The TAUS DQF model

The Translation Automation User Society (TAUS) created its own translation quality assessment models, seven of them, in fact. To determine which of the seven should be used in which cases, the reviser must answer the following four questions:

– What category does the content fall into (user interface, online content, marketing materials, user manuals, instruction manuals, online help, audio/video content, social media content, etc.),

– Is the content regulated by the industry (yes or no), – Was it produced for in-house use (yes or no), – Is the communication channel B2B, B2C or C2C.

The application of the relevant model is determined based on how important one of the following three parameters is: time, utility and sentiment.

Out of the seven DQF models, we will briefly examine the following six:

1. Adequacy/fluency: TAUS determines the meaning of adequacy and fluency based on the definition given by the Linguistic Data Consortium.

They say adequacy means “how much of the meaning expressed in the gold standard translation or the source is also expressed in the target translation”, while according to them a fluent translation is “one that is well-formed grammatically, contains correct spellings, adheres to common use of terms, titles and names, is intuitively acceptable and can be sensibly interpreted by

a native speaker”. Revisers tend to go over the text sentence by sentence and evaluate adequacy and fluency on an even-numbered scale. After this, inter-rater reliability is used to eliminate randomness from the evaluations of the individual revisers.

2. Adherence to regulatory markets: when a  translation is intended for a regulated industry, the industry’s checklists must be used to see if the translation complies with sector regulations.

3. Community-based feedback: translations are evaluated by the end users based on any dimensions (for example, adequacy or fluency). Although it may seem like an inexpensive evaluation method, this is actually the most expensive, because the process needs to be inserted into the product’s documentation process and users need to be motivated to evaluate the translation.

4. Customer feedback: if a customer requires an unusual amount of support or has too many questions, it is possible that there is a translation error in the document. Customer feedback will usually make its way to the people responsible for the translation, who can then make the necessary corrections.

5. Readability: this model measures the readability of the translation with subjective feedback. Although there are automated quantitative methods to evaluate a translation in this way, feedback from native speakers is probably more valuable.

6. Usability: this model uses comprehension tests, which involve users having to answer questions after reading the product information. The evaluation can involve a test in which a user is asked to perform a certain task based on what the text says, while the evaluators check to see if the user succeeds.

Other methods can be the use of screen recordings, think aloud protocol or eye tracking.

5. Conclusion

My goal with this text was not to give a detailed presentation on error typologies, but rather to simply get the reader to understand why there is a need for objective and language-independent quality assessment, from the point of view of the client.

Something else I wanted to demonstrate is that there is no better solution than

the use of error typologies if there is only one reviser revising the translation.

I presented the four most well-known quality assessment models, since knowing about these models can help the reader in introducing error typologies if there is ever a need. It is important to know, however, that all error typologies are based on observation, and that all successful evaluation methods were developed under unique circumstances. Finally, it should be noted that not everybody favours quantifying translation assessment. We should never force these methods onto others without first listening to what they have to say about them.

References

Chesterman, A. 1993. From ‘is’ to ‘ought’: Laws, norms and strategies in translation studies. Target 5(1): 1–20.

Dróth J. 2001. Formatív értékelés a fordítás oktatásában [Formative assessment in translator education]. Unpublished PhD dissertation. Pécs: Pécsi Tudományegyetem.

Hansen, G. 2010. Translation ‘errors’. In: Gambier, Y. & van Doorslaer, L. (eds) Handbook of Translation Studies. Amsterdam: John Benjamins Publishing Company. 385–388.

Hegde, V. & Pielmeier, H. 2014. The Top Industries for LSPs. Lowell. MA: Common Sense Advisory.

House, J. 1997. Translation Quality Assessment: A Model Revisited. Tübingen: Narr Verlag.

Klaudy K. 2007. Fordítástudomány az ezredfordulón [Translation Studies at the Turn of the Millenium]. In: Klaudy, K. (eds) Nyelv és fordítás. Válogatott fordítástudományi tanulmányok [Language and Translation: Selected Papers in Translation Studies]. Budapest: Tinta Könyvkiadó. 33–43.

