The Value of Effective Training in Translation and Localization
By: Stephen Doherty (Centre for Next Generation Localisation)
03 September 2013
Training translation and localization professionals in the critical use of cutting-edge technologies not only adds value to their work, but has many positive knock-on effects for all stakeholders.
High-Quality Training for High-Quality Translation
In times of lean and agile workflows and reduced resources for translation and localization departments, we ask ourselves: What is the rationale for training? How can training expenditure be justified? What is the return on investment?
Training can be a benefit for buyers and vendors and their employees in terms of productivity and quality gains, and the overall enhancements in work culture and innovation. While most translators will already have some type of formal and/or on-the-job training, being given the opportunity to learn and use state-of-the-art methods and tools can be a great attraction.
Approaches to learning outcomes typically fit into three categories: cognitive, skill-based, and affective. In the case of translation, cognitive outcomes take the form of knowledge and application of translation strategies, and adopting different rules to editing content given its source (e.g. post-editing MT, see below). Skills relate to the understanding and usage of software programs such as CAT tools, where a critical knowledge in applying tools is essential, so that both the translator and the organization overall know what a tool can and cannot do, and, more importantly, how the translator can make effective use of the tool to make gains in productivity and quality. Lastly, affective outcomes concern the attitudes and beliefs that the translator and organization may have about a certain method or technology. A case in point is using the construct of self-efficacy, or empowerment, to teach translators about how to best use statistical machine translation in their work if it fulfils their needs from job to job. Again, with an emphasis on learning what they can and cannot do with SMT and how, by being ahead of the technology curve, translators and their employers can gain tremendously, much like early adopters of translation memories in the early 90s.
Assessing and Designing Training Content
A needs analysis will give a systematic and comprehensive understanding of the gaps that exist in the current training program (if any), and what key skills and knowledge will add strategic value to the organization. Sometimes it may be difficult to strike a balance between learning a new technology and the adopter finding value in return. However, subscribing to a student-centred approach of training allows input from the translator into what it is missing from their perspective, e.g. a small technical modification to a translation workflow may result in higher throughput, but only the translator may spot the loophole and know this. Building upon the core competencies of excellent reading, writing, translating, and quality assessment, the presence of a meaningful and rewarding evaluation of training outcomes adds further value. Perhaps the case study for the translation project comes from an upcoming client request, perhaps the translator can now teach others what they have learned, perhaps the practice materials for quality assessment are based on an existing job, or perhaps the takeaway is the creation of empowerment to critique existing methods and technologies.
On a micro-level, each individual's job performance benefits directly from relevant and effective training. Performance is the interaction between declarative and procedural knowledge and the self-motivation moderated by the work environment, and we express it in the balance between the effort put into the job at hand and the (potential) rewards to be found. Mismatches where needs and expectations are fulfilled lead to decreased performance and higher work stress, etc. In other words, training can help to support this balance, and encourage the engagement of knowledge into application with tangible returns for the individual translator. On a macro-level, effective training creates a more positive and inclusive organizational structure. The process of inclusive training engages and values individuals within the organization and their contributions to its development. Obviously there are limitations to this, but the locus of training provides an excellent mechanism for education and innovation at many organizational levels while not being resource heavy for management, e.g. a translation project manager allowing a translator with expertise in terminology to lead a workshop for others on this topic.
Training Translation and Localization Professionals
Translator training institutions and programs increasingly recognize the importance of technology. In the last decade there has been a rise in the number of courses for translators focusing on CAT tools and beyond. Some of these lead up to certification endorsed by translation technology vendors, which enrich CVs and demonstrate that translators are market-ready. Translator trainers and students are increasingly aware that it is vital to invest in these areas, so that new professionals entering the marketplace are equipped with the skills required to successfully join the production workflows of LSPs and localization providers or to establish themselves as freelance translators.
With CAT tools now widely accepted both in the profession and in educational institutions, the uptake of MT training courses has followed a similar pattern in response to the expectations of the market. The growing relevance of statistical MT as a viable option to produce quality translation for an ever expanding set of language combinations, content types and domains opens up a range of training needs. Translation graduates must now be able to readily identify which content could be translated effectively with MT, which can be easily post-edited, while taking traditional factors into account: purpose, perishability, visibility for intended readers, and standards/legal requirements, etc.
