LINGUIST List 23.2559

Thu May 31 2012

Review: Computational Linguistics; Translation: Depraetere (2011)

Editor for this issue: Rajiv Rao <rajivlinguistlist.org>



Date: 31-May-2012
From: Daniele Russo <danijel.russogmail.com>
Subject: Perspectives on Translation Quality
E-mail this message to a friend

Discuss this message

Announced at http://linguistlist.org/issues/22/22-4952.html
EDITOR: Ilse DepraetereTITLE: Perspectives on Translation QualitySERIES TITLE: Text, Translation, Computational ProcessingPUBLISHER: De Gruyter MoutonYEAR: 2011

Daniele Russo, Department of English Studies, University of Milan

SUMMARY

This book is a collection of articles on translation quality edited by IlseDepraetere. The authors of the contributions are both academics and industryprofessionals who share their experience and expertise in their fields ofspecialization. The articles are all in English, but in the Acknowledgements theeditor states that several articles were translated from French.

The edited volume is divided into four sections, and a total of eleven articlesdiscuss translation quality from different perspectives. More specifically, theaim of this volume is to share insights which are training-oriented orindustry-based. This broad approach includes the quality of human translationand machine translation, specialized translation (also called ‘pragmatictranslation’ in the volume) and literary translation, as well as translationsdone by both students and professionals. The methodological approach in thisvolume is based upon concrete translation evaluation contexts supported by dataanalysis and hands-on experience, and the contributions can be read through apragmatic notion of translation in which a translated text (and hence theevaluation thereof) must be adapted to the communicative context of use.

In the Introduction, the editor states that “there are three issues that areimportant when it comes to translation: quality, quality, and quality” (p. 1),which means that regardless of the translation theory we are pursuing, the mainobjective of our research effort should be the improvement of the quality of thetranslation output. Depraetere emphasizes the importance of the variety ofcontributions included in the volume and its connections with current studies;also, the editor stresses the need for building a bridge between theory andpractice and provides a gist of each paper in the volume.

Part I: “Translation quality in the translation training context”

The first article, “A global rating scale for the summative assessment ofpragmatic translation at Master’s level: an attempt to combine academic andprofessional criteria”, by Anne Delizée, discusses the challenges faced byuniversity teachers when evaluating student translations. She presents theevaluation model used in the Master’s degree program in Specialized MultilingualTranslation at the University of Lille 3, which is aimed at making studentsaware of industry standards while giving necessary attention to the pedagogicalrequirements of the training context. The goal of the model is to evaluatepragmatic texts, i.e. technical, scientific, legal, economic, financial orcommercial texts, with the primary goal being communication. The evaluationmethod combines the assessment of the translation output and the assessment ofprofessional performance. The rating scale is thus based on the followingcriteria: linguistic skills, translation skills, discipline, linguistic skillsin the target language, professional skills, methodological skills, andtechnical skills.

The second article, “Comparing formal translation evaluation andmeaning-oriented translation evaluation: or how QA tools can(not) help”, by IlseDepraetere and Thomas Vackier, is based on a collaborative project between theApplied Languages Department of the University of Lille 3 and Yamagata Europe, aservice provider for translation and localization. The project employs a corpusof Spanish into French student translations and aims to compare the scoresattributed to the students’ translations by a human evaluator (using theassessment model presented in the previous article) with the results of anautomated formal quality check done via a computer-aided tool (QA Distiller).The purpose of the study is to check to what extent formal translation qualityis indicative of overall quality.

The third article, “Number and gender agreement errors in student translationsfrom Spanish into French”, by Carmen Núñez-Lagos and Nathalie Moulard, is basedon the same corpus as the previous article and focuses on a frequent error type-- number and gender agreement -- in Spanish to French student translations andexplains the reasons underlying the different kinds of mistakes that the authorshave identified in their corpus. In fact, number and gender errors can be verycommon between “gendered” languages such as Spanish and French. This study showsa relative unawareness of the syntactic hierarchy in noun phrases in thetrainees’ mother tongue. In order to avoid these errors, the two authorsmaintain that it is necessary to raise the trainees’ awareness of the impact oflinguistics courses in a translation curriculum.

