Translation Evaluation : The Suitability of the Argument Macrostructure Model for the Assessment of Translated Texts across Different Fields

Even though there exists an undeniable need for an acceptable translation among translators, translation scholars and translation teachers, the question of acceptability and the criteria against which this acceptability can be determined are still controversial. There is a lack of generally agreed criteria against which translation can possibly be evaluated, despite the fact that international as well as local standards of translation are clearly witnessed. In spite of the fact that some scholars suggest certain parameters that can be utilized for the purpose of translation evaluation, such as quality of TL, accuracy, register, appearance of TT, situationality, and so on, there seems to be no parameter on which evaluators may rely to arrive at an overall quality assessment for the TT. The present paper argues that the argument macrostructure model should be utilised as a translation evaluation parameter to determine the translation quality. This model should comprise several standards and grades peculiar to different translation contexts and situations, so that it can successfully be applied in the case of assessing translation materials within both academic as well as professional settings.


Introduction
Translation is deemed an intellectual and complex product characterized by heterogeneity (Williams, 2009, p. 7).It can be viewed as an ongoing evaluative process as it involves continuous check of TL equivalents that can stand for SL lexical items until opting for the TL most appropriate equivalents (Munday, 2012, p. 155).Even though there exists an undeniable need for an acceptable translation among translators, translation scholars and translation teachers, the question of acceptability as well as the criteria against which this acceptability can be determined are still controversial (Williams, 2009, p. 3).There is a lack of generally agreed and accepted criteria against which translation can possibly be evaluated, despite the fact that international as well as local standards of translation are witnessed (Williams, 2009, p. 3).Indeed, translation evaluation has scarcely received any attention in translation studies (Munday, 2012, p. 12) in spite of the clearly de facto interest in the role played by translation as intervention (Munday, 2007) and the importance of translation in relation to ideology (Lefevere, 1992;Alvarez & Carmen, 1996;Tymoczko & Gentzler, 2002;Calzada Pérez, 2003;Cunico & Munday, 2007;Munday, 2008).
The present paper chiefly discusses the concept of translation evaluation; it presents at the outset some definitions of evaluation coupled with certain explanations to further elucidate the concept concerned.The functions fulfilled by evaluation are succinctly outlined, along with the specification of the tests that should be passed by any evaluation model prior to being approved.The paper then uncovers a number of problems that result in unsatisfactory evaluation, showing how each of these problematic areas negatively affects the process of translation evaluation.A complete section is devoted to address the argument macrostructure model, its components, the grid proposed for the model, its application to text evaluation and its suitability for various texts across different disciplines.The present paper argues that the argument macrostructure model should be utilised as a translation evaluation parameter to determine the translation quality.This model should comprise several standards and grades peculiar to different translation contexts and situations, so that it can successfully be applied in the case of assessing translation materials within both academic as well as professional settings.

Definition and Concept
Evaluation is crucial to translation (Munday, 2012, p. 11).One general definition of evaluation is that "evaluation is taken to mean the determination of merit, worth, or significance" (Scriven, 2007, p. 1).It is evident that the above definition accords special importance to the concept that evaluation is tightly linked to the specification of value and significance.In other words, the primary task that should be performed by the evaluator is to demarcate and present the worth and importance of certain work as opposed to the demerit and insignificance of that very work.Such work with no doubt demands special interpretive skill (Lemke, 1998;Munday, 2012).Within the same line of thought, Volosinov (1929Volosinov ( -1973) ) claims that all utterances, which are structured, are unquestionably subject to evaluation.He goes to argue that each utterance is deemed an 'evaluative orientation' and has ipso facto elements, which are not only meaningful, but also valuable (p.105).This clearly indicates that utterances are not value-free, but consist of components which refer to specific entities and can concurrently be evaluated.
Three crucially substantial functions of evaluation may notably be identified.These primarily reside in conveying the views of speakers and/or writers, thus reflecting the system of values of these people and their community.Furthermore, evaluation helps in creating relations between writers and readers and/or speakers and hearers, a matter which facilitates Authors' points as well as investigates the truthfulness of particular statements.Also, evaluation plays a crucial role in organizing the discourse (Thompson & Hunston, 2000;Fairclough, 2003).Viaggio (1999, p. 125) claims that even if Toury's views on translation, which lie in the notion that a translation is whatever a particular given culture sees it to be, are to be completely accepted, prescriptivism can never be avoided.If there is no method of improving translation, time, money and breath will then be clearly wasted.The same point is also made by Scriven (1993), criticizing the principle of values-free, who argues that it can never be supposed that the evaluation process and results are all made at random, such as expressions of preference and taste.However, evaluative statements are principally derived from solid definitions and established facts (p.13).Translation and evaluation may also be viewed as a process of complex interplay.Based on her analysis of certain English articles, their translations into Korean and their final revised translations, Kang (2007) draws the conclusion that the translators' choices and the editors/evaluators' ones can be viewed as a result of a complicated interplay of translators' and editors/evaluators' subjectivities, which are founded on their interpretations of the original text, institutional objectives and well-defined norms (p.240).
