Tool Box Logo

 A computer journal for translation professionals


Issue 17-2-271
(the two hundred seventy first edition)  

Contents

1. Let's Share!

2. Explorer CliffsNotes (Premium Edition)

3. Making Machine Translation Pay

4. This 'n' That

The Last Word on the Tool Box

Doing the Right Thing

Almost exactly a year ago I shared my pride in being a translator because of this:

"The New Zealand Society of Translators and Interpreters has launched the "Treaty Times Thirty" project in which more than 90 translators will translate the English and Māori versions of the Treaty of Waitangi into 30 languages. The final versions will be presented on International Translators' Day in September.

"Why is this such a remarkable gesture? Because the original (rush-job) translation of the treaty was so severely botched that it caused countless injustices and has so far cost the government of New Zealand 1 billion NZD in reparations. You can read one retelling of that story right here."

It took a little longer than expected, but the translation is now complete and was officially presented on February 17. The translation was done into New Zealand Sign Language, Afrikaans, Arabic, Bislama, Danish, Dutch, Esperanto, Farsi, Fijian, French, German, Greek, Gujarati, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Malay, Nepali, Portuguese, Romanian, Russian, Simplified Chinese, Spanish, Tagalog, Thai, Traditional Chinese, Turkish, and Vietnamese. You can find more information right here (where you can also download a PDF version of the resulting book), or in this news report.

And just in case you don't quite understand why this is such a powerful story, make sure to re-read the retelling of the original story of 1840 that I linked to above. It was likely the most costly and perhaps most unfairly carried out translation ever, and we should celebrate this meaningful attempt to finally get it right.

ADVERTISEMENT

TQAuditor translation quality assessment and monitoring system

Track every mistake your translators make and give detailed feedbacks to them.

Get the full picture of translation quality inside your team, try it now for free

 

1. Let's Share!

FIT is the International Federation of Translators, and chances are that you are already a member -- even though you might not be aware of it -- indirectly through your translators' or interpreters' association. FIT has just released a position paper on the "Future of Professional Translators" that is really very well-thought-out and most definitely worth your time to read. (Don't be confused about the many misleading dates that are mentioned around it -- it was published just a few days ago.)

One of its statements is this:

"The traditional image of the solitary translator is definitely changing. Specialisation, a team-oriented approach to the work and the willingness to constantly refine the knowledge of tools will be essential for a successful career in the translation industry."

And then it continues with this:

"In fact, translators should seek to influence the development and become co-creators of the tools they will be using in the coming decades."

I could not agree more with either statement.

We're already dealing with at least one way to find a voice in the development of our technology elsewhere in this journal, so let's talk about the team approach. I had a talk with Luis Lopez and Daniel Brockmann about SDL's new GroupShare 2017, SDL's now five-year-old product for LSPs and translation buyers to distribute access to centrally hosted translation projects that can then be worked on in Trados Studio. The current version of GroupShare is not available in its cloud edition (so it has to be hosted on your own server at this point), but that should change within a month.

Luis mentioned that only 15-20% of all LSPs currently use a collaborative solution -- which honestly surprised me and made me wonder whether those numbers are influenced by an SDL customer-centric view. On the other hand, SDL prides itself on having spent much of last year gathering feedback through personal interviews with project managers and conducting the SDL Translation Technology Insights Research that involved almost 3,000 translators, LSPs, and translation buyers (you can download the various resulting reports right here), so maybe it is an accurate perception.

Not surprisingly, much of what you can see in the new version of GroupShare is based on the results of that feedback. But those of us who have followed SDL in the last few years know that they are not about to rush things and are very methodical about introducing changes. Both impulses can be seen in GroupShare 2017.

The completely redesigned GroupShare website -- which is also fully functional on mobile devices (thanks, HTML5!) -- is now not only a place to view projects but also to manage and set them up, preferably those that are often repeated and based on existing templates. More complex projects still have to be set up within Trados Studio.

