Translation
Monday, March 22, 2010
Machine Translation (New York Times)
I, Translator, a thoughtful, and reassuring look at Google Translation, appeared in today’s (Sunday, March 21, 2010) New York Times Opinion section.
Sunday, February 22, 2009
Mark Stevenson on the Pros and Cons of In-house Translation
An article Mark Stevenson originally wrote for the Tsuyaku-Honyaku Journal titled The Pros and Cons of In-house Translation was recently posted on the JAT site. As someone who has done a bit of both in-house and freelance translation, I found myself nodding in agreement as I read his description of the relative benefits and drawbacks of each approach. The article is worth reading if you are currently working in either of those capacities and are interested in how people on the other side of the fence earn their living.
Sunday, December 14, 2008
Review: Improving Translation Quality
A colleague at my office recently lent me her copy of Kevin Morrissey’s Improving Translation Quality (the Japanese title is ?????????????—??????????), a very useful guide to measuring and improving the quality of technical translations that was produced with help from Hitachi Technical Communications Inc.
There is, of course, a great deal of technical information produced in Japan that needs to be translated into English for consumption in markets overseas, but there is not always a clear understanding of what constitutes a good translation and what distinguishes it from a merely passable or even an unacceptable one. This book addresses that need by providing a general overview of the issues facing technical translation in a Japanese-to-English context, providing specific examples of common problems in the field, by introducing Hitachi’s Translation Evaluation System (TES), and by offering suggestions on how Japanese source documents can be prepared to facilitate the process of translating them into English.
The book is intended for J-to-E translators, but also for native English editors and translation managers. It also thoughtfully acknowledges the fact that, although organizations like SWET and JAT tend to emphasize that J-to-E translations are best left to native speakers of English, the reality in Japan is that often non-native speakers of the target language are responsible for producing translations in English; and so the book offers recommendations accordingly, including the common arrangement of having a native speaker of English edit the translation prior to publication.
The author emphasizes the importance of having a system for making judgments about the quality of translations, because without such a system of making distinctions based on quality, cost alone tends to become the deciding factor when purchasing translations, which can lead to undesirable results. For this purpose, the author introduces TES, a system that breaks down common errors in J-to-E technical translations into several categories that are represented by various error codes. Incorrect or missing articles (a, an, and the), for example, are TES error code 01, and problems with subject/verb agreement are TES error code 02, and so on for a total of 27 types of errors in nine categories (grammar, spelling, unnatural English, meaning, clarity, poor verb usage, terminology, inconsistency, and conformity with stylistic requirements).
A native speaker of English would use these codes as a rubric for arriving at a quality score based on the number of errors per 1000-word sample, which would then be used to assign a grade to the translation. The author recommends grades ranging from “A” for a very good translation to “D” for an unacceptable one. Under this system, a translation with ten or more errors per 1000 words is deemed to be of insufficient quality and must be fixed. For translations of less than 1000 words, the author also recommends a simplified “Pass/Fix” grading system, in which case only translations with no errors get a passing grade. (This system is somewhat different from the one I have used in my own work, which seeks to evaluate errors by how critical they are, but I have no doubt that this approach is very effective at providing a basis for making comparisons about the quality of the work produced by different translators.)
The book also offers standard technical writing guidelines for the Japanese authors of technical documents, who quite possibly have even more influence over the quality of the final translation than the translator does (after all, even a fantastic translator cannot be expected to produce a good translation from a document that was not very well organized in the original Japanese). This is the part of the book that filled me with a desire to put a copy in the hands of all the “clients” in our office, the people who prepare the original documents for translation. Fortunately, the entire book is written in both English and Japanese—including the recommendation from the chairman of the Japan Technical Communicators Association in the front and the indexes in the back—which makes it easy to share even with people who are not translators themselves.
The translation of technical documents from Japanese to English is certainly a fairly specific niche in the overall translation market, but within that niche, this book addresses many of the important considerations in producing high-quality translations. For this reason alone, this book should be regarded as required reading by anyone getting started in the field or anyone who needs to be able to evaluate the quality of J-to-E technical translations.
Saturday, December 13, 2008
You can’t polish rubbish
Werner Patels has an excellent post titled “Bad translations cannot be salvaged” over at his weblog, Translation - Language - Culture, in which he laments getting requests to edit translations with an excessively high “rubbish factor,” a request I imagine most SWET members have had to deal with at some point.
