ToolkitSmall

A computer newsletter for translation professionals

Issue 10-12-180
(the one hundred eightieth edition)
Contents
1. Choose One, or Two, or All of Them?
2. Moses Never Got There (Premium Edition)
3. Holiday Fonts
4. The Windows 7 Bag of Tricks (Premium Edition)
5. T-Index
The Last Word on the Tool Kit
Quantum Leaps

Here is one aspect of technology -- or "the history of technology" -- that really fascinates me. Once a certain technology has truly taken hold, it's here to stay. Or so it seems. It's here to stay unless a) someone with great financial backing comes up and methodically tries to push it out (think of what Internet Explorer did to Netscape); b) when a competitor takes a gigantic step forward and releases something so vastly superior that it elevates that particular technology to a whole new level of advancement, giving users no choice but to switch (what Google did to Altavista or what email did to the fax); or c) when a completely new value proposition is made (for instance, free versus paid, or software as a service vs. paid application).

One new tool (not a language tool, mind you) that I played with this week might just have made this quantum step. Qwiki presents a new concept of browsing. Rather than listing links to webpages that contain the information you're looking for, it distills the information in presentations with voice, image, and video. Really, really clever.

Jeromobot may also have made a quantum leap this last week with his completely new Android version.

1. Choose One, or Two, or All of Them?

Just last week somebody in a newsgroup devoted to translators asked whether she should use a second translation environment tool, one that is more commonly known as a tool for freelance translators -- Wordfast -- in addition to the tool that she was using for a specific client -- Transit.

Without wanting to discuss either Wordfast or Transit, I thought this was an interesting question.

First of all, the answer seems obvious. If she has already purchased Transit, has learned how to use it, and is happy with how it performs, there's no reason to buy another tool. With the exception of some very rare formats, translation environment tools all support the same source file formats and the exchange of translation memory and terminology database data, and the vast majority also supports each other's translation file format, such as Trados .ttx files, bilingual Word files, Transit package files, Wordfast .txml files, and of course XLIFF files.

So, if there is any way for you to work in a single tool that generates and stores your supporting data, that is what you should do. This will allow you to keep the loss of data to the very minimum (as in when you might have to convert a translation memory from an external format to the format of your tool); you'll only have to become an expert in one tool; you'll have to purchase and maintain only one; and you'll have a realistic understanding of what your tool can and cannot do (which sometimes helps before you agree to do a certain project . . . ).

Of course, there are exceptions. There are those very rare source formats that only one tool supports, translation-environment-tool-specific formats that competing TEnTs cannot support, and (and most common and annoying) server-based projects that third-party tools cannot hook into. But except for these cases, use one tool only.

Now, if you have not decided on a translation environment tool or if you're so frustrated with your existing tool that you're looking for a change, there is help! Many of you know that I've partnered with Intrawelt, an Italian translation and eLearning company, to build TranslatorsTraining, a site that introduces the vast majority of translation environment tools in a no-frills and very objective manner through videos of each of the tools, all prepared by their respective vendors and edited by us. It's good, it's free, and it seems to me just about the perfect thing to look at on a slow Advent Sunday afternoon or Monday morning.

And if you make a decision for one or the other tool, we even offer special deals for them (you will have to get to the end of the respective videos to see the special link and/or discount code). Here are some examples: memoQ translator pro normally costs €620, but at TranslatorsTraining you pay only €496; Déjà Vu X Professional is regularly €660, but we offer it for €528; and SDL Trados Studio Freelance costs €549.25 instead of €845. Good deals? We think so, too.

While we're talking about translation environment tools, here are a couple of news items and amendments:

Star Transit NXT has just released its latest Service Pack that includes Microsoft Office 2007 support (beats me why Office 2010 is not officially supported yet -- but that's true for Wordfast Classic as well), InDesign CS5, and QuarkXPress 8 support as well as import and export capabilities of TMX files from its terminology database component TermStar.

In a recent newsletter I wrote about subsegment matching abilities of some tools, and some folks from SDL pointed out that Trados Studio also supports subsegment matching with its AutoSuggest add-on. This is true and hereby duly noted.


ADVERTISEMENT

SDL End of Year Sale now on!

 

Enjoy exclusive savings when you buy SDL Trados Studio 2009 SP3 Freelance, plus get SDL AutoSuggest Creator free! Are you an existing customer looking to upgrade to the latest version? Upgrade online and receive 25% off for a limited time only.

 

To learn more about our products and to take advantage of our offers visit -- www.sdl.com/toolkit_12

 

Hurry, offers end 31st December!

