Terminology management in practice: Real world example (Part 1)
Our third SDL Trados webinar summary comes from Silvia Cerrella Bauer, Managing Director of CB Multilingual Ltd. She has been freelance translator, conference interpreter, and certified terminologist, with a background in corporate communications and business administration and extensive experience in the translation and terminology fields.
I have to say that one of my first contacts with terminology was made via webinars that I signed up for and one of the first ones was Silvia’s. I have shared several of her resources in my blog, so I was very happy to see another familiar face. Special thanks also to Fiona Merwood, SDL Senior Marketing Executive, who did a great job coordinating this webinar. It was a very comprehensive lesson, so I had to divide it into two parts. I know this is a long post, but I am sure you will find it very useful and insightful.
In her presentation she talked about how systematic terminology management can efficiently support a new business process and explained how terminology management helped to save time, reduce costs, and increase quality in a localization project, which had English and German as source languages and French as target language.
She started by providing some background information about the translation department that was in charge of the localization project. It included five translator-terminologists who were in charge of translating publications (website content, brochures, graphical user interfaces, etc.) They had to deal with a number of issues such as complex jargon, tight deadlines, limited resources, big volumes, uncontrolled language, and partial consensus. In addition to the five translator-terminologists, other stakeholders included specialists who were in charge of compiling technical documents such as manuals and software.
Their short and long term terminology objectives from the beginning of the project were to achieve (i) higher translation quality and consistency, (ii) higher quality and consistency in technical documents in English and German which they had to proofread and review, and (iii) higher productivity. The overall goal was to achieve quality over quantity, which was planned to be achieved through regular updates, using recognized sources and a 6-step quality-assurance process in which every record was checked twice. They also used specific standards and wrote and documented the processes in the form of terminology guidelines.
They used ISO standards including those for term formation, although they rarely had to create terms, and it was mostly about achieving consensus on the preferred and deprecated terms in order to promote corporate language use. They also used a mixed descriptive/prescriptive approach. It was a translation oriented process (that is, terminology was managed “after the fact”: terms were entered into the database as translations were being made).
The process was organization specific. They included topics from all business areas using internal sources. Term entries were quite comprehensive and included product and service designations. Quality and user orientation were the key elements of the terminology management process.
When writing the terminology management guidelines, they made sure that the five translator-terminologists were identified as the “owners and instruments of knowledge transfer”. It was a compendium of knowledge acquired on a learning-by-doing basis. They specified the processes, the tasks, the roles and responsibilities. The guidelines were constantly updated to allow for changes, for example, to the system configuration or structure. The descriptive/prescriptive approach used to manage the terminology was also included. They provided a detailed explanation of entry structure, clues for handling multilingual terminology issues such as terminology gaps, methodology (entry validation, content management, etc.), and included several annexes related to relevant topics such as data element format rules.
Silvia presented an example of a term entry that included (i) at entry level: creation date, name of creator, entry class (see more details below), entry number, subject, and (ii) at term level: the term in the three languages, and for each language term, the source, definition, definition source, and a note that included related terms as hyperlinks. However, after a survey among staff members, the final database came down to fewer categories.
The terminology workflow included several steps: entering the term (done by a terminologist), validating the term using their 6-step quality-assurance process (done by terminologists and subject-matter experts), and disseminating and controlling the terminology (done by the terminology manager). These steps were recurring, that is, every time an update was made the process started again. Corrections and updates were related to system, model, user interface, and user feedback. As for user interface, decisions were made on what information was presented to the user and what information could only be accessed by the translator-terminologists. Once they had a “critical mass” of terms that was representative of all business areas, they made the first publication (dissemination), so that staff members could start using it as soon as possible and provide feedback to improve termbase output.
The tools and tasks used for term entry and validation included checklists, guidelines, term recognition tools via the translation memory, and reviewing tools. The tools and tasks for term dissemination and control included importing and exporting, preparing statistics, doing backups, monitoring use, and developing communication strategies to raise awareness among staff members on the work being done by the translation department to help them save time.
The validation process for each term was very comprehensive and there were seven Entry Classes (EC) assigned to every step process. For every new or incomplete term entry, the system automatically allocated EC1, and then the administrator allocated EC2 (entries imported from glossaries), EC3 (entries imported from term extraction program) and EC7 (entry proofread twice and fully validated). The translator-terminologists manually allocated EC4 (complete but not yet proofread), EC5 (same as 4 plus checked by expert), and EC6 (only proofread once).