Alexander Künzli, Stockholm University

This report gives an overview of the results obtained in a research project on translation revision. Translation revision refers to the scenario in which a person other than the original translator checks a draft translation for errors and makes any necessary changes (see also Mossop 2001, p. 169).

Data was collected from 20 professional translators, 10 native German and 10 native Swedish speakers. The participants were asked to revise three German and three Swedish draft translations of three French source texts, while thinking aloud. The texts used included (1) a judicial decision, (2) an instruction manual for an avalanche safety net, and (3) a direct-mail advertisement. In addition, the quality of the revised translations was assessed by a subject-matter expert for each text in the French to German part of the study. These subject-matter experts were asked to identify instances of: (1) justified changes-resulting in a quality enhancement of the text; (2) hyper-revision-unnecessary changes; (3) over-revision-the introduction of errors into the translation; and (4) under-revision-failure to correct errors in the draft translation. To link product with process, changes introduced by the revisers in the draft translations or their failure to make a change were then analyzed together with relevant verbalizations in the think-aloud protocols (TAPs). The overall aim of the project is to investigate the following questions: How do revisers define their task? What kinds of changes do they make and how many? To what extent do they make the draft translations better? How quickly can a good revision be done?

The grant approved by the Bank of Sweden Tercentenary Foundation contributed to the writing of seven articles. The findings are presented here, followed by suggestions for future directions in translation revision research.

Results

1. Künzli, Alexander (2006). Die Loyalitätsbeziehungen der Übersetzungsrevisorin. In Michaela Wolf (Ed.), Übersetzen-Translating-Traduire: Towards a "social turn"? (pp. 89-98). Münster/Hamburg/Berlin/Wien/London: LIT-Verlag.

This paper deals with the revision of the advertising text from French to German. The focus is on exploring the usefulness of think-aloud protocols for the sociology of translation revision, in particular the revisers' loyalty towards themselves and towards the other parties involved in a translation project. The results reveal different types of loyalty conflicts and, in particular, the difficulty of being equally loyal towards oneself as reviser and towards the other parties at the same time. It is shown that if revisers try not to invest more time than they believe they can bill the client, their revised translations often contain a relatively high amount of under-revisions-failures to detect and correct errors in the draft translations. The findings also illustrate the role of power relationships in the production of the target text. Overall, the observations suggest that think-aloud protocols can be used to study the difficulty of coordinating the diverging expectations of the different parties involved in translation projects and to identify strategies for better cooperation among them.

2. Künzli, Alexander (2006). Teaching and learning translation revision: Some suggestions based on evidence from a think-aloud protocol study. In Mike Garant (Ed.), Current trends in translation teaching and learning. Helsinki Department of Translation Studies Publication III (pp. 9-24). Helsinki: Helsinki University.

This paper investigates the relevance of the research data for real-life translation revision and for teaching translation revision. It is argued that this type of investigation is particularly relevant as it is very common in the translation business that translations are revised by someone other than the original translator, and because the demand for translation revision will probably grow. Thus, under the aegis of the European Committee for Standardization, the European Association of Translation Companies has proposed a new standard for providing good translation services (www.cen.eu). This standard recognizes separate roles for translators and revisers; each translation must be revised first by the translator and then by a second person. This means that most translators-to-be will work in parallel as revisers. They need to know the principles and develop the skills on which professional revision practice relies. Based on evidence from the think-aloud protocols of the French to German part of the material, the paper examines three subcompetencies involved in an overall-and yet to be defined-translation revision competence: (1) strategic competence, (2) interpersonal competence, and (3) professional and instrumental competence. Overall, the data highlights the need for revisers to develop a task definition for the revision job at hand, to apply relevant evaluative criteria, and to decide what to do after a problem has been detected. Moreover, revisers need good knowledge of available linguistic information sources, such as contrastive grammars and style guides, to be able to justify their changes. The findings also shed light on the possible consequences of a lack of positive attitude towards the translator's work on the written translation. Finally, the results reveal that although search engines such as Google are a very useful information source for translators and revisers, their use is not entirely unproblematic. Indeed, revisers often stop searches with a feeling of having gotten off the track and wasted time, with no noticeable benefit for the task at hand. Based on these observations, the paper then sketches a possible outline of a course module on translation revision: overall aims and learning objectives, course content, learning and teaching activities, assessment activities and criteria.