From a translator’s point of view, it is pretty widely acknowledged that you cannot do much to improve a poor translation. There’s really no point is trying to polish up something that is clearly rubbish to begin with; you really need to start over and redo the whole translation from the beginning in cases like that, but it can often be difficult to persuade clients that a retranslation is what they really need, rather than just a bit of editing, or worse, a “native check” (an all-too-common request in the translation business in Japan). What strategies have SWET members developed for (A) avoiding or declining requests for editing work when a retranslation is clearly required or (B) persuading clients to bite the bullet and pay for a retranslation?
Wednesday, October 15, 2008
Translation Blog
BookTrust: Translated Fiction is offering a new blog in which Daniel Hahn, translator of three novels by the Angolan writer José Eduardo Agualusa (one won the 2007 Independent Foreign Fiction Prize), will blog the process of his translation of a fourth Agualusa novel beginning September 29, taking “us, week by week, through his translation process.” He promises something different from “a day in the life of”; he promises the whole translation process. He will be “working through a novel from my own first launching into a first draft, right up to publication. It’s not a blog about the life of a translator – musings about translation generally, reports of events I’ve attended or readings I’ve given, people I’ve met at launch parties, books I’ve read – but intimately about a single piece of translation work, which I hope will bring you closer to the experience, to the pleasures it brings and the questions it raises.”
The first entry, September 29, was “What this blog is for”; the second, October 7, was “Anticipating general problems”; the third, October 10, wondered “To footnote or not to footnote.”
Perhaps the comments in the SWET Weblog and discussion in the Forums (General Discussion, Thread: Translation Process in Real Life) will reflect facets of daily J-E translation work and share translation details as we read about Hahn’s approaches and techniques and questions.
Saturday, August 23, 2008
Google to offer services for translators?
I first noticed this article at CNET News before the recent Obon holidays, but details seem to have leaked out that would suggest Google is preparing a document translation service that would put people who need translation services in touch with translators who can provide them. Although the service itself is not available yet, it seems to include tools for translating and reviewing documents, managing glossaries, and viewing one’s previous translations, all of which are features typically found in computer-assisted translation (CAT) tools.
Google’s motive for providing these services is not yet clear, but it’s not difficult to imagine that Google would be interested in using the translations stored in this system to improve their automated translation capabilities. Whatever the motivation, if Google’s translation center does come to fruition, it will be interesting to see what effect it will have on sites like proz.com, on translation agencies and other middlemen in the translation business, and quite possibly on the makers of CAT tools like Trados.
Thanks to Philippa Hammond at Blogging Translator for the heads up and to Philipp Lenssen at Google Blogoscoped for more details.
General • News • Noteworthy Finds • Translation • (0) Comments • Permalink
Tuesday, June 03, 2008
Murakami Haruki on translation, writing
Last month the Mainichi Shimbun published a series of articles based on a lengthy interview with Murakami Haruki. In addition to being a famed writer, the man is an accomplished literary translator, with Japanese versions of works from Truman Capote, Raymond Chandler, F. Scott Fitzgerald, and J. D. Salinger to his name. (See the end of this post for links to all five articles, in both Japanese and English.)
One thing I found interesting was his statement in the first article that old translations have an expiry date. The changing nature of the Japanese language means that classic foreign books benefit from retranslation every half-century or so:
Murakami says the “use-by” date on translations means they have a “50-year limit” of effectiveness because of changing writing styles in Japanese. Murakami says that the flood of works translated into Japanese during a literature boom here in the 1960s are now reaching their “use-by” dates.
On my shelves I have a number of translations of ???? [Hyakunin Isshu], the 100 poems compiled by Fujiwara no Teika around eight centuries ago. Murakami’s words ring true when you compare a relatively modern translation from Steven Carter or Joshua Mostow with H. H. Honda’s 1957 One Hundred Poems from One Hundred Poets, which turns the 31 syllables of each Japanese poem into not-so-elegant rhyming quatrains in English. (Of course it’s possible to go back to Clay MacCauley’s 1917 Hyakunin-Isshu (Single Songs of a Hundred Poets), available online, and see that traveling still farther back in time doesn’t necessarily mean a journey deeper into the world of English poetic devices inappropriately applied to Japanese verse.)
