Quality Evaluation in Localization: Top Three Factors Creating Change

MiguelMiguel Gonzalez is Quality Director at Welocalize. In this blog, Miguel talks about recent changes in localization and how they are affecting our approach to quality evaluation (QE). He also selects his three must-see sessions for the upcoming LocWorld 2014 in Dublin.

During the last few years, the localization industry landscape has undergone profound changes. Traditional approaches have been challenged for matching content, talent and QE models, highlighting the need for more sophisticated, targeted and integrated approaches. Why? What factors have caused these changes?

  1. The rise of social media and its adoption by companies as a key sales and marketing tool has sparked an exponential growth in community and user-generated content (UGC), which is way beyond the capacity of the traditional translation supply chain.
  2. Emergence of new translation production methods like machine translation (MT), crowdsourcing and community translation has given companies the ability to translate mid to low-impact content in a faster, more cost-efficient manner. It has also allowed the appearance of a “parallel” supply chain of non-linguistic, low-cost resources.
  3. Quality requirements for high-impact content have become more stringent. Companies expect translated content not only to capture their particular voice and style, but also to elicit the same emotional and commercial response from the end-user as the original text. We are seeing a rise in demand for transcreation, not just translation.

As a result, the increasingly transactional nature of localization cycles means that traditional QE models are too cumbersome and detached from today’s fast-paced, on-demand production environment. The traditional classification of content into components such as UI, UA and .doc has lost most of its relevance. Conventional QE and resourcing models are becoming obsolete as they were devised with only a few different content types in mind and considered only a relatively homogeneous supply chain, exclusively based on human, professional translation.

New, modern localization frameworks need to take into account a multiplicity of factors:

  • There is a broad range of content types and not all content is created equal
  • What is the content domain, sub-domain and similar metadata?
  • How is the content created?
  • How will it be translated?
  • By whom?
  • How will it be delivered?
  • How frequently and how fast?
  • What is the target audience for each type of content?

A structured way of answering these questions would be the creation of a content categorization taxonomy that can guide your production processes and decisions. Create a matrix where each combination of the factors above is assigned the appropriate QE level and assessment instrument, as well as the precise skill profile which best suits each category.

The ultimate challenge would be to dynamically integrate the content categorization and talent profiling matrix into your production processes and build them into your TMS/CMS systems. This would allow the content to be:

  1. Accurately and automatically tagged and categorized from the outset
  2. Channeled through content-aware workflows to the appropriate translator with the right skill-set, experience, level of domain specialization and other factors
  3. Routed to the most suitable QE validator via the most appropriate QE level and instrument for error typology, adequacy and fluency, readability score, end-user acceptance test and more.

Achieving the right alignment of content, talent and QE levels will go a long way to addressing some of the more pressing challenges of our industry, including the efficient processing of vast amounts of diverse content and the management of an ever-expanding supply chain. It will also help your company obtain better, more consistent results in terms of quality performance and return on content.

Welocalize will feature our expertise and share findings at Localization World 2014, in Dublin June 4-6. Quality, quality evaluation and talent management are hot topics. Here are three must-see sessions for LocWorld 2014:

  • Translation Quality Evaluation Summit: Welocalize’s Olga Beregovaya and Lena Marg are featured speakers at this Localization World pre-conference session on June 4. A panel of experts will be discussing four main topics relating to QE: Quality Estimation, Crowdsourcing, Readability/Usability Evaluation and Sampling for QE.
  • Language Quality ManagementModels, Measures, and Methodologies: How Do Others Do It?: Experts panelists from Snap-On Inc. and Microsoft will discuss the challenges of assessing translation quality on June 5.
  • Attracting and Developing Talent for our Growing Industry: At this half-day pre-conference session on June 4, a team of experts look at how the industry needs to proactively manage the supply of professional staff or face with a severe talent shortage. Panellists include Welocalize´s language program architect, Katharina Zimnoch.

Ready to talk quality?