Tool Box Newsletter Logo

 A computer newsletter for translation professionals


Issue 13-8-225
(the two hundred twenty-fifth edition)  
Contents
1. Studio 2014
2. The Long Fight to Stay Relevant
3. Continental Shifts? (Premium Edition)
4. Chicago Ain't No Sissy Town
5. New Password for the Tool Box Newsletter Archive
The Last Word on the Tool Box
The Face of Translators and Interpreters

The wooden ship pitched and heaved in the storm, and Jo�o automatically lifted his cooking pot to protect it from the resulting slosh of stinking excrement around his feet. Here below deck, surrounded by the groans and smells of sick and dying men, even the daily water ration stank so badly he had to plug his nose to drink it. Bands of rats prowled the ship during the night, gnawing on the feet and faces of the sleeping men and boys. As a Jesuit novice and one of the youngest on board, Jo�o cooked for the sick and warded off rodents when he could, but he wondered how many of the one thousand men who had started the voyage in Portugal would still be alive at the end. He had heard tales of ships landing with only 200 sick souls still alive. And India could still be months away.

The year was 1574, and conditions like these would have been common for Jo�o Rodrigues, a 14-year-old orphan, on his long and arduous way to Japan to become a Jesuit missionary.

In the sixteenth century, the Portuguese Jesuits had worked out a system to locate and recruit the most promising young boys from Portugal's well-stocked orphanages. These boys like Jo�o were educated in Jesuit institutions and then sent on to one of their many mission outposts around the world.

Of all the mission societies, it was the Jesuits who most eagerly looked for ways to achieve inculturation, or culturally appropriate communication of the Christian message. A central component of this goal was the study of language. Once Jo�o had landed in Japan and regained his land legs, he channeled that same determination that had kept him alive on his grueling three-year odyssey to Japan into learning the Japanese language.

As the Jesuits turned their focus onto Japanese culture in the late sixteenth century, it quickly became apparent that there was no one in the mission who could match Jo�o's linguistic abilities. His chief task soon became that of interpreter, working in particular as an intermediary between the Japanese court and Western merchants, sailors, and religious and governmental officials.

In fact, his skill set was so extraordinary that before long he had assumed the position of imperial confidant and interpreter to the rulers of Japan, interpreting language as well as all things cultural. Fifteen years after arriving in Japan, Jo�o attended at the deathbed of the ruler Toyotomi Hideyoshi, whom he had served as interpreter and adviser. He then remarkably maintained his highly influential status with Hideyoshi's successor and greatest rival, Tokugawa Ieyasu, the founder of the last great Japanese shogunate.

Despite Jo�o's influence, life for missionaries in Japan was perilous during these years. Jo�o's Jesuit brothers were alternately persecuted, forced into hiding, and killed in mass executions, and Jo�o himself walked a tightrope back and forth between the halls of power and the cells of the persecuted. Though he was not able to prevent the infamous crucifixion of 26 European missionaries and Japanese followers in Nagasaki, he did receive permission to stay with them and make sure they were killed quickly. Most remarkably, he himself remained untouched, having become so indispensable to the Japanese leadership that he was able to maintain his influential position throughout all the upheaval and bloodshed. Eventually he even became the personal commercial agent of the Shogun, serving as the gatekeeper for all commercial activities between Japan and European countries.

However, when the relationship between the Japanese and Portuguese governments finally deteriorated in 1610 to a point where the Jesuit mission was given the choice of expelling Rodrigues or all its missionaries, Rodrigues left Japan, never to return. The remaining 23 years of his life he spent in a kind of exile in China. Though he tried to resume his position of authority in matters of language and culture, he could not recapture the excellence he had displayed in Japan.

 

What a rich life story, multi-faceted, mysterious, and melancholy all at the same time (something that novelist James Clavell recognized when he used Jo�o Rodrigues' biography to craft the character "Martin Alvito" in Shōgun).

An obvious and relevant aspect for us is the unusual role he played as an interpreter -- an orphan from halfway across the world who became a confidante to kings and a policy maker for potentates. Unlike the impulse behind this year's Face of Interpreters and Translators Photo Contest,  there is no known portrait of Rodrigues: he didn't need one, for unlike many of us, he was certainly not invisible during his time.