The fourth article “A Lexicogrammar approach to checking quality: looking at oneor two cases of comparative translation”, by Christopher Gledhill, shows how alexicogrammar approach can be used in order to check the quality of atranslation. The core of this approach is built on the basic assumptions thateach sign in the language has its own particular lexical and grammatical nichein the language system, and empirical data, such as corpora, as well asconcordances and contextualized examples, are to be used in order to analyze thehabitual use of signs in discourse. In this sense, the lexicogrammar approach isless demanding than other statistical approaches, and more rewarding than amanual and unsystematic approach, as it requires the use of an on-line orcorpus-based concordancer and the ability to observe and compare typicalpatterns of signs in different textual contexts.

Part II: “The evaluation of machine translation”

The fifth article, “A contrastive analysis of MT evaluation techniques”, by IlseDepraetere, gives a survey of methods to evaluate machine translation. Machinetranslation is a technology that is more and more employed in the translationindustry, which is why there are continual attempts to search for more adequateevaluation techniques. In this study, a corpus of 2,250 source words wasmachine-translated with a rule-based system and a statistical system and thecustomized output was evaluated through four techniques: human evaluation ofadequacy and fluency, automated evaluation, evaluation based on error analysis,and evaluation based on post-editing time. The correlation and non-correlationof the results offer new insights into these cutting-edge applications.

The sixth article, “MT evaluation based on post-editing: a proposal”, byNathalie De Sutter, presents a new evaluation technique of machine translationbased on post-editing time and post-editing distance. The article shows that thetime spent on correcting the mistakes of machine translation and the textualsimilarity between the machine translation output and the post-edited version ofthe machine translation output correlate with the general quality of the outputas assessed by human evaluators.

Part III: “Quality Assurance in the translation workflow”

The seventh article, “Quality Assurance in the translation workflow -- Aprofessional’s testimony”, by Nancy Matis, gives a survey of various qualityassurance processes that can apply to standard translation projects or tosoftware, websites, and other e-content localization projects. The authordistinguishes two main levels of quality check: linguistic quality assurance andtechnical quality assurance. The choices regarding revision processes depend onseveral factors, such as type of project, subject of the project, customer,volume of the project, resources, schedule, and budget. The quality assuranceprocess may be very quick and general or, on the contrary, extensive and veryprecise.

The eighth article, “A contrastive analysis of the automated QA tools (QADistiller 6.5.8, Xbench 2.8, ErrorSpy 5.0, SDL Trados 2007 QA Checker 2.0 andSDLX 2007 SP2 QA Check)”, by Antonia Debove, Sabrina Furlan and Ilse Depraetere,shows that quality assurance tools constitute a welcomed technique in thetranslation industry, since they offer a means of performing automated checks ofthe consistency, terminology, number values, tags and punctuation, and in thisway, make the reviewers’ task easier, leaving them more time to focus onsemantic errors. This contribution offers a comparison of five formal qualityevaluation tools on the basis of the most relevant features: interface anduser-friendliness, performance, and cost. Finally, the article emphasizes theimportance of integrating such tools into the translation workflow.

The ninth article, “Management of translation memory quality in the SpanishDepartment of the Directorate-General for Translation of the EuropeanCommission”, by José Luis Vega Expósito, is aimed to show, using examples fromexperiences in the Department of the Directorate- General, how translationmemory management applications can contribute to terminological andphraseological consolidation, and consequently, to the stability, clarity andconsistency of the administrative language.

