Over the past decade, DITA, an XML-based open standard, has gained increasing popularity in technical communication. DITA enables businesses to publish smart and consistent technical content while saving costs. Companies adopting DITA improve their process efficiency by reusing content across platforms. In turn, this decreases localization spending. Benefits of migrating to the structured-content standard also include enhanced user experience, and optimized content management.
However, some companies fail to meet their expected ROI at the end of their deployment. For example, WhP, a localization services provider specialized in technical content and operating worldwide, found that out of 100 prospects and customers, 75% said they had not reached the projected results 2 years after their migration project was over, and processes were mature. In this blog you will learn why companies fail to successfully move to DITA, and how to remedy.
What Is Localization and How Does It Work?
For a company to efficiently benefit from DITA, its content strategy might need adjusting. Documentation managers and technical writers might redesign some content during migration. Content reuse is also low and translation memories can be reduced throughout the process. Translation costs may increase in the first year, but this will balance itself out after the first translation round is completed.
Let’s first look at the localization process and how it works.
Translation is the process of transferring content into another language. Localization not only implies translating, but also consists of adapting content to targeted markets in the context of globalization. Companies operate and market their products and/or services worldwide.
Localization relies on a few major principles:
Filters and segmentation
The content to-be-translated is selected and extracted out of the verbose XML/DITA content. It is subsequently broken down into a list of segments with context.
Translation memories (T.M.)
Source segments are saved in data with their corresponding translation and context. Previously completed translations are reused, and only updates will be translated.
Term bases
They gather key terms with their context and translation in one place.
Workflow
This is the sequence of tasks to translate and review and review a set of files.
Translation Management Systems facilitate these translation processes when working with large volumes of files. They automate workflows and ensure maximum translation efficiency.
Using the right assets to successfully manage localization is also important. A style guide will define a tone to follow for target languages. In addition, subject matter and product expertise ensures proper management throughout the process with all stakeholders.
Why Do Companies Fail To Meet Their Expectations and How To Remedy
Information Model
How you design your DITA information model determines the success of the localization process. You should always be mindful of the structure and quality of your source language content, even when not localizing. Here are a few ways to do it using DITA and MadCap IXIA CCMS.
1. Reuse Strategy
Content reuse and associated cost savings are companies’ main incentives when migrating to DITA. To implement it properly, writing teams need a clear strategy to define what content can be reused, and how.
- DITA content is modular; you can reuse sections, known as “topics,” in multiple deliverables. For instance, a topic can be added to different products, versions, or customers’ documentations. Libraries of reusable content can serve for notes and safety information.
- Conditions allow to personalize content for different audiences, adding attributes within a topic instead of creating several versions of it. Conditions should be implemented at a topic level. If used on part of a sentence, the translator will not be able to account for the other possible translations of the sentence in the target language.
- With variables such as a product names, a topic can be used for several products. Variables should be used carefully to avoid quality issues in translated content— incorrect gender use, for example. They should not be preceded by articles or contain declinable nouns.
2. Beyond the control of the technical writer
Part of the content exceeds technical writers’ control. It is created, managed, and translated outside of the CCMS, including brand and product names or UI terms. With DITA, this external content can be referenced and managed independently from documentation using keys. Technical writers can translate documentation before external content is translated.
3. Country-dependent content
Content may vary based on the country where it is delivered. Even when different countries use the same language, elements such as addresses, measurement units, links, and other standards differ from one country to another. For example, if “Canada” in an English version is translated to “España” in the Spanish one, an error is introduced in the translation memories and may cause issues in translation.
To curb these issues, you can apply conditions to select the appropriate topic when content varies. You can also use keys to inject country-dependent variables into the content. Another trick is to exclude country-dependent content from translation, such as addresses, using the @translate attribute with the value “no”. The content in question will not be translated.
4. Redundant content
Redundant content occurs when the same content is found at two or more instances. It makes it harder to maintain source content, increase translation costs, and generate inconsistencies between source and translated content. Here’s how to avoid it:
- In links between topics, the title of the topic being linked is often hard-coded in the link. In DITA, internal links are populated automatically, so hard-coded links are rarely needed.
- References between maps can be managed easily by creating a centrally maintained list of publication titles when it can’t be automized. These titles can be referenced using keys.
- Navigation titles in table of contents should not be hard-coded.
- To create summaries of topic sets, you can use a content reference instead of copy-pasting short descriptions.