But the part of the story that touches me the most is the last part, beautifully and painfully described in Michael Cooper's biography (Rodrigues the Interpreter, Weatherhill, 1974). After his forced exit from Japan, Rodrigues assumed that he could replicate his tremendous success as a master of language and culture in China -- and pitifully failed.

I wonder whether it's fear of this same failure that is somehow ingrained in us when we look at changes in our workplaces and feel inadequate to grasp the new technologies and processes. It's essential to remember, though, that unlike Rodrigues, we aren't being forced to give up our excellence in the field of our first calling, nor do we have to become masters of technology -- we just have to be able to employ it adequately. 

1. Studio 2014

I had a chance to speak with some folks from SDL last week about their upcoming end-of-September launch of SDL Trados Studio 2014 and MultiTerm 2014. In the very back of my mind I had hoped to see some kind of game changer in the new incarnation of the software (I'll come back to that later), and I was disappointed with its absence. Still, overall the new version seems like something that most Trados users would eventually want to upgrade to. It's a bit like when Microsoft upgraded Office 2003 to Office 2007.

This comparison to the Office versions seems particularly appropriate because the most immediately noticeable change is the demise of the menus and the rise of the ribbon. There has already been a bit of excitement in the blogosphere about the ribbon. I like that particular blog's author (hi, Riccardo!), but I don't share his passionate dislike for this feature. It's been a very long time since I've actually even given any thought to the ribbons in the different post-Office 2003 versions that I've been working on for years. Well, that's not completely true: I do actually appreciate them when I'm clicking on an image or a table and the appropriate ribbons are displayed automatically. Have they slowed me down? Not really. Have they sped me up? Not really, either.

The truth is that power users like you and me who work in the translation environment tool(s) of our choice are not likely to be heavy menu users, anyway. At least I'm not. I use keyboard shortcuts wherever possible -- and if those are not available (or known to my fingers), I typically find them on the ribbons (provided that they are organized in a logical manner, and that remains to be seen with Trados 2014).

It was interesting to hear what SDL's Daniel Brockmann had to say about why they introduced the ribbons: to a) create a more fluid feel in the software (that's the part that Riccardo and others will disagree with) and b) to expose little-used but powerful options that might otherwise be hidden in the menus. The second point especially makes a lot of sense to me.

What else is new? The most prominent new feature is that the Trados 2007 application WinAlign has finally been replaced with a Studio-native alignment tool. There are two features that set the new alignment tool apart: First, it allows you to align a text and retain in-context matching information so that you can get context matches (aka 101% matches, ICE matches or guaranteed matches) when it comes to using the translation memory for translation (I really liked that). Second, there is a little more nebulous option called "alignment quality value." This is a setting that allows the user to adjust the confidence level of the match. For example, if you have set a high quality level and there is a huge difference in the number of words in the source and target segments or there are numbers in the source but not in the target, the translation units would be rejected. This is all well and good, but it would be very helpful if those kinds of settings could actually individually be really fine-tuned by the user, � la AlignFactory where you have six different categories with three settings each and can also create a report on the rejected sentences.

(I'm going to go out on a limb and predict that one of the first few service packs for the new version will have something like that.)

Once the alignment is done, the aligned segments are not shown in an alignment editor where you can modify the alignments but right in the translation memory view. In fact, there might not even be an alignment editor in the core tool; that would be available only on the SDL OpenExchange. Such is the faith in the new alignment. We shall see whether it's warranted or not.

Other improvements to both MultiTerm and Trados Studio include quite a few of the faster, easier, and smarter variety -- since there is no final product yet, I did not have a chance to test it so I can't report on it. The "private beta" has been running for five months, according to Daniel, and he is quite confident that "stable" will also be a characteristic of the expected features.

A couple of other features that I like are the possibility to view any combination of files within one project in a single view. This is something that other tools have been offering for some time, but it will doubtlessly delight existing Trados user.

Are there newly supported formats? The hearts of competitors like Wordfast Pro, memoQ, and Star Transit will go pitter-patter when they find out that some of their formats are supported directly (either directly through the core product or through an app in OpenExchange). I guess if a "small" tool supports the format of the market leader (as has been the case for most Trados formats by "the other tools" for many years), it's seen as an attempt to eat into the marketshare. If the market leader supports the formats of other tools, it's an acknowledgment that the market might not be as monolithic as is sometimes assumed.

