[IEEE IPCC 2005. Proceedings. International Professional Communication Conference, 2005. - Limerick, Ireland (July 7, 2005)] IPCC 2005. Proceedings. International Professional Communication Conference, 2005. - Project management and quality assurance in cover-to-cover translation of a medical journal

  • Published on

  • View

  • Download


2005 IEEE International Professional Communication Conference Proceedings0-7803-9028-8/05/$20.00 2005 IEEE. Project Management and Quality Assurance in Cover-to-Cover Translation of a Medical Journal Mary Ellen KeransFreelance authors editor and translator; Instructor, English for the Specific Purposes, Universitat Internacional de Catalunya, Barcelona, Spain mekerans@telefonica.netAbstractExperienced medical translators are fully employed in Barcelona, so when a well-indexed medical journal began cover-to-cover bilingual publication in 2003adopting a policy of international communication in two languages rather than a single lingua francaa new team of freelance translators had to be formed, trained, supported, and monitored. This case study spans the negotiation, early development, and later solutions that resulted in feedback loops that allowed the publisher, editorial board, and translation team to form a community capable of learning and enhancing product quality. In addition to discussing practical issues, I analyze why bilingual scientific journal publication is a viable strategy for non-native English speaking scientists who want to participate in international discourse in an organized group while retaining use of the language of local epistemic or clinical activity.Keywords: scientific translation, non-native English writing, social construction, knowledge societiesIntroduction: bilingual scholarly publicationan emerging model Cover-to-cover bilingual or multilingual journal publication, until recently, has been of interest for the commercial dissemination of high-profile journals like the British Medical Journal, Lancet, Journal of the American Medical Association, andJournal of the American Dental Association,among others. Concurrent multilingual book publication has also become a common commercial practice as publishers have reorganized repeatedly since the 1980s to seek wider markets. A new trend toward full bilingual publication of small scholarly journals originally in languages other than English is different, however. It has proceeded quietly at great expense on the initiative of independent scientific societies that own internationally indexed national journals, particularly in clinical subspecialties. Examples areArquivos Brasileiros de Cardiologia and Jornal de Pediatria, both from Brazil, various Chinese journals,[1] and the Spanish journals RevistaEspaola de Cardiologia and Archivos de Bronchoneumologa (Archivos).This paper provides a description of the Archivostranslation project, which has been running smoothly since the spring of 2003. I will discuss 1) problems foreseen based on looking at translations from a previous full-journal translation project, 2) new protocols developed to match the goals of bilingual publication, 3) recruitment, support, and monitoring of translators, and 4) unforeseen problems that have arisen and how some have been solved. Issues of quality assurance are emphasized and work load is considered. Finally, the significance of bilingual publication of such journals is discussed in terms of the social concerns that motivated the Archivos project and the growing awareness of paralinguistic obstacles that continue to make it difficult to disseminate 2202005 IEEE International Professional Communication Conference Proceedingsscientific research from non-English language cultures in spite of indexing, quality translation, and open access. The purposes are 1) to describe a procedural model whose features might be applicable in similar contexts and 2) to suggest theoretical frameworks that can be applied to analyze such enterprises. In a case study of an ongoing project whose participants cannot be fully concealed, and of which the author is a participant, issues of confidentiality and conflict of interest must be addressed frankly. To assure that treatment would be fair but that data would be available for use, the following measures were taken: 1) permission to use examples of internal correspondence among translators was obtained from team members; 2) permission to use examples of drafts, errata, and negotiation of changes was obtained from the publisher and editor in chief; 3) measures to maintain the anonymity of authors were agreed upon; and 4) other participants were promised the opportunity to append their response to a draft of the manuscript. Translators were also given a short open-ended questionnaire asking about their motivations in joining and staying with the project. Case Description: translation management for open access bilingual publication Spring-Summer 2002concerns about quality assurance, defining the problem and the goals Archivos, a monthly respiratory medicine and chest surgery journal, was the second well-indexed title the publisher, Ediciones Doyma, proposed to translate fully for an open-access web version. The journals editors were motivated by a desire to increase readership, citation, and impact factor without abandoning Spanish, the clinical language of members of the two medical societies that supported the journalthe Spanish Society of Pulmonology and Thoracic Surgery (SEPAR) and the Latin American Thoracic Society (ALAT). In doing so, the editors also sought to attract more and better manuscripts by serving authors better. What the editors asked of the publisher were translations that accurately reflected the original texts and were sufficiently well presented and written to inspire the confidence of the intended readers. How to obtain such translations was not immediately clear, given that accomplished medical translators are generally fully employed by the existing market in Spain. Furthermore, the publisher expressed concern over translation quality and missed deadlines in an earlier cover-to-cover translation project that had been running for several months behind schedule. New solutions, including how to choose translators, schedule production alongside translation, and monitor it, were wanted. Estimates were sought from the usual suppliersagencies and freelancersbut confidence in simple outsourcing strategies was low and the publisher was now willing to rethink them. As one of the possible suppliers, I first reviewed the existing translations the publisher had posted for the earlier bilingual journal. My purpose was to assess whether the criticisms of past work were justified or unreasonable. I quickly found errors as gross as heard that should have been heart in an article title, but most problems were more subtlesentences that were difficult to read and atypical terminology choices by translators. I also saw evidence of abstract and full text information mismatches that pointed to process quality assurance questions that were not caused by translation error but that should have been detected at that point and at earlier ones. Occasional translation errors suggesting lack of understanding were noted. On the other hand, stretches of text that were problem-free were also present, suggesting that problems might stem from text management problems rather than translator incompetence. In the process I noticed web page problems outside the scope of translation that would interfere with document access. I concluded that the project was promising, provided the editors and publisher would integrate translation and publication processes so that the needs of both were met. I proposed a budget for quality translation as conservative as possible, but with a view to real conditions in the Spanish