The actual setup of projects has been made a lot easier through a system of "Dynamic Resource Access," which will temporarily give participants (translators, editors, proofreaders, etc.) all the access they need to fulfill their respective tasks without having to assign those rights manually.

There is a dashboard now that gives a nice overview of ongoing projects -- in future versions it's likely to be more customizable by allowing users to remove and add widgets. As far as integration into and with other tools, an API is available to do just that. Naturally SDL services division would be happy to help you with that (as a paid service), or you can do it yourself or ask a third-party developer to do it for you. The latter is the case for management systems like Plunet and XTRF, at least for GroupShare's last version; so far there is no official word on whether the old connectors work for the new version.

Here's what I like -- and I know many of you will, as well: If a translator receives a virtual server-based "package" through GroupShare, the client can permit or allow all kinds of things concerning that package, but (presently) they cannot prevent you from attaching your own translation memories or termbases to the project. I consider this a very translator-centric approach.

If you are already using GroupShare, you might want to take a look at a rundown of all the new features right here.

Let's talk about who is using this (not on the receiving end but proactively on the managing end). They're primarily LSPs of various sizes but also translation buyers who, according to Luis, are relatively high on the Localization Maturity Level (a concept that was introduced by Common Sense Advisory), meaning clients who are willing and able to proactively engage themselves in the translation processes. Yet another maturity level (and amount of translatable data) is found among users of other SDL products such as SDL WorldServer or other comparable monster tools.

Translators are really not anywhere on the radar as active users of GroupShare. In a way that's a shame because that was the original intention, but there is apparently a product in the works through SDL's Language Cloud that will allow translators to form workgroups and share resources in real-time. For now, if you and your co-translators are all using Trados Studio, the work-around solutions that Paul Filkin described awhile ago might work just fine.

Of course, you can purchase and use GroupShare, but methinks that's rather unlikely with a starting price of about 2,000 euro/year for the cloud version and 5,000 euro/year for the on-premise version.

 

ADVERTISEMENT

Enhanced productivity in your work

memoQ will release at least 3 amazing, rich feature packs in 2017 to support your work.

Stay tuned and download the most recent release -- memoQ Adriatic -- from www.memoQ.com on 22 February!

Why memoQ Adriatic? Click here to learn more...

 

2. Explorer CliffsNotes (Premium Edition)

Here is an advance list of those of you who will not be interested in this section: non-Windows users, users of more advanced file management systems than Windows/File Explorer, and those who are more computer-savvy than I. That leaves exactly five of the 11,000+ who receive the Tool Box Journal.

So this is for you, you, you, you, and you!

 

. . . you can find the rest of this article in the Premium edition. If you'd like to read more, an annual subscription to the Premium edition costs just $25 at www.internationalwriters.com/toolkit. This will also give you access to the archives the of Tool Box Journal going back all the way to 2007.

 

ADVERTISEMENT

The best way to translate PDF

TransPDF converts your PDFs into high-quality XLIFF, ready for translation using your own tools. It then flows your translated XLIFF back into the original PDF to give you a finished PDF.

PDF translation is simpler, quicker and more accurate with TransPDF. 

Get your free TransPDF account now and try it for yourself. 

 

3. Making Machine Translation Pay

I've been in conversation with STAR's Nadira Hofmann and Ralph Benz for some time about their machine translation solution, and they shared some interesting insights with me that I would like to pass on to you.

When I say "STAR" I'm referring to the large Swiss language service provider that also develops and maintains a host of translation-related products, with Star Transit as their flagship product. Transit's fans have regularly complained over the years that I don't report on it regularly enough, and in a way they're probably right (my last lengthy review is right here). The fact is that I do use Star Transit quite a bit for a large client, and I am continuously amazed by its reliability and performance. So the reason I don't write about it more often is not that I don't think it's worthwhile to mention; instead, it has more to do with its product philosophy. In contrast to most other tools, Transit does not release many "versions" -- Transit 2.7, the much-loved and very stable version, was introduced in the late nineties, followed by an ill-fated and faulty successor (Transit 3) that was quickly replaced with Transit XV in 2001 and with Transit NXT in 2008. Instead of releasing new(ly hyped) versions all the time, Star releases fairly major "Service Packs" -- nine in total so far for the NXT version -- and these don't always lend themselves to writing new reviews about each (though this admittedly might be my own perception and fault).