2. Moses Never Got There (Premium Edition)

At a recent conference Achim Ruopp of Digital Silk Road, a machine translation consultancy, mentioned Moses for Mere Mortals to me, a specially packaged version of the open-source machine translation engine Moses. Moses for Mere Mortals was developed by two Portuguese EU translators in coöperation  (I love that spelling!) with a computer developer. As the name of the product indicates, it is significantly easier to use than the original Moses (remember that Moses had to die before he reached the promised land?)

Moses is the machine translation engine at the basis of the home-grown statistical MT development efforts of some of the larger language service providers as well as products such as Asia Online.

A few weeks ago I asked whether any of you had any experience with using Moses for Mere Mortals, and I got some interesting responses and leads. Here is what Ignacio Garcia, a professor at the University of Western Sydney who teaches translation technology, said:

I've been using the Moses for Mere Mortals these last weeks -- not for translating, though, but for research/training-related reasons.

I tried to install Moses first from the step-by-step guide in statmt.org and ran into all sorts of problems (the guide is good, but my Linux literacy is limited). With much less effort I was able to put Moses together following the quite clear MMM guides, and gathering some available data from the statmt.org site. And it works, although not as good as Google Translate -- yet.

MMM also has some Windows add-ins that allow you to export Moses output into TMX and to import TMX into the format Moses reads, so, in theory, yes, it can be easily integrated into our TEnT's workflow.

Would it be really useful, right now, for freelance translators? Perhaps for those who over the years have collected (or are able to access) BIG memories closely related to the type of texts they still translate. For me and for the few freelancers I'm in contact with, that's probably not the case.

But even if you had the memories, my guess now is that it will take too many hours to fine-tune the system(s) -- finding out what the right proportion of "general" versus "in-domain" data should be -- to get results that are significantly better than those of the free online systems.

Having said that, I loved MMM and I'm sure the future of technical translation will end up integrating TM (ok, TEnTs) and MT in that way. But for the freelance (as against corporations and big LSPs) to benefit, clean "general" and "domain-tagged" monolingual and bilingual data need to be made available first. And for this, projects such as those of TAUS Data Association need to succeed.

Two things were very notable to me in Ignacio's retelling of his experience with Moses for Mere Mortals. One is the need to use a Linux computer (Moses for Mere Mortals runs on the Linux distribution Ubuntu). It's probably fair to state that the majority of translators have had no exposure to Linux, and while this does not have to be a major obstacle if one is truly willing to plunge into building a customized machine translation to raise productivity, another point that Ignacio raises is probably even more noteworthy. He says that the output of Moses for Mere Mortals is not as good as that of Google Translate.

When I first read this I chalked it up to him being a single guy working on training a system with a presumably smaller corpus. But when I found this article by Russian translation provider Logrus, a company that has invested major data and human resources to train the full-monty Moses for the translation of Microsoft documents, I found that their results, while reasonably good, did not reach the quality of the Microsoft Bing Translator. This truly made me wonder at all the effort in building customized statistical machine translation engines when the two free giants (Microsoft and Google) actually perform better. I realized that the data-crunching abilities of those corporations far exceed anything that you and I can even dream of, and yet both Google Translate and Microsoft Bing Translator are generic machine translation engines and apparently compete quite well with engines that are trained with a very specific corpus of data. (I am quite aware of the problems connected with data privacy when using either Microsoft or Google, but here I'm looking at functionality only.)

While these two cases should not be generalized, it's something that's worth being aware of and continuing to watch: How much of an investment is necessary to build a statistical machine translation engine that is better than one of the two online giants? I have a strong feeling that the answer depends on both language combination and subject matter. (Chances are that it's no coincidence that it was Microsoft's Bing Translator that performed well on the translation of Microsoft's documents, for instance.)

And then there is another effort that is somewhat comparable to Moses for Mere Mortals. Tom Hoag, who is spearheading this endeavor, used to work for machine translation developer Asia Online (it probably helps that both Tom and Asia Online are located in Bangkok). At some point Tom decided to go independent, first with consulting and now with a company called Precision Translation Tools.

Like Moses for Mere Mortals -- which, according to Tom, is no longer being actively developed (i.e., the version of Moses that is presently wrapped up in the installer will be the last one) -- Precision Translation Tools also offers a packaged version of Moses; in fact, it offers various packages. PTT's homepage leaves no doubt what's on the mind of its owners. It says:

Moses = waste of time.

Do Moses Yourself™ Community Edition (DoMY™ CE) transforms Moses into a simple do-it-yourself project so you can immediately enjoy the benefits of translation software that works. Precision Translation Tools created DoMY™ CE to give you a fast start that saves you thousands of dollars.

Now, DoMY™ CE = translation productivity.

If you think that these statements sound a little over the top, I'd have to agree (the machine translation community has suffered from this kind of hyperbole for much too long). Still, the offering of PTT is interesting. Its basic version, the Do Moses Yourself Community Edition or DoMY CE, is available for free (and a suggested donation) and includes all components of the Moses machine translation engine plus a couple of tools that were specifically developed by PTT, including the CorpusFiltergraph, a tool for aligning large amounts of data for machine translation corpora.

(Forgive me this side note: Am I alone in thinking that MT developers tend to lean toward strange names: Moses for Mere Mortals? Do Moses Yourself Community Edition?)

vPresumably the installation, training, and first translation excursions can be done within a day compared to the several months of development time required with the unpackaged version of Moses. Noting the hype of DoMY's makers above, this might well be an overstated time frame, but then a major component of their business model consists of offering (very inexpensive) consulting and training programs (aside from the slicker paid-for version DoMY Pro -- $50 -- and upcoming paid-for server editions), so you should be able to get up and running with minimal investment. (That is, if you don't count the high-powered Linux computer with Ubuntu installed as part of the investment.)

Precision Translation Tools was only launched this year, so there are a number of things that are still in the "Coming Soon" phase, including the above-mentioned server versions and a workflow that allows round-tripping between Windows and Linux machines (so you can still use your Windows computer to do the actual translation or post-editing after the machine translation on the Linux machine has been performed).

Since the large, big Moses itself is in a constant state of development, the plan is to update the different versions of DoMY every 6 to 12 months. It will be interesting to see where PTT's efforts will be in 12 months.


3. Holiday Fonts

Hey, if you don't like to read about fonts, feel free to skip right over this! (And please stop giving me a guilty conscience for writing about something I really love.)

When Windows Vista introduced a whole new slew of attractive fonts (the Clear Type fonts C-series -- Constantia, Cambria, Corbel, Calibri, Candara, and Consolas -- as well as the Segoe family and Nyala), most were geared toward everyday use by the Western user. Windows 7 had the non-Western user community as its main focus with a large range of new Asian and African fonts. One notable exception that almost escaped my notice is Gabriola, a font that is packaged into both Windows 7 and/or Office 2010.

Back to Gabriola in a second. Let's first talk about the improved font management in Windows 7. If you open the Fonts folder in Windows Explorer (just press the WinKey, type Fonts, and press Enter), you can immediately see the first difference: each font file's icon (visible with Medium, Large, or Extra Large icon views) shows a preview of the font.

Second, you might notice that it looks as if you have a lot fewer fonts in your Fonts folder. Chances are you don't, but the fonts are now combined into families (such as all Arial or all Garamond fonts), which can be expanded by double-clicking the icon. This makes it a lot easier to manage.

Third -- and this may be the most interesting new feature -- you can now activate and deactivate fonts by right-clicking a font icon and selecting the Hide option. What is this good for? Ever noticed that some applications take forever to load? And, if the starting screen shows you what is currently happening, loading the fonts takes particularly long? You can avoid this, and at the same time have Windows start up much faster as well, if you disable all the fonts you commonly don't use and just re-enable them by right-clicking again and selecting Show once you need them. Before Windows 7 there actually were specific applications that you could purchase to do this with, but now it's built into the system.

But now back to Gabriola. We are in the midst of the holiday season. Chanukah is in full swing right now (Happy Chanukah! Check out how Jeromobot celebrated it this year); Christmas is around the corner. Wanna send some really nicely styled holiday greetings? You might want to have a look at the new Gabriola font. (Note that you will have to have Word or Publisher 2010 or a "real" desktop-publishing tool like InDesign for the fancy stuff I'm going to show you.)

Gabriola is a free-flowing OpenType font that allows for the use of different stylistic sets (seven to be exact) as well as for contextual alternates. Contextual alternates are letters that are displayed differently depending on where in the word they are located. Imagine the letter r, m, or n with a little hook on its left side when it's by itself, but the hook is gone when another letter intrudes into its space (see the m in the examples below.)

This is not unique; in fact, it can be done with a good number of fonts, including most of the "C"-fonts mentioned above. (The example of the proportional character treatment of the Japanese Kazuraki font that I mentioned a couple months ago certainly goes way beyond this.) The specialness of this in relation to MS Office applications is that it can be activated or deactivated by launching the Font dialog (Ctrl+D) and selecting Use Contextual Alternates on the Advanced tab.