Any evaluation models including translation quality assessment (TQA) models should pass the validity and reliability tests.Validity deals crucially with the elements measured by evaluation.For instance, the evaluation is designed to measure translation skills, translation performance, etc.On the other hand, reliability is mainly concerned with the evaluation being able to yield the same results when carried out again on the same type of people and under the same conditions (Williams, 2009, p. 5).Such tests are of utmost significance and should be passed by translation evaluation models in order to be approved.The validity test addresses the areas in translation that should efficiently be assessed by the translation evaluation models and whether or not such models have effectively managed to assess these areas which are indicated above.The reliability test, however, examines the consistency of the translation evaluation models in question.Are the results produced by such translation evaluation models the same if these models are carried out again and again on the same people and under the same circumstances?

Problems of Translation Evaluation
There are a number of problems that often hinder the translation evaluation process.Amongst these is the lack of the linguistic knowledge required and knowledge of the subject-matter by the evaluator.Another problem is the difference amongst evaluators with regard to elegant style; some evaluators consider it crucial for the achievement of an acceptable translation, while others do not accord it special importance.Another difference among evaluators resides in the treatment of punctuation errors, spelling mistakes and typos.While some evaluators ignore this type of errors in their assessment, other evaluators view them as major errors as they are detectable by the target reader.A clearly noticeable problem of translation evaluation also lies in the difference amongst evaluators concerning the level of accuracy and precision in the transfer of the ST content into the TL.Some evaluators often overlook slight shifts in meaning particularly if the message content is precisely transmitted to the target text, while other evaluators will adhere to the concept of fidelity to the source text, and they will never allow even a slight change in the place of a particular concept in the target text (Williams, 2009, pp. 5-6).
Another pivotal obstacle in the way of proper translation evaluation is that some translation evaluators resort to sampling analysis instead of full-text analysis.Indeed, sampling analysis may affect the validity of translation evaluation process as it results in certain negative consequences.Amongst these is that the evaluator may not be aware of any compensations made by the translator in the unsampled parts of the target text.Another defect of sampling analysis is that the evaluator may not take into consideration the whole text, which may lead to the lack of comprehensibility of the collective meaning of the target text as one unit (Williams, 2009, p. 6).Sampling analysis may also give rise to the unawareness of the evaluator to major errors committed by the translator, albeit in the unsampled parts of the target text.This may particularly be relevant in the case of newly established or novice translators, who can easily deviate from the intended meaning of the original text (Gouadec, 1989, p. 56). a number of translation services as well as translation teachers work with micro-textual analysis of samples as it is useful translation assessment technique for it does not only save time, but also enables evaluators to count errors and use them for translation assessment.However, a crucial problem of this technique lies in having a translation which includes more errors than those allowed by the grid developed by the evaluator to be a good translation.Although some evaluators classify errors into different categories of seriousness, such as major, minor, etc, the problem with this technique crucially resides in the lack of mutual agreement or consensus among evaluators on what constitutes a major error, as opposed to what represents a minor one (Williams, 2009, p. 6).
In spite of the fact that scholars, such as Nord (1991) and House (1997) suggest certain parameters that can be utilized for the purpose of translation assessment, such as quality of target language, accuracy, register, appearance of target text, situationality, and so on, there seems to be no parameter on which evaluators may rely to arrive at an overall quality assessment for the target text (Williams, 2009, p. 7).
The features that should characterize translation quality assessment as a tool used for translation assessment within academic context may considerably differ from those required for building translation quality assessment utilised for the purpose of translation quality control by a particular translation agency (Williams, 2009, p. 7).Along similar lines, Hatim and Mason (1997) assert that a clear distinction should be made between assessment of translation quality, translation quality control and translation criticism on one side and assessment of translation performance on the other (p.199).It goes without saying that it is not possible to propose a quality standard that can meet all requirements and can ipso facto be employed for the sake of specific translations evaluation (Williams, 2009, p. 7).