Part IV: “Domain-specific quality: legal translation and literary translation”

The aim of the tenth article, “Quality issues in the field of legaltranslation”, by Patricia Vanden Bulcke and Armand Héroguel, is to tackle theissue of quality assurance in the field of legal translation. Firstly, thiscontribution discusses the nature of the legal text and the role of its specificfeatures in translation by proposing a taxonomy of sub-genres according tofunction, translation strategies and genre characteristics. After showing anumber of examples from English, Dutch, French and Spanish, the author concludesthat the translation of a legal text is bound more to comparative law analysisthan language. As soon as the legal text is defined with regard to category,genre, and system, the translation is to align with the function that thetranslated text must fulfill within the new legal context (according to Nord’sSkopos Theory, 1997. Finally, the authors state that concrete quality criteriafor legal translation should go beyond the realm of terminology and includeother considerations such as phraseology and structure, style, and syntax whichare peculiar in legal texts.

In the last article, “The problem of self-assessment in literary translation”,by Olivier Vanwersch-Cot, the author proposes to analyze the process ofself-assessment on the basis of concrete examples drawn from his experience as aliterary translator. He specifically focuses on his translation of the novel ‘Dejoodse messias [The Jewish Messiah]’, by the writer Arnon Grunberg. The authorconcludes that self-assessment is not an accessory but rather an essential partof the translation process of literary works because it helps recreate meaning,rhythm, images, and connotations.

EVALUATION

This volume presents a collection of eleven articles exploring various topics ontranslation quality from both theoretical and empirical points of view, whilealso introducing the state-of-the-art technologies used in this field. It isvery informative and, on the methodological level, puts forward new hypothesesthat can prove useful both to trainers and language professionals. In fact, thisvolume can be of interest both to the layman interested in translation qualityand to the language specialist interested in evaluation techniques; it isactually free of the excessive use of technical jargon typical of other similarpublications. The language use is thus very precise, scientific, and clear butnot parochial.

The articles are all in English, but the copious examples of translation are inEnglish, French, Spanish, and Dutch (in some cases Spanish into French and Dutchinto French translations include English translations for those who do notunderstand those languages). Many of the articles focus on the role oftechnologies in the translation process and in the translation evaluation ofspecialized texts, while other contributions explore diverse approachesinvolving literary and legal texts. Compared with previous work in the field,this collection of articles shows a very pragmatic approach and providessensible solutions to the task of assessing a translated text, especially foreducational purposes.

Those interested in translation quality models will find valuable insights inthe first section of this collection, in which Delizée presents the evaluationmodel used in the Master’s program in Specialized Multilingual Translation atthe University of Lille 3, and Depraetere discusses the use of a cutting-edgecomputerized translation quality checker to implement formal quality intranslated texts. Both contributions illustrate the necessity of interactionbetween human quality assessment and computer-assisted technologies. Gledhillintroduces another empirical hands-on approach that is particularly feasible andsystematic, as it is based on deep-rooted lexicogrammar notions.

The editor, who also authored and co-authored three articles of the volume, hasdone an excellent job of selecting and compiling the contributions, which show awide range of themes and approaches. Even with such variety, the cohesion of thecollection is maintained through frequent links between the articles of thebook, thus highlighting interdependencies among apparently distant fields, suchas literary translation and technical translation.

The best aspect of this book is that it is the result of the collaborationbetween academics and industry professionals. Therefore, it synthesizes varyingviewpoints based on real professional experiences. The volume stands out fromother books on translation quality because it does not oppose machinetranslation on ideological grounds. Conversely, it shows that it is bound tobecome a common support tool in translators’ daily practice, and thus, fullyintegrated within the human translation (''transediting'' in Ulrych’s words, 2011)process.

REFERENCES

Nord, Christiane. 1997. Translating as a Purposeful Activity. FunctionalistApproaches Explained. Manchester UK, St. Jerome Publishing.

Ulrych, Margherita. 2011. Transediting and its relevance to medical discourse.In Loiacono, A., Iamartino, G., and Grego. K. (eds.). Teaching Medical English.Monza, Polimetrica Publisher, pp. 75-93.

ABOUT THE REVIEWER

Dr. Daniele Russo is a faculty member at the University of Milan, where he teaches English Language and English Linguistics to undergraduates. His research interests include translation criticism, diachronic linguistics, medical specialized language and translation. He is also a translator of fiction and specialized literature.


Page Updated: 31-May-2012