freelance market. I asked that the publisher in turn develop a tight production schedule that showed when translation of an issue would take place in the context of other steps. On the assumption that translation quality assurance needed to be part of other types of monitoring, I asked that document retrieval problems be studied and corrected. I also argued they should target a schedule that allowed posting translations and MEDLINE abstracts simultaneously so English texts would be available as soon as the first potential international readers2212005 IEEE International Professional Communication Conference Proceedingswanted them. Finally, I asked that translators be treated as authors of the English versions in the editorial process, in the sense that no changes be made in their texts without their approval. Table 1 summarizes the values and expectations expressed by the different parties during negotiation. Related topics are grouped in rows. Initiating points, in bold face, should be read first. Responses are in italics.Table 1. Issues discussed in project negotiation, decision-makers responses, and outcomes* Managing editor (publisher) Translation manager Editorial board Outcome Main negotiating pointsProvide overall quality on time. Respect the translated textsif changes are wanted, translators must approve them. Response: Were not the New England [Journal of Medicine]. Agreement that we are targeting correctness, clarity and readability. A text should inspire reader confidence. Predictable turnaroundmeet production schedule deadlines at all key points. Predictable work flowpublisher and editors should follow a strict production schedule.First response unknown; conflict developed in the second year as editors complained that issue closing deadlines were too early for end-of-year-issues.Apparent agreement, shared goals Response: agreement on the value of simultaneous bilingual posting, when abstracts are sent to databases Translations should be posted and linked to databases at the moment the Spanish journal is indexed. Response unknown Agreement that the production schedule would assume this goal Response to editors request for freeze: not possible. Fees must be at the high end of market rates. Response: protest at overall cost. Ask for 3-year freeze. Agreement to try to contain costs. No 3-year promises made. Lowering of project management costs is predicted. Process issuesTranslation starts when the page proofs are available. Response: Good way to assure that texts have few if any Spanish language or information errors. Response unknown. Agreement Response: agreement, although the page proof (pdf file) was named as the source text. Eventually the publishers assistant undertook conversion. Work should be submitted on Microsoft Word files. Response unknown. Agreement. Editors idea of routing full translations would delay posting and not be of clear benefit, based on previous experience. Editors idea would potentially lead to a great deal of time spent negotiating final English form. Authors should approve the translations. In the past, authors received translated abstracts on their galleys. Pending quality assessment, translators retain author-type control over English version. 2222005 IEEE International Professional Communication Conference ProceedingsManaging editor (publisher) Translation manager Editorial board Outcome The translation project manager should route abstracts to authors for approval and negotiate form (step added during piloting of first issue). Response: Acceptance as a good opportunity for a manageable feedback loop between translators and authors. Response: Wait and see if this is sufficient for author approval of translations. (Accepts to try publishers proposal.) Agreement on the usefulness of routing abstracts to authors. Translators should proofread the English galleys (added after budget had been accepted). Response: Excellent way to monitor compliance with translators insistence that submitted translations be respected. Symbolic fee is added. Response unknown Agreement * Bold face indicates initiators of points; italics indicate responses expressed. Related points are grouped in rows.The mediator was the publishers managing editor, although at different times others present included the editorial assistant, the production manager (responsible for page making and other physical processes), and a page maker; the principal freelance copyeditor assigned and the head in-house copyeditor; the manager of computer services; members of the scientific societies editorial board; and one translator. Direct contact with the editorial board was intermittent, and messages were usually conveyed through the managing editor and assistant, who met with them weekly. The importance of the managing editors mediation was great, as the project held surprises that were potential sources of either conflict or learning. A budget was based on the assumption of a 2-week turnover for translation, which would have meant that post-translation editing could not be accomplished. Rather, individual translators would have to guarantee quality, supported only by terminology consultation through the project manager. Post-translation editing and feedback would be provided only on abstracts. Overall quality could only be spot checked randomly on each article and translators would be evaluated accordingly. Another assumption was that, although a new team of scientifically inexperienced freelance translators would have to be formed, they would have a very steep learning curve if given support. Sporadic monitoring would identify problems, and editors would judge whether the system was succeeding or not. SEPAR and the editors accepted the publishers budget, after I had submitted the translators, and the project was scheduled to run for 3 years. Fall-spring 2002-2003focus on matching processes to goals, recruiting and piloting Even as the budget was accepted, the project became more ambitious than originally conceived. Two changes played a major role in the translation processes eventually developed. One was the publishers switch to a 4-week turnover period for translation and the other was finding the local freelance market to be even tighter than anticipated. Table 2 shows the effect of the first change in terms of time budgeted and time actually used in the first year, during which translation only took place for the July through December issues. Face-to-face meetings between the translation project manager and the publisher or publisher and editors in the first year took up 10 more hours, so start-up time was therefore about 5-fold more than expected. Reasons will be analyzed in the next section. Table 3 shows the 7 steps in the publishers production schedule that affect translation scheduling directly, alongside the management and quality assurance steps necessary. In addition to the 7 steps shown, in-house cycles included copyediting between steps 1 and 2 and 2 galley cycles for author and editor approval along with proofreading before step 4. Before step 1, peer and editorial review has taken place. Table 3 should be read by first understanding a single issues cycle from the translators and publishers point of view (columns 1 through 4). Then, the final columns show the project managers actions and overlapping issues. Issue+12232005 IEEE International Professional Communication Conference ProceedingsTable 2. Work load for project management of cover-to-cover biomedical journal translation Concept Hours budgeted for 6 months Hours spent in 6 monthsRecruitment 6 30 Orientation and monitoring: journal and genre style, drafting processes, resolution of translators doubts by the team leader, establishment and revision of terminology databases; spot checking Post-translation quality assurance: editing of titles, abstracts and key words; checking full translations against source texts 50 250 TOTAL related to translation quality assurance: 56 280 Table 3. Translation and quality assurance steps in the bilingual publication production cycle (spacing is not proportional to timing) PublishersscheduleReference issueWorkingintervalTranslator Project manager Overlapping issues 1. Table of contents sent to project manager Day 1 12 days State com-mitment for the issue. Negotiate assignments. 