Anyway . . . STAR also offers customized machine translation engines. Unlike SDL's proprietary engine, STAR's engine is based on the open-source statistical machine translation engine Moses. Another differentiator to SDL's BeGlobal/Language Cloud offering is that presently STAR's only target group for the MT offering is corporate clients. And this matches the offering's structure, which requires a relatively intense and ongoing involvement of STAR's MT development team. That team not only does an initial assessment of the customer's need but also trains the engine with the customer's data and then continues to retrain it at regular intervals until the customer-specific settings are fine-tuned and the task of retraining can be turned over to the customer (so it's not a dynamically trained engine like SDL's or Lilt's).

The data that is used is naturally the translation memory data (either coming from Transit "reference material" or through translation memory exchange TMX files by third-party tools), but there is a much stronger emphasis on high-quality terminology data and inflected terms than in competing products. The terminology data comes from STAR's TermStar product or exchange formats from other terminology management programs. For the inflected data, the Transit-internal morphology engine (available in 15 European languages) analyzes the TM data and matches the inflected forms that it finds with the existing terminology.

At the last tekom conference in Stuttgart, STAR and the Deutsche Bahn (DB) co-presented about the experience that DB had with introducing STAR MT. In that case, the training corpus consisted of 602,678 TM units, 30,140 terminology records from the termbase, and 38,754 records of inflected terminology from the TM (the TM and termbase originated in Star Transit/TermStar since the DB has been a long-standing technology customer of STAR). You can find the presentation that was used at the tekom right here (German) along with an English write-up right here.

What's interesting about the MT introduction at the DB -- aside from the numbers that, not surprisingly, indicate a successful implementation -- is the very careful and hype-free process with which this was achieved. It involved all stakeholders and, most interesting for many of us, this resulted in overwhelmingly positive feedback from freelance translators (note that STAR is not a service provider for DB).

One other likely reason for the successful implementation is that MT suggestions are used in Star Transit in various ways. One is a display alongside TM matches in the respective pane on the translator's screen. Another is the so-called "TM validated MT match" that I described in Edition 250 of the Tool Box Journal. Here's what I said back then:

"Traditionally we have looked at MT as something that should come into play if there is no perfect or fuzzy match within the TM. This makes sense because the TM is, of course, the gold standard, created as it is by us (or our team). What if, Star thought, we also displayed MT suggestions alongside fuzzy matches? They might be of as good or even better quality, especially if it's just a terminological difference that makes the TM match fuzzy. And what if we evaluated the MT suggestions (I always hate saying "MT match") on the basis of our fuzzy TM matches?

"Here's an example:

    • The source sentence is "Pressure increase too slow when filling reservoir"
    • The fuzzy TM match is "Druckanstieg zu schnell bei Füllung des Tanks" ["Pressure increase too rapid when filling reservoir"]
    • The MT suggestion is "Druckanstieg zu langsam bei Füllung des Tanks" ["Pressure increase too slow when filling reservoir"]

"The program is able to compare the fuzzy TM match (for which it "knows" that there is only one unknown term) with the MT suggestion to find out that there is no other difference between the two than that particular term. It then concludes that the MT suggestion in all likelihood is correct -- the worst it could be is to have one term incorrect -- and it becomes an "Advanced MT match.""

This is what I call a translator-friendly use of machine translation.

Aside from the specific implementation at DB and the feature set of STAR MT, I was also interested in the profitability of offering MT as a product. It has often been said that despite all the big talk about MT, the only ones who have been able to actually make money with it are the small handful of folks who have been able to sell their MT companies (if they were indeed able to make money after paying back their investors) and Google and Microsoft. And the latter two are profitable not so much through licensing fees as advertisements.