Genji Monogatari is one work that’s been translated numerous times over the years. Maybe there are other Japanese books out there whose English translations are nearing their read-by dates, and translators should start preparing updated versions of them for a twenty-first-century readership. Well, at least translators who have the financial luxury of wrestling with words and phrases for hours on end, rather than maximizing their output per unit of time. As Murakami points out in the second article:
“Translating from English into Japanese is like solving a math problem,” he says. “Just like there are some math problems that people can spend an entire day trying to work out, it’s possible to spend a long time thinking about why particular words are used the way they are. Some people are suited to this and others aren’t. But I like that kind of thing.”
Murakami’s interview, written up in five parts:
- ????????????????????????????????????????????????? (translated as Haruki Murakami opens up about translating America’s literary giants)
- ???????????????????????????????????????????? (translated as Murakami says American contemporary classics ‘really significant’ for his writing)
- ??????????????????????????? (translated as Murakami’s next epic poised to become his biggest ever)
- ???????????????? ?????????????????????????????????? (translated as ‘Breakfast at Tiffany’s’ helped inspire Murakami to write for his daily bread)
- ????????????????????????????????????????? (translated as Murakami aims to find harmonious balance between globalism and regionalism)
Translation • Writing • Fiction • (1) Comments • Permalink
Sunday, June 01, 2008
Meet Felix, a clever new CAT tool made by GITS
If you use computer-assisted translation (CAT) tools in your work—particularly if you translate between English and Japanese—you owe it to yourself to take a look at Felix, a new application released last month by Okinawa-based Ginstrom IT Solutions (GITS). Although Felix is technically not new in the sense that it is the reincarnation of TransAssist, the developer’s vision for Felix and ambitious development roadmap leave me feeling comfortable thinking of Felix as a new development in the CAT market, even if the application itself has already been under continuous development for a number of years.
Because the friendly folks at GITS have already made a demo showing how Felix works in Microsoft Word, and because there is a trial version that you can download and try on your own for translations of up to 500 units, I will not attempt to show how it works here, but will instead focus on some of the things that I think make Felix unique.
Overwriting the Source Text
As far as I know, Felix is the only CAT tool that works by overwriting the source text as you translate. Although this approach is perhaps not as cautious as the ones used by tools like OmegaT and Deja Vu X (both of which import the source text into an external editor and then export your finished translation as a separate file, leaving the original document unchanged) or Wordfast and SDL Trados (both of which create what are commonly called “bilingual” or “uncleaned” files that include both the source and target text until the translation is finished and the document is “cleaned up”), it does have the advantage of being both very easy to work with and very fast. Once you have finished working through all of the source material, you’re done. There is no export step or cleanup process at the end of the job—which, as any translator who has used other tools can tell you, can be the most panic-inducing part of any CAT-based process if it does not work as expected. Felix cleverly sidesteps potential problems in this area by eliminating these steps entirely.
The only real caveat here is that you will want to work on a copy of the document you are translating, rather than the original document, just to make sure that you have a backup of the original text in the unlikely event that something goes wrong while you are working. This is a common sense rule that should be observed when using any CAT tool, however, so it does not stand out as something that you would need to pay special attention to when using Felix.
Unobtrusive Control Over Formatting
Many clients may express a strong preference for translations to reflect the same formatting as the original document, but actually formatting is every bit as translatable as the rest of the document. In technical documentation, for example, it is common for the commands in the menus of software applications to be enclosed in brackets in Japanese, but those same commands are typically written without the brackets in bold text when translated into English. Formatting changes like this are important to ensure that the translation reads naturally in the target language. Some CAT tools make the mistake of assuming that formatting present in the original must also be present in the translation (this is particularly true of the ones that import the source text into an external editor), but Felix avoids making any assumptions in this area and instead allows the translator to control the way the text is formatted as an integral part of its translation workflow.
Works in the Application in Question
SDL Trados and Wordfast both work directly in Microsoft Word, but when it comes to Excel or PowerPoint documents, the translator is required to either switch to a different application (Trados) or attempt to bring text from those types of documents into Word for translation and then export the finished product back to the original files (Wordfast). Neither of these approaches are as convenient or as intuitive as Felix’s approach of opening the file in question and doing the translation right in the application that created it.