3. Künzli, Alexander (2006). Translation revision-A study of the performance of ten professional translators revising a technical text. In Maurizio Gotti & Susan ?ar?evi? (Eds.), Insights into specialized translation (pp. 195-214). Bern/Frankfurt: Peter Lang.

This paper looks at the data from the revision of the technical text. The French to German part of the material contained an assessment of the quality of the draft translation and of the revised translations by a subject-matter expert. The paper focuses on how the revisers dealt with a terminological problem, where the draft translation had four possible equivalents as alternatives. The quality assessment reveals that only one of the ten revised translation solutions is correct. The analysis of the sequence of operations performed by the revisers to produce the German translation equivalent for the technical term left untranslated by the original translator on the one hand, and the analysis of how the revisers' uncertainty manifested itself in the revision process on the other hand, reveal that operations performed to solve comprehension and expression problems often stop before relevant terminological knowledge has been acquired. It also indicates that available information is not always efficiently processed. These findings suggest that a lack of pre-existing terminological knowledge undermines revisers' overall feelings of certainty, preventing them from regarding the source text as an important source of knowledge acquisition and from establishing effective intratextual links. Suggestions for future directions include the study of how the different types of competencies and knowledge possessed by a reviser interact and how dictionaries are used.

4. Künzli, Alexander (2007). The ethical dimension of translation revision. An empirical study. The Journal of Specialised Translation, 8, 42-56 [special issue on translation revision].

This paper deals with the process data from the revision tasks of the French to German part of the study. The focus is on the ethical dimension of translation revision, and on a further examination of the revisers' (sense of) loyalty to the different parties involved in prototypical freelance translation revision jobs mediated by translation agencies. The paper studies how loyalty relationships express themselves in translation revision, to what extent revisers are loyal to themselves and in what respect they feel they have to make compromises concerning the values and ideals they strive for in their work. The think-aloud protocol analyses reveal a number of ethical dilemmas and loyalty conflicts between the different parties involved in translation (revision) projects: source-text author, client, translation agency, target-text reader, translator, reviser. They deal, for example, with the conflict between the economic demand for speed and the ethical demand for thoroughness, reliability, or quality. The verbalizations also highlight the need to consider the revisers' responsibilities to themselves, their own legitimate interests and expectations. The large number of verbalizations indicating that revisers must compromise with respect to the values or ideals they pursue in their work due to time constraints raises the question of whether, in translation research, the effect these situational factors have on the translation process and on the quality of the final product have been sufficiently considered. Thus, a not-so-good translation might often be the result of lack of time or lack of access to an information source rather than of insufficient linguistic or extralinguistic skills.

5. Künzli, Alexander (2007). Translation revision-A study of the performance of ten professional translators revising a legal text. In Yves Gambier, Miriam Shlesinger & Radegundis Stolze (Eds.), Doubts and directions in translation studies: Selected contributions from the EST Congress, Lisbon 2004 (pp. 115-126). Amsterdam: Benjamins.

This paper deals with the data from the revision of the legal text (French-German). The revised translations were evaluated by a subject-matter expert. The focus of the analysis is on investigating the changes the revisers made in the draft translation and the principles that guided their thoughts and actions. The results show that the proportion of changes that do not lead to a quality enhancement, as well as that of failures to correct errors in the draft translation, is relatively high. They also suggest that time is necessary, but not sufficient to guarantee quality in translation revision and that there is a discrepancy between verbalized principles and actual behavior. Thus, several revisers declared that their job was not to retranslate, but to make any necessary changes. An analysis of their evaluative utterances shows, however, that in some cases they tend to impose their own linguistic preferences on the text at the expense of the choices made by the original translator. The data also illustrates the role played by the reviser's (lack of) motivation in his or her performance as a result of tiredness. This last observation is also of practical interest, as it raises the question of the extent to which revision jobs should be interspersed with other tasks (see also Mossop 2007, p. 10).

6. Künzli, Alexander (forthcoming). Address pronouns as a problem in French-Swedish translation and translation revision. Babel.

