CATs, TEnT and all that jazz

You are currently browsing the archive for the CATs, TEnT and all that jazz category.

Yulia Akhulkova, ITI Lzd., Moscow, Russian translation company

Back at the time when IBM punched cards and magnetic tapes started giving way to floppies, Russian software developers used to joke about the most typical signs of a Russian software program. Typically, it refused to run if the programmer was not around.

Those days are long gone. If failing to run at that time could mean eliminating any computer security problems from the very start, things have certainly changed now. Kaspersky Lab, a Russian multi-national software giant seems to have found a less radical, but far more profitable way to deal with security issues.

Since Mayor Michael Bloomberg announced his New Year’s resolution to take a free coding class in 2012, everyone has been speaking about programming. But Russia, I believe, belongs to those countries to which the leading American corporations have been outsourcing their software development projects since years, along with other offshores like India, Bulgaria or Moldavia (where programmers probably also speak Russian among themselves).

Recently, I was experimenting with terminology management tools like Interplex (a must-have for conference interpreters with iPads) and QA programs like the new Beta version of ApSIC Xbench (I think Verifika is also worth a try, at least for its support of MemoQ 6 format). So it came in handy to learn and test a software program from Russia called MultiQA, “a complete web-based terminology management and quality assurance solution for localization and translation projects”, as the website claims. I don’t know if the system admin somewhere in Moscow is always around, but this program does run smoothly indeed, no worries about that.

MultiQA was developed by the Moscow-based translation company ITI Ltd. first for its own needs, but now it is underway to be used by and offered (as SaaS) to many other language service providers.

Term management and online glossary software MultiQA

MultiQA is an online terminology management tool with quality assurance functions. It can be used to compile bilingual or multilingual glossaries or import an existing term base (as an Excel file, i.e. XLS, XLSX, Excel 2003 XML, CSV or TBX, it is also possible to export MultiQA glossaries in Excel-based, TBX and MultiTerm XML formats).

According to a recent survey by Joanna Gough from the University of Surrey (“From translator’s black box to translator’s tool box: Tools and Resources for Translation Professionals”, November 2012), the most frequently used terminology management tools are SDL Multiterm (14,7%), ApSIC Xbench (4,5%) and TermStar (3,8% of some 600 respondents). MultiQA is probably the least known software for online glossaries and terminology management (it wasn’t included in the survey), but it certainly doesn’t deserve to remain obscure. It doesn’t only run smoothly, but has a well-thought-out, clear GUI and immaculate documentation (user guide, FAQ, workflows and use cases etc.) in English.

My idea was to test its features and functions within my own workflow, e.g. try to use MultiQA with an existing term base generated with MemoQ. I found out that a CSV format with a limited number of fields (source, target and status as default fields) works best for importing a term base in MultiQA. The question is why I should do it? What are the advantages of leaving a comprehensive translation environment like MemoQ with term management and QA functions for a standalone web-based tool?

In my opinion, MultiQA has at least two specific features which make it stand apart from other functionally similar language tools and justify its use on its own merits. The first feature is a set of assignable attributes to define a term status:

Term management with MultiQA - for multiple translator teams

Considering that available user-generated glossaries are becoming increasingly “overcrowded” and ever more “clouded”, a certain restrictiveness might be a benefit. It helps to sort out the mess which you occasionally find when looking up a term in e.g. Multitran.ru (currently the most popular web English-Russian and German-Russian dictionary) or dict.cc. There are simply too many options. They cannot help but produce the feeling of Qual der Wahl (agony of choice, like the Germans say).

On the other hand, the possibility to selectively approve or prohibit use of certain terms in the target language provides a consistent, client-oriented approach.

May we call it censorship, in the extreme case? Paired with a hierarchical user administration, the assignment of term status (as well as some other functions like “glossary freeze”) makes MultiQA especially suitable for project managers who have to supervise external translator teams. Project-wise, especially when dealing with large projects (technical or software documentation with repetitive phrases and words), it may be well justified to impose rigid rules and restrictions on the terminology used. So the answer is “it depends”.

The second specific, if not downright unique, feature is called “parsing”. This picture from Grammarly.com may give an idea of what “parsing” is about:

Grammarly.com picture - English and German definite articles

Words in languages like Russian, other Slavic and Baltic languages, have a plethora of forms and vary according to the language-specific morphological rules. In German, you have several forms for the definite article only, but in Russian you have declinable forms for most words (at least you don’t have any articles, if it may serve as a consolation).

According to Anton Soldatov from ITI Ltd., a specialist in Norwegian and the architect of MultiQA software project, the “parsing” is a truly unique feature of this terminology management tool. MultiQA automatically generates the correct morphological forms based on language-specific rules and helps to eliminate false positives (“noise”), inevitable with other quality assurance (QA) tools.

For those who are curious about the linguistic basis of “parsing”, I suggest you contact Anton Soldatov,  head of IT department, or Yulia Akhulkova from ITI Ltd. (the lovely lady from the photo above, in case you were also curious, is head of localization department at ITI Ltd.). Anton and Yulia will be happy to help with everything related to MultiQA.

I found the software quite nice and usable. Certain things are going to be improved or changed. TermCheck, the key quality assurance feature, as the user guide claims, clearly belongs to such things (currently, TermCheck works like a black box that automatically generates and sends a QA report per email).

For many potential users of MultiQA, it will be necessary to deal with security concerns and data confidentiality issues, especially in view of a potential conflict of interest. Storing proprietary terminology databases on a server controlled by a Moscow-based translation agency can turn out to be a much more grave issue than usability and the technical stuff.

For ITI, it means more work to get done. But “success comes before work only in the dictionary”, as the late Vidal Sassoon said. The Russian developers of MultiQA are ambitious to make their “dictionary” a success. I wish them luck.