Oh, and to come back to the groundbreaking features that I had hoped for. Awhile back I wrote this column in which I outlined possibilities of different data resources "talking" to each other -- which in my opinion truly would present a breakthrough. I sent a copy to all the different tool vendors, inluding SDL, and Daniel assured me back then that this was exactly what they were working on. Well, he still assures me of that, but says it just wasn't quite ready. . . . Oh, well. Maybe in Studio 2016? 

ADVERTISEMENT

Why do people work with memoQ

'The thing I really love of memoQ is that it always offers you (at least) 2 different ways to face the everyday issues, so that you experience "safety" - because you know you are relying on a tool that will not let you down - and "fun" - because it seems a video game with so many options and levels to discover!'

(Renato Renno - The Foreign Friend srl)

Download the full-featured trial version, register for the free online training, and attend our educational webinars. Learn more at www.kilgray.com.

2. The Long Fight to Stay Relevant

No, I'm not talking about Anthony Weiner, the mayoral candidate in New York (if you don't know what this refers to - please do not look it up!). Instead, I'm referring to QuarkXPress, the desktop-publishing product that was caught napping for too long (they actually weren't sleeping -- they were busy thinking up evil plans to squeeze money out of their non-English-speaking users!). Once a serious contender showed up in the form of Adobe's InDesign, the market share of QuarkXPress crumbled and its makers finally were forced to reinvent themselves, and then reinvent themselves again and then again. You get the idea.

Still, there are some designers and some print shops that are still using Quark. If you work for one of these you might be asking yourselves what the translation support will be like for the new and upcoming version 10. Translation environment tools have long stopped supporting QuarkXPress directly (with the notable exception of Star Transit, which offers an optional plug-in that supports Quark up to version 9.2). Most TEnTs refer users instead to an additional tool called CopyFlow Gold for QuarkXPress, which allows you to export the translatable content into a tagged file that can be processed by most TEnTs. I checked with the company that sells CopyFlow, and a version of the tool for the new version of QuarkXPress will be available in late September.

There you have it. 

ADVERTISEMENT

Introducing SDL Trados Studio 2014

We are pleased to reveal the latest version of the world's #1 CAT tool is coming soon.

Featuring...

  • A smart new design with an intuitive and easy to use menu bar
  • Powerful alignment technology to leverage past translations in 'one-click'
  • Faster project preparation such as batch processing and project analysis performing up to 5 times faster
  • Accelerated file handing with new and improved file filters and faster opening, preparing and saving times

We're announcing more features and innovation over the coming weeks, so stay tuned!

 www.translationzone.com/studio2014 | #Studio2014

3. Continental Shifts? (Premium Edition)

Pangeanic is a mid-sized Spanish language service provider that started relatively early (2008) to use machine translation for its own production processes. Like most other statistical machine translation systems, it was based on the open-source Moses system. (By the way, have you ever wondered whether the makers of Moses realized the irony that their product's namesake had a speech impediment?) Pangeanic kept on refining the product with a more user-friendly interface and other options and started to market it in 2010 to other companies (under the product name PangeaMT).

About a third of their customers are other language service providers, and the other two-thirds are corporate customers. That's something Pangeanic would like to change -- by specifically focusing on a new segment of the market: freelancers. Freelancers are clearly the group in the translation industry that has been least targeted by MT vendors (and with some freelancers, that's quite all right!).

There are rules-based machine translation systems like Systran and PROMT (both of which actually have a statistical machine translation component as well, but only in the high-priced server editions that are essentially not available for freelancers to own); rather involved packages like Do Moses Yourself by Precision Translation Tools; companies like KantanMT that offer statistical machine translation on top of already pretrained machine translation engines (but start at 500 euro per month); SDL's BeGlobal at a minimum cost of $1,500 per year (for three million translated words); and lastly Microsoft Translator Hub, which is free in exchange for donating your (or make that: your client's) data. (Premium subscribers can read about all these different tools in the indexed Tool Box newsletter archives.)

So, pretty slim pickings for a freelance translator. The solution that PangeanMT now offers to freelance translators is -- as far as I can tell -- strictly paid on a per-word basis (.002 euro cents per word) with no cost for training or retraining the cloud-based machine translation engines. Since the training of the machine translation engines in the available language combinations (EN<>ES, FR, IT, DE, DU, PT, NO, SV, and JA, plus Chinese starting in September and Arabic and Russian at the end of the year) is done on top of already-trained machine translation engines (based on TAUS and EU data), there is no minimum amount of data that needs to be added to achieve some quality improvement -- but clearly you won't achieve much with a couple hundred translation units. (For reference purposes: the makers of KantanMT recommend a minimum of one million words -- I wouldn't be surprised if that's about the same here.)

