Volume 2

Issue 4

December 2005

Feature Article Up-front translation strategies save big time downstream Ben Martin Partner, Industrial Wisdom bmartin@industrialwisdom.com Effective delivery of content across languages requires up-front attention to three main areas: 1) translation-friendly authoring strategies, 2) leveraging of content management technologies, and 3) aligning translation activities with an international market strategy. This article pulls from the author's experience and explores best practices in each of the three areas. Too often, translation is an "oh-by-the-way" activity emerging as a consideration late in the product life-cycle. The content is "thrown over the wall" to localization vendors with little understanding of how to minimize costs and accelerate time-to-language processes. Several content management technologies can augment the management and processing of source content into target languages. These technologies support management of all levels of content from the terms, to the translated sentences, to the chunks to the information architecture, to the workflow of the content through the translation processes. All optimization of content should be aligned to the international market strategy. This strategy should drive the translation investment, set the criteria for what is "success" in a particular market, and drive an enterprise view of the content. On the information highway to translated content, you can pay as you go and get huge savings at your language destination, or you can ignore language factors in your content strategy and pay big time at the end when it comes to translation and localization. I learned these lessons the hard way on the road of experience. I served as Vice President of Global Content Management for JD Edwards, responsible for all the authoring activities around product for training and documentation and the translation of that content into 21 languages for the online help and UI and eight languages for the documentation and training materials. We were under a mandate to figure out how to write it once, reuse it many, translate it once, reuse it many. We ended up down many cup-de-sacs before we pioneered a throughway to single-sourced content and optimized language processes. Part of this initiative was defining the requirements for a technology that could support our vision. Not only did we re-engineer how we authored and translated our content but we developed our own content management solution that ultimately was purchasable by our customers and a revenue generating product for JD Edwards. To maximize your savings, my experience says your best bets are focusing on: Writing strategies that are translation friendly from the beginning, with built-in accountability for the writer Leveraging content management technologies to manage: versioning relationships between source and target language translation change translation workflow terminology taxonomy Aligning translation activities behind a clearly articulated international market strategy Translation-friendly writing strategies Innocuous decisions by source authors can have exponential cost impact when it comes to translations. For example, the decision to repeat a procedure throughout a set of information every time it is relevant ratchets up translation costs. The decision to vary your language to sustain interest in the reader rather than standardize on one term per one concept can significantly slow down the translation process and up the costs. The non-decision to include random graphics and sports metaphors in the content without attention to whether those examples are relevant outside the US market can present a significant speed bump to completion. Deciding to cut and paste graphics rather than reference them and the decision to tolerate embedded text that cannot be extracted significantly slow-down translation downstream. Choosing to allow separate authoring communities to author according to departmental standards and guidelines with no accountability to a corporate content strategy invites personality-stamped content that says all over again what has already been said only in an unnecessary, costly way. The hidden choice to embark on a linking strategy that does not factor in language equivalents will force a major scavenger hunt in your target languages as they attempt to mirror the original source. What is often missing in the writing process is accountability for the cost downstream. At JD Edwards, we trained our English writers that they were not successful until their content had successfully been translated into seven languages. The translators rated the writer's content on ease of translation and captured in a comments audit trail those concepts, graphics, or writing styles that were particularly problematic. The translators at JD Edwards were internal and were motivated to provide feedback, but you might encourage your translation vendors to give you feedback on the content you send. You will be surprised by some of the things they report to be problematic and, most likely, you will have the chance to correct yourself next round and shave days or even weeks from the translation process. Not only should accountability be measured in terms of ease-of-translations but also in amount of change. We found that while our software was changing between 30 and 40% each release; a manual was changing 100%. Doing a little investigation, we found that a new writer inherited another writer's manual and decided to perform a 100% improvement on the content. I never met a writer who didn't think they could "up" the quality of the inherited content with a significant rewrite. Even on their own content, writers were guilty of what I call "willy-nilly" change: injecting their preferred wording to be more user friendly. Always performing change in the best interests of the user, they lost the perspective of the impact they were incurring downstream. There is a change saturation point from which a company cannot afford to sustain and still deliver release-current, profitable products. I often talked of implementing a change audit that would leverage our translation memory tool. I envisioned that each time a writer checked in their content, they would receive a pre-translation analysis letting them know the number of segments/words they changed, multiply the amount changed by the going per-word rate (for example, 25 cents), and deliver the message, "The changes you have made will cost JD Edwards $XXX amount, are you sure you want to continue?" Knowing my writers to be good corporate citizens with a vested interest in its success, I could trust them to make the right decision at that point. Another frontier of accountability is measured in the sheer volume of words. It becomes a time, capacity and money issue when looking at the number of words you need to push through the translation pipeline and to your markets. The more words, the more time and money required. A deep content analysis of what is getting authored should surface opportunities to single-source, reduce wording, and right-size your messaging. A short, concise set of content is often the best strategy for your English content regardless of target languages but when you are looking at translating the packaging, brochures, websites, training, and documentation you need to be on the lookout for how to streamline your volume of content. It is common to see a 40% word reduction in the enterprise corpus without seriously impacting critical communication. 