Across - software - CAT tool

The ever-popular genre of consumer guides and reports is ever more popular before Christmas. The holiday season is also a favorite time for best-of lists of all kinds. What is true of vacuum cleaners and ebook readers is also applicable, to a certain extent, to the translation industry. There are no direct comparisons of languages, services, translators and other language service providers (perhaps it were time), but there appear, once in a while, lists of software tools. I remember a special edition of BDÜ’s (German translators association) magazine or various comparison charts of translation and localisation programs like an euphonious tabella comparativa di strumenti CAT by Marco Cevoli. As complex and detailed such confronti can be, there still has been something missing in the software reviews. As a technical translator I am used to the routine structure of a handbook. One of the first chapters of a typical operating manual in German is called “Sachgemäße Verwendung”, i.e. Intended Use or, sometimes, Purpose. This is the particular criterion which I have been missing when dealing with available reviews of CAT tools.

It may seem that the intended use of any CAT or TEnT (translation environment tool) software is obvious, in any case such programs don’t differ much in purpose and intent. However, there is a particular program which stands out specifically in terms of its purpose and target group. I am talking about Across, or rather Across as a “Free CAT Tool for Freelance Translators”. This is a translation program known for both being heavily advertised by its manufacturer, Across Systems (“manufacturer of the Across Language Server, a market-leading software platform for all corporate language resources and translation processes”, according to its own statement). and just as regularly and heavily bashed by fellow translators with practical experience of using it. A couple of weeks ago Jerzy Czopik, one of the most respected colleagues in our international community and a well-known authority in CAT tools, gave Across short shrift in his statement: “Keine Across-Jobs mehr – bitte alle mitmachen!” (no more Across jobs – join in everyone!). In my opinion, this call is long overdue.

The seemingly obvious purpose of Across, that is to serve as a CAT tool, is a sham. Across CAT tool is a means for a higher purpose. It is a management tool, specifically intended for use by project managers of translation agencies to better control individual suppliers. The perfidious nature of Across lies in its “free” Personal Edition for freelance translators. Due to its infamous and obnoxious lack of interoperability, Across as a stand-alone application is simply of no use. And, “as a client for accessing customers’ Across servers” (its second purpose according to the manufacturer’s statement), its intended use is to deprive the freelancer of both his/her freedom and his/her “lance”. Working on a customer’s Across server, an individual translator is expropriated of translation, translation memory, labor and skills. Working as a freelance translator with Across “Free CAT Tool for Freelance Tranlsators”, you give up your freedom.

This may sound too harsh, but I am saying it from experience. About a year ago I was coaxed by a translation agency to “give it a try” and take up a “review job”, i.e. editing a third party translation, with Across. The translation agency seemed decent enough, some faces on the website even looked like they were not just pictures bought at Fotolia or from any other royalty-free stock photography site. However, the personnel turnover gave alarm signals all over the place. I lost count of how many project managers and contact persons came and went within a rather short span of time. My jobs were getting ever more sophisticated, serious translations for corporate and scientific publications where both subject and style matter. Translations that require special knowledge and outstanding writing skills. On the other hand, I was increasingly requested to “review” (in this case, just estimate and tell my opinion about) other translators. Apparently, the agency was running low both on project managers and translation “vendors”. In line with the grand scheme of changes, the agency switched to Across and started applying serious pressure on its then loyal suppliers. In so doing, it obviously (logically and fittingly, I would add today) lost its most skilled and experienced translators, but was left with better (?) prospects ahead. That is how come I was asked to edit a third party Russian translation and give Across Language Solutions a chance. So I did.

Suffice to say, I was appalled at both the cumbersomeness and inefficiency of the software, and the mediocrity of translation it produces. The third party Russian translation I was asked to edit was pathetically mediocre, not much different from other Russian (sample) translations which I previously reviewed for this agency. Only in this case, its mediocrity didn’t result only from poor skills and inexperience of the unfortunate Russian translators alone, but was also due to the limiting systemic nature of Across CAT Tool with its restrictive segmentation rules and counterproductive logic. Fighting against software flaws and crashes can drive anybody mad, so those who “persevere” are either those who have stopped to care about quality and consistency long ago or some poor buggers who just have no other choice (beggars cannot be choosers, but they do choose their way to serfdom, “choosing” Across). Perhaps, the so called Quality Assurance module of Across would more or less serve its purpose when translating lists of spare parts with long numbers, but for well-written, quality technical, corporate or medical documentation it is not a quality assurance tool, it is a quality deterioration assurance tool. All in all, working with Across feels like a fight which you could never win. Working with (against) this unbeatable program is a tremendous waste of time, the only dubious consolation being no results left on your side – everything goes to your client and stays on your client’s server. (If the worst comes to the worst, I am afraid it would be a problem even to prove that you did the work altogether. But if you already agreed to give up your freedom, nobody owns you nothing, so it is only to be expected, isn’t it?)

Apart from quality, this software also stifles your productivity. On his Facebook page, Jerzy Czopik wrote: “The software is simply there to make us all slaves… And on top of this: due to the famous performance of Across I have spent one hour to translate two CWU with a 127 untranslated words in total. Having this very high output I will certainly be very rich in my fourth or fifth life, IF I ever be able to get the current one done…”.

I remember a discussion of Across on a translation forum where a colleague compared switching to Across from Trados or any other CAT tool with changing from driving a Mercedes to driving a Trabant (a famous racing car from the German Democratic Republic). But my real issue with Across is not that much about technology, performance or productivity, it is rather of ideological nature. Kevin Lossner in a recent interview said: “Across is the worst offender I know of… – their strategy of marketing incompatibility as a corporate asset disgusts me. