Nuclear Technology—

a Translation Testing Ground

t first glance, the commercial nuclear industry may not seem to hold much appeal for today's environmentally-sensitive technical translators. In years to come, however, it will generate gargantuan volumes of work in the United States, China, France and Finland, where new and safer nuclear power plants are being planned, have reached the bidding stage, or are already under construction.

The complete nuclear energy cycle, which extends from mining, uranium enrichment and fuel production (front end) to spent fuel reprocessing and site cleanup (back end), also covers the multiple engineering specialties required to build, operate, maintain and decommission electricity-generating and fabrication facilities worldwide. Nuclear R&D provides further translation opportunities, as do waste management studies, quality assurance manuals and the safety analyses used by regulatory agencies. A single stage in an advanced reactor project can produce up to two million pages of potentially translatable equipment specifications and related documents.

The European nuclear giants currently draw on a translation supplier base that spans several continents and includes both large service providers and smaller, highly specialized operators or freelances living in or traveling regularly to their source language countries. In China and Russia, western firms outsource translations in situ or use branch office staff to localize their publications. In the former Eastern block, teams of mother-tongue technicians are often recruited to translate directly at nuclear jobsites.

With some of the toughest contracted-out texts on the market, nuclear engineering challenges the core competency of translators in a growing number of language combinations. These linguists help their customers to create wealth, through language-sensitive technology transfers, which are part of day-to-day business in the world's emerging economies.

English has long been the lingua franca of nuclear power specialists everywhere, and that includes India, Japan and South America. Use of native English-speaking translators should thus normally raise the level of international speak across the globe. But nuclear concepts are a minefield for unsuspecting amateurs. French I&C (instrumentation and control) documents written for reactor upgrades and third-generation NPPs offer striking examples—like "carte de flux", mistakenly rendered as "flowchart" in translations for a large power utility, or "protection channel", a treacherous faux ami for "voie de protection" when relating specifically to in-core flux detectors.

Whatever the target language, the risk factor inherent in such complexities is compounded by the increasing visibility of multinational groups that regularly post double-digit growth figures...and the stiff competition this has fostered. The nuclear industry is poised to become a global testing ground for 21st century technical translation.

In the late 1990s, translation purchasers for some nuclear OEMs and government-subsidized research institutes sought to improve bottom line figures by promoting use of MT software packages, then had to ramp up budgets to cover the need for human revision of less than perfect computer output. Other hope-inspiring panaceas, such as wholesale tendering for each and every project, or multiannual price freezes have since yielded mixed results, confirmed by negative feedback from final users. Sadly, this trend has tended to lower work giver quality expectations and encouraged decentralization of outsourcing responsibilities to company departments or individual employees, ill-qualified to verify foreign-language prose. While the root causes of such situations usually lie within the original purveyor's organization, enterprising translators with the right reflexes could make things change for the better over the next decade.

Good nuclear translation requires detailed understanding of the technical elements underlying a source text—e.g. exactly how a particular reactor system works! Just as importantly, it means willingness to reproduce the author's reasoning throughout the target version. A reviser recently hired by an architect-engineer to check translated design reports for a cutting-edge plant series found that few outside suppliers had researched the chronology of its development. They could not legibly translate even introductory paragraphs on its evolutionary features. Subsequent translation content likewise reflected the failure to find and follow this historical guiding thread. In such cases, the readability of a translator's style depends heavily on attention to (and where necessary, addition/deletion of) transitions. For the same reasons, document conclusions in the final language must "gel" with objectives appearing in the introduction.

Linguists seeking to make their mark in nuclear fields need up-to-date, authentic background material—whether from Internet, the trade press or engineering handbooks—as aids for credibly phrasing intricate technical descriptions. This is all the more true as the rare hardcopy or electronic dictionaries available neglect newly-coined nuclear services terms and the vocabulary preferences of competing plant vendors. Wordworkers with scientific diplomas should definitely also surf nuclear websites to avoid overuse, for instance, of rod dropping (German: Einfallen der Abschaltstäbe), which, in English, is commonly shortened to a much more readable rod drop.