Nord, C. 1997. Translating as Purposeful Activity. Functionalist Theories Explained.

Manchester: St. Jerome Publishing.

Papp Gy. 1982. Miféle kritika a fordításkritika [What does Translation Criticism Criticise]? Híd 46(9): 1034–1043.

Reiss, K. 1971. Möglichkeiten und Grenzen der Übersetzungskritik. München: Max Heuber.

Sources

SAE J2450. Society of Automotive Engineers, Vehicle E E System Diagnostic Standards Committee. Warrendale, PA, 2011. http://standards.sae.org/wip/j2450/

LISA QA model 3.1. User’s Manual. Localization Industry Standards Association.

Romainmotier, 2007.

QT Launchpad MQM model. DFKI. Berlin, 2014. http://www.qt21.eu/launchpad/

TAUS DQF model. Translation Automation User Society. De Rijp, 2013–2014.

https://evaluation.taus.net/resources/error-typology-guideline, last accessed on 15 December 2015.

Ildikó Horváth

E-mail: horvath.ildiko@btk.elte.hu

1. Introduction

The language service provider profession has evolved and undergone considerable changes over the course of the past several decades. One of the most important changes in fact was that it started to evolve into a profession. Professionalisation involves several individual and collective factors, such as the attainment of a certain social status, suggestions to define standards of best practice, to control access to professional knowledge, and to control education and work opportunities (Wadensjö et al. 2007: 2). The language service provider profession today has its own professional organisations, codes of ethics, different training courses, post-graduate training courses, advanced training courses for teachers, online message boards, blogs, etc.

Language service providers can be divided into two broader categories:

professionals who mediate between languages in writing, or translators, and those who mediate between languages orally, or interpreters. The profession, however, is nowhere near homogeneous, since both translators and interpreters can work in a variety of contexts: they can be freelancers or employed at companies. They can also specialise in a variety of fields or work for international institutions. They may have formal qualifications or they may have taken up the profession without having received any formal training.

The translation profession has more and more variety. Instead of ‘traditional translators’ the market now looks for versatile language service providers who have a good grasp of the different phases of the translation process, meaning that besides being skilled in translation in the traditional sense, they also know how to prepare, revise, post-edit translations or create and manage termbases. Interpreting is an equally diverse profession: depending on the mode of interpreting, we can distinguish consecutive interpreting, simultaneous interpreting and chuchotage.

We can also distinguish a variety of interpretation types based on the subject and/

or venue of the interpreting event: these types can be judicial, medical, church, community or conference interpreting. Furthermore, many freelance interpreters also regularly take up translation jobs.

Alongside the diversity, it is clear that there is a kind of uniformity to the everyday jobs of translators and interpreters. What Gouadec (2007) says about translation applies to interpreting as well. He says that all language service providers face the exact same challenges: a lack of appreciation for their work, complexity of their tasks, information and communication technologies impacting their work methods, the pressure for cost efficiency and the battle for professional status (Gouadec 2007: XVIII).

The process of professionalisation seems to be contradicted by the spread of non-professional language services, in other words translation and interpreting services that are not paid for. Pérez-González & Susam-Saraeva (2012) say there is such an abundance of these services that professional translation and interpreting is now considered a  subcategory of translation and interpreting and not its norms-defining prototype. These services are spreading so fast that there are now online data management systems to support the work of volunteer translators and interpreters (Bey et al. 2006; Boitet et al. 2005; Utiyama et al. 2009).

The reasons for the proliferation of volunteer translation and interpreting are manifold: the perennial priority of cost saving; the ever accelerating development of the internet and ICTs; the rise of online communities. In what follows, I will define the concepts behind volunteer translation and interpreting and illustrate several fascinating examples. Then I will discuss who undertakes translation and interpreting on a voluntary basis and the motivation factors for them doing so.

Finally, I will turn to the questions this issue raises for the profession itself.