From Training for CAT Tools to Training for MT
With the evolution from one-size-fits-all free online MT services to tailored statistical MT systems, technology-oriented translators can advise us on many topics, including: legacy resources that can be deployed to create domain-specific MT solutions (e.g. translation memories or parallel documents held by a company or LSP), as well as suggest ways to acquire further parallel data for the training of customised MT systems, for instance from the Web. The successful deployment of statistical MT relies on further stages of fine-tuning and quality benchmarking according to the specific contents requiring translation, here too, the expertise of the translator provides valuable inputs.
A number of universities specializing in translator training provide courses covering these aspects and simulate the whole cycle of applying MT to texts in a variety of domains. Graduates who benefit from these innovative learning experiences can collaborate with companies and LSPs in coming up with creative technological solutions to meet the challenges of translating more efficiently when faced with increasing volumes of content. This, in turn, increases the need for a skillset in post-editing that includes: bringing MT output to publishable standards, checking the overall accuracy of MT for gisting purposes, and feeding back into the MT process itself - all of which cannot be achieved without effective training.
Training for MT Post-Editing
Until recently, post-editing skills had to be acquired on the job by translators or linguists working in institutions or companies that were early adopters of MT. Similar to what has eventually happened for CAT tools, many universities have now recognized the importance of providing this training for their students, so that employers and professionals do not have to carry the basic costs of further training after graduates start their careers. Hence, there is a trend for translator training programs to also incorporate specific elements of MT post-editing in their teaching. A by-product of this is that students become more sensitive to issues of quality requirements, time pressures and streamlined production workflows that are important in the industry today. Cutting-edge approaches now even make use of post-editing data and translator feedback and evaluation outputs, so that MT can be further improved by incorporating the human in the loop. Thus, acquiring skills leading to the successful use of MT during training, including familiarity with post-editing techniques, goes hand in hand with teaching students the fundamentals of quality assessment and evaluation techniques in translation.
Training to Take Home
In short, the above has provided a range of approaches to the provision of training for translation and localization professionals, where the scope for customization is limitless. Key aspects to effective and valuable training interventions are the alignment of the organizational and practitioners' needs, the adoption of a student-centred approach with meaningful and rewarding outcomes, and the wider fostering of a more inclusive and innovative work culture. In this way, high-quality training can easily translate into high-quality translation and localization services with increased productivity, throughput, and longevity throughout. On a more personal level, the translator gains empowerment in the face of ever-changing technologies, critical thinking and problem solving skills, and augmented interpersonal communications with stakeholders of different natures, e.g. tool designers. In this sense, the translator as the user of a tool can provide honed and valuable user feedback and recommendations to developers and managers in an environment where they are self-motivated to contribute as their views and feedback are welcomed and sincerely valued.
There are several sources of information and training materials that focus on the translation and localization industry. One such outlet is QTLaunchPad, a European Commission project that has been working to provide training materials in cutting-edge approaches and technologies including: multidimensional quality metrics for translation quality assessment, the use of quality estimation for effective MT quality prediction and sampling, and the integration of user-centric and open tools for the translation industry - these can be found on the project's website at: http://www.qt21.eu/launchpad/content/training
Further reading can be found in the articles below, all of which are freely accessible via their links:
- Doherty, S. & Kenny, D. & Way, A. 2012. Taking statistical machine translation to the student translator. Proceedings of the 10th Biennial Conference of the Association for Machine Translation in the Americas. San Diego, CA.
- Doherty, S. & Moorkens, J. 2013. Investigating the experience of translation technology labs: pedagogical implications. Journal of Specialised Translation, 19, pp. 122-136.
- Kenny, D. & Way, A. 2001. Teaching Machine Translation and Translation Technology: a Contrastive Study. Proceedings of the MT Summit VIII Workshop on Teaching Machine Translation IAMT/EAMT, Santiago de Compostela, Spain, pp. 13-17.
- Lagoudaki, E. 2008. The value of machine translation for the professional translator. Proceedings of the 8th Biennial Conference of the Association for Machine Translation in the Americas. Waikiki, Hawaii, pp. 262-269.
Stephen Doherty, BA, HDip, PhD, MBPsS, is a post-doctoral researcher in the Centre for Next Generation Localisation in Dublin City University. He conducts research on topics of language and cognition, human-computer interaction, machine translation, and translation technologies. He is currently working on QTLaunchPad, a collaborative European research initiative dedicated to overcoming barriers in machine translation and language technologies.