| ADVERTISEMENT | |
The Words You Want. Anywhere, Anytime.
Let WordFinder open a new world of opportunities- get access to millions of words and translations from the best dictionaries, on your computer, via a web browser, on your smartphone or tablet. Stuffed with lots of smart features. WordFinder has what you need as a translator in your everyday work - anywhere, anytime!
Read more at www.wordfinder.com
|
|
1. Fair Trade Translation
| |
Gert van Assche and Daniel Marcu are what you would call veterans in the world of translation. Gert has worked independently and for SDL (or companies that were eventually acquired by SDL) for many years; Daniel Marcu was the brain behind LanguageWeaver, the first viable commercial statistical machine translation system that was eventually also acquired by SDL and is now, after many rebirths, the engine that drives the SDL BeGlobal and Language Cloud solutions.
They have now teamed up to offer something rather interesting that they call FairTradeTranslation. I agree, it's a funny name -- reminds me of the smell of Nicaraguan coffee beans from my youth -- but it seems to be a relatively accurate description of what their system actually does.
Let's start from the beginning, though.
In a recent article on LinkedIn, Daniel states that "in spite of massive progress, MT does not yet deliver consistent or high quality. However, low-quality MT is often shoved onto the workbenches of professional translators who are asked to create high-quality products by post-editing bad MT while being paid less for 'the privilege of being more productive.'"
I know many would agree with this, but to appreciate it completely you have to understand who's talking. This is the person whose company was the first to commercialize statistical machine translation.
When I talked with Gert and Daniel a couple of weeks ago, Daniel said something similar but almost more striking. While some big enterprises have built good machine translation engines, he said, the percentage of the market that is served with those is "a drop in the bucket."
I know! Right!?
I'm writing this at the tail end of the MT Summit 2015 in Miami where I was asked to be on a panel on translators and machine translation. Those who have read my newsletter for a while know that I have been invited a number of times to speak at MT conferences as a quasi-representative of translators. Each time I felt honored and always tried to represent you adequately (knowing all along that I was inevitably doomed to failure because there is no "collective you"). It was interesting this time, though, that some of the responses I received were a little edgier than before. I had suggested that machine translation is a very welcome technology, especially as we've been able to work with it in the last year or so - as an integral component of the existing translation process rather than exclusively as a source for post editing. One fellow from IBM in particular was exceedingly unhappy with the suggestion that post-editing might not be the most productive way of dealing with machine translation, pointing to the IBM translators who worked with IBM's high-quality internal machine translation and how little they have to change to the output. So why would I suggest otherwise?
He might be completely right -- but what he is saying has nothing to do with you or me (minus the folks who do indeed work for IBM and a handful of other comparable customers with a similar setup).
And really, that's one of the areas where relatively isolated discussions on machine translation are just that: isolated or, to quote Daniel again, "a drop in the bucket."
So, back to FairTradeTranslation. They provide you with a platform that allows you to upload MS Office or text files or (SDL)XLIFF files. The next steps:
- machine translate the file (if the system thinks that the three connected engines -- Google Translate, Microsoft Translator, and SDL Language Cloud -- have a "good enough match"),
- evaluate existing translations (if the file is a pretranslated XLIFF file),
- replace those translations if it can find better ones, and
- give you an estimate of the quality of the machine translation so you know how profitable it will be to work on that file or project.
The underlying software for the tool is a statistically-based system that attempts to evaluate the quality of the MT proposals (even in comparison to low fuzzy TM matches) and, according to a setting determined by you, simply deletes or replaces poor MT suggestions (or TM matches) rather than passing on the bother and frustration to you.
Gert and Daniel know that the system is not infallible; in fact, in the FAQ section they say this: "Q: How can I fool the system? A: In many ways. For example, if you choose to submit a file with human translations, we will reject some as being not good enough -- translations produced by machines and humans are different, after all."
That's right, but if a) you're using machine translation as a base translation that you then want to post-edit; b) you know that your clients have no problem with their texts going to public resources (especially the Microsoft Translator is an issue here); and/or c) you need a quick way to look over a pre-machine-translated project to see whether it's worth your time before you commit to post-editing it, FairTradeTranslation might be a very interesting solution for you.
As a rare and casual user (up to 5,000 words a month), and if you're only interested in the basic features, you can use the system for free. If you want to process higher volumes and want the system to learn from your editing choices and selected options, you can pay $149 a year (for up to half a million words).
I'm not sure whether this system is going to completely change how most of us work -- it might mean some changes for some -- but what I really, really like about it is this: It's just so creative and it shows that we are doing very well if we have smart folks coming up with such smart solutions for us.
The complaint that the world of translation is not vibrant enough and does not come up with enough creative solutions is just wrong. We are doing quite well on that front.
|
| ADVERTISEMENT | |
Webinar with limited places!
