informations bibliographiques

et lexicographiques
bibliographical and
lexicographical information

Williams, Malcolm, Translation Quality Assessment: An Argumentation-

Centred Approach. 2004. xix +188 pages. Ottawa: University of Ottawa
Press, 542 King Edward Avenue, Ottawa, Ont. Canada K1N 6N5. ISBN
0–7766-0584-4. Price $29.95, £20.00. Perspectives on Translation Series.

Reviewed by Ahmed  Seddik Al-Wahy, Associate Professor, ­ Faculty

of Languages (Al-Alsun), Ain Shams University, Cairo, Egypt. Email:
­[email protected]; [email protected]; [email protected].
edu.eg.

This book proposes a model for translation quality assessment (TQA) based on ar-
gumentation theory, as the theory developed within the framework of discourse
analysis. Justification for this approach to TQA is adequately provided in the
book’s Introduction, which states that, in the field of translation, there is a state
of “assessment chaos” (p. xiv), and lists at least ten problems and issues that pre-
vent consensus in TQA, ranging from disagreement on basic notions such as qual-
ity and accuracy to problems with rating and quantification. In addition, most
existing TQA models are microtextual, i.e. they focus on the sentential/subsen-
tential level rather than evaluate coherence and the logical development of the
text as a whole. The proposed model overcomes such shortcomings by prioritizing
the macrotextual, discourse level in the assessment process, without ignoring the
smaller units such as words and phrases.
Attempts to incorporate the macrotextual level into the assessment process
have generally been confined to translation training, especially with reference to
journalistic and literary texts. By contrast, Williams shows with authentic examples
that his proposed model is suitable for assessing instrumental translations – i.e.
translations that serve short-term communication purposes and in which aesthet-
ics play a minor role. In this context, translation is regarded as a language indus-
try, to which industry-related concepts such as total quality, quality assurance, and
professional standards apply. Unlike other discourse-based models, the argumen-
tation-centred model is designed to assess translations produced in an institution-
al context. Williams makes it clear from the start that his model is not intended for
literary, religious, or philosophical texts and that he will not ­tackle issues such as
Babel 56: 1 (2010), 95–99.  © Fédération des Traducteurs (fit) Revue Babel
doi 10.1075/babel.56.1.07wah  issn 0521–9744  e-issn 1569–9668
96 Informations bibliographiques et lexicographiques

fidelity, adequacy, or acceptability that are commonly raised in translation stud-

ies. Apart from such issues, the model is based on the conviction that a translation
should ideally be accurate and should read as if it was originally written in the tar-
get language.
The book falls into two parts, the first of which develops the argumentation-
centred model, while the second tests and refines the model and defines translation
quality standards. The review and comparison of the main TQA models in Chap-
ter One show that all the main approaches share one common feature, which is the
categorization of errors, though they differ in their concept of categorization. Some
models are quantitative and microtextual, while others are more textologically-ori-
ented. Similarly, some models are standards-referenced, while others are criterion-
referenced.
Chapter Two presents the theoretical background of the argumentation-cen-
tred approach, which is based on the belief that argumentation is an essential com-
ponent of all texts, even those that may seem to be purely informative or that are
traditionally categorized as narrative, expository, or descriptive. The model has two
components: argument schema and rhetorical topology. The argument schema has
four obligatory elements: the claim (the main argument toward which other elem-
ents converge), the grounds (the information supporting the claim), the warrants
(statements showing how the grounds are connected to the claim), and the backing
(the overarching principle governing the issue at hand). In addition, there are two
optional elements: qualifiers (or modalizers, which enhance or mitigate the force of
the claim) and rebuttals (statements that contradict the supporting arguments).
The rhetorical topology component, which is discussed in detail in Chapter
Three, is concerned with the macro-elements of the argumentation structure of the
source and target texts. There are six elements of rhetorical topology: organiza-
tional relations, connectives, propositional functions, types of argument, figures of
speech, and narrative strategy. Organizational relations are concerned with the ar-
rangement of speech acts to achieve the purpose of the text. Connectives are con-
junctions and other inference indicators that help the progression of ideas and sig-
nal logical relations between them, thus giving the text its cohesion. Propositional
functions, on the other hand, give the text its coherence and are represented by lo-
gical relations between propositions. Types of argument, such as definition, com-
parison, relationship, circumstance, and testimony, play a key role in determining
the reader’s response to the text and, therefore, occupy an important position in a
full-text TQA. Figures of speech, which are traditionally viewed as ornamental, aes-
thetic elements, are considered to be essential to various kinds of writing and to
play a key role in argumentation. Finally, the narrative strategy determines how the
author of the text reveals or hides his/her presence, through personalization/deper-
sonalization and the use of qualifiers.
 Bibliographical and lexicographical information 97

