turnaround times and reduce human error.

Collaborating and evolving with linguists

Project managers are responsible for setting the right tone from the onset to empower and enable linguists to communicate with them, among each other, and with the customer. To do this, project managers set clear expectations, ensure availability of reference materials, creative briefs, access to source context, setup of query management, and other resources. Following are some essential traits in linguists that ensure a healthy collaboration and successful web localization project outcomes.

• Responsiveness and timely deliveries
• Attention to detail
• The ability to translate with user experience in mind – a willingness to “go beyond
• translation,” to render a target text that sounds and feels like it was created in the target language.
• Agility to work on and respond to work on multiple batches simultaneously, each with their specific requirements.
• A willingness to communicate with other team members, clients etc. Proactive query raising and intra/inter team communication is expected, to address source text ambiguities and ensure a coherent approach.
  An openness to work with new tools

Website projects and large batch-driven projects do not differ essentially from any other project in terms of what a project manager looks for in linguists to guarantee a successful project flow. It is a very close collaboration

*This article first appeared on Slator’s Tool Box newsletter,  This monthly newsletter focused on technology and productivity for linguists and project managers. Read the full issue and subscribe here.