 |
 |
 |
 |
 |
 |
 |
CAT Qualitat EUS Kalitate GLG Calidade POR Qualidade SPA Calidad |
contents
Introduction | The origins of the discussion | Quality criteria | The quality of the product | The quality of the process | Research potential
The study of quality is hotly debated, perhaps because of its subjective, relative, vague, and fickle nature.
The practice of translation was always accompanied by the reflection on what it means to translate well. Over time, there has been no loss of momentum: on the contrary, since the second half of the twentieth century, there has been a significant development in both theoretical reflection and the publication of professional standards, partly explained by an increase in the volume of translations (Xueting, Espín-García and Presas 2018: 236) and, above all, by the definitive start of Translation and Interpreting Studies (TIS).
In TIS, the relevance of quality can be noticed in the space it occupies in the realm of specialised literature: between 2000 and 2017, around 6% of the articles published in some of the leading journals in the field researched quality (Akbari 2018: 553). Similar figures are provided by the Bibliography of Interpreting and Translation (BITRA, Franco Aixelá 2001-2020), if a search for the term quality is conducted for the years 2018 and 2019.
While interest in the concept continues to expand, there is also the feeling that quality is a thorny problem (Jiménez-Crespo 2017: 478), for many reasons: from the theoretical and philosophical complexity of the concept itself to the enormous variety of environments in which the activity is present, as well as the fact that quality in translation plays a role in different tasks (Koby and Lacruz 2017: 11). Moreover, there is no such thing as a prefixed perfect translation; on the contrary, several translations of the same original text may be considered equally acceptable.
It is also difficult to define quality because of its mutability: translations that were taken as good at one time are later disdained; or else, translations that were criticized become applauded. An example of the former are Milan Kundera’s French translations, corrected by the author himself; an example of the latter is Omar Khayyam’s now classic poetry collection Rubaiyat, the English translation of which (by Edward FitzGerald) was initially misunderstood.
Finally, there are concepts that sometimes overlap with that of quality. This is the case with ethics, which deals rather with the translators’ loyalty (Nord 2016: 571) to the original author, to the client, to the reader, and to themselves. The path of ethics intersects with that of quality as, on the one hand, quality is perceived as an ethical obligation and, on the other, ethical behaviour constitutes a hallmark of quality. Professional standards are actually based on both concepts.
However, the most related issue is that of academic “evaluation”, for that which is usually evaluated is (the) exact translation quality. In order to approach the study of these concepts, it is useful to delimit the aspects that directly involve each of them. Thus, the tools for assessing translations and translators or the particularities of this activity in different environments belong to the field of evaluation. Meanwhile, the study of quality in translation is more linked to the criteria that define it.
![[título del epígrafe]](2.2-X.png "[título del epígrafe]"){kind=small}
Historically, the debate on how to translate well was mainly linked to the notion of equivalence. This concept − analysed in dual terms of free versus literal − was used by different authors to defend one of such two positions as the only valid one.
By the 1st century BC, Cicero (Conde 2009: 63) and Horace (Melby, Manning and Klemetz 2005: 407-408) criticized overly literal translations; their position marked the beginning of a pendulum swing concerning the relationship between quality and literalism.
 |
| Saint Jerome as Scholar, by El Greco [Source] |
Saint Jerome also assumed that absolute literalism did not work in some cases; in his Letter to Pammachius (395), he mentioned for the first time the translation of the sense. In the Middle Ages, free translation continued to be the orthodoxy, but the contrary tendency did not disappear completely.
This coexistence of positions continued during the Early Modern Period; in fact, the pendulum swung faster than ever: while outstanding figures such as Étienne Dolet stressed the distancing from the original language as characteristic of good translations (Hurtado 2001: 108-109), the tendency named by Gilles Ménage as les belles infidèles was close to adaptation − in the sense of a strategy to freely reconstruct the impact or the purpose of the original text (see Bastin 2009: 5). In addition, John Dryden and his commitment to paraphrase or sense-for-sense translation (Melby, Manning and Klemetz 2005: 407-408) advocated a middle ground between “word-by-word” translation and free adaptation. The same was found in critical currents that, in the heat of the Counter-Reformation, demanded both greater accuracy and fidelity (Hermans 1997: n.p.).