This paper deals with the French to Swedish part of the research project on translation revision, together with an analysis of relevant material collected within the framework of an earlier study on translation (Künzli 2003). The focus is on the handling of address pronouns. In French, there are two parallel address systems, with either unmarked T (tu) or V (vous). In Swedish, on the other hand, the T form (du) has been the usual form since the 1960s. In recent years, V (ni) has started to reappear, at least in service and business situations. The choice of the appropriate address pronoun may therefore constitute a problem in French-Swedish translation and translation revision. An analysis of the Swedish translations of the French source texts in which direct address was used reveals both interindividual variation in the choice of the address pronoun, and intraindividual variation, with several translators and revisers showing inconsistent address use. Process data from the think-aloud protocols highlights the effort even experienced translators and revisers invest in finding the appropriate address pronoun. The participants' spontaneous negative reaction towards the Swedish V form, even among participants who decided to maintain V after all, further highlights the fact that contemporary Swedish still lacks a V form able to be used with the same function as the French vous.

7. Künzli, Alexander (forthcoming). Think-aloud protocols-A useful tool for investigating the linguistic aspect of translation. Meta.

This paper looks at data from both the research project on translation revision and from an earlier study on translation (Künzli 2003). It deals with both French-German and French-Swedish language pairs. Think-aloud protocols (TAPs) of translation processes have traditionally been used to study the cognitive aspect of translation. The focus of this paper is to demonstrate their usefulness for investigating linguistic and language-pair related aspects of translation. Deborah Cao's (1996) model of translation proficiency is used as a theoretical framework for classifying translational language problems verbalized by the participants. The model is based on Bachman's (1990) model of communicative language ability and includes three main variables: translational language competence, translational knowledge structures (world knowledge and subject matter knowledge) and translational strategic competence (planning, executing, and assessing a translation task). The translational language problems discussed in the paper fall into the following categories: (a) grammatical (the interpretation of French participial clauses), (b) textual (the use of connectors), (c) functional (different realizations of one and the same language function in French on the one hand, and in German and Swedish on the other), and (d) sociolinguistic (the rendering of the formal vous into Swedish by the more formal ni or the less formal du). The findings reveal that translational language problems play an important role in the translation process-even in that of very experienced translators. The results also suggest that the identified language problems may at least partly be explained by Cao's (1996) model of translation proficiency, adapted from Bachman (1990), and more specifically in terms of variations in specific components of translational language competence. However, since greater experience in one language pair does not seem to correlate with better performance in every respect, the findings also stress that more research is needed to shed light on the correlation between translational language competence, experience of translation or translation revision, and quality.

Outlooks

The empirical studies reported above deal with a variety of revision issues. Still, they can cover only a small portion of the extensive material collected within a research project that makes use of the following data sources: (1) think-aloud protocols, (2) written translations, (3) quality assessments, (4) logbook documentation (such as documentation of dictionary use), (5) questionnaires, and (6) interviews. It therefore appears useful to finish by outlining some suggestions for future directions that appear particularly relevant for both theory and practice.

From the theory point of view, the development of a model for successful translation revision seems particularly interesting. Such a model will have to take into account not only fundamental revision subprocesses (evaluation, strategy selection, realization of changes), but also a control structure (such as the task definition), the reviser's knowledge base, his or her motivation and, finally, situational factors. The role played by affective factors such as motivation has received little attention in translation process research (see, however, Jääskeläinen 1999). Likewise, Alamargot and Chanquoy (2001, p. 14-18) have shown that in writing research, revision models have neglected the writer's motivation. Their suggestion of more research into the role played by motivation in the activation of revision processes therefore certainly also applies to translation revision. In a similar vein, it appears important to gain a better understanding of which aspects of the different subcompetencies involved in translation proficiency are relevant to translation revision and to what extent they have to be modified to reflect the specificity of translation revision. That way, teaching translation revision will be even better adapted to the trainees' learning needs.

A second important issue is the possible correlation of certain revising procedures with quality in translation revision. This aspect appears all the more important as the French to German part of the material, which included quality assessments, revealed in all three tasks a relatively high proportion of revisions that did not lead to any quality enhancement, and of failures to correct errors in the draft translations. It seems worthwhile to focus in particular on the questions: In what order do revisers check the different revision parameters such as meaning transfer, facts, language and style, and typography (see Mossop 2001, chap. 10)? In what order do they read the source text and the draft translation? Do revisers conduct a systematic comparative reading of source text and draft translation or do they only return to the source text at specific points? If so, how often and why?