When I asked about this (I can be very dis(ch)armingly direct), I was told that, yes, offering machine translation as a technology vendor can be profitable if it's part of a larger offering. Of course, as mentioned above, STAR is heavily involved in (paid) training and retraining of the MT engine. And there are also other MT-related offerings that most of Star' MT clients use (and pay for), including STAR MT Translate, a browser-based solution that can be used by any authorized user for their everyday translation needs.

Conclusion: It's possible to make money with MT -- if it's part of a whole technology infrastructure. And that should be a cautionary tale for many who either are offering or are thinking about offering MT as a standalone technology and/or service.

 

ADVERTISEMENT

Leave the office 20 minutes earlier today!

Using MindReader for Outlook, you can write e-mails more quickly and more consistently.

Watch the short video for more information on MindReader for Outlook functionality and usage: https://www.youtube.com/watch?v=YAPLHSvVrBc.

 

4. This 'n' That

Awhile back I listed a number of search-and-replace "tricks" for Microsoft Word in the Tool Box Journal, and these were also published in the ATA Chronicle (by the way, if you haven't taken note of the newly designed online and offline Chronicle, you should -- it's beautiful). I said this about one of the tips: "I would be hard-pressed to make a good case for why this next tip is of particular benefit for a translator. It's not. But I think it's cool anyway." And, boy, was I wrong, as Matthew Kushinka from RedLine Language Services let me know. He has published two blog posts that give two excellent use cases for a translator to re-sort listings the way I described it, one for converting date formats and one for converting number formats. Good ones!

 

Another article I was alerted to was by Jeroen Tetteroo from Language Solutions. This article deals with the translation file exchange standard XLIFF and its supposed lack of flexibility when dealing with paragraph-based content coming from content management systems and embedded HTML. Knowing that (XLIFF) geeks like Twitter, I knew that the best way to reach them was by posting a link to the article. Lo and behold, "localization tools enthusiast and XLIFF grump" (his own words -- I would never have come up with "grump" myself) Chase Tingley showed up in force and had much of the rest of XLIFF geekdom "liking" it. According to them, it's the tool developers and not the standard itself that's falling short. (Personally, I think the whole concept of XLIFF has been hijacked by tool developers as an easy way to represent bilingual files within a tool rather than aiming at exchangeability.)

Be that as it may, it looks like it would be good if we as a community could find our voice in this and not just leave it up to the standards and tool developers. Of course, you could make your opinion known at the public review for XLIFF v2.1 which ends February 24, but knowing that this takes a bit of expertise, it might be good to combine voices, which brings us to . . .

 

. . . the committee devoted to organizing a platform where translators (and interpreters!) can collect and refine ideas about where we really need improvements in our technology. I floated the idea in the last Tool Box Journal, and just like that a group consisting of Iulianna van der Lek-Ciudin, Barry Slaughter Olsen, Alexander Drechsel, Tom Alwood, and Martin Kappus (and, where applicable, their respective students) has formed to tackle this. I will keep you posted. (And join me in thanking these folks for engaging in this!)

 

ADVERTISEMENT

Teamwork made easier with the new SDL Trados GroupShare 2017

Managing projects has never been easier than with GroupShare 2017 as Project Managers have increased transparency over the projects they are managing, better security and ability to track progress in real time.

Further updates to GroupShare 2017 include a new modern look and feel and lots of new integration possibilities. Read more >

 

The Last Word on the Tool Box Journal

If you would like to promote this journal by placing a link on your website, I will in turn mention your website in a future edition of the Tool Box Journal. 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 journal with more than one email address, it would be great if you could unsubscribe redundant addresses through the links Constant Contact offers below.

Here is a website that recently added the link:

Giolester.com

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

© 2017 International Writers' Group    

 



 

Home || Subscribe to the Tool Box Journal

©2017 International Writers' Group, LLC