But the nuclear translator's biggest asset is raw instinct—that magical ability to grasp the sense of an obscure or poorly written passage for which models are absent. In an industry with a patchwork of subdisciplines, this analytical bent of mind comes from asking the right questions, not just through a secretariat or agency interface, but in regular back and forth exchanges with document authors. Direct dialogue likewise puts to rest pesky misconceptions about CAT: no computer can sit down with an engineer to explore the real gist of a statement that sounds wrong when translated verbatim.

The nuclear industry has not escaped market pressure for translation memory discounts or the "revise-all" trend now being set in stone by certification standards. Yet truly effective revisers with nuclear know-how remain the exception; and paying non-specialists to revise at the expense of translator fees can be counterproductive. The fallacious idea that "analogies" and "repetitions" in the TM sense are somehow separate from the rest of a text has already dangerously limited the linguist's financial stake in certain nuclear documents. By negating his potential contribution to overall language unity, vast TM-generated reductions also discourage quality initiatives in an area where correct usage can be safety-critical.

Smart translators partially combat these tendencies by training themselves to perfect copy as they produce it—thus alleviating need for multiple edits. They address competition from computers as part of a customer awareness strategy that combines translation with extended value services like rewriting or conference synthesis. For some nuclear disciplines, there are no reliable English documents available on the Internet (the popularity of this universal language has led even respectable publishers to print books and articles in substandard versions of it). Linguist-readers can help halt the accumulation of strange syntax with inaccurate terminology quoted from such publications and keep defective English from spreading, through retranslation, into other languages.

Educating customers also implies teaching them the value of information. Some tips for getting the message across: Use excerpts from nuclear chemistry reports to illustrate how wide-ranging subject research is hindered by inadequate translation rates (especially now that high-tech literature is being locked up on paid subscription websites). Explain why failing to accompany outsourced work with like texts already vetted by editorial committees/overseas subsidiaries wastes time and money. Then point to the risks of translating names of key reactor control signals without suitable context and/or person-to-person dialogue...

Nuclear conglomerates, with their long-standing QA culture, need to reconsider technical translation within a less rigid, consistency-driven management approach—by far the best road to enhanced quality/price ratios. As managers in their own right, translators should challenge bottom line-based supplier approval procedures and strict reliance on quality standards that assess translation processes, not their results. Much more fundamental to effective purchasing are streamlined in-company document flows that eliminate duplication and partitioning (and help do away with different renderings of the same words/phrases seen in many a company showcase!).

There is no doubt that automation—of nuclear vocabulary mining, for example—can have positive financial impact. The human mind nevertheless maintains an edge over information technology. From 1983 to 1985, long before "corpus extraction" became a household word, one world-class NPP builder set up joint engineer/translator working groups to compile bilingual glossaries on lesser-known aspects of its activity. Twenty years later, this seminal collection, the result of networking by translation practitioners, is still a reference for various language combinations and online lexicons.

The ability to transpose such team efforts to a 21st century environment may make or break nuclear translation; and the impetus for change will probably come from freelances. Over the past two decades, salaried linguists have been largely unable to sell themselves as document adapters or lexicographers and thus to shape industry foreign-language policy. Their closed-circle attitudes toward customers and subcontractors have sometimes isolated them from mainstream decision-making (the current fad of recruiting native-tongue, monolingual copywriters is symptomatic of declining translator input) and have contributed to price stagnation at the end of the outsourcing chain.

It can be tricky to interface with nuclear engineers who overrate their own linguistic talents. But most end-users are quick to see the cost-effectiveness of jointly-crafted, target-language models, even for nuts-and-bolts specs or basic test procedures. Proactive linguists should carry dialogue one step further by alerting customers to advances in translation recycling software and promoting constant terminology control for cheaper, more consistent quality. Only loyal partnering and thorough specialization can, in the future, position nuclear translators to augment their added value, negotiate higher fees for both non-repetitive and TM-driven operations, and ultimately revive the practice of...getting raises!