At the beginning of the Late Modern Period, the three laws of Alexander Fraser Tytler seemed to respond to the romantic current that prevailed at that time (Pegenaute 1996: 23), according to which − for political and identity reasons − translations maintaining the exoticism of the originals were preferred: examples of this thread are Friedrich Daniel Ernst Schleiermacher or Wilhelm von Humboldt. Nevertheless, the 18th and 19th centuries were also varied in theoretical reflection; in fact, Tytler’s own position can be understood as one of the meeting points between the two extremes dominating until then (Pegenaute 1996: 26).
The discussion between literalism and naturalness persisted and evolved in the 20th century: until the middle of the century, most authors agreed that there is only one way to make a good translation, that is to say, that quality is an absolute notion (Melby, Manning and Klemetz 2005: 408). This view, influenced by such currents as the Vienna Circle, changed when the theory began to focus also on other translation types and not only on literary translation, which had been the most recent focal point up until that point in time.
In a context marked by the rise of international trade, the technological revolution and a notable increase in literacy among the population (and coinciding above all with the failure of linguistics as a framework for the nascent TIS), German translatological functionalism, followed by its skopos theory, emerged. As is widely known, with Vermeer, Reiß and Holz-Mänttäri, this theory places the addressee and the purpose of translation at the centre of the debate.
The main consequence of embracing functionalist postulates is the replacement of the static notion of quality with a dynamic approach, sensitive to initiators and purposes, to addressees and specifications (Melby, Manning and Klemetz 2005: 416). In short, from this point on, it is accepted that there is no ideal translation (either literal or free) for all contexts and situations; on the contrary, a translation is of sufficient value as long as it respects the purpose − the function − given to the target text.
Contemporary to the German school emerged descriptivism (with such authors as Toury, Hermans or Lambert), a theory rejecting the prescriptive legacy. According to this viewpoint, a translation does not have to fulfil any prior characteristic − such as the presence or absence of literalism − to be seen a (good) translation; instead, the historical group that receives it must simply consider it as such (Pym 1995: 170). Consequently, the most that the researcher can aspire to is to discover what each society understands by a quality translation, in other words, to approach its study from a socio-historical point of view. Not only does this school imply relativism, but it also stresses the temporal variability of the concept of quality.
In short, the historical vascilations show the concept of quality to be relative, liable to changes over time and dependent on contexts, restrictions and situations.
Criteria are standards that allow us to make value judgments; in translation, as in other disciplines, these factors must be clear, explicit and well-founded (House 2001: 156). In studies on translation quality, the dilemma of literalism is not the only criterion under consideration.
In order to define the criteria for translation quality, at least three factors should be taken into account: the object on which quality is measured; the very conception of translation and, finally, three fundamental aspects or axes of translating.
To begin with, it is necessary to specify the object, the quality of which is being assessed. Translators themselves are sometimes singled out as the very object of analysis. However, it is not possible to estimate the quality of translators directly, that is to say, without taking full account of what they do (the product) or how they do it (the process), which is why quality criteria are often proposed for the products or processes involved in translating.
Process and product are interconnected: the product is the consequence of the process and, in turn, the process is the way in which the will to create a product is resolved. The borderline separating one element from the other (Vandepitte 2017: 17) leads to some confusion in the proposal of quality criteria, as sometimes it is not known exactly to which of the two objects they refer.
To continue, there are at least two relevant conceptions of translation here, explaining why misunderstandings and new criteria constantly arise in the study of quality thereof.
The classical conception concerns the prototypical task, according to which, a translator works on a text written in one language to recreate the same text in another language, with equivalent meaning and length (Melby, Fields, Hague et al. 2014: 397). This view is rather academic, static, and strict. Moreover, it is limited in scope, as it excludes many tasks associated with or integrated into the activity itself, as well as less prototypical phenomena, such as group and collaborative projects, localization, transcreation, synthetic translation, audiovisual translation and even interpreting in all its forms.
In contrast to the classical conception, Melby, Fields, Hague et al. (2014: 395) propose a broader and more dynamic notion in which translating means creating content according to agreed-upon specifications and which corresponds to other previous content. Such a definition − perhaps also more closely linked to professional practice − includes all the aforementioned phenomena and entails an alternative view of quality, in which there is room for multiple criteria adapted to projects very different to the classical model.
Finally, translation quality can be approached by taking into account the three fundamental axes from which specific criteria emanate.
First, the focus can be placed on the conventions relating to the environment, system or context in which the product is integrated or in which the process is carried out. According to this normative axis, for a product or process to be good, it has to fit:
Translating is an interpersonal—as opposed to interlinguistic or intercultural—cooperative text-production activity where production is constrained by social and cultural factors and guided by a principle of creative imitation.