40% reduction in English is 40% reduction in each target language so your savings are in multiples equal to the number of languages you tackle. The final accountability should be realized in a quality measure. A large portion of quality is measured in adherence to standards, consistent voice and terminology, appropriate acronyms, and accuracy. There are some tools that I class in the content management space that help authors stay on course when it comes to authoring in compliance with corporate standards. Acrocheck is such a tool. It integrates into the authoring environment, learns your style-guidelines and terminology, and then dynamically provides feedback to the author when slipping into passive voice, unapproved terminology, wrong tagging, or an acronym violation. The Director of Translations at JD Edwards was willing to fund the purchase of this tool for the Technical Publications department because she was convinced she would get a return on her investment first round. Despite a dedicated editing staff, content was still slipping through that was in violation of the agreed upon standards. Because our editors were focused on the substantive edits of new sections and significantly rewritten sections, no one was really monitoring the seemingly inconsequential changes until they hit translations. Too often, inconsequential changes were significantly slowing us down. Leveraging Content Management Technologies Just employing decent writing guidelines is not enough to optimize your content for translations. There are relationships that need to be managed in your content management system so that change can be synchronized across languages. Mechanisms using workflow and parent-child relationships are necessary to manage parallel universes of websites and documentation and training. Understanding relationships of your product content to the product (for example, strings in the software interface or link titles on your website) can augment that ever elusive simultaneous shipment of product and product content. At JD Edwards, we gained huge time-to-market efficiencies by requiring user guides authors to use references to field titles so that when the title was translated it was automatically pulled into the target language content. We wrote a custom "fetch" into a database for the title and then resolved the title at publish time. This content management technique saved us the embarrassment of two different translations (one on the screen and one in the guide) as well as eliminated our dependency on a translated UI before we could finalize the documentation. At JD Edwards, we chose to own and manage our linguistic assets (translations and terminology) to ensure their integrity. Without a clear strategy to manage your assets can result in a mishmash of memories that compromise your messaging in your target markets. We integrated our translation memory tool (Trados) into our authoring environment. When the author felt the content was stable and ready for translation, the author would launch a process that first involved cloning the structure and metadata of the content into the seven target languages. The system converted English links to the target languages, changed the appropriate metadata (such as the language tag) and added a "child of" metadata reference to the parent chunk. Once the cloning process completed, we ran the content against our translation memories to determine how much real change had occurred and to automatically replace unchanged English sentences with their already translated counterpart. We had a resource per language team own the translation memory and terminology. Their primary job was making sure the memories didn't get corrupted with bogus translations and that the terminology was being appropriately implemented. We had one set of memories for each of our three domains: Financials, Manufacturing, and Distribution. In today's market, companies engage several translation service vendors who each maintain their own memory, which almost without exception results in different translations for the same content, unless someone is assigned to manage a combined set of memories. An often overlooked content management tool is a terminology management utility. Too often, terminology management is relegated to a translation productivity tool. Consequently, the organization is not in a position to stem the tide of new words that describe the same thing over and over. It only provides a snapshot of uncontrolled variations. When properly put to use, a terminology management system should be the vetting system for insurgent lingo and confusing "synonyms". Too often, the writer's code of creativity is at odds with a rationalized terminology. This creativity code might read as follows: If I can find a better term than the existing term, I will use it. If I can flavor the writing with terms that mean the same thing but inject interest into the writing, I will. If someone else wrote it before me, I can write it better. I must vary the language, rather than bore the reader with consistency. Managing terms must be a core competency of content authoring groups. Prospective customers make buying decisions based on website content and customers experience the product based on training content. A confusing array of product titles and acronyms, a departmental vocabulary that disregards the conventions of another department, and the creative flavors of terms undermine clear business transactions. All of this confusion is compounded during translation. Web metrics reveal the power of an optimized terminology for searching. Companies are made and lost on how they surface in a web search. An ambiguous, scattershot set of terminology will find you on the 15th page of 16 pages of hits. The days where terminology management is a back-office translation productivity tool are over. Companies must step up to a managed vocabulary that is synchronized across the product content throughout the product lifecycle. Another content management offering that optimizes management of your translations is a metadata management tool that helps you single-source your metadata or corporate taxonomy. The one I am most familiar with is Schemalogic but there are others. This tool should manage in one place a consistent