A man who needs no introduction, Renato Beninatto writes in his January post about the zeitgeist of the industry, as he does every year. As Renato puts it, 2013 might be a year for evolution, rather than revolution, when things are settling down. While we are standing in anticipation for the next big thing, let’s put on our mythbusting cap, and look into the tropes of today’s localization world. In the upcoming series of posts, we check on the evolutionary state of the most prevalent, already mature concepts. And when the revolution comes, we hope they won’t be the first against the wall either.
The concept of collaborative translation is old as time, but still not trivial to pull off, keeping quality, productivity and value/cost ratio high at the same time. Even a well-orchestrated project may fall short of the expectations if all the usual elements of linear localization are treated as a given.
While a well-defined structure and an online environment is indispensable to ensure that terminology, meta-data, background and context information are available for everyone to discuss or refer to in the project, the challenges of collaborative translation cannot be addressed by relying solely on such predefined methods. With all the automation inherent to today’s processes, it’s easy to think of the ones pulling the weight, the translators as resources.
Gábor Faragó, wizard and mastermind of the fanciest things at espell has a spot-on analogy. Let’s assume that a translator has 10 coins of attention that (s)he can apply during the translation process. Until our tools become smart enough to understand intentions, the translator has to deal with technical and administrative tasks as well, which can be considered a net loss. 1-2 coins are spent on firing up the translation tool and other technical tasks, 1-3 coins on research as well as reading and understanding the references/instructions, and it takes another 4-5 to translate. This is the best case scenario, usually another 1-2 coins are needed to solve issues, understand new requirements, communicate with the project manager, collect queries, make comments, implement changes, and so on. The available resources already exhausted, any further work taxes the translator’s attention.
When it comes to collaborative projects or content that requires imagination, not just experience (later on this in another post), liberating a few coins is essential. The default linear setup doesn’t take into consideration the extra workload the translators need to face, especially if there is no editor, or multiple editors are involved.
There are several tweaks, though, that go a long way in terms of quality, transparency, attention to detail as well as productivity.
The linear model assumes that it is the translators’ responsibility to deal with inconsistencies in style and terminology on-the-fly, even if in the form of retrofitting their translation. Apart from alienating the participants of the project, issues of this type are usually time hogs and hard to unscramble, moreover, they cannot be prevented even by the most rigorous style guides or termbases. Therefore, it is usually worthwhile to shift not directly translation-related work to a dedicated person, and to involve translation management more.
A localization lead should be someone who is not tasked to translate, and preferably a linguist with expertise in the subject. His/her task during the process can be quite complex:
- Preparing a brief overview to complement the generic style guide. Most issues crop up during translation because the translators are not in a position to grasp the entirety of the content. Needless to say, when they are in for a translation of 100-200k overall, they don’t even have a chance. The localization lead doesn’t have to read every word of course, but needs to understand the big picture enough to be able to convey it to the participants. Moreover, having a localization lead on board who is familiar with all the aspects can be conducive to meaningful communication and can help the translator feel more involved.
- Working together with the client (if necessary) to author a style guide.
- Organizing a kickoff call. Instructions on paper/screen are as good as forgotten. A brief (really) kickoff call will not only sort out misunderstandings, but more importantly, renders the project personal. Sensitive translation work is still done by people, and giving projects a face and purpose is an important, often overlooked soft issue.
- Answering/merging/processing queries in a single channel. Centralizing and deduplicating questions is a quantifiable boon, because it prevents overburdening the client, minimizes backtracking, and unveils dependencies in time.
- Managing and compiling the terminology database(s). If (s)he is a subject matter expert, (s)he can be responsible for compiling the source and manage/oversee the terminology translation once the set is approved.
- Managing validation. 3rd party or client validators feel more involved if a dedicated person is in touch, which, again, is a soft issue, but arguably important to prevent having a vicious circle of discussions on preferences or evaluating quality.
Language leads are involved in the translation process, and responsible language-specific tasks, such as:
- Translating compiled glossary
- Localizing the style guide
- Answering queries that don’t need the client’s engagement
Agile content setup
It’s not easy to set large volumes of content in stone – and it is bound to be modified, reviewed, cross-referenced, sent back, queried, lost, found, subjected to inquiry, lost again, and finally buried in soft peat for three months and recycled as firelighters. While there is no known solution for resurrecting such material, some measures can be taken to prevent a change disrupting the whole localization process. The actual implementation varies, but a way to manage rolling updates and resolve dependencies is to prepare all content to reflect meaningful, unique and coherent context information. This is a crucial prerequisite to flawless pre-translation (by leveraging existing translations), as the context may affect inflection and the same source may have several different valid translations. Moreover, customizing a connector that plugs into the system can take the edge off of the delta updates by diminishing manual work.
And finally, the usual suspects are all lined up:
- Integrating quality control. Fine-tuning the QA engine for the different content types provides instantaneous feedback for the translators. Moreover, additional QA checks need to be run after each batch.
- Terminology framework. The terminology is best plugged into the online translation framework, allowing every party to see changes and additions, while restricting update and review rights for the terminology lead only.
- Environment setup. To facilitate teamwork, high translation output, and consistency, a server-based translation environment cannot go amiss, where all resources were available on the fly. Such a setup also provides an online validation mechanism for the client to review the most sensitive parts.
Collaborative translation is a buzzword, and it is thrown around not as a method, but as an advantage. Naturally, it can be both, but it isn’t the Holy Grail of localization. While this whole post laboured the point of implementing it in a specific way, the myth really is that collaboration can single-mindedly offer all-encompassing solutions.
Prepare for more point-labouring in two weeks!