The Argument Macrostructure as a Parameter for Translation Assessment
According to Williams (2009), every text includes an argument, even if the text is informative, the writer is viewed to have an intension behind giving certain information in his/her text, which mainly resides in attempting at persuading the reader to accept this information and act accordingly.With this in mind, every source text comprises a particular argument macrostructure, and that the translator should keep this structure intact in the target text.Other ST functions and linguistic characteristics are also crucial, though the preservation of the ST argument macrostructure in the TT is deemed a cardinal element for the achievement of high quality translation (p.11).Toulmin (1964) examines the notion of argument in several fields, and reaches the conclusion that the elements that make up the argument are principally identical in different texts of various fields.Likewise, the assessment made in texts as well as the argument force are also alike across diverse fields of knowledge.Ideas discussed in a particular text can be categorized as 'good', 'appropriate', 'satisfactory' or 'unsatisfactory' regardless of the text type or the field of knowledge to which the text in question belongs.The main differences are represented by the criteria against which the judgement concerning 'goodness', 'appropriateness', etc of an idea may differ across areas of specializations.In Other words, all the rules that primarily govern the assessment of an argument within a particular text are contingent upon the field to which the text concerned is related, while the terms employed for the purpose of assessment are not affected by the difference in fields, and are ipso facto invariant (p.38).The general components of argument are unchanged, while the specific elements of these components hinge on the field to which the text concerned belongs (Williams, 2009, p. 12).Toulmin, et al. (1984) propose two crucially important components of an argument in any field of knowledge; these are claims/discoveries as well as grounds (p.25).These two elements are substantial to translation quality assessment.The claim or the discovery refers to the end result or the conclusion drawn from the argument.In other words, the claim/discovery of a particular argument is the core point shared by all elements of the argument.The claim/discovery needs to be supported by one or more pieces of information, such as statistics, facts, common knowledge, etc.Such information represents the grounds of the argument.It is worth pointing out that a particular claim may be founded on more than a single ground (Williams, 2009, pp. 12-13).For instance, the claim of an argument can be represented by the following sentence: the announcement of a new MA Programme in translation in the university.The grounds on which such claim is based may possibly be: the willingness of a number of students to pursue their MA degree in translation, the pressing need for translators and translation researchers in the market place, and so on.Based on the foregoing, it is clear that linguistic analysis plays a crucially important role in the process of translation evaluation (House, 2001, p. 254).
Having accepted Toulmin's macrostructure, it is possible to define the concept of error both on the basis of scientific and philosophical theory along with professional practice.With this in mind, it is evident that the implementation of the argumentative mode, i.e. the argument macrostructure would doubtlessly enable the translation evaluator to understand and be fully aware of a new concept, which lies in the translation defect.This concept of translation defect severely affects the validity of the translation as it hinders the flow of reasoning, particularly the claim/discovery and grounds in the text concerned.The implementation of such parameter shows that not all defects, which are regarded as major defects in view of this parameter, hinder the pivotal elements of the argument macrostructure (Williams, 2009, p. 13).
The central point resides in the concept that any translation violates or does not transfer the argument macrostructure of the original text to the target reader hinders the mission of the text in question, thus dispensing with the argumentative function of the text concerned.The defects that cause mistranslation of the argument macrostructure are described as critical defects.Transfer defects are deemed major defects, rendering the translation unusable, while other major defects, such as contresens, charabia, etc do not affect the usability of the translation.Other transfer defects are viewed as minor defects (Williams, 2009, p. 13).Critical defect is crucially concerned with the misinterpretation of the text, as pointed out by Bensoussan and Rosenhouse (1990), who regard the specification of critical defect as a valuable criterion in the process of students' translation evaluation through emphasising the extent to which the student's translation has conveyed the main message content of the source text, rather than being involved in microtextual error analysis.This also offers us a clear definition of the acceptable level of valid translation, which primarily lies in fully transferring the argument macrostructure of the original text in the target text, thus producing a translation deemed void of critical defects.Taken on board the nature of the argument macrostructure parameter, the translation evaluator is better of examining the whole translation, rather than carrying out a sample analysis, in an attempt to locate the acute translation areas and assess translation quality of all parts of text, which is deemed part of the translation and which comprises elements of argument macrostructure (Williams, 2009, pp. 13-14).
An evaluation grid can be proposed based on two sets of parameters.These are core, which is the argument macrostructure and field or use-specific, which resides in the area of specialization and the intended use of the translation.The core parameters can apply to all translations, whether achieved by professional translators or students, and regardless of its purpose and the field to which it belongs (Williams, 2009, p. 14).This makes the core parameters valid for the case of translation evaluation within the academic sphere as well as within professional settings.The field or use-specific parameters are implemented on the basis of the evaluator's view since their significance varies across areas of specializations.For instance, the evaluator often decides the application of these parameters in view of any work statement or contract concluded between the translator and the client.The field or use-specific parameters comprise figures of speech, terminology as well as target language quality and formatting.Target language quality can further be divided into grammar, style, typography and usage (Williams, 2009, p. 14).