5. Graphic format
Localizing bitmap images is lengthy and costly. Images containing text should ideally be created as vector graphics, enabling localization of the text without modifying the image.
6. Screenshots
Managing screenshots in different languages is time consuming and delays documentation delivery. Storing screenshots as vector graphics is an efficient method. They are created once in the source language and populated using translation memories to generate target language versions.
Content quality
Consistent content composed of short sentences and a direct style will facilitate localization.
Concise and direct content is more likely to impact translation memories positively, and leaves less room for error. Translators will understand and translate content faster.
Conditional content and content inclusion can help increase content reuse but should be used parsimoniously. Otherwise, they make it impossible to translate content and cause issues, queries from translators, and unforeseen costs to handle exceptions.
Writers should also stay away from language specifics, including puns, acronyms, and culture references, which do not translate in other languages or rely on standard knowledge.
Using English as a source language is preferable when it is possible to find competent translators at a reasonable cost (90% of translators work from English into their mother language).
Here are some best practices to maintain high quality content. These can be implemented when content is updated or during scheduled maintenance.
- Simplified technical English generally seen in the aerospace industry can be applied in other industries. Added up to glossaries, direct style and short sentences, it can go a long way.
- Schematron rules enable performance checks and ensure content follows certain criteria.
- Working with corporate solutions such as Acrolinx and Congree to improve content quality can provide valuable insights.
- Peer reviews offer consistency and are highly instrumental.
Quality content not only reduces the risk of errors in translation; it also decreases its costs and helps provide a better user experience.
Localization
Localization is often considered a separate process from DITA migration as it requires resorting to an organization different than documentation management. However, this investment is oftentimes necessary. Most companies have limited to no knowledge in DITA, which can affect content quality and processes efficiency.
The most frequent mistake is the lack of multilingual reference. Content is translated in a TMS, and the reference is captured in the translation memories. It is then delivered to the CCMS where it is reviewed and corrected, without the changes being updated in the TMS. When new content has to be translated, the translation memories and translator will deliver erroneous translations.
To prevent this, it is critical to perform linguistic review so translation memories are up to date, and the translator learns from the reviewer. This requires a WYSIWYG environment such as WhP’s Augmented Review solution. This helps limit the corrections of published content, and maintain up to date translation memories. In addition, implementing a translation memory deprecating process contributes to maintaining content quality.
As opposed to continuous localization, where content can only be sent to localization when approved, and delivered when ready, agile localization is a dynamic process. Localization projects are initiated at each new sprint, completed and delivered within it, thus accelerating workflows.
Although agile documentation presents proven benefits, many companies show restraints as they rather have a final budget before authorizing localization, and because content localization requires preliminary work.
How IXIA CCMS Can Help Facilitate Your Localization Process
Agile, incremental, waterfall
In IXIA CCMS, localization coordinators can send content for translation in phases or topics as soon as they are finalized, and don’t need to wait until a document is completed to do so. Authoring and localization can occur simultaneously. Translators receive a package including an output of the source language document to give them context.
Snapshot localization
Created at customer delivery, or at the release of a new product version, snapshots capture a map at a point in time, and list all the topics it contains along with their respective revision numbers. Snapshots allow you to easily locate a previous documentation set and update it, generate output of an earlier release, or to translate it to an additional target language.
Review and translation
IXIA CCMS allows users to perform a final review of localized content before it is delivered to the customer. If it is not approved, content can be flagged and sent back to the language service provider to fix, thus ensuring translation memories are up to date.
Pivot Languages
Using a pivot language such as English from which target languages can be translated if your content is not authored in English can help reduce localization cost and complexity.
Graphics
Finalized graphics can be tagged and extracted automatically for translation using an image localization kit. IXIA CCMS allows users to manage several versions of an image in localized content (resolutions, for print or web, etc).
Automation
IXIA CCMS can be connected to a Translation Management System (TMS) to send and receive content. IXIA CCMS Localization Scheduler allows one to schedule content exchange at regular intervals.
It can be concluded that localization process issues originate from unclear business strategy and objectives. Localization should be an integral part of content strategy. Otherwise, companies may have to rethink it all down the line, and end up spending more time, money, and resources.
Reviewing and questioning your content will allow you to create the appropriate strategy to get the best out of your localization processes. IXIASOFT and WhP can assist you in this initiative.
This blog was originally presented as a IXIAtalks webinar by Dominique Trouche from WhP and Henrietta Taylor from IXIASOFT.