Right word, right place
How do I use and maintain terminology dictionaries?
In this webinar, you will learn how easy it is to get started with terminology work using the TermStar terminology management system as an integrated component of Transit.
Webinar for beginner translators and terminologists
November 24, 4:00 p.m. - 5:00 p.m. (UTC+1)
To register, simply send an e-mail to [email protected]
www.star-group.net
|
|
2. ~ (Premium Edition)
|
Forgive me for quoting a few paragraphs from a newsletter of almost exactly two years ago when I first wrote about TaaS or Terminology as a Service:
TaaS is a project that has received major funding from the European Union Seventh Framework Programme. It has five collaborators: Fachhochschule K�ln, Kilgray, University of Sheffield, TAUS, and -- as the coordinator -- Tilde. (...) It will eventually be a large, cloud-based terminology resource in all official working languages of the EU for translators, interpreters, terminologists, and technical writers.
Presently, Bulgarian, Croatian, Czech, Danish, Dutch, English, Estonian, French, German, Greek, Hungarian, Irish, Italian, Latvian, Lithuanian, Maltese, Polish, Portuguese, Romanian, Russian, Slovak, Slovene, Spanish, Swedish, are supported. Note that "support" for one language does not necessarily mean the same for every other, but before we go into this, we should probably see what this tool does in the first place.
Once you've registered, you can upload one or several files in various formats (PDF,
MS Office, RTF, TXT, XLIFF, XML, FrameMaker, or HTML), have terminology extracted from the file(s), apply content within existing terminology resources to those terms, select from the suggested translations and/or translate the terms, and then export it so that you can use it within your terminology database or glossary.All this would still not be too overwhelming -- after all, there are plenty of tools that extract data -- were it not for a number of advanced tools that are (optionally) applied to this process. These include Tilde's wrapper system for CollTerm ( here is some more detailed information about that), which performs a linguistic analysis (part of speech tagging, lemmatizers, morpho-syntactic patterns, etc.) as well as statistical analysis; Kilgray's terminology extractor, which also performs a language-independent statistical analysis; and a tool to normalize terms, which brings terms into their canonical forms (typically nominative singular or infinitive). This latest, very cool feature is unfortunately only available for English and Latvian at this point (thus the aforementioned different levels of support).
Once that is done, the extracted list of terms will be run against a number of (again, optional) resources in the following order: 1. your own personal resources that you might have collected on the site; 2. other users' terminology (I'll explain in a second); 3. the EuroTermBank; 4. the EU's inter-institutional terminology database IATE; 5. the TAUS corpus; and 6. the TaaS statistical database (SDB) that consists of aligned web data. Once these databases have been queried for translations, they will be shown as suggestions from which you can choose by just clicking on them and/or you can enter your own translation.
Of course, one of the ideas behind this project is to make it possible to share terminology data. At the outset of each project you can enter a whole lot of optional data, but you will need to make a decision on the language combination, the domain of your text, and whether you want to share the data with other users. The shared data will not include the complete texts that you upload but only the term pairs that you will end up with in your termbases (and only on an individual term pair level rather than complete lists of term pairs).
As I said, two years have passed since I wrote this, and a number of things have happened with TaaS as well. For one thing it does not exist anymore - at least, not as a workable platform under the name TaaS. It is now called Tilde Terminology and can be found as a service on Tilde's website. Presently there are a respectable number of 1,500 registered users on the platform and approximately five million verified terms.
But just like that, the applicability of the Tilde Terminology service might now sky-rocket with the release of a Trados Studio plugin for Studio 2014 and 2015.
With the plugin, the emphasis is not on term extraction and glossaries translation, but on terminology translation of your current project via your resources that you might have built up within the Tilde resource (maybe even using the term extraction feature for the very project you are working on) and/or the publicly available data (the aforementioned five million terms). Once you download and install the Trados plugin and connect it to your Tilde account, found terms are displayed on a per-segment basis (by default in a new pane underneath the translation pane). I really like that these are also displayed as AutoSuggest terms (i.e., terms that pop up as suggestions as you type and which you can then accept). It's really quite cool, especially if you work in an area with a domain that has a good match with one of the Tilde domains, provided that it is filled with appropriate data. I tested it with English into German data, and as you can imagine there was plenty of data to be had.
This tool becomes particularly useful (just in case you hadn't seen much use for it so far) through sharing terminology with colleagues. How? Well, there is not only the plugin for Trados Studio but also connectors for a number of other tools, including memoQ, Wordfast Anywhere, and OmegaT, all of which can access the same shared terminology as you translate. (Of those, Wordfast Anywhere is the only one that also offers the AutoSuggest capability with the Tilde resource.)
Presently all of the plugins are fully functional and free, but there will be paid editions eventually that will allow for continued unlimited access whereas the free edition will only give access to something like 1000 terms a day.