While you're on that dialog, you can also see that there are a number of other options, including one for Stylistic Sets. This option won't do any good with most fonts, but do me a favor (again, you will have to have the correct version of MS Office), put some text into the Gabriola font, and then apply stylistic set 1 through 7 to see what happens -- it should be something akin to my example below. (In case you're not impressed, any 10-year-old girl will be -- and I've even chosen the right term to impress her further.)

Gabriola Magic

One more thing about the holidays: I had a great idea this year.

Freelance translators: Wouldn't it be great if your favorite language service provider that you've worked for all year sent you a year's subscription to the Premium edition of the Tool Kit newsletter instead of the same old boring Christmas card? Why don't you mention that to your project manager as a suggestion?

LSP representative: Wouldn't it be great to give a much appreciated gift to your freelancers as well as one that you will actually benefit from (because your translators will become more technology-savvy)? Send me a note at jost@internationalwriters.com and I'll even be able to offer you a special price for each of the subscriptions.

Jost: Wouldn't it be great if I could communicate these things a little more smoothly?

Anyway, I truly think this is a good idea. A couple of LSPs have already jumped on board, and it would be great to see more of you joining this program.


4. The Windows 7 Bag of Tricks (Premium Edition)

The Jump List feature in Windows 7 is the function that allows you to right-click on any icon in the Taskbar and select from a number of options, including, and most helpfully, the most recently opened instances of documents with that particular program.

This also works with Windows Explorer. However, since you typically use Windows Explorer so frequently throughout the day and there is space only for the last seven visited locations, chances are you won't see the place you need to go to once a day to, say, make a backup of your current project at day's end.

The good thing is there is a special feature for Windows Explorer: You can manually bookmark some favorites to the top of the Recent list by pinning folder locations. Just click on any folder and drag that folder icon to the Explorer shortcut on the Taskbar. You'll see the message Pin to Windows Explorer before you release the mouse button. The folder will now appear under a Pinned section of the Jump List, and you can remove it by clicking the Unpin from this list icon on the right side of the panel.

And because it's so much fun to do this in Windows Explorer, you can also do it in a couple of browsers, including Firefox and Internet Explorer, so that you have webpages pinned down in their Jump Lists, allowing you to open the pages without opening the browser first.

And here is another great trick if you want to start work on a specific project for which you need to open more than one application (such as a browser, a voice recognition program, and a translation tool).

All you need to do is open Notepad and type something like this:

start /d "C:\Program Files\Mozilla Firefox\" firefox.exe

start /d "C:\Program Files\Nuance\NaturallySpeaking10\Program\" natspeak.exe

start /d "C:\Program Files\ApSIC\Xbench\" XBench.exe

Once you entered the text, save the file as a *.bat file. The .bat extension tells Windows that this is a batch file that contains a stack of commands that it needs to execute when the file is opened. To open it you'll only need to double-click on it and (in this case) Firefox, Dragon NaturallySpeaking, and XBench are automatically started. Clearly you are probably interested in other programs, so all you need to do is replace the path to the program within the quotation marks and enter the file name of the program afterward. Save it as a *.bat file and you'll be happy for the rest of your life. (Ah, if it were only be that easy. . . .)

ADVERTISEMENT

Translation in a team has never been easier!

 

Take advantage of reach collaboration features of Text United Translation Hub and register before year's end as a team leader providing a promo code: TeamDec10

 

You will have full access to translation memory, terminology management and collaboration functionality for free for a month!

 

Learn more at www.textunited.com

 

Register: https://www.textunited.com/Registration.aspx

5. T-Index

No, T-Index is not the name of one of the dreaded T-Windows applications of Trados versions of old; instead it's a clever new site by Italian LSP Translated (these guys are also the folks behind the ultra TM MyMemory). Unlike MyMemory, T-Index is primarily geared toward translation buyers who can examine a massive chart (which will be updated on a semi-annual basis) that displays the presumed purchasing power of Internet users within a certain language group and country. It arrives at that number by looking at the total number of Internet users of a given country as well as the presumed average GDP per capita for Internet users (the authors assumed that Internet users belong to the richest segment of the population). And while the chart is organized by languages, these can be drilled down to arrive at per-country numbers.

Translated built this as a sales tool for their translation services -- to communicate into which languages their clients' websites should be translated -- so you might not want to link to their particular site for your own sales and marketing purposes (unless you want to carry a copyright notice that promotes someone else's service), but it's a clever tool nevertheless that might give you a good idea for a creative marketing approach.

The Last Word on the Tool Kit

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 Kit. 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.

Here are two readers who recently loaded the code:

www.catherine-muir-translator.com

www.german-translation-tips-and-resources.com

© 2010 International Writers' Group