2. Copyedited manuscripts sent to translators Day 10 Translatetitle,abstract, and key words; route to projectmanager. 3. Deadline, translatedabstractsDay 20 2-3 days, translator+ 8 days for projectmanager and authors Answer translator queries, revise and negotiate the translation to route to author, with queries. Negotiate final version between author and translator.Prepare English table of contents. Translation of issue-1 is taking place (30 days). 4. Spanish source text arrives as page proofs (pdf file)Day 55 5. Deadline, full translations Day 85 30 days Translate full texts. Help with translators research and interpret journal style. Give at least 1 post-translation edit, spot-checking against source. Negotiate final version with each translator.6. English galleys sent to translatorsDay 92 3 days 7. Deadline, proof-read galleys Day 95 Proofread. Start full article translation, issue2Deadline, full translation, issue3Proofreading, issue2Assignments given for issue+1Deadline, full texts issue2Start translation, issue1 Abstract Deadline, issue+1translation, issue-1 Assignments for issue+2Start translation, issue+1Abstract translation, issue+22242005 IEEE International Professional Communication Conference Proceedingsdenotes the one following the reference issue described in the first column. Issue 2 is the one started 2 cycles earlier, and so on. Thus, 6 issues overlap. The complexity of journal production surprised many translators and had to be adjusted to. For example, the need to translate abstracts early in the cycleso they could be inserted into Spanish page proofs for the editors final approvalat first led to translators requests for changes in them later on. Such changes generate publishing costs, but they also generate serious information transfer problems if publishing process quality assurance does not guarantee that the right abstract is posted to databases. Soon, however, translators gave a different type of attention to abstracts, as they learned to comb manuscripts for information gaps like unspecified units for lab test results that could be queried when abstracts went to authors for approval. Overall, the publishers allowance of a 4-week turnover for translation (steps 4 to 5) signaled that expectations were high and meant that post-translation editing and feedback to translators could take place much more fully than originally planned. Proofreading (steps 6 to 7) had not been planned originally and likewise contributed to raising product quality. Together, those changes meant that the translation team had a reasonable framework that encouraged taking full responsibility for English version quality, including tasks usually handled by copyeditors in publishing settings where English is the local language.Recruitment and retention of the first team members took longer than anticipated because experienced scientific translators are well-supplied with work at rates they consider adequate and lacked incentive to join a project. This was the case even though the publishers agreed to pay 33% more than their common rate and about 22% more than well-paid nonscientific translation rates earned in the local market. Although 7 or 8 experienced translators had been planned for (6 or 7 plus the project manager), to handle between 35,000 and 40,000 words per month, the full team has included 10, handling closer to 45,000 words per month. We chose to have a large team so as not to leave the journal vulnerable to illness or burnout and not to leave translators overcommitted to one client and vulnerable to sudden loss of income. Translators were recruited mainly from among lightly experienced translators interested in a new challenge or income source and willing to learn. Three original translators besides the team leader could be described as highly experienced, though only the project manager was experienced with science and the genres of medical journal articles. Attrition has been a problem. Of the original team, 5 remain besides the manager, 3 as core translators of longer articles and 2 as regular translators of shorter ones. Two soon left: one highly experienced translator found the medical content upsetting and another lightly experienced translator realized that learning medical translation would keep her from other priorities. A third from the original group left after the third issue, explaining the fees did not compensate for the time that had to be spent. Of 6 translators (4 lightly experienced, 2 experienced) who joined the team between the third and fifth issues to replace original translators or replacements, 3 continue. One of the 6 (lightly experienced) joined and left after doing excellent translations for 2 issues, explaining that translation was less satisfying than teaching. Apart from that person and the first translator who left for personal reasons, the other 4 lost translators realized the task of working to the standard required would take more time and effort than they had originally expected to give. Of the 6 translators who tried and quit, the manager judged that all were capable of learning the work. Instead, they preferred other sources of income.The 5-fold increase in recruitment and monitoring load shown in Table 2 can thus be attributed to freelance market conditions, to the originally unforeseen task of routing abstracts to authors for approval, and to post-translation editing and feedback loops. This increase in time gave much stronger support to product quality but could not be billed without re-starting negotiation. Addition of a proofreading step, however, led to an 8% increase in fees that helped motivate translators to assume sometimes unfamiliar editorial responsibilities over the coming months. The first issue was started 3 months early to pilot the feasibility of the production schedule and process, after which routine translation started. Spring-summer-fall 2003, winter 2003-2004learning in the early stages of implementationThe premises underlying recruitment, support, and 2252005 IEEE International Professional Communication Conference Proceedingsquality assurance were that 1) work would proceed mainly through the Internet, with few meetings; 2) virtual contact among translators and with the project manager would be practically daily, whenever the translator needed it, by way of a group e-mail list; 3) post-translation editing would take place but the translator was ultimately responsible for accuracy as the editor/project manager would be unable to check every word of every translation against the source text. Post-translation editing, therefore, identifies problems and raises issues, but editing must always be re-evaluated by the translator, who is also the last person on the team to see page proofs before web posting.Training consisted of an orientation meetinghandled individually for late-arriving team membersand informative documents. One document provided a translation self-test for lightly experienced persons to decide if they were attracted to the task. Another described the goals of good-enough translation in terms of readability for scientists of whatever language background who needed texts in a style familiar to them from the genre and as culturally neutral as possible. Advice on drafting and controlling types of attention at different stages in translation process were given