- Translation Quality Assessment: An Argumentation-Centred Approach
Translation of 'instrumental,' or non-literary, texts is booming, and the competence and reliability of practitioners vary greatly in a largely unregulated profession. There is consensus among the stakeholders in this activity on the need for quality assessment but little agreement on the specifics of standards, definitions, and methods. To address this need, Malcolm Williams, a veteran of Canada's federal government Translation Bureau, proposes an ambitious model of quality control in Translation Quality Assessment: An Argumentation-Centred Approach, based on his University of Ottawa doctoral dissertation.
The quality control systems in place in large translation-providing organizations, such as the Translation Bureau, as well as those used in many training programs, tend to be quantificational (they count mistakes) and microtextual (they examine the words in a sentence in a paragraph, without much regard for the whole text). Like the enduringly popular criticism of journalism and other writing produced to deadline, this kind of assessment is fairly simple, in that the scope of errors is small and predetermined (comma splice, dangling participle, misused word); it is also often superficial and irrelevant and ignores major strong and weak points.
Under Williams's system, the evaluator must first identify the Argument Schema of the source text - its claims, their grounds, the connection between claims and grounds; the backing, qualification, and any rebuttal, exception, or restriction - and its 'rhetorical topology,' the linguistic means by which the argumentation is achieved. The extent to which the translation successfully renders the Argument Schema of the text determines its quality, its success as a translation of a purposeful text. A misplaced comma, a clumsy or ambiguous formulation, or even a mistranslation may be unimportant if it does not impinge on transmission of the text's argumentation; on the other hand, incorrectly rendering the rhetorical force of a conjunction or misrepresenting a qualification as part of the grounds can be fatal.
Williams illustrates his method thoroughly by going through the evaluation of several well-chosen examples, using an Argumentation-Centred TQA (translation quality assessment) grid, convincingly backing up his claim that argumentation analysis of a source text and argumentation-centred assessment of the translation 'cover all aspects of the message(s) [End Page 188] and purpose of a text. They bear on the full text, the microtext, and their interdependencies. As such, they meet a key criterion of TQA validity: measurement of a sufficient quantity of the object of evaluation so that the results … may be applicable to the object as a whole.'
Clearly, quality assessors will have to possess a high degree of analytical skill and discernment to perform this type of evaluation - not to mention a great deal of time. If assessment is to bear on the whole text and its translation, there is no way around an analysis of the source text in its entirety to determine its Argument Schema; thereafter, certain key portions of the translation can be identified for evaluation and a sample selected for microtextual assessment.
It seems unlikely that translation services would be willing to devote to quality assessment resources perhaps equal to or greater than those required for the actual translation. Williams's model provides the means to do a thorough, objective, meaningful evaluation; it does not aim to be cost-effective as well.
The latter part of the book is devoted to testing and refining the model and to discussion of a standard for translation quality. In appendices, Williams provides samples of an evaluator's mark-ups of a text and filled-in grids, as well as a glossary. The bibliography and lists of further readings are helpful (although Jane Grimshaw's book on the argument structure of predicates is not relevant, and Noam Chomsky's Rules and Representations is out of place under the heading 'Documentation and research on norms, standards, quality control, and management'). [End Page 189]
Sarah Cummins, Department of Linguistics, Université Laval