set of categorization (taxonomy) and hierarchy of concepts and terms that apply to all content systems in your company. Too often, departmental content solutions grow up in silos and their tagging strategies are not coordinated so a two character code in one system might be the same thing as a three-character alpha code in another system. What needs to happen is "meta management" across all systems using meta-data. This allows for a federated view over the systems and a way to synchronize change to the information architecture. This taxonomy needs to migrate to all language sites and content stores. Allowance needs to be made for translation of language-dependent metadata values. For example, the values for job role would need to be translated for each language but the values for document type could remain in the source language. Ensuring that you have a mechanism for stamping your target language deliverables with core corporate metadata will streamline queries and unify your treatment of product content worldwide. Aligning Translation Activities with an International Market Strategy Too often, translation activities are driven by knee jerk reactions to customer and prospect demands rather than in alignment with a holistic approach to the market opportunity. I have seen the "tale" of the sales rep wag a company's language strategy. I've watched a convincing salesperson say, "If we just had this product in Lithuanian, I could blow away my quota numbers?" and suddenly, the company is jumping through hoops to deliver Lithuanian content products. I've seen where a strategic customer holds a company hostage by withholding upgrade and maintenance money unless the company delivers the product in Mongolian where the customer has an operations office. I've even seen it where a company did its first business assessment of its language strategy after ten years of translating and discovered it had been investing in the translation of Danish when all the Danish customers were using the English version of the product. You not only need accountability on the authoring side of the equation but accountability for the content/translation strategy. You need a way to snapshot the "R" in your ROI for languagesælike how many customers are using your product in language and how much does a single language cost? Your company strategy team should define what cost-to-revenue relationship justifies going into a new market. They need to determine what timeframe is reasonable/allowable and what factors would necessitate the difficult decision to pull out of an area. A good target after the initial investment is 5-10% revenue of large markets, 10-20% of small markets. Your company's scope will obviously determine costs. How much you invest in real estate, infrastructure, support, and how many levels of content you translate should correspond to the market opportunity over the last two years and the projections for next two to five years. You should set a minimum revenue bar before going into a market. The strategy team needs to come to terms with the fact that each market might require some unique deliverables. They have to balance what can be decided centrally and what needs to flex with the regional market. Who decides (corporate headquarters or the regions) what gets translated, internationalized, localized, globalized - what becomes base functionality? Who determines the budget for each of these tasks and ensures that all impacted organizations are appropriately resourced? As you work through each of these challenges, you will need to document your decisions. This will protect you as you drive future sales and fund ongoing support. Your decisions per market should include your: Market entry and exit criteria Product, release, bug-fix, and update commitments Language and localization commitment criteria Legal verbiage that allows changes to made downstream to key policy areas, and what commitments can and cannot be made contractually Traveling on the information highway to translated content requires that you know what the end-game is about. This requires a translation strategy so all other activities can line up behind it. No need to single-source for single-source sake but once you realize what your market opportunities and goals are, it becomes compelling to get control of your content - not just in authoring accountability or content management enabling technologies- but also in aligning your content/translation approach to your international market strategy. The opportunity for huge savings is to never authoring another word until you to integrate it with a translation-accountable approach, manage it with content management technologies, and drive it with an international market strategy. Summary The opportunity for huge savings in your content processes (especially translation) is to never author another word until you define an authoring process that includes a translation-accountable approach. Writers need to understand the cost that their changes incur and need to be held accountable for the impact their changes and choices have on the translation process. We reaped a 290% ROI the first year we implemented our single-source solution at JD Edwards. You can reap these kinds of savings as well by retraining your authors and reengineering their processes. Getting this level of return also requires leveraging content management technologies which support the optimized authoring processes. Our biggest gains were in managing relationships between our source and target chunks. In addition, intelligently managing linking within source and the links at cloning time, were huge time and money savers for us. Lots of maturity has occurred in the content management arena and the path to big savings is having the right vision and driving a successful implementation. The often overlooked factor in successful multi-lingual content applications is having a translation strategy aligned to the international marketing strategy. It took several iterations before we at JD Edwards were able to get the right leadership team assembled and knowledgeable enough to drive priorities and directions for translations and localizations. Without this executive-level strategy, you run the risk that all the savings you achieve are nullified by the inattention of your market analysts. You could have efficient and cost-effective processes when it comes to delivering translated content but end up with content products that no one in your language markets wants. There are lots of cul-de-sacs on the highway to translated content and there are more spin-outs and stalled-initiatives than successful initiatives. You would do well to seek out those who have been successful and learn from them. When exploring their success, take note of their authoring accountability practices, their content management technologies, and their executive-level, translation strategy.