A Computer Journal For Translation Professionals
(the three hundred fifteenth edition)
The older I get the more vividly I
seem to dream. My dreams are so vivid, in fact, that I sometimes have a
hard time distinguishing between what was dreamed and what was real. (On
one hand, it's really annoying; on the other, it's very useful as an
always available excuse -- "I had no idea I promised that in real life,
honey. I was sure I'd dreamed it!")
The last couple of nights I've been
having impostor dreams. One night I dreamed that I went from person to
person only to find out that each was smarter than me. The next night it
was essentially the same, only this time no one liked me. (Insecure
much?) It is truly no fun to wake up in the morning with that taste in
This reminded me of another kind of
impostor syndrome that I do actually suffer from in real life, the one
where I compare myself unfavorably to literary translators. Now, I'm a
technical translator (with that I refer to a kind of translation that
requires a specialized vocabulary, so anything between legal, medical,
business, government, and technology), and I'm well established there
and generally happy to be working in my particular niche.
But then I find myself reading
interviews with literary translators (such as the wonderful series of
interviews in the
Los Angeles Review of Books
and I marvel at the polished and profound thoughts they are able to
articulate about their translation work. When I tweet quotes from these
interviews, I see that you (those who follow me on
like them as well. And yet, those quotes describe an area of
translation in which a) I don't feel qualified and b) -- ahem -- I
likely couldn't afford to work.
So, what gives?
Well, I've come up with a few
conclusions. First: Of course literary translators have more beautiful
things to say about translation! After all, they are
literary translators whose work is supposed to change minds and hearts.
Second, though I might not be quite
as eloquent in my expressions about transformations and metamorphoses
achieved via translation and while translating, I (should) know that I'm
actually performing the same task -- even though my product is less
likely to reach hearts as it is to protect limbs in safety warnings and
teach brains in user instructions. The same is probably true for any
other organ or body part for your work.
And third, I get to see a side of
translation that is typically not seen by literary translators: how
technology intersects with translation. Although I realize that literary
translators might not find this side attractive, I love it -- the
"machine in the loop" (see below) allows me to optimize what I'm good at
while relying for everything else on something that's better than me in
other tasks: the machine.
So, on second thought, perhaps I
should shed my impostor syndrome as I picture myself standing alongside
literary translators. And as we stand together, I'm thankful for the
wonderful products they create and the powerful words they find about
our common profession.
translate5 Coming into Its Own
Continuous Release Cycles
Machine in the Loop
MultiLingual Summer Series
The Last Word on the Tool Box Journal
Translate your Office documents or CAT bilingual files using the MT of your choice
GT4T is an intuitive MT add-on to
all your Windows/Mac apps. It is now also a file translator that
supports a great number of file formats. Get your Office documents or
CAT files pre-translated without having to upload to a server. Click
here to get a free license:
translate5 Coming into Its Own
I would like to spend some time
talking about an ever-evolving tool that -- so far -- has found a
passionate user and supporter group particularly in German-speaking
. But before I do, two quotes. One from myself of six years ago when I did a lengthy review of the then-
in issue 234 of the
Tool Box Journal
"), followed by a more lengthy quote by Marc Mittag, the lead developer behind
, also from the
Here's what I said as a summary of sorts about translate5:
"[T]his is where
into play. It offers a super-easy-to-use and completely configurable
web-based interface in which bilingual files can be edited, commented
upon, and evaluated to be brought back later into the originating
translation environment tool so the translation memory will reflect the
last state of the project and the project owner will have a very clear
idea about its quality."
And here is what Marc said about the underlying concept of developing the tool in 2016:
"Since I entered the language
industry in 2002, I have observed two major trends: Word-prices on the
market are under pressure. And the involvement of IT in translation
processes is on the increase. Both trends are increasingly accelerating
in recent years.
"Yet another trend has also recently
become apparent, however: The big LSPs in the market are trying to take
control of translation processes through technology. They are trying to
own the process -- and thus make suppliers and customers dependent on
"All three developments are
interconnected: Building their own technology enables the big players to
control the processes and keep the pressure on prices. Thus they are
making it more and more difficult for small and medium-sized LSPs to
stay profitable -- and to keep their independence. Without their own
control over the processes, these LSPs tend to become mere pawns of the
big LSPs, with the danger of being kicked out of the market.