In order for the evaluator to be conversant with the aforementioned sets of parameters and reflects the appropriate importance of each parameter in the overall evaluation, the multicriteria model propounded by Larose (1994) can be of use here.Such model demands that a specific weight is given to each parameter prior to evaluation.Likewise, the quality level for each parameter is also weighted on the basis of the evaluator's specification in the development of an overall rating (p.369).It is evident that Larose's multicriteria model is clearly influenced by Nida's criteria-based analysis where three different translations of the same text are evaluated on the basis of six main parameters known as isomorphs.These include elements, such as rime, referential meaning, concision of poetry lines, and so on.Nida views isomorphs as a particular method through which major issues and problems of equivalence may be identified.Also, isomorphs serve to oblige the analyst to determine the semantic aspects of the text such that they can be described and measured in light of conformity degrees.Owing to the fact that isomorphs consist chiefly of sets of features, they often oblige analysts and translators to consider such features patterns rather than disjoined similarities and differences (Larose, 1998, p. 179;Williams, 2009, p. 15).
As indicated earlier, according to the argument macrostructure parameters, a satisfactory translation is the one that involves full transfer of the argument macrostructure into the target language.If this requirement is met, translation for information purposes will be deemed satisfactory due to the lower weights given to receptor language parameters (Williams, 2009, p. 19).However, one question can be raised here.How would the model in question fit the assessment of trainees and/or students?It goes without saying that the target quality level is primarily contingent on learning objectives and outcomes to be completely accomplished, and the rendition of all aspects of the argument macrostructure by students and/or trainees may be too noble to arrive at.Nevertheless, the argument macrostructure parameters can be incorporated into translation rating grid specific to students and trainees provided that it is weighted in a different way from that of translation for information purposes.This means that in order for the argument macrostructure parameters to be more comprehensive and flexible, they should comprise several standard grades.Moreover, the narrative rating, such as 'acceptable-unacceptable' 'satisfactory-unsatisfactory' should be wholly grounded on the relevant standard (Williams, 2009, p. 19).
Based on the foregoing, Williams (2009) suggests different sets of standard specific to divergent translation contexts and situations.In maximum/publication standard, the target text should present all claims and grounds of the source text and satisfies all non-core parameters, with no critical or major defects.This standard should be applied to translations for publication.A relatively less accurate standard is called minimum professional standard according to which the target text should precisely manifest all claims and grounds of the source text, with no critical defects.This standard may be applied to translations for information purposes.A third relevant standard is known as student/trainee standard where the target text should include all claims and grounds of the source text and satisfies non-core parameters in light of the assigned learning objectives and outcomes.As the term suggests, this standard should be applied to academic settings.In substandard, the target text shows failure in transferring the claims and grounds of the source text into the target language and also fails in satisfying any of the non-core parameters appropriate to its particular function or the specified learning objectives and outcomes (p.19).

Concluding Remarks
Despite the clearly paramount importance of acceptable translation in this globalised world and the pressing need for specific criteria against which this acceptability can possibly be determined, the corpus of translation studies patently lacks generally agreed measures that can be used for the specification of translation acceptability and validity.This de facto dilemma has ipso facto urged certain translation scholars, researchers and translators to put some noticeable efforts and exercise their independent reasoning in an attempt to explore specific criteria against which translated texts may be evaluated.
It is evident that evaluation is crucial to translation as it uncovers the merits and demerits of the translation in question in addition to the fact that it ameliorates the translated work as every work is unquestionably susceptible to improvement and development.However, there are a number of obstacles that can predominantly impede the process of translation evaluation.Amongst these are the lack of linguistic knowledge and acquaintance with the subject-matter by some evaluators in addition to the difference among them with regard to the importance of elegant style and its impact on translation evaluation as well as the disagreement among evaluators concerning the treatment of issues, such as punctuation errors, spelling mistakes, typos, and so on.The difference in the level of accuracy and precision with regard to the transfer of the ST message content into the TL forms an important problem of translation evaluation along with sampling analysis with its negative consequences, as elaborated above.Micro-textual analysis of samples, albeit a useful evaluation technique, creates problems of unawareness of potential translation errors found in the unsampled parts of the TT.The lack of mutual agreement among evaluators in what constitutes a major error, a minor one, etc, is also another obstacle in the way of satisfactory translation evaluation.
The present paper argues that the argument macrostructure model may be adopted as a translation evaluation parameter to determine the translation quality.It is a collective yardstick that takes into consideration the argument of the ST as a whole and can equally be applied to texts of diverse areas of specialisation, taking on board that all texts contain arguments regardless of their types.The evaluation grid proposed for the argument macrostructure is composed of two sets of parameter, which are the core and field or use-specific.The former is the argument macrostructure, which can be applied to both academic and professional settings, whilst the latter represents the area of specialization and the intended use of translation, whose application is primarily contingent upon the evaluator's discretion since the importance thereof differs from a particular field of knowledge to another.Finally, the paper in hand claims that in order for the argument macrostructure model to be properly valid as an evaluation criterion for both the academic as well as professional spheres, it should encompass numerous standards specific to diverse translation contexts and purposes.