in that document and at the orientation meeting. Problems found in the publishers earlier project with another journal were discussed in another document. Yet another document dictated a file-labeling system to allow drafts and editing stages to be tracked at a glance and described the publishers production process to translators unfamiliar with it. A method was prescribed for handing abstract translations over in a 2-column table (Figure 1) that would allow the project manager to evaluate them and dialog quickly with the translator until agreement was reached, and then forward them to a corresponding author, who could also evaluate them and respond quickly without having to Figure 1. An edited draft translation of an abstract, with comments2262005 IEEE International Professional Communication Conference Proceedingshandle several files. Translators who needed help learning to use more advanced word-processing features, such as creating tables or tracking changes and adding or deleting comments, were counseled individually by e-mail or phone. Some needed to learn how proofreading was done, while others only needed to be apprised of the differences between Spanish symbols and those used in the UK or USA. Only one translator, who joined at the third issue and left after the sixth, used a translation memory program, so a system for maintaining a team glossary was established. Tools consisted of one encyclopedic medical dictionary and a spell checking program for medical and pharmaceutical terms. At the end of the pilot issue, the publishers head copyeditor also unexpectedly mandated adherence to the style manual of the American Medical Association (AMA),[2] rendering unnecessary previous discussions among the translators to come to consensus on style. The style manual immediately proved a valuable resource to have on hand for a variety of reasons, however. Copies of articles were provided to translators interested in learning more about the interface between writing, publishing, translation, and editing.[3-7] How to build a corpus for use with a concordancer was also described and free software was recommended (AntConc 3.0, http://www.antlab.sci.waseda.ac.jp/antconc_index.html), but this tool was not widely used until early in 2005, when a 215,000-word corpus proved sufficient for answering many queries. By that time the Archivos web page was being used as a parallel corpus and 2 translators were switching to computer-assisted translation. The pillars of ongoing translator development were personal feedback through editing and revision, as shown in Figure 1, and group querying. A typical query example is shown below:Can I say experimental study for unestudio de intervencin. It doesn't say much more: Bicker et al21 comunicaron, tras un estudio de intervencin de 24 meses de duracin..... The following reply was given: No, I think experimental study would be odd for a clearly clinical context. Trial is used for studies in which an intervention is being applied. Look at the title of the reference.However, the author probably chose estudio de intervencin because the relevant chunk is intervencin de 24 meses rather than estudio de intervencin. The treatment period was 24 months. If you think thats the case, you might consider something like Bicker et al21 evaluated an antiparasitic treatment over a 24-month period, finding that an association between.... Queries involved terms, natural expressions for the genre, as above, or apparent or real ambiguities or contradictions. Many concerned how to apply AMA style. In addition, translators themselves organized a meeting early in the pilot issue, and topics covered were mainly general questions of style and file exchange protocols. A third meeting was scheduled after 6 months of work to teach explicit editing criteria, because group list querying and personal feedback were proving insufficient to guarantee a steep learning curve for all translators, some of whom were learning both scientific and editorial concepts at the same time. Specific tasks to complete and discuss, as exemplified in Table 4, constituted the agenda, with the following stated purposes: 1) to develop the ability to see where revision is needed and 2) to talk about the range of ways to solve a text problem that has been detected. Not all problems could be solved within the translation team. A system of author querying was therefore proposed that would collect information without involving the author in prolonged discussion or review of full translations. The essence of author querying was to keep the exchange shortasking questions that could be answered with a yes, a no, a limited choice, or a very short clarification or explicit information. A good query gave enough context to allow the author to answer based on the e-mail alone, without looking for the full text. Examples were provided to show that all queries (written in Spanish) needed the following elements: 1) a subject line to identify the journal, 2) personal introduction (I am translating your article ....),2272005 IEEE International Professional Communication Conference ProceedingsTable 4. Examples of tasks for training translators to be critical revisers of their own work Punctuation tasks: Detect the problem and edit before looking at the comment.Sentence Comment Patients usually present with respiratory symptoms, fever and weight-loss. Use American Medical Association style. (Put the Oxford comma after fever.)In patients infected by the human immunodeficiency virus (HIV) multiple, cavitated nodules can be seen. Commas are unusual between adjectives in English. When they occur its because theres a list or because there are two of the same type of adjective. Remove. Optional comma after HIV if you think readability is improved. Another more recent study,25 evaluated the diagnostic yield of CT, PET, and endoscopic ultrasound in the staging of lung cancer in candidates for surgery. Why put a comma between a subject and its verb? Remove. (And move the reference number to the end as no author is named.) Revising/Editing tasks: Cover the Spanish column until you have a sense of where the problem is. Then, use the Spanish to help revise. Make a decision about the draft before looking at the comment. English draft Spanish source Comment Bronchoscopy often reveals stenosis due to infiltration or caused by an exophytic mass.6A menudo la broncoscopia objetiva estenosis infiltrativa o por masa exoftica6.Parallelism in lists or for concepts that should be read as similar to each other: create it if necessary. The translator has used a simple adjective (due) and a derived one (caused) and as they both mean the same thing, the reader pauses to wonder why. Suggestion: ... stenosis due to caused by infiltration or caused by an exophytic mass.6Amphotericin B is the drug most widely used in the treatment of IPA.43 The initial dose is 0.6 to 1.2 mg/kg/d, infused intravenously. In severe infections, however, higher doses are often needed, which increases adverse effects, mainly nephrotoxicity, electrolyte disturbances, and hypersensitivity reactions. La anfotericina B es el frmaco ms usado en la API42. La dosis inicial es de 0,6 a 1,2 mg/kg/da, por va intravenosa, aunque en infecciones graves a menudo se precisan dosis ms elevadas, con lo que aumentan sus efectos secundarios, fundamentalmente la nefrotoxicidad, los trastornos electrolticos y las reacciones de hipersensibilidad.Grammar provide a specific referent for each relative pronoun, preferably adjacent to the pronoun. (The structure ...needed, which... is not strictly correct in written academic prose.) Suggestion: When higher doses are needed to treatIn severe infections, however, higher doses are often needed, which increases adverse effectsmainly nephrotoxicitykidney damage, electrolyte disturbances, and hypersensitivity reactionsaremore common. Did you notice the incorrect reference number?Results of the various studies in terms of efficacy, however, have been very variable.62-64... pero los resultados de los diversos estudios en cuanto a su eficacia son muy variables61-63.Tighten the prose by expressing it directly. Did you notice the reference number error? Suggestion: Results of the various studies in terms of eEfficacy, however, hasve been very variedgreatly from study to studyable.62-6461-632282005 IEEE International Professional Communication Conference Proceedings3) a statement of the problem (I have difficulty interpreting the phrase xxx xxx in the following sentences from page xx.), and 4) a direct request (Can I interpret this to mean...