(Muñoz 2017: 564)
Second, the emphasis may be on the use of the product or the process, motivated by an intention − usually the creative imitation of a previous communicative product, as well as by the application of the text to a given situation (Koby, Fields, Hague et al. 2014: 414, on implicit and explicit specifications or requirements). According to this functional axis, for a translation to be good it has to work.
Third, the return that the participants in the communicative act receive from the product and the process should be considered; this value is a perception connected to the relationship between cost and benefit. According to this economic axis, for a translation product or process to be good, it must be worth the effort.
The following sections explain and illustrate how the quality criteria depend on the three factors mentioned above, that is: the object, the two conceptions of translation, and its three main axes.
Historically, the debate on quality in translation has focused on the result of the activity, i.e. translations. However, TIS have evolved similarly to the industrial sector: first, it was solely the estimation of the product, and only over time was the evaluation of processes included − as for example in the “total quality” (management) paradigm.
As regards the product, and considering the above mentioned axes, three main quality criteria may be taken into account: appropriateness, accuracy, and return.
Appropriateness
The first criterion − relevant also to the assessment of monolingual products − is appropriateness. This term is sometimes related to fluency, although they are not exactly synonymous. When a translation is appropriate, it fits in with the conventions of the language in which it is created; thus, this criterion includes readability, style, naturalness, and grammatical correctness.
| Quality metrics for machine translation (SDL Community) |
The problem, however, is that there is no single way of expressing oneself appropriately. In a broader, more dynamic conception of translating, there are multiple circumstances affecting the criterion of appropriateness: certain types of literature, thematic areas, more or less hybrid modalities or the most specialized professional communities entail particular views of what an appropriate communicative product in a given context means.
Therefore, a translatological decision considered suitable for a simple text may be wrong for a dubbing, where specific restrictions and conventions may be employed. In the same way, a poor or overly stilted expression for a standard literary text may be suitable for experimental works. In short, the criterion of appropriateness implies different interpretations and specific quality indications.
Some examples of this specificity were documented in simultaneous interpreting (see, e.g., voice pleasantness, in Collados 1998: 155), telephone interpreting (sound quality, in Wang 2018: 101), dubbing (isochrony, in Chaume 2016: 2), live subtitling (character identification or speed, in Romero and Pöchhacker 2017: 150), and so on.
Also noteworthy in itself is quality in machine translation, where products are not analysed in order to evaluate the subject (translator), but the machine. In this − more pragmatic than theoretical − field, quality can be determined by a person or by an automatic system, among which GTM, TER and above all BLEU stand out (Doherty 2017: 133-135). Automatic systems, which are increasingly common, use a translation as a model and yield metrics that show the degree of similarity of the text translated by the tool with respect to the chosen models, or else the number of steps that the translation produced by the machine needs to become the model translated by the person. In other words, the translation produced is more appropriate the closer it is to the reference model.
Accuracy
The second criterion is accuracy. It refers to the functional axis and is connected with the concept of informativity proposed by Muñoz (2007: 259) and with that of fidelity. A translation is accurate when it fulfills the intention for which it was created; the question is, then, for what is the activity of translation.
According to the strict conception, a text is translated in order to create another text that represents the former in another language, in a relationship of variable similarity. This objective − that of creative imitation − is the one that is taken for granted when there are no other specifications, ergo a translation is all the more accurate the more it resembles the original text, generally speaking.
However, a translated text can resemble an original one in a variety of ways, all a priori defensible if they are coherent. One example is poetic translation (see Holmes [1970: 94-99] and his “forms” of translating poetry), where priority may be given to content or to formal aspects; the latter, in turn, may aim at a philological resemblance (such as rhymed translation) or at a functional resemblance (a rhymed classic becomes free verse, in search of a new modernity of the original in the target context). There are other valid examples, such as slang, sentence periods, lexical repetition, and so on.
Moreover, absolute accuracy is not always desirable, as is the case when the translation corrects errors that were present in the original text; when the source information is reduced − such as in synthetic translation − or, simply, when theory comes across professional practice (see, for example, Lambert [2018: 284] and his critique of codes of good practice in interpreting), often because in the real world intentions are always multiple and never unambiguous.