Finally, it seems worthwhile to describe in more detail how the changes revisers introduce into a draft translation affect text structure. As already mentioned, the subject-matter experts assessing the quality of the revised translations in the French to German part of the material were asked to apply a taxonomy consisting of four categories: (1) instances of justified changes-resulting in a quality enhancement of the text, (2) instances of hyper-revision-unnecessary changes, (3) instances of over-revision-the introduction of errors into the translation, and (4) instances of under-revision-failure to correct errors in the draft translation. Evaluating a reviser's work is essential. The identification of successful strategies for improving a draft translation is relevant for both teaching translation revision and real-life translation revision. However, this type of classification does not give any information about the length or the object of a specific revision. Changes can be of more local nature (such as spelling) or of more global nature (text coherence). Here, translation research might benefit by testing the usefulness of taxonomies developed in monolingual writing research (see, for example, Alamargot and Chanquoy 2001). Many of the latter taxonomies build on categories that also appear relevant for translation revision. They categorize revisions according to the level of language, operation, and meaning. In other words, they take into account what (word, phrase, clause, etc.) is done (addition, deletion, substitution, rearrangement) and with what effect on text structure (meaning preserving or meaning changing). This type of linguistic analysis can be conducted independent of any quality assessment, and could therefore be applied to the French to Swedish part of the material. One particularly interesting aspect is the identification of a possible curve in the revisers' work, a shift from working on the level of the meaning of the text to working on the form. The existence of such a curve is not unlikely as formal revisions are cognitively less demanding. This type of question appears very relevant for professional translation revision as well.

Additional works cited

Alamargot, Denis and Chanquoy, Lucile (2001). Through the models of writing. Dordrecht: Kluwer.
Bachman, Lyle F. (1990): Fundamental considerations in language testing, Oxford, Oxford University Press.
Cao, Deborah (1996): "Towards a model of translation proficiency," Target 8-2, pp. 325-340.
Jääskeläinen, Riitta (1999). Tapping the process: An exploratory study of the cognitive and affective factors involved in translating (University of Joensuu Publications in the Humanities 22). Joensuu, University of Joensuu.
Künzli, Alexander (2003). Quelques stratégies et principes en traduction technique français-allemand et français-suédois (Cahiers de la recherche No. 21). Stockholm: Stockholm University, Department of French, Italian, and Classical Languages (www.diva-portal.org).
Mossop, Brian (2001). Revising and editing for translators. Manchester: St. Jerome.
Mossop, Brian (2007). Empirical studies of revision: what we know and need to know. The Journal of Specialised Translation, 8, 5-20 [special issue on translation revision].

Publikationer

1. Künzli, Alexander (2006). Die Loyalitätsbeziehungen der Übersetzungsrevisorin. In Michaela Wolf (Ed.), Übersetzen – Translating – Traduire: Towards a “social turn”? (pp. 89-98). Münster/Hamburg/Berlin/Wien/London: LIT-Verlag.

2. Künzli, Alexander (2006). Teaching and learning translation revision: Some suggestions based on evidence from a think-aloud protocol study. In Mike Garant (Ed.), Current trends in translation teaching and learning. Helsinki Department of Translation Studies Publication III (pp. 9-24). Helsinki: Helsinki University.

3. Künzli, Alexander (2006). Translation revision – A study of the performance of ten professional translators revising a technical text. In Maurizio Gotti & Susan Šarčević (Eds.), Insights into specialized translation (pp. 195-214). Bern/Frankfurt: Peter Lang.

4. Künzli, Alexander (2007). The ethical dimension of translation revision. An empirical study. The Journal of Specialised Translation, 8, 42-56 [special issue on translation revision].

5. Künzli, Alexander (2007). Translation revision – A study of the performance of ten professional translators revising a legal text. In Yves Gambier, Miriam Shlesinger & Radegundis Stolze (Eds.), Doubts and directions in translation studies: Selected contributions from the EST Congress, Lisbon 2004 (pp. 115-126). Amsterdam: Benjamins.

6. Künzli, Alexander (forthcoming).  Address pronouns as a problem in French-Swedish translation and translation revision. Babel.
 
7. Künzli, Alexander (forthcoming). Think-aloud protocols – A useful tool for investigating the linguistic aspect of translation. Meta.