As far as the Trados plugin, there are some glitches. Capitalization is non-existent at the moment in the target language. (Did I mention that I tested it in German, which happens to be a real Problem Language?) I also wish there was more documentation, but I was already promised that this will be fixed.
I also think it would be fantastic if the term extraction that now can only be done right within the Tilde Terminology platform could be done within the translation environment tools. But maybe that's something for a second generation of plugins.
|
3. The Tech-Savvy Interpreter: Choosing a USB Headset for Remote Interpreting (Column by Barry Slaughter Olsen)
| |
With a growing number of platforms that provide remote simultaneous interpretation for a plethora of use cases (e.g. Interprefy, VoiceBoxer, ZipDX), interpreters now find themselves needing to equip their offices with the right equipment for the job. And since we don't have a sound technician on the payroll (or at least I don't), knowing a bit about sound and headsets becomes a real asset.
For this month's installment of the Tech-Savvy Interpreter I want to focus on something absolutely crucial for anyone planning to work remotely as a simultaneous interpreter. It's called "full-duplex audio." Put simply full-duplex audio allows you to receive and transmit audio signals at the same time-something absolutely essential for simultaneous interpreting, of course. This simultaneous two-way flow of audio is often called "doubletalk" by audio engineers (No, it is not the sinister cousin of Orwellian "doublespeak"). For the tech geeks out there, you can read more about it here.
Unfortunately, as we discovered when building a remote simultaneous interpretation platform a few years ago, finding a high-quality USB headset that is capable of simultaneously receiving and transmitting high-quality audio is not a simple task. The reason? Many of the USB audio chips used in headsets today can't deliver a consistent audio signal from the microphone while sound is being played in the headphones. And that is a big problem if you plan to use the headset for simultaneous interpretation.
Most retail electronics stores do not stock more than one or two models of USB headsets-usually low-end models for the occasional Skype user and high-end models for hard-core video gamers. To complicate things further, sales associates usually don't know enough to provide the guidance to ensure you get the USB headset you need. So what's a hardworking simultaneous interpreter to do?
Fortunately, there are a handful of full-duplex-capable models out there that handle doubletalk very well. A big "thank you" to Michael Graves from ZipDX for compiling this list. All headsets have been tested and can send and receive high-quality audio at the same time. For convenience's sake, the links offered below are to Amazon. The headsets, however, can be purchased from multiple sources.
VXi Passport 21V: This is a great, inexpensive headset from a company that only makes headsets. It's light, comfortable, durable and sounds great. They are available in single or dual-ear models. VXi sells refresh kits that include new foam for the earpieces and pop filter for the mic. That gives you some sense of how long they expect the headset to last.
The headset cable ends in a "Quick Disconnect" (QD) fitting. You must add a suitable lower cable, of which there are three:
VXi Passport 21P: This is the same headset as the first option, but equipped with a Plantronics type Quick-Disconnect connector. This allows it to be used with a less expensive Plantronics DA40 USB adapter cable.
Jabra BIZ2400 Duo: These are light and sound good. The round control in the cord has a rotary volume control with mute function.
Plantronics Blackwire C720: On a pure audio quality basis this headset works adequately. It's not the most comfortable. It connects to the computer via USB. It has a control in the middle of the wire that's actually a Bluetooth radio, so it can also be paired to a cell phone.
Logitech H650e: These USB headsets are available in single or dual-ear models. They're not as durable as the VXi. They have a flat cable that some may find to be stiff.
You'll notice that all of these USB headsets are wired. Wireless technology introduces a whole new set of complications and is best avoided for remote simultaneous interpretation.
Do you have a question about a specific technology? Or would you like to learn more about a specific interpreting platform, interpreter console or supporting technology? Send us an email at [email protected].
|
| ADVERTISEMENT |
Do your clients need simultaneous interpretation for conference calls?
ZipDX has pioneered proprietary technology that makes high-quality over-the-phone simultaneous interpreting a reality for both the public and private sectors. A "multilingual conference call" or "multilingual virtual meeting" works like a regular conference call, connecting several people in different locations. ZipDX enables a multilingual conversation or discussion by making it possible to connect simultaneous interpreters to the call. Interpreters can be located anywhere and use a web-based virtual interpreter console.
Our pricing model makes it simple and cost effective for interpreters and language service companies to offer remote simultaneous interpretation for virtual meetings.
Want to learn more? Watch our video: https://www.zipdx.info/solutions/multilingual/
Contact us by email at: [email protected] | Phone us at +1 888 ZIPDX LLC (1-888-947-3955)
|
4. Clean Slate?
| |
Almost exactly five years ago, I wrote about Tom Hoar and Precision Translation Tools for the first time. At the time he had just launched his first Moses-based machine translation product, called DoMY. It was a prepackaged Linux-based command-line tool that allowed you to build your own machine translation engine. Two years later he released the engine with a graphical user interface (but still on Linux), and another two years down the road, in 2014, his company released the program as a server-based application with connectors to tools like the TEnT OmegaT or Welocalize's translation management system GlobalSight.