These situations make it advisable to view the criterion of accuracy in a more relative or flexible way, i.e. the criterion of accuracy has less to do with fidelity than with effectiveness (Muñoz 2007: 261), usability or even the expectations of the final addressee who, in the end, can check whether the translation works or not
Return
The third criterion is return, which refers to the relationship between cost and benefit when creating a product. A translation returns when the result is perceived to be worth the price.
The most relevant aspect of this quality criterion − which depends on the economic axis − is that the same translation can be considered acceptable or not depending on its price. A paradigmatic case is that of illegal translations of video games and other audiovisual products carried out by fans of the originals: the products are consumed for free in this context but would not be evaluated positively by the same addressees if they accessed them through the commercial circuit
In the business world, return is not only related to the price of products, but also to the economic impact. Thus, the extent to which the translation is used publicly or privately can be decisive, so that mediocre texts may be taken as valid if they do not directly damage the company’s reputation, for example, in the case of draft reports or internal communications.
Finally, business models in which clients themselves decide how much they want to pay on the basis of the quality they demand − interactive practices that are increasingly common in the industry − pose a challenge to those who conceive of translation only according to its strictest conception:
The initial globalization and outsourcing of translation services has been expanded exponentially through new crowdsourcing workflows with different specifications, crowds, technology support etc. This has allowed opening up the provision of quality with different quality tiers related to different production mechanisms rather than internal features of the final translated product, shifting the ultimate responsibility of the quality of the translation to customers who select the level of quality through a wide range of considerations, such as the available budget, permanence of the translation, potential risks involved, receiving audience, etc.
(Jiménez‑Crespo 2017: 489)
People became interested in describing the translation process shortly after business environments accepted the idea that in order to control the quality of the product the entire production cycle had to be taken into account.
Within TIS, the novelty of this approach means that the terminology is not yet settled, with a distinction sometimes being made between the process and the service (e.g., in Kockaert and Segers 2017: 148), although finally both terms may refer to the manner in which translation is actually carried out. While the former is more linked to the classical conception (the process performed by a single person, with a standard text), the latter involves more stages and stakeholders.
In addition, the study of quality in the translation process may include two scopes: the individual and the social. For either of them, three main quality criteria can be proposed, depending on the three respective axes: compliance, effectiveness, and profitability
Compliance
The normative axis in the translation process takes the form of the criterion of compliance with professional standards in a context with specific conventions. Accordingly, a translation process is good to the extent that it complies with the most widespread standards and conventions in force.
As in many other professional services, the growing importance of the process in the study of translation has encouraged the emergence of quality standards. Norms or standards are guidelines designed by committees of experts and aimed at harmonization (Heard 2017: 747), i.e., at the conformity of products and processes. In practice, they are also advertising claims used to dignify the profession; likewise, they bear some similarity to codes of ethics and best practice, another type of standards related to professional ethics rather than to quality.
 |
| Cover page of the UNE-EN ISO 17100:2015/A1:2017 standard, provided by the UNE |
Several quality standards have been proposed for translating, such as UNE‑EN‑15038:2006 (CEN 2006) or the most recent − SAE J2450_201608 and ISO 17100:2015 (ISO 2015). Most of these norms comprise different aspects, for example: originals and translations; cooperation between clients, workers and suppliers; their rights and duties; the selection of translators; the organisation of the process (preparation, project management, revision, proofreading, etc.); as well as a section where the most important terms are defined.
Like many other industries, translation is increasingly bound by internationally agreed standards for service provision, through bodies like the CEN (European Committee for Standardization) and ISO (International Organization for Standardization). Establishing objective quality criteria has traditionally been seen as contentious, if not impossible, in translation studies; but in the real world, such criteria have indeed been defined and are increasingly applied to LSPs’ work.
(Drugan 2013: 1)
The individual scope includes the guidelines that affect reading, writing, revising, and the mental operations developed for each task. Conversely, the social scope deals with issues related to service provision (from hiring translators to implementing internal quality controls), along with rules and tools for specific stages, e.g. translation memories and machine translation systems, not to mention documentation or term extraction and selection (see, for example, Chiocchetti and Lušicky 2017: 171).
Effectiveness
The functional axis applied to the translation process corresponds to the criterion of effectiveness. Irrespective of the conventions and standards in force, the translators and the translation companies may act according to their own convictions. Compliance with quality standards does not necessarily lead to acceptable results in all possible cases: this lack of effectiveness is, in fact, one of the most common criticisms about quality standards, also in translation (Heard 2017: 747-748).