There are several tools that PangeanMT offers for cleaning the TM data before you use it to train the machine translation engine. This is an interactive process where you are prompted to look at potentially offending segments and either delete them or use them to improve your glossaries (which are also a component that is used in the training process).

Manuel Herranz, whom I talked to about this, was refreshingly realistic about some of the limitations of the system. For instance, he was very cognizant of the fact that not all language combinations always produce the same kind of quality. German lags behind the Scandinavian and Romance languages, Russian is even further behind, and Japanese is a whole different matter altogether (here is an interesting paper on the work that Pangeanic did with Toshiba for EN<>JA). He also conceded that the output quality of statistical machine translation has almost reached a ceiling -- if it were not for the higher success through better domain-specific data.

(I agree that better training data will make the translation output better, but I also think that there is a ceiling that will stop the improvements. From a translator's perspective, I think that an even higher ceiling could be established if we had the TM, the termbase, and the MT systems "talk" to each other as I mentioned in the column linked to in the Trados article.)

So it's not surprising that PangeaMT users are encouraged to continue retraining their data with ever better materials to achieve better output. Right now this has to happen relatively manually: While a number of monolingual file formats are supported (TXT, DOCX, ODT, HTML) or are about to be supported (XLSX, PPTX, OpenOffice ODS/ODP, XML, CSV, InDesign IDML, FrameMaker MIF), translators typically pretranslate the supported bilingual files (TMX, TTX, XLIFF) in their translation environment tool, upload the files to the PangeaMT cloud to have the rest translated by MT, download the files, and post-edit them in their TEnT. Once the file is post-edited it is again uploaded as new training data for the MT.

Coming in September for several tools (presently Wordbee, MemSouce, and XTM) should be a more integrated process where the manual processes described in the last paragraph essentially would all be done tool-internally. There also will be an app in the SDL OpenExchange that will integrate PangeaMT into Trados Studio with a less thorough but still largely automated interaction with the MT cloud.

When I talked to Manuel about the system he insisted that it's a hybrid system (a combination of rules-based and statistical MT), and strictly speaking he's right because there are some rules about syntax that can be used (as described in the PangeaMT/Toshiba paper linked above). But since those seem the pretty much the only rules, I'm not sure I would call this a hybrid system.

But then that might simply be a matter of semantics. And isn't that it's about anyway!?

4. Chicago Ain't No Sissy Town

CHICATA, the Chicago Area Translators and Interpreters Association where I had the honor to speak a couple of times in the past, has invited Peter Schmitt of the University of Leipzig to speak on September 21, 2013, about "New Trends and Risks at the Translator's Workbench." He'll cover (and rip) things like free dictionaries on the Internet, suggest new and better ways to provide professional dictionaries (he has authored a bunch), warn of some risks of "translation memory systems" (I think he means translation environment tools), and sing the praises of machine translation. Well, kind of. In short, I would really love to be there for this presentation! You can find all the details on CHICATA's website.

On the following day, September 22, Peter will be elsewhere in Chicago to talk about the precarious position of freelance translators and the lack of correlation between the price and the quality of a translation as seen in the recent "scandalous results of a translation test among LSPs in Germany." You can find information about this event in the same flyer on CHICATA's website.

5. New Password for the Tool Kit Archive

As a subscriber to the Premium version of this newsletter you have access to an archive of Premium newsletters going back to 2007.

You can access the archive right here. This month the user name is toolbox and the password is lupus.

New user names and passwords will be announced in future newsletters.

The Last Word on the Tool Box Newsletter

If you would like to promote this newsletter by placing a link on your website, I will in turn mention your website in a future edition of the Tool Box newsletter. Just paste the code you find here into the HTML code of your webpage, and the little icon that is displayed on that page with a link to my website will be displayed.

If you are subscribed to this newsletter with more than one email address, it would be great if you could unsubscribe redundant addresses through the links Constant Contact offers below.

Should you be  interested in reprinting one of the articles in this newsletter for promotional purposes, please contact me for information about pricing.

� 2013 International Writers' Group