Summary of “Defining Translation Quality”
The article “Defining Translation Quality” (Koby, Fields, Hague, Lommel, and Melby) is the third article in a series of articles. This particular article starts by referencing the first two articles. It mentions that the first article talked about the scope (how broadly one defines translation) and specifications (how explicitly requirements must be stated) concerning definitions of translation and that the second concerns discussions of quality and five approaches to it (transcendent, product, user, production, value). In this article, the authors seek to address translation quality and the ramifications of definitions.
The authors repeatedly state throughout the article that they disagree on the issues being discussed. However, they start with points of agreement. These points include the definition of “customers” as “requesters” and “end-users.” In addition, the authors agree that translations need a degree of accuracy (agreement between the source text [ST] and the target text [TT]) and fluency. Perfection is not a reasonable goal; it is maximal fluency and accuracy that translators should aim for. They also agree that providers should understand requesters’ purpose and consider the audience and that there should be objective measures of translation quality.
The authors then describe a broad view of a quality translation as one that “demonstrates accuracy and fluency required for the audience and purpose and complies with all other specifications negotiated between the requester and provider, taking into account end-user needs” (p. 416) and includes “full translation…summary translation, localization, and even gisting” (ibid). The narrow view describes a quality translation as one that transfers the ST message completely (denotation, connotation, nuance, style, etc.) into the target language (TL) in a culturally appropriate way and generally reads as an original TL text. Summarization, transcreation, gisting, and some aspects of localization are not considered translation.
The authors then describe specific examples of areas in which the debate between broad and narrow views come up (the explicitness of specifications, the existence of minimum service levels, and the inclusion of specific error categories). Concerning whether or not the goal of fluency and accuracy should be achieved through agreed-upon specifications, advocates of a broad view say yes while advocates of the narrow view tend to say that it is enough to mention the text type and maybe the target audience.
Concerning minimum levels of service and whether or not these exist regardless of what requesters ask for, the authors note that customers often don’t know what a project actually requires. One question they pose is whether or not, in light of the possibility requesters may underestimate how raw and hard-to-understand lightly-edited machine translation (MT) can be, providers should meet a minimum level of fluency. Another is whether or not providers who provide a higher level of accuracy and fluency are ethically required to inform requesters. The authors suggest that, since requesters often change their minds upon receiving the final translation, providers make their translations as fluent and accurate as possible.
Another area of debate is the inclusion of specific error categories in every translation quality (TQ) metric. The authors note that there is disagreement about which specifics are necessary but agree that the existence of such categories is useful. The narrow view says specific categories (namely, meaning transfer, terminology, domain-specific writing quality, and domain-independent TL accuracy) are necessary, whereas the broad view favors negotiation of which categories are necessary. The authors suggest that the two views can be satisfied by defining general minimum requirements and allowing for parties to renegotiate those requirements according to the needs of specific situations.
The authors give two examples in which a broad view would be useful: time-sensitive intelligence intercepts (as quality control takes time that the requesters do not have) and the use of MT to match user queries to support databases (as identifying database content is all that is needed, and anything else costs time and money). They note that proponents of the broad view think that quality should reflect the scenarios in which translators may work, while proponents of the narrow view believe their view is more in line with how professional translators operate.
The authors conclude by noting that the one thing they agreed upon in the first article is that people who discuss translation quality should identify where on the broad/narrow continuum their definitions lie. They also note that another thing they do agree on concerning the applicability of quality management to the translation industry is that whether a quality management or transcendent position is taken matters and is inseparable from how one defines translation. They also agree that the definitions of “translation,” “quality,” and “translation quality” need to be laid out. The authors conclude with two statements; namely, “Translation quality metrics must be built on a well-defined foundation including at least clearly stated definitions of translation, quality, and translation quality,” and “The important debate about which kind of definition to use, broad or narrow, is not over.”They express hope that their articles will contribute to a “meaningful and beneficial debate” (p. 420).
Source: Koby, G. S., Fields, P., Hague, D., Lommel, A., Melby, A. (2014). “Defining Translation Quality,” Revista Tradumàtica, pp. 413-420; número 12