[restatement]?). An example, translated from the Spanish that was used in the real query, is shown in Figure 2. Some might argue that such matters involve rewriting and are not for translators to solve, while others might consider that a competent medical translator should be able to correct the problem stated without author help. In the Archivos project, we insist the translators be alert to breakdowns in text coherence, as described elsewhere,[6, 7] yet remain highly sensitive to when authors need to be consulted. In the query shown, there was a possibility that the figure 21% was of the whole patient sample or of a subsample, and looking at data tables in the article did not settle the question. In fact, thanks to the query, the author was alerted to a discrepancy between a table and his textwhich also explained why the translator could not solve the problem aloneand wrote to say, Theres an error in the percentage of hypertensives in the whole patient sample, which should be 44.6% as in Table 1, not 21%. I forgot to make that correction [on the galleys].... Such queries were produced at a rate of around 10 to 15 per issue, but many articles required none. Translators expressed difficulty framing such questions, deciding on how much context they needed to provide, and sensing when a query could not be put off; so, as project manager, I often became involved in helping to draft them. At first, some expressed frustration or even resentment over problems they felt authors, peer reviewers, and copyeditors should have detected before them but which they were expected not to reproduce in the English version blindlysuch as inconsistent presentation of findings that made interpretation difficult, plainly wrong numbers like the one in Figure 2 (or 16/9 mm Hg instead of 160/90 mm Hg for blood pressure in another text), forgotten references, and contradictory statements in the texts. As manager, I framed such problems as a matter of understanding writing and editing process as discussed in the articles provided during orientation.[3-5] Translators were asked to see writing as being unfinished until a text was published. It sometimes had to be mentioned that peer review has an uneven record: it can lead to manuscript improvement but does not necessarily do so.[8] In short, scientific journal publication was presented as an imperfect enterprise that needed work. They were also asked to take an interest in studying how editorial process affected writing process and to see the translators role as necessarily involving an editorial function in a publishing context where English is not the local language. As the team came to work more confidently and understand scientific publishing more fully, frustration seemed to wane as new habits took hold. Detection of source text problems also led to the start of feedback loops between the translation team and the editors and the publisher during this period. By the third issue, an emergency meeting with the publisher was scheduled to look at problems found by translators during the 3-week summer holiday during which translation was taking place but editors, publisher, and authors were unavailable. Rather than courting conflict by stressing traditional domains of action for peer and editorial reviewers, copyeditors, and translators, we focused on how to feed back information emerging as a result of the type of close reading of a text that a translator is obliged to give. We wanted to know if feedback could be used to improve details of the Spanish article in page proofs and also if it could affect the way future articles were being prepared. Some, but not all, problems suggested a need to change the publishers notions of the scope of copyeditingfor example the finding in a source text of a proofreading error, FC/ VO2, a meaningless reference to a ratio of true sampling errors for 2 variables, instead of FC/VO2, a meaningful reference to a ratio of changes in those same variables (FC is heart rate in English). Others were writing problems that could have been detected during copyediting, but which also should have been detected by authors, peer reviewers, or editors even earlier. An example has been mentioned above in the context of author querying (Figure 2). In another example clearly indicating inattentive writing and reading on the part of all who had seen the text in earlier stages, the translator found a Discussion paragraph that began by stating, XXXX10 evaluated...in an uncontrolled trial..., but soon stated that the improvement in an outcome measure in that same trial was significantly different in the treatment and control groups. Author queries had gone unanswered during the summer and the article being cited was very old and not readily available, so the translator attempted to camouflage the contradiction in the 2292005 IEEE International Professional Communication Conference ProceedingsFigure 2. Translation to English of a typically short, highly specific query to which the author can answer yes or no and add information optionally. Even though the translators understanding was still imperfect at this point (see 79% of hypertensives), the query and later revision led to correct translation. English version and reported doing so. When author contact was eventually made, an erratum slip was issued for the Spanish edition and the English version was corrected accordingly. Early spring 2004 to the presentsolutions and smooth processing Given that translators were detecting textual problems that sometimes involved scientific information, the publisher was motivated to postpone the moment of printing the Spanish edition as long as possible into each scheduled translation period and establish in-house procedures to make use of information coming in about each paper. Another editorial process change was the commissioning of a post-acceptance technical reviewer-editor. The wording of acceptance letters was also eventually changed to take into account that revision, even after acceptance, might have to be negotiated. Another effect was tighter quality assurance measures by the head copyeditor. We rejected periodic meetings as impractical because they were time-consuming and often came too late in the production process to be useful for making corrections on specific texts. End-of-translation handover notes about the text were rejected for the same reason. Instead, translators continued to work problems through by way of team and author queries, but as soon as a specific error was detected, an e-mail was sent to the publishers editorial assistantthe single person who always knew the status of all Archivos manuscripts in production and how information could best be channeled. That assistant stored all error notices and transferred them to the copyeditor or page makers at the right moment. The practical advantages of this system for the translators were that they solved their text-related problems promptly in a way that circumvented frustration and eliminated the worry that they might be 2302005 IEEE International Professional