In the case of HTML files, Felix also requires the use of a different application (the WYSIWYG TagAssist editor), but even in this case the translator is still working in an environment that makes it possible to edit the content of the document while translating, which is an important thing to be able to do in many cases.
Unlimited TMs/glossaries
The ability to draw from an unlimited number of translation memories and glossaries is not unique to Felix (OmegaT and Deja Vu X offer this as well), but as far as I know Felix is the only CAT tool that works directly in the application in question that offers this capability. Wordfast, although it comes close, is limited to three glossaries, one active TM, one background TM, and one “very large translation memory,” which is a kind of Web-based shared repository for translations that Wordfast users are able to draw from. Felix’s approach to managing TMs and glossaries is flexible, powerful, and best of all uncomplicated.
Felix is Scriptable
If you know what you are doing, most CAT tools can be scripted to some degree, but aside from OmegaT (the source code for which is available to anyone who is interested), Felix is probably the most amenable to user-provided extensions of its native capabilities. As far as I know, Felix is the only CAT tool that provides property and method specifications along with examples in VBA/Visual Basic and C++ that show you how to build on Felix’s functionality. Although many translators might never need to do this, it is nice that Felix is designed to help you do so if you choose to (it is also a refreshing acknowledgement that the translator’s skills may extend beyond simply the linguistic ones required for translation work).
Complementary Tools from GITS
In addition to Felix and TagAssist, GITS has been busy producing a collection of complementary tools, including Count Anything, a word-counting utility that supports a variety of file types; Analyze Assist, a program that analyzes the documents you want to translate and compares their contents with the translations already stored in your memory files, making it possible to estimate how long the translation will take; and most recently Jamming2Felix, a utility for converting glossaries from the popular Jamming format into the Felix format. All of these things, when used together, make it easy to get started using Felix right away.
Conclusion
For a single translator working primarily on translations of Microsoft Office documents or HTML files in a Windows environment, Felix is an attractively priced, compellingly robust offering. The 1.0 version does not yet support sharing TMs and glossaries over a network, so it is not yet suitable for use by translation teams of the sort you might find in an agency or corporate environment, but the development roadmap suggests that this kind of capability is coming soon. Once this is in place, I would recommend Felix for anyone who needs a translation tool that is simple yet powerful, easy to learn and easy to use, and lets you work the way you want to, without getting in the way.
Tools of the Trade • Translation • (1) Comments • Permalink
Monday, May 19, 2008
Facebook gets translated, saves a ton of money
Booming social networking site Facebook has managed to get its entire interface ready to go in Japanese at what must have been a fraction of the cost of using professional translation services for the entire process. This article (in Japanese) tells how they added Japanese to the list of nine languages available for the interface:
????????Facebook??????????????????Facebook???????????????Facebook?????????????????????????430??3??????1500?????????????????????????????????????????
To sum up: Facebook added a “translation application” to its services. Users could sign up to use this app and start providing their glosses for the various bits of the interface—the “click here for XX” and “last updated on YY” and other little chunks of text that guide you around the website. A group of 430 users made their way through the 1,500 or so phrases, translating them into Japanese, over the course of three weeks, voting on the best options when there were more than one. At the end of this “democratic process” the site operators had a pro translator (translators?) look over everything before signing off on the new language.
With Google trying to get users to improve its automated translation tool and Facebook throwing the gates open to the masses in this way, it looks like the “getting people to do the grunt work for free” approach is becoming a popular one, at least among large Internet operations with a broad international user base. Just call what you do a community effort and away you go.
(If you’re on the site, feel free to add me, assuming I got my URL right there.)
Thursday, May 08, 2008
Corinne McKay on translating in OmegaT
In my day-to-day translation work, I get a lot of mileage out of a number of different computer-assisted translation (CAT) tools, one of which is the very impressive OmegaT, an open source translation memory tool that has an innovative approach to the way translation tasks are handled. One of the most impressive things about OmegaT, however, is that although it has many robust features, it is developed by a team of volunteers who make it available to anyone who would like to use it absolutely free of charge.
Because the developers do not ask for any sort of compensation for their efforts, I am always pleased when I see someone write a nice review of OmegaT, like the one Corinne McKay recently wrote in her blog, Thoughts On Translation. I hope other translators who have tried OmegaT and found it useful will do the same, because the word-of-mouth marketing helps call well deserved attention to this useful application.
Tools of the Trade • Translation • (2) Comments • Permalink