Since all this was still happening exclusively on Linux, it was accessible only to folks who liked to work in that environment or who felt adventurous enough to invest in a Linux computer and the necessary learning curve that comes along with any new computing environment.
The numbers proved this was not the most attractive offer to the translation community. According to Tom, maybe 300 individual translators use the system, many of whom are using the free open-source version.
Long story short, Tom hired a new engineer and ported his machine translation program to the Windows environment to make his program more palatable to you and me (or, as many very aggravatingly would say, "you and I"). This is available now, but only in the form of a command-line tool (meaning that you essentially have to control the tool by typing commands into Windows' Command Prompt application -- not the way we like it with clicking on buttons, etc.), with no connectors to existing translation environment tools (so you have to convert and pretranslate your texts in the form of a TMX file and then use that in Trados, memoQ, Wordfast, or whatever).
To find funding for this, Tom and his (small) team started a crowdfunding campaign that they just now successfully finished (though you can still sign up and claim one of the tools that is slated for release with a 40% discount). The new tool is called Slate Desktop and will have a fully functional graphical user interface and connectors to some translation environment tools (though I'm not sure which will be operable on the January release date).
So what to think about this tool? I think Tom is doing us a favor by trying to communicate statistical machine translation in very down-to-earth terms (sometimes even a little too down-to-earth for my taste) and giving us (as a community) the notion of accessibility. I also think that his tool will work well for some translators who work almost exclusively for one client, or at least within one relatively narrow area of specialization. I could imagine that the productivity increase of 25-30% that Tom claims could easily be realized by some of those.
Others will have a hard time finding the required amount of their own data that they need to feed to the tool. Most of us have an estimated minimum of 130,000 translation units in our TMs, and many of us have much, much more, but is it all within one narrow area of specialization? You see, if I'm working in "heavy machinery" translation, for example, it's typically not good enough to mix data from different clients or different applications of the machinery, let alone data from completely different fields.
I know there are colleagues who work in very narrow domains and mostly for one client, and I think for them this tool might be a real productivity boost, but I'm not sure how much that's true for others. In its promotional materials, Tom talks a lot about the style of the individual translator, claiming that the tool will learn my unique style. But I have to say that I change my "style" all the time as I translate. I might have a certain style as I write this newsletter, but when I translate I try to respect the style of the original. Am I still going to add my own imprint to every translation I do? Sure I am. But I suspect that's not distinctive enough across all projects and clients to train an MT engine adequately.
I will continue to watch Slate's developments, and I will be very ready to change my mind about a wider-than-expected application if I find out I'm wrong.
|
| ADVERTISEMENT | |
Across v6.3: Translators Have Made Their Choice
The new Across version for translators will be available in late November. Many optimizations are based on suggestions from the Translators' Advisory Board. In the future, translators will be able to integrate their own TMs and terminology lists. Additional file formats, such as PDF and JSON, will also be supported. The new version will facilitate increased customer proximity and enable the translator to highlight his skills.
More information will be available at www.my-across.net from November 23.
|
5. This 'n' That
| |
I've been a grump recently when it comes to conferences. Yes, I'm at the ATA right now, and, yes, I will enjoy it (grumble, grumble), but my frequent flyer status next year will most likely go down by a category since I haven't been at very many conferences this year.
So it comes from the bottom of my heart when I say that I can't wait for the Nordic Translation Industry Forum, November 19 and 20 in Reykjavik. When I booked my flights a week or so ago, they were shockingly cheap. And can you think of a better place to be in November than in Iceland? While actually learning in high-powered sessions and making new friends in the process?
Mats Linder just sent me the latest version of his Trados manual, which now covers everything up to version 2015 with the current Service Pack 1. It's a mighty 500-page tome at this point, and you can find it right here.
Another new publication is the Out of Office newsletter that deals with everything regarding MS Office that could ever come in the way of someone "professionally working with text." It's published in four languages (EN, DE, FR, ES) and is written and translated by distinguished colleagues. You can download the first edition (000) for free on the website and subscribe to the paid edition for later issues.
|
6. New Password for the Tool Box Archive
| |
As a subscriber to the Premium version of this journal you have access to an archive of Premium journals going back to 2007.
You can access the archive right here. This month the user name is toolbox and the password is kingtide.
New user names and passwords will be announced in future journals.
|
| 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 listed the Tool Box Journal last month:
www.thefutureprooftranslator.com
Should you be interested in reprinting one of the articles in this journal for promotional purposes, please contact me for information about pricing.
� 2015 International Writers' Group
|
|