Communication Conference Proceedingsmerely guessing at meaning. Errors were communicated in a way all considered useful before printing to guarantee the Spanish and English versions would contain the same information. No one felt vulnerable to eventual accusations of inaccuracy and long-term advantages could potentially come from the publishers analysis of errata issue by issue for patterns suggesting the need for other process changes.As translators became more adept, the project management burden (over 40 hours per issue in the first 6 months) became less but, even after nearly 2 years, has not fallen below about 25 hours per issue for support duties, well above the predicted time. The reason is that while simple queries have decreased, post-translation editing, originally not budgeted, has continued. Such editing serves several functions: 1) it adds polish and consistency at the same time accuracy is being checked; 2) it helps produce translations in stable house style (AMA style), so copyeditors are not tempted to input changes, potentially introducing language errors; and 3) feedback for on-going translator development is individualized. New translator orientation eventually tapered off, as the team became a stable group of 10 personsout of 16 who had at some time participated. However, as the learning curve and usefulness of members varied, differences in per-word fees developed, and as manager I charged a symbolic difference between the negotiated rate and some translators rates. The differences introduced were slight, but encouraged improvement and training for relatively less experienced medical translators. The extra invoice processing work for the publishers editorial assistant was tolerated, partly because having a large team of 10 leaves the journal less vulnerable to sudden changes. The most important recent improvement has been the publishers development of a production schedule that is relatively free of overlapping of full-text translation periods caused by holidays. Theoretically, overlapping should matter little, the publisher had originally assumed, given that the same translation period of 4 weeks is available per issue for every article. However, overlaps meant that quality assurance oversight was threatened, if for example abstract editing for a large issue happened to coincide with full-text editing of another large issue plus a large part of the translation period of a third. (See Table 3 for inevitable overlaps.) Once serious unnecessary overlaps disappeared, and translators of shorter texts also complied with a slightly shorter turnover time so that post-translating editing could be staggered, translation ran much more smoothly. The most striking feature of this bilingual publication project is depicted in Figure 3, which shows a wealth of feedback loops through Figure 3. Flow chart showing information transfer and stable feedback loops created during translation process within the publishing processAuthorPeer Reviewers PublisherEditorial BoardTranslationProjectManager TranslatorTranslationgroupTranslationProjectManagerPublisher11223344manuscript source textEnglish translation English article posted on the web translation processpublication processcreation2312005 IEEE International Professional Communication Conference Proceedingswhich participants formed a system capable of learningfrom information sent back to sources. Enhanced sensitivity to dissonance in source texts was naturally generated through the intense attention to local and overall meaning that is required for careful translation. However, without participants willingness to receive, interpret, and act on the information, feedback loops would probably not have prospered and become strong. The ability to perceive textual dissonance as something that can be and needs to be rectified has been described as a key feature that distinguishes proficient writers from novice ones,[9] but as translators are not the intended revisers of texts, there is potential for a conflict of roles if the situation is not managed wisely. Translators do in fact revise them in pre-publication processes [6, 7] just as other non-authors revise, especially in non-English language settings.[10, 11] Observation of the Archivos bilingual publication project suggests that an analogous ability to sense and correct problems underlies successful editorial process if feedback loops are operating. Response to feedback on textual dissonance sensed by translators has been positive enough to affect long-established editorial habits. Recently, at the start of the third volume of bilingual publication, the role of translators as sensors was recognized explicitly by the outgoing editor,[12] while the new editor proposed creating an additional formal loop between the translation team and the new post-acceptance technical reviewer empowered to suggest negotiation of further author revision if needed.Discussion To the well-known pressures to choose English for professional interaction and scholarly discourse, editors of biomedical and related scientific journals have another powerful incentive: English abstracts are essential for inclusion in the great bibliographic databases. The importance of indexing by the US National Library of Medicines MEDLINE has been reinforced by its free availability on the Internet. English is also one of the keys to inclusion in the prestigious databases of the Institute for Scientific Information, the compilers of impact factors and other bibliometric indices often used to evaluate researchers contributions. Nevertheless, indexing is an insufficient condition for reaching international readers, who seem to face many obstacles to discerning the relevance of information, especially if it seems to be published on the periphery. The problem is important enough to have attracted researchers attention. Gregoire et al [13] found that article choice based on language of publication has an impact on the conclusions of systematic reviews of randomized controlled trials. For example, trials published in English have been found by some researchers to report statistically significant treatment differences more often than non-English trials,[14] a pattern that would indeed affect conclusions if language influenced a reviewers choice of reading. Juni et al [15] found, on the other hand, that it was non-English trials that reported more statistically significant results, but that the overall effect of omitting them from a meta-analysis would be only 5%. The design of those trials tended to be poorer, they said (a situation that would prevent data from being pooled and that would therefore lead to omission of the non-English trial anyway). However, another study of English and German trials found that design quality was similar.[16] The last authors summed up the situation wisely, saying that non-English research should not be ignored because it is clearly difficult to ascertain the effect on any particular literature review. Still, there is obviously a tendency toward separate discourses based on language in spite of the increasing inclusiveness of MEDLINE and other databases. Scientific societies attempting to attract readers through open-access translations are therefore responding in a heroic way. Even so, even as they close the language barrier, they meet additional ones. One barrier is the visual flagging of originally non-English articles in square brackets on database search outputs, even when free full text links will lead first to an English translation of a Brazilian, Spanish, or Chinese journal, while the original language versions will be hidden behind several additional hyperlinks. Citing conventions also undermine bilingual publication efforts: when authors cite these journals, they must properly refer to the original language version, to which the international standard serial number belongs, and thus no trace is left of an English translation having been read. In this way, natural dissemination among readers is not reinforced. Citing problems have been particularly problematic for Chinese journals.