"This is especially true for single
and regional language vendors. But it is true as well for smaller multi
language vendors (MLVs) who simply cannot afford to invest as much in IT
as the big ones. So these smaller MLVs are not and will not be able to
compete with the big ones in efficiency, in the new upcoming technical
trends (like automatic linguistic quality checks, self-learning MT, or
business analytics for translation processes and quality), and in
setting up specialized processes for their customers.
"As I perceive it, to cope with
these trends, small and mid-sized LSPs have to invest in IT and thus be
able to compete and stay independent and profitable. They have to own
their software -- instead of owning licenses.
"Of course, these LSPs cannot afford
to do all of this on their own. They must cooperate with others. Yet
this is no handicap: Analyses show that many of the most successful
companies are those who cooperate with their competitors.
"This leads me to the main topic: To
cooperate between competitors on IT, you need an easy solution from an
organizational and legal point of view. This is exactly what open source
is made for and has served for numerous times."
a tool that was originally created for the translation review process.
It follows an open-source model where a group of language service
providers have pooled their funds to pay Marc's company to build a tool
that can compete (hopefully favorably) with other large commercial
tools. Not surprisingly, much in what Marc is saying in regard to large
LSP-owned technology is primarily reminiscent of SDL. (At the time of
Marc's writing, Lionbridge was unsuccessfully trying to push its
technology to the larger market; other tools, such as Wordfast, might be owned
by companies like TransPerfect but are decidedly independent.) In
today's landscape, however, there are newer attempts to connect services
and technology again, such as with Lilt or Smartling.
Here is what has happened to
2016. Successful fundraising campaigns that raised 63,000 euros plus
numerous features that are specifically supported by one or more parties
be used by LSPs in -- as mentioned above -- mostly German-speaking
Europe. Interestingly, it is not only for the review process anymore.
Where previously the actual translation and file management process was
done in tools like
translate5 now increasingly encompasses the entire translation process
workflow. How is that possible? By adding a number of features that
round out the tool as a complete translation environment tool.
Here are some of those:
- A web-based interface that relies either on an installation on your own server or a hosted cloud service
- A strong terminology tracking component (this, by the way, is the only major part that was taken on from the OpenTMS project of the 2000s and early 2010s in which many of the companies involved now, including Marc's, also participated)
- A visual translation editor that
shows you the translation in real-time in the original format (any of
the supported, visually displayable file formats) as you translate it in
the translation table below
- A highly customizable translation interface with simultaneous multi-filter features
- Track changes that are comparable to MS Word (and importable to SDL Trados and XLIFF 2.x)
- Support for all file formats supported by the open-source Okapi tools plus bilingual files from SDL Trados, memoQ, Across, and Star Transit
- Access to a large number of content management systems via Xillio (see the report in TBJ 293)
- Integrated high performance TM models via OpenTM2 or NEC-TM (a tool-agnostic TM system, developed by the EU and based on Pangeanic's ActivaTM, which in turn was formerly known as ElasticTM -- see TBJ 265) and API-based support of SDL Trados TMs via SDL GroupShare
- Integration with a number of MT engines
- Workflow management including
job types, simultaneous multiple user support, automatic job
notifications for every workflow participant, customizability of every
workflow (component), etc. -- but no client management
- Automated QA checks for
terminology, tags, segment length (based both on number of characters
and pixels for a selected font), and linguistic features for the 20+
languages that LanguageTool supports -- and of course the manual QA checks that translate5 was originally built for (see my report in the TBJ 261, mentioned above)
- Integrated risk prediction scores via ModelFront (see TBJ 312)
The frequent references to earlier editions of the
Tool Box Journal simply highlight the philosophy of
translate5, which seeks to integrate with other technologies as well as integrating readily with others through its complete REST API.