There is no single correct way of doing things: whilst the criterion of compliance is imposed from the top down, effectiveness emerges from the bottom up and is connected with the character of the individual participants, with their preferences and vices, which can lead to success even if they do not conform to standards.
A translator may perform better in inadvisable conditions, such as in stressful situations (De Rooze 2003: 65) or working in the early hours of the morning. Similarly, a company may adapt a standard to its own idiosyncrasies and, based on these, design a system that, thanks to the involvement of its employees, works at every link in the chain. Furthermore, the practice is adapted to the client, who may make demands, sometimes − but not only − to reduce costs, such as: assigning a task to apprentices; using defective glossaries or avoidable anglicisms; or saving stages of the process, e.g. the postediting of the machine translation product.
Another difference between compliance and effectiveness is that the latter is always specific, dependent on the goals of each project, the procedure of which is considered acceptable as long as it functions. To measure effectiveness, checks must be conducted at every stage, often by establishing in advance controls between the client and the company, and between the company and its employees.
Dialogue is important when implementing flexible, customisable and modifiable quality systems over time, taking advantage of the best of the norm and the best of innovation, such as the hybrid experiences proposed by Drugan (2013: 180-181): initiatives mixing (top-down) hierarchical and traditional approaches with (bottom-up) open and collaborative approaches. The former bring a greater capacity for investment or implementation of good practices; the latter, in contrast, tend to be more dynamic and rely more on end-user feedback (Drugan 2014: 115-116).
Profitability
The criterion of profitability is circumscribed to the economic axis in the translation process. A process is good if the translator or the company perceives that its implementation pays off. Every organization has costs, not only in terms of money, but also in terms of effort and time invested (“time is money”).
A working scheme may incorporate successive revisions of a translation by different subjects, which can lead to a high quality product. Nevertheless, if the process is too costly, either regarding money or resources, it may be discarded by the translator himself or the company. Conversely, a plan can be complex, but at the same time profitable for the translators who, despite the effort, can learn lessons that help them to improve, automate processes, etc. According to the criterion of profitability, experience is relevant: the process is good if it is an improvement with respect to the previous step, and bad if it is a setback.
This criterion is somewhat related to those of productivity (Sánchez‑Gijón 2014: 438) and efficiency (Muñoz 2007: 261). It implies attending to any action that seeks to improve internal mechanisms or business success as a whole, such as, marketing campaigns, impact and branding, promoted values, the hiring of more effective workers, or the commitment to continuous training.
Finally, as regards the social or collective scope, profitability also includes issues related to corporate social responsibility and employee potential and health, since a profitable process compensates the entire network involved in the creation of the products, especially the translators themselves.
With respect to the product, the gradual transformation of the activity, incorporating new modalities and tasks into the practice of translation renders the broadening of the criterion of appropriateness unavoidable in order to better define the acceptability of products within each field and at each moment in time.
Regarding the criterion of accuracy, empirical and experimental studies (based on increasingly common technologies, such as eye-tracking, keystroke logging, mental chronometry, and heart rate) should still be applied so as to verify, for instance, how the replication of previous information becomes more flexible in situations in which a reduction or amplification of what the original suggests is precisely what is to be expected. The conscious or unconscious discrimination of certain elements could point to a hierarchy in the importance that translators give to the information extracted from the texts from which they translate
As for the criterion of return, studies may be based on the reception of the target addressees − taking into account different types of quality, such as those of the Kano model − or the acceptability levels arising in the various environments in which translators participate (Muñoz 2010: 151-152). An example of the latter is how innovative tools and methodologies, such as machine translation or non-professional collaborative translations, have become more and more accepted.
There are even more possibilities connected with the process, as there are still many untouched areas in this relatively unexplored concept. The compliance criterion will be fruitful as long as the industry continues to evolve, thus emerging new, increasingly specific standards while refining and replacing the existing ones.
Nonetheless, in order to research the criterion of effectiveness further, it is necessary to learn about the processes that − being different − work equally well; for this purpose, it could be useful to refer to studies related to translator profiles or translation styles (such as Dragsted and Carl 2013), and to the processes performed in the workplace (see Risku, Rogl and Milošević 2017: 3).
Lastly, as regards the profitability criterion, it would be useful to discover (from a cognitive approach) how aspects that may represent an added or tacit value for the translator or for the company itself affect the translation process: for instance, the role played by the remuneration or the perceived impact with respect to reputation and business opportunities.