[1] The present case study has focused on how one groupcontributors to a journal owned by two scientific societiesis managing the work of 2322005 IEEE International Professional Communication Conference Proceedingsadding a new language to their discourse without shedding full use of their native one. How individual teams of authors whose native language is not English cope with publishing in English has been the topic of research in applied linguistics and language teaching [3, 4, 9, 10, 17, 18] and has also been discussed by scientists and editors. [19-22] How commercial and learned publishers cope with the added difficulties of managing their texts has been little discussed, however. Nevertheless, studying the coping strategies of bilingual groups or the services that support them is probably as important to understanding changes in language use and participation in knowledge societies over time as are the strategies of individual bilingual writers. Bilingual publication is of particular interest because it potentially supports natural local language growth and epistemic activity at the same time it facilitates international readership. This is especially important in medicine because English is unlikely ever to become a global clinical language, and if lingua franca use by the powerful evidently reinforces a social elite [23] or there is intellectual isolation of those who share a geographic area but not an epistemic language,[24] health care discrepancies are likely to increase. Furthermore, monolingual English publication within a non-English context may do little to attract international readers even as local readers are lost, a possibility suggested by a finding that the national impact factors of Spanish journals published only in English were lower than those published in Spanish, even though the international impact factors of the English-only journals remained modest.[25] Little is understood about bilingual reading behaviors in these contexts, however, even though sustaining English publications in order to gain readers from abroad is fairly common [26-28] and the Netherlands has published science mainly in English from the seventeenth century while only a few Dutch language journals have been sustained.[28] Cover-to-cover bilingual scholarly publication, then, is an interesting development in the epistemic activity of what sociologists have termed knowledge societies [29] and what linguists have termed discourse communities.[30] Furthermore, if scientific writing is viewed within the framework of social construction of ideas and their representation in discourse, the possible effects of translators or authors editors need to be examined, a point made strongly by Burrough-Boenisch [10] in an analysis of the persons who shape the form of texts by many writers who are not native speakers of English, and also mentioned by Flowerdew.[11] The present description of translators early roles in a particular case provides a background for later analyzing the effects of translation on other participants, including authors, in other similar cases. Certain features in the Archivos projects quality assurance design are of practical interest for journals wanting to take a similar direction. First, it was set up quickly, using the existing freelance market structure and translators currently available in their present state of expertise. Barcelona is a privileged setting in which many educated native English speakers offer their services, two university translation schools train undergraduates, and several universities run postgraduate translation programs with specialized modules. Nevertheless, accomplished medical translators and authors editors are fully employed by small research groups, as Spain is a country where many investigators use the help of language mediators to one degree or another. Therefore, a fee structure and conditionsresearch support and oversightwere provided to motivate lightly experienced translators to cooperate in producing the required quality in a new field. A second feature worth replicating is the willingness of a publisher and editors to devise new protocols and schedules to accommodate translators needs in the interest of producing quality translations to deadline. To do so, all participants had to learn aspects of each others values and ways of working. Translators, for example, had to learn about why abstracts were translated much earlier than the full texts were, inconveniently so from their point of view. Editors, meanwhile, had to learn how long quality translation and assurance takes and the real costs involved. The third noteworthy feature is the feedback loops which were initially established to support and monitor translation quality but which proved useful for guiding changes in other editorial processes. The willingness to take time to work closely with translators may be a defining feature of the successful client looking for quality. At a round table discussion among demanding clients of financial translations organized by the French society of translators, it was seen that top-end clients rely on skilled translators to point out holes in their source texts, to the extent that input from a skilled translator will almost always lead to changes and clarification of the original text (p. 2332005 IEEE International Professional Communication Conference Proceedings6).[31] That this happens between translators and individual writers is not surprising, but the Archivos case shows how it can also happen in a more complex group driven by strict deadlines. The outcome of changes for editors, authors, and the publisher is still unknown, although the hypothesis that quality will be enhanced over time is a testable one. Finally, stable translators in a stable system have made this project possible and a profile has emerged of the person likely to be motivated to stay. A questionnaire revealed that personal learning was a main motivator. Some mentioned learning about the content itself or problem-solving skills that carried over into other work and finding ways to bring knowledge from other work to bear on the Archivos project (synergy). Others mentioned personal pleasure in doing a demanding task well and surprise at liking to have work edited. They expressed satisfaction at being expected to give quality and having success noticed and appreciated. Proper pay was mentioned by most, always balanced with other sources of satisfaction, and several noted that although the same overall amount could be earned on easier jobs at lower rates, the Archivos one was preferred. A liking for associating with other translators on a team was also expressed. The profile that emerged was of a person who thrives on intellectual challenge, knows how to turn difficulties to advantage, and appreciates excellence. The ability to take pleasure in making an effort and researching content and language is apparently more important than academic background, as the present team includes people with studies of widely varying levels and types. Future challenges include finding ways to assess the results of, and hence the utility of, cover-to-cover translation, although subjectively only satisfaction has been expressed by everyone involved to date. No specific outcome measures other than increasing readership, possibly reflected by impact factor, were set for the Archivos project. Similarly, the editors of another journal defined their goals in terms of a definition of the product itselfto create a vehicle to make their research available worldwide, free, in two languages.