Defining and measuring linguistic quality

Quality is a word which is thrown around loosely, in many different contexts. What one person considers to be quality, another may not.

This is especially true with something as subjective as translation. People interpret language differently, and translation quality is often judged on subjective criteria such as style and choice of terminology. Some aspects of translation are are objective, however. ‘Yes’ translated as ‘no’ is clearly wrong, for example.

The Project Management Institute’s A Guide to the Project Management Body of Knowledge (4th edition) defines quality as “the degree to which a product meets the specified requirements.” The International Organization for Standardization says quality is “determined by comparing a set of inherent characteristics with a set of requirements. If those inherent characteristics do not meet all requirements, a low or poor level of quality is achieved.” Both of these definitions include the concept of quality as a scale, with varying degrees, and also state that there must be a set of defined specifications to provide a point of reference on the scale.

Just because a translation accurately conveys the intended meaning of the original source text does not necessarily make it good quality – a quality translation is more than just maintaining meaning – it has to meet the defined specifications and be fit for purpose.

Linguists cannot always guess the preferences of the end client, so they need defined specifications to know what ‘quality’ is for every particular project. For example, does the translation need to use specific terminology (white blood cells or leukocytes)? Does the translation need to be concise so that it will fit into a pre-set brochure layout? Do foreign acronyms need to be translated in footnotes? Do currencies need to be changed? The list goes on and on. Translators also need to know the context of the translation in order to ensure it is fit for purpose. How will it be used? Who is the target audience?

Clients can provide these specifics through source language glossaries, bilingual translation glossaries, style guides, reference materials and notes including the context, target audience, and any other quality requirements.

Because quality is a scale of varying degrees, we also need to be able to measure quality, as well as defining it. There are a handful of systems measuring linguistic quality, including the Localization Industry Standards Association (LISA) QA model, and the Canadian Language Quality Measurement System. A good system should include error types (spelling, grammar, omission, style, terminology, etc) and severity of the error (with corresponding definitions of severity levels) as well as weightings for the different types of errors. Assigning the severity of an error can be somewhat subjective, and proper specifications are necessary for reviewing style and terminology, but such scales can help us to determine what is excellent / good / poor.

At Web-Translations, we support the Quality in Translation initiative. Quality Control steps are included within the framework of our proactive Quality Assurance strategy. All translations are proofread by a second native linguist, and everyone in the team, from marketing and sales to the project management team, is focused on providing excellent quality translations.