Language Quality Management: 7 rules to ensure quality in your global contentcotuong
Quality is a concern for anyone who processes and produces texts. However, there are often discrepancies when it comes to the definition of quality. An individual’s predispositions and expectations play a decisive role. In this article, we want to demonstrate which traceable criteria can make quality verifiable.
Thucydides once said that beauty is in the eye of the beholder. Any kind of subjectivity, however, is problematic in the “industrial” process of producing global content. So, customers, service providers, and translators should opt for predictable quality with calculable risks. The seven rules set out below make quality measurable, and thus manageable.
It is obvious that the linguistic requirements of a key brand message are higher than those of an internal memo. It would not make economic sense to demand the same quality standards. Therefore, different benchmarks with error rates and expectations on translation and review should be defined in “content profiles” based on the text type. To apply these benchmarks to the translation quality, they must be defined and made clear before the translation is completed. This leads us on to the first rule:
Set out the expectations by means of different content profiles, which include quality factors, key performance indicators, objectives, and benchmarks. Numerous industry standards and initiatives, such as TAUS DQF, can be used as a guide for defining content profiles.
In addition, other instructions like style guides, key messages, and terminology are also required as linguistic, stylistic, and branding guidelines. Terminology helps ensure to deliver a clear and comprehensible message in all languages. By providing different ways to access terminology and always keeping a feedback channel open, you will ensure collaboration is ongoing. This brings us to the next rule:
Encourage active and sustainable terminology work and ensure that all stakeholders can use the terminology quickly and easily.
During translation, questions are regularly raised, as good translators frequently pinpoint ambiguities and even errors in the text. These queries have enormous potential for improving the translation itself and the entire process associated with the creation and use of the text. To exploit this potential, all queries must be consistently and systematically gathered, validated, and made available for future projects as quickly and easily as possible. Here it is important that insights from the queries are communicated in a structured manner to all those involved in creating and translating the text. With this information, you add a very insightful link in the quality chain.
You must ensure that all stakeholders can access and process the queries simultaneously and in real time. Two key features, on which you should place high importance, make this possible: collaborative, online access and the ability to delegate all tasks via a real-time, traceable workflow. This is a much more efficient way of working compared to conventional Excel or email-based systems. This brings us to the third rule.
Exploit the quality potential in translator queries using a well-thought-out, structured management process.
Once the translation is complete, it can be evaluated based on the factors and expectations laid out in rule 1. At this stage, reviewers should assess to what extent the translation meets the requirements based on the customer- specific, predefined criteria. Every edit a proofer or reviewer makes to the text is documented with a quality criterion. One basis for this would be adapting and simplifying the harmonized DQF/MQM Framework in line with the essential criteria determined in the respective content profile. The exact weighting of criteria, such as mistranslation, terminology error, grammar etc., is then adapted to the content profile; for example, for internal documents, “style” can be considered “irrelevant to the evaluation”. The different weighting given to the various criteria results in a clear quality figure extrapolated from the amount of text assessed. This forms the basis for process decisions going forward.
This process yields strategic, analyzable, and historic key performance indicators, giving you a precise and effective way of managing and monitoring quality, broken down by language, text type, translator, reviewer, etc. as well as strategic input for short-, medium-, and long-term process improvements.
Turn the frustrating review process into strategic quality management with “business intelligence” and associated control mechanisms for linguistic and translation quality.
Review and proofing are time-consuming. To bring further process improvements that enable you to achieve your translation objective more quickly and to a higher quality standard, we have three further rules for you.
As not all texts need to meet the same quality requirements, content profiles can be used to define the different criteria and benchmarks for the respective text types. One major advantage of this approach is that for some types of text, no more than 10% needs to be reviewed to yield a meaningful figure. A cleverly calculated spot check is often sufficient and reduces the review requirements hugely. Modern review systems possess this functionality.
Use content profiles and sampling to save time, frustration, and money during the review!
So far, we have consciously left out supporting technologies. However, only the assistance of software enables company-wide processes to be managed and, above all, scaled. This starts with expert systems for collaborative, company-wide terminology management; takes us on to browser-based systems for collaborative working on texts, including communication, feedback, and proofing; and ends with visualization and evaluation of the key performance indicators yielded from the entire process. Targeted data analysis allows us to identify weaknesses in the process at an early stage and to manage risks accordingly. For example, “internal” quality figures generated from a review can highlight lacking processes, unclearly defined responsibilities, or even problematic suppliers.
“External” key performance indicators, like access rates to certain content in different languages, combined with data from terminology management, for instance, can also provide valuable content-related feedback on the use of various terms. Targeted A/B testing can give some indication of which style works best or what level of localization is required. These interesting and relatively unexplored approaches lead us on to:
Gather and visualize process data to optimize the entire system.
Software solutions and enabling their compatibility in the process play a key role as transferring content from a CMS, PIM or ERP system into formats suitable for translation requires many different steps which are cumbersome, time-consuming and error-prone. Therefore, it is extremely helpful to analyze the process and precise needs for automation and specify details like linguistic components, such as translation memories, termbases, file and folder handling.
Here, it is an especially smart move to use existing standards, such as the German COTI, or even contribute to defining them. The GALA initiative TAPICC stands out at an international level, whose goal it is to simplify the exchange of data between all those involved in the localization process.
Automate routine tasks. By standardizing and speeding up projects, you will relieve the project management burden and save on expenditures as well as time.
Following these seven rules makes quality measurable, transparent, and predictable. With the quality results yielded, you can actively manage all process steps in a targeted manner. Quality is no coincidence, but an interdisciplinary matter that must be systematically planned, monitored, and managed. By providing you a way to fine-tune specific parts in your process, the investments in active quality management also reduce the overall costs of assured quality.