How then would one use
? Well, it's an open-source tool, so you can download it for free (minus some select features, including the visual editor, the
integration, and track changes) and install it on your own premises. If
you would like to have some support along with that, you can purchase a
support contract (which makes you part of the translate5 community --
you can get an idea of what kind of companies are also part of the community using
Or you can purchase a hosted license at
at various price points that you can all find
When I talked with Marc a couple of
weeks ago, he said a few interesting things. First of all, he is not
trying to make the tool do everything. For instance, you might have
noticed in the list above that there are no sophisticated client
management features. While workflow is an important aspect, Marc's
emphasis is primarily on building a well-rounded translation environment
tool (CAT tool) that is able to deal with all aspects of file
management, translation/editing in its various forms and phases, and
efficient processing. Still, if the community or even single members of
the community (which could even consist of an individual or small groups
of freelance translators) want his company to develop features, they
are free to ask and pay for that. Once developed, the features would --
true to the open-source concept -- likely then be available for
Another product that is also offered
-- either as a standalone product or as part of the package -- is
which can provide a company-branded interface for automatic
translation. This translation will not only be based on the machine
translation of your choice (in fact, it does not have to be based on MT
at all), but also on any translation memory and termbase data within the
client's infrastructure -- an interesting proposal for translation
buyers who need a real-time feature like that but are very security- and
PDF Translation for Professionals
Translating PDFs is easier and quicker with
- Integrated into Memsource, memoQ, ONTRAM
- Compatible with all CAT tools.
- Fast log-in for Proz members.
for your next PDF project.
Continuous Release Cycles
It's hard to do up-to-date coverage of
Why? Because of its continuous release cycle. Four releases a year
theoretically means that I could write four articles a year on one
product alone. Since that really would not be particularly helpful, I
try to reach out to the folks at memoQ every couple of years or so to
talk to them about what they're up to at that particular point in time.
A couple of weeks ago, I met with
Veronika Pándi and Zsolt Varga (both Product Owners) and Product Manager
Ádám Gaugecz to talk about the latest 9.4 release and its emphasis. COO
Brúnó Bitter chimed in later about business matters.
First of all, you may have noticed
that none of those four was a founder or member of the board. It's not
that I wouldn't have enjoyed talking with any of those original folks,
but I was really happy with the group I talked to because it reminded me
once again that memoQ has truly evolved into a mid-sized translation
technology provider (actually, given the landscape of translation
technology, more like a "mid-sized company that looks really large").
Long gone are the days when it was just the founders, a (very) small
handful of full-time developers, and a lot of freelance coders. Today,
90 people work for memoQ.
Still, the themes we talked about
were very familiar from memoQ's inception. One theme is the emphasis on
the "balance between satisfying the different user groups," either with
each release or with alternating releases that satisfy both groups. (I
think this particular release favors the multi-person company over the
freelancer a little, though memoQ disagrees).
The other familiar theme reflects a
fairly good sense of what their customers are looking for (though I have
to say that many of their competitors have this understanding as well).
This explains why there is sometimes an emphasis on features that seem
relatively insignificant to the outsider. ("Just like with
Google Translate, it's now possible to exclude certain segments from automatic translation by
Microsoft Translator." Hmm, I would not really classify that as particularly important, nor would I emphasize those fixed Norwegian problems with
Tilde Translator -- but, hey, some users will be pleased.)
In my view, the two main emphases of
this release are the ability to batch manage (probably more like:
"batch-tame out-of-control") resources and the ongoing work to find some
kind of balance between the desktop interface and the web interface.
And at least the former of the two shows that
memoQ is a tool that has been around for a while -- why otherwise would you have to deal with too many resources.
In this release, it is now possible
to analyze the usage history of resources (termbases, translation
and manage them accordingly by deleting, exporting, merging, etc. This
is all done via an elegantly implemented and overhauled
Resource Console. While I think it's a very useful addition to
I would love to have seen an added feature here: a semantic analysis
that would determine how useful which resources are for which project,
or how un-useful for any project ("zombie resources" in memoQ speak), by
looking at the content rather than only the metadata such as date
We had a nice little side
conversation about a kind of resource that used to be quite popular
awhile back, the so-called "big mama TM," a TM containing everything a
translator processed, regardless of subject matter, client, etc. In my
mind this was a good technology for a time when we looked at TMs only as
repositories for perfect and fuzzy matches (which meant that 90+% of
all content was used exactly once and then never again), but with
today's more sophisticated ways of using TM content it has lost its
usefulness and sometimes can be more harmful than helpful. So I was
surprised to hear that, in Veronika and Zsolt's estimation, this is
still a technique some (albeit a smaller number of users) consider
I have written about
strategy of ensuring feature parity between their browser-based and
desktop interfaces by first developing it for the (more
difficult-to-develop-for) browser and then implementing it for the
desktop. I think that's really smart -- if you want parity, that is.