Chapter Four begins with defining critical, major, and minor defects as the
terms are used in industrial quality control, and then discusses how these can re-
late to TQA.Williams suggests that judging whether a certain defect is critical or
not depends on a number of factors, including the text type and purpose and the
expected readership. Generally, however, a critical defect in his view is one that af-
fects the core argument of the source document.
Chapter Five, the first in Part II, tests the model on two instrumental texts. Af-
ter analysing the source text in terms of its argument schema and rhetorical topol-
ogy elements, the target text is read to see if such elements have been accurately
rendered. Each element is discussed separately and then the results are shown in
a grid. The discussion shows that errors that may be regarded as critical or major
under other quantitative, microtextual TQA approaches can be regarded as minor
in the argumentation-centred model (and the reverse is true).
Based on this analysis, Chapter Six adds further refinements for optimizing
the model and incorporating a rating scale in it. For instance, the analysis shows
that there is a need to define the degree of seriousness of error. Williams, therefore,
distinguishes among critical, major, and minor defects. A critical defect is one
that affects an element in the argument schema (thus impairing the usability of
the translation). A major defect is one that affects an important part of the micro-
text, but not the argument schema. Other transfer defects are considered minor.
The analysis also shows that the parameters of arrangement, organizational rela-
tions, and figures of speech can be dropped and reserved for specific purposes and
that some parameters can be grouped together at one level of evaluation. In add-
ition, the analysis indicates that the model needs to incorporate other parameters,
such as terminology, style, usage, and syntax, depending on the text type, field,
and purpose. Williams finalizes his model by distinguishing between two sets of
parameters: core and field/use-specific. The former (comprising argument schema,
propositional functions, connectives, arguments, and narrative strategy) apply to
all kinds of instrumental texts, while the latter (including terminology, figures of
speech, and target language quality) apply only when needed, at the evaluator’s
discretion. As for rating, Williams begins with a preliminary scale with two evalu-
ation grades: “satisfactory” (for translations having no defect affecting the argu-
ment schema) and “unsatisfactory” (for translations having at least one defect af-
fecting the argument schema). This rating scale is further developed by weighting
factors on percentages. For example, the argument schema is assigned 30% of the
total evaluation, but any translation must be free of argument schema defects in
order to be acceptable. Other parameters are usually given lower percentages, de-
pending on the type and purpose of the text being examined.
Chapter Seven develops a translation quality standard based on the proposed
model. Williams uses the term “standard” in more than one sense. For instance, he
98 Informations bibliographiques et lexicographiques

uses the term to refer to the levels of quality to which professional translators are
expected to conform. He proposes a three-grade standard scheme: (1) publication
standard, (2) information standard, and (3) minimum standard. In all three stand-
ards, the text should accurately render all components of the argument schema
(which is the only requirement of the minimum standard). In the information
standard, the text should further meet the requirements for selected core and field/
use-specific parameters. The publication standard, which is the maximum stand-
ard, requires the text to comply with all the above in addition to meeting all the
requirements for target-language parameters. Another sense in which the term
“standard” is used is the evaluator’s judgement of the quality of the translation. In
addition to these standards, there is also the label “substandard”, which refers to
texts that fail to render the argument schema and/or do not meet the requirements
for at least one core or field/use-specific parameter. This is not a level of quality
that professionals are expected to conform to, but the evaluator’s judgement of the
quality of the translation.
One of the advantages of the book is that it takes the reader step by step through
the development of the model, from its embryonic stage to its final, refined shape.
The critical review and comparison of TQA models highlight the need for an ar-
gumentation-centred model that can overcome their shortcomings by focusing on
the discourse level. After the theoretical background is provided, the model in its
preliminary form is developed and then tested on a number of instrumental trans-
lations of different types. It is then refined and streamlined in the light of practical
application. Finally, a rating system is developed and a quality standard is set. The
Terminology Appendix and the Further Reading section make the book equally
useful to translation students, professionals, and evaluators who may consider ap-
plying the argumentation-centred model.
There are a few misprints in the book, such as the use of or instead of of in
“Concern for excellence in translation or [sic] literary and religious works dates
back centuries” (p. xiii). There are also typographical errors related to cross-refer-
encing, especially in Chapter Three. For instance, Williams promises to discuss the
narrative strategy in Section 3.10. (p. 41), which never appears in the book, but he
discusses it in Section 3.7. The reader is also referred to Section 3.3.6 for a “preced-
ing reasons” list (p. 43), while the actual list appears in Section 3.3.3. As a reader,
I would also appreciate it if acronyms (such as “ARTRAQ”, which Williams uses to
label his model) were explained with reference to their original phrases. Such mi-
nor misprints do not at all belittle value of the book and can be corrected in future
reprints.
The TQA models reviewed, the texts analysed, and the examples given may
suggest that Williams is mainly concerned with the language pair French and Eng-
lish in a Canadian setting. However, the theoretical framework is universal and the
 Bibliographical and lexicographical information 99

model can be applied to texts from any language. One problem, however, could be
that the relevant French sections in the analysis are not glossed, which may make
them difficult to follow for readers who do not speak French.
As presented in the book, the argumentation-based model has many advan-
tages in comparison to other TQA models. To begin with, the proposed model
covers both the macrotextual and microtextual levels and attends to the macro-
structure (the content) and the superstructure (the schema) of the text. It also
solves the problems of sampling by selecting passages that include elements of the
argument schema. The model is also theory-based and therefore has the relative
advantage of being objective, consistent, and reliable. In its refined form, the mod-
el is flexible enough to assign different weights to different elements according to
the nature of the text under investigation and is not as complicated in its calcula-
tions as other models.
In my view, the book is an important contribution to the literature on TQA.
It is a valuable resource for trainee and professional translators, translation eval-
uators, and researchers with an interest in the evaluation of non-literary transla-
tion. The book has shown with authentic examples that the argumentation-cen-
tred model is appropriate as a tool for assessing the translation of any kind of
instrumental text. However, full benefit of the book can only be gained by prac-
tical ­application of the model by translation quality evaluators in real institutions.