Going global: Getting started with content localization

Have you been asked to deliver your content in another language but don’t know where to begin? The decisions you make early on when designing and developing your content can make or break your translation and production processes. It’s very hard (and expensive) to make changes as you run into problems during translation or production. It’s even worse when the problems are discovered by the consumers!

Let’s begin with some definitions and then take a look at what you can do to prepare for localization and how translators perform their work.

Localization (abbreviated as L10N, or L – the next 10 letters – N) is the process of adapting a product for a specific international market or locale. It involves file analysis, translation, proofreading, reformatting, and testing for appropriateness in the target locale. This term also describes the general practice of producing products for various locales.

Translation (sometimes abbreviated as T9N) is the process of converting from one language to another. Translators often use software to expedite the process. Software can also perform the entire translation without human intervention, albeit with varying degrees of accuracy. In many cases, translation also includes proofreading to catch any mistakes made by humans or software.

Internationalization (I18N) is the practice of designing a product to be as culturally neutral as possible, accounting for issues such as language, design conventions, and tool limitations. In the case of software, it would mean not hard-coding menu and button labels into the source code, but using a separate text file to store the label text. That file can then be translated without touching the source code itself.

Why does the industry use numeronyms like L10N? I have no idea.

Transcreation (sorry, no acronym) is the complete re-creation or adaptation of a product for a specific locale. In this case, content is authored from scratch in the target language using locale-specific conventions. Transcreation is most commonly used in marketing, where the message needs to really resonate with the target audience and cultural context is key. Imagine an ad that uses a U.S. sports football celebrity. That person is probably unknown in Europe, where “football” is soccer, so you’ll need to re-create the ad with someone else.

How do you localize?

As a content developer, step one is to look at your authoring tools and intended output types and figure out what your capabilities and constraints are. For example:

• Can you internationalize your templates?
• Can you internationalize your outputs or how you produce them?
• What educational level do you need to write for?
• Are there cultural expectations to meet or taboos to avoid? For example, particular uses of imagery can be perfectly fine for some audiences and problematic for others.

If you don’t already have one, create a robust style guide that clearly defines the tone, voice, and structure for your content, and also include:

• A glossary that defines important terms and concepts
• Terminology rules with reasoning that clearly describes how and when to use the terms and why one term is approved for use but other similar terms are not
• Common content structures and rules for using them (tables, figures, procedures, etc.)
• Iconography and the meaning behind them
• Units of measurement, currency, and such
• Output formats and their specifications

Essentially, document every aspect of what and how you intend to develop your content. This not only helps all of your authors create reliable, consistent content, but can be shared with your localization team to prepare them for the translation work. Standardizing how you create your source content will make that content better overall, and will make the translation work easier as well. The localization team should also be asked to provide feedback and propose changes that will improve quality in other locales.

The mechanics of translation

While not all translation processes work the same or involve the same tools, there are a few common elements. First, there is the question of who or what will be performing the translation. These days, translators could be humans or software, or a blend of both. 

Human translators, sometimes referred to as linguists, are usually fluent in at least two languages (the source you are writing, and the target they are producing). But language proficiency isn’t enough for all cases. Sometimes they need to also have subject matter expertise. Whether you are hiring translators on staff, using freelancers, or are engaging with a localization service provider (LSP), consider whether language aptitude is enough, or if they need to know about the subject they’re translating. Someone who has never worked in the medical field should not translate instructions for a dialysis machine.

On the software side, there is machine translation (MT). Machine translation has been available for a few decades now. It uses pattern-matching algorithms to find the best probable matches for your content from previously translated material. Recently AI has entered the equation, but the process is still very much the same with some deeper logic applied. While machine translation is much quicker than a human translator, translation quality varies.

Sometimes it may make sense to use a hybrid approach, and it’s not uncommon for human translators to use machine translation in their work. In this scenario, the translator uses machine translation initially and edits the translation afterwards. Often, you’ll see this called “machine translation with post-editing.”

Most professional translators use a computer-assisted translation (CAT) tool to expedite their work. The CAT tool ingests a file to be translated and parses the text into strings, usually by sentence or phrase. The CAT tool then presents the source strings on the left and leaves an open field on the right for the translation. 

The translator may (should) also use translation memory (TM) in their work. Translation memory is a database of prior translations. The CAT tool can leverage the TM by pulling in previously translated strings that exactly match the strings being translated, and by providing options for the translator to choose for non-exact matches (fuzzy match). Once the file is completely translated, the translator can update the TM with new strings.

A translation management system (TMS) combines the capabilities of translation memory and CAT tools with other efficiencies such as machine translation, workflow management, project histories, status dashboards, and file transfer. With a TMS, you can assign translators to a project, allow them to share a centralized TM, edit each other’s work, and more, all while monitoring the status of the project in real time. Most language service providers (LSPs, or translation agencies) use a TMS internally. Some larger companies also have an in-house TMS.

For more information about localization—and how to maximize your investment in it—check out our previous series of posts as well as our white paper, Localization strategy: Your key to global markets. 

More questions? Contact us!