[32] However, is sufficient visibility gained by using this expensive strategy? Are visibility and use growing? Downloading of articles can be monitored and so can impact factor, but both are imperfect measures of readership and reader response. Do readers perceive quality of translation the same way the editors do? Comments by participants in the feedback loops indicate that another line of inquiry would be to examine whether members of the scientific societies perceive benefits either from the product itself or participation in the interactive process that produces it, but those are not the primary outcome measures suggested by the editors own objectives. A wealth of peripheral research questions come to mind for linguists or translation theorists given that large bilingual corpora are being generated by Archivos and other journals. Meanwhile, the present paper has offered one model for integrating translation and publication processes in a way that supports overall quality for scientific societies unwilling to sacrifice discourse in their native language as they seek readers abroad. Acknowledgments I. K. Patten and K. Shashok gave appreciated critical comments on an early draft. J. Herranz and R. M. Fernndez provided valuable advice on the arrangement of Figure 3. Journal editor P. Casan kindly consented to use of documents, provided the privacy of authors of unpublished material was protected. J. Alonso, for the publisher, kindly consented to analysis of internal correspondence. Translators agreed to analysis and use of their drafts and correspondence: A. King, C. Law, J. Lewis, B. Londres, G. Morley, A. Oltra, I. K. Patten, and I. Temkin. References [1] S. Ren, G. Zu, and H. Wang, Statistics hide impact of non-English journals, Nature, vol. 415, p. 2, 2002. [2] C. Iverson, A. Flanagin, P. B. Pontanarosa, R. M. Glass, P. Glitman, J. C. Lantz, H. S. Meyer, J. M. Smith, M. A. Winker, and R. K. Young, American Medical Association Manual of Style. A Guide for Authors and Editors, 9th ed. Philadelphia: Lippincott, Williams & Wilkins, 1998. [3] M. E. Kerans, Eliciting substantive revision of manuscripts for peer review through process-oriented conferences with Spanish scientists, in Trabajos en lingstica aplicada, C. Muoz, Ed. Barcelona: Universitat de Barcelona, 2001, pp 339-347.2342005 IEEE International Professional Communication Conference Proceedings[4] M. E. Kerans, Close to home: notes on the post-publication withdrawal of a Spanish research paper, Ibrica, vol. 4, pp. 39-54, 2002. [5] M. E. Kerans, Pre-publication translation: What's wrong with the process? in II Jornades Catalanes sobre Llenges per a Finalitats Especfiques. El llenguatge cientfic: edici, tradcci i implicacions pedaggiques, F. Luttikhuizen, Ed. Barcelona: Universitat de Barcelona, 1999, pp. 163-169. [6] K. Shashok and M. E. Kerans, Translating the unedited science manuscript: Who fixes what shortcomings? in Proceedings: First International Conference on Specialized Translation, J. C. Chabs, M. Cases and R. Gaser, Eds. Barcelona: Universitat Pompeu Fabra, 2000, pp. 101-104. [7] K. Shashok, Translators as editors, EuropeanScience Editing, vol. 20, pp. 14, 1999. [8] R. H. Fletcher and S. W. Fletcher, The effectiveness of editorial peer review, in PeerReview in Health Sciences, F. Godlee and T. Jefferson, Eds. London: BMJ Books, 1999, pp. 45-56.[9] N. Sommers, Revision strategies of student writers and experienced adult writers, in TheWriting Teacher's Sourcebook., G. Tate, Ed. New York: Oxford University Press, 1988, pp. 119-127. [10] J. Burrough-Boenisch, Shapers of published NNS research articles, Journal of Second Language Writing, vol. 12, pp. 223-243, 2003. [11] J. Flowerdew, Discourse community, legitimate peripheral participation, and the nonnative-English-speaking scholar, TESOLQuarterly, vol. 34, pp. 127-150, 2000. [12] M. Perpi-Tordera, Cuando Archivos de Bronconeumologa se hizo mayor, Archivos de Bronconeumologa, vol. 41, p. 1, 2005. [13] G. Gregoire, F. Derderian, J. Le Lorier, Selecting the language of the publications included in a meta-analysis: is there a Tower of Babel bias? J Clin Epidemiol, vol. 48, pp. 159-163, 1995. [14] M. Egger, T. Zellweger-Zahner, M. Schneider, C. Junker, C. Lengeler, and G. Antes, Language bias in randomised controlled trials published in English and German, Lancet, vol. 350, pp. 326-329, 1997. [15] P. Juni, F. Holenstein, J. Sterne, C. Bartlett, and M. Egger, Direction and impact of language bias in meta-analyses of controlled trials: Empirical study, Int J Epidemiol, vol. 31, pp. 115-123, 2002. [16] C. A. Junker, Adherence to published standards of reporting: A comparison of placebo-controlled trials published in English or German, JAMA, vol. 280, pp. 247-9, 1998. [17] M. J. St. John, Writing processes of Spanish scientists publishing in English, English for Specific Purposes, vol. 6, pp. 113-120, 1987. [18] C. P. Casanave, Transitions: The balancing act of bilingual academics, Journal of Second Language Writing, vol. 7, pp. 175-204, 1998. [19] J. R. Benfield and K. M. Howard, The language of science, European Journal of Cardio-thoracic Surgery, vol. 18, pp. 642-648, 2000. [20] K. Shashok, Author's editors: Facilitators of science information transfer, Learned Publishing,vol. 14, pp. 113-21, 2001. [21] M. Lagnado, Professional writing assistance: effects on biomedical publishing, Learned Publishing, vol. 16, pp. January, 2003. [22] R. Coats, B. Sturgeon, J Bohannan, and E. Pasini, Language and publication in Cardiovascular Research articles, Cardiovascular Research, vol. 53, pp. 279-285, 2002. [23] F. Salager-Meyer, The hidden dimension and underlying connections of LSP teaching: Who's pulling the strings? in Linguistic Studies in Academic and Professional English, I. P. Fortanet, Juan Carlos; Posteguillo, Santiago, Ed. Castell de la Plana: Universitat Jaume I, 2004, pp. 9-31. [24] S. Montgomery, Of towers, walls, and fields: Perspectives on language in science, Science, vol. 303, pp. 1333, 2004. [25] R. Aleixandre Benavent, J. C. Valderrama Zurian, M. Castellano Gomez, R. Simo Melendez, and C. Navarro Molina, [Archivos de Bronconeumologia: Among the 3 Spanish medical 2352005 IEEE International Professional Communication Conference Proceedingsjournals with the highest national impact factors,] Arch Bronconeumol, vol. 40, pp. 563-569, 2004. [26] A. Marusic, Editing biomedical journals in Croatia, European Science Editing, vol. 30, pp. 10-11, 2004. [27] H. Maisonneuve, Editing health care journals in France, European Science Editing, vol. 29, pp. 109-111, 2003. [28] J. Burrough-Boenisch, Culture and conventions: writing and reading Dutch scientific English, [Thesis online]. Utrecht: LOT, 2002. Available:http://www.lotpublications.nl/publish/issues/Burrough/[29] K. Knorr-Cetina, Epistemic cultures: How the sciences make knowledge. Cambridge, MA: Harvard University Press, 2000. [30] J. Swales, Genre Analysis: English in academic and research settings. Cambridge: Cambridge University Press, 1990. [31] C. Durban, Demanding clients state their case: comments on the client round table at La Rochelle, (SFT universit d't, July 2002). 1(1), pp. 3-7. Available: http://www.jostrans.org/issue01/issue01toc.htm [32] J. P. Piva and P. C. R. Garcia, [Eyes on the present and looking into the future,] Jornal de Pediatria, vol. 77, pp. 1-2, 2001. About the Author Mary Ellen Kerans is a freelance authors editor and translator in Barcelona. She also coordinates a program of English for the health sciences at the Universitat Internacional de Catalunya. Her current projects include forming the journal translation team described in this paper and developing an online writing course focusing on biomedical research articles (funded by the Fundacin Espaola para la Tartamudez). 236