However, I now see other traditionally desktop-based tools doing it a
little differently and maybe just as smartly. In the last
Tool Box Journal, I wrote about
and how its makers view these two interfaces as complementary and yet
fundamentally different. I think that's how memoQ sees this as well. As a
result, there are some features in the 9.4 release, such as the ability
to change delivery dates, that are implanted only within the browser
environment rather than the desktop -- because they seem more helpful
for the kind of job that is being done there or the kind of user who
An across-all-users estimated 10-20% of time working with
memoQ is spent in
with a definite upward trend, especially (and interestingly) in North
America. (Methinks that even 20% still seems low, but that's probably
because of my young age -- no, wait!)
I asked Brúnó (at some point a
Hungarian speaker will have to explain to me where and what to emphasize
in that accent scenario) about how business has developed during the
crisis. Here are some things he shared:
"It's been uneven. Some months are
actually better than last year (March, April, July), others worse
(February, May) or about the same (June). What's great is that there are
key areas that seem to be unscathed by the pandemic.
memoQ cloud is
thriving (30% y/y growth vs. last year same period) and we're going
strong in the US (+30% y/y growth) and Japan too (+10 y/y growth). The
freelance translator segment is surprisingly resilient (+3% growth in
sales vs. last year same period). Most of the disruptions seem to be in
the form of delays affecting larger deals. Our clients seem to be more
open to SaaS and operational expenditure type costs (e.g.,
or server subscriptions) when external conditions are uncertain while
all of the delays seem to be around perpetual license deals that are
capital expenditure costs."
I think that's interesting and certainly a lot rosier than I would have expected.
Process management and control, high cost transparency,
networked teamwork, higher quality – a business solution.
Corporate Language Management powered by STAR CLM.
this short video
for more information on STAR CLM functionality and usage.
is a term that's been around for 20 years or so. Originally it was used
mainly for things like flight or drive simulators, but in the last
couple of years it has become the term of choice for translation
technology providers, especially when used in relation to self-learning
machine translation. Lilt, Lionbridge, and Gengo started to use it in
2018, Transifex, DeepL, and Unbabel in 2019, and I'm sure there are many
others I've overlooked.
I'm not sure what image you conjure
up when you hear that term. To me it sounds like the human/translator
being subjected to the machine (which provides content to correct) or
else at some parity with the machine. Neither scenario is one that
sounds particularly attractive or true to me. So I was thrilled to
use the term "machine-in-the-loop," which sounds a lot better to my
ear. I reached out and asked him about the source of the term, and he
referred me to the
introduction of a book
by Sharon O'Brien and Owen Conlan where it says:
"One term that has been used to
describe translator interaction with MT is 'Human-in-the-Loop' automated
translation, but this places the human translator in a peripheral
position in relation to the machine process and reduces her to the role
of verifying or fixing the machine errors. A more desirable description
from the translator's perspective might be 'Machine-in-the-Loop'
translation, where the human agent is served by the machine, and not the
other way around."
SDL Trados Studio 2021: Save up to €260 with our July Super Sale!
Get 30% off new licenses, upgrades and pre-orders of
SDL Trados Studio 2019 AND get Studio 2021 FREE on release. PLUS get access to a FREE Studio 2021 training session!
MultiLingual Summer Series
The beloved magazine
was recently purchased by Renato Beninatto and Tucker Johnson. They are
also the owners of consulting firm Nimdzi but have vowed to keep the
two companies as separate entities with little or no cross-business
They are also adding some services to the MultiLingual portfolio, including the
MultiLingual Summer Series: Meaningful Conversations with Thought Leaders
The first of these four sessions will be conducted on Thursday, August 6
(on machine translation); the second session, on August 13, will deal
with diversification. I will host the second session, and I'm honored to
be joined by
, Mila Golovine of
We'll be talking about how both freelance translators and LSPs can
become stronger by diversifying their service offering -- a trend that
many of you will be either thankfully or painfully aware of during this
time of crisis. I'm genuinely excited for this event, and I think that
you will enjoy it, too. You can register for free
We are halfway through 2020, and you
still have time to turn this year into the greatest translation year
ever! Manage your translation projects effectively, count words in 70
file formats, and earn more with smart software!
For 3 days only, save 50% on AnyCount, TO3000, Projetex, AceProof, and Exactspent!
The Last Word on the Tool Box Journal
If you would like to promote this
electronic 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
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
If you are interested in reprinting
one of the articles in this journal for promotional purposes, please
contact me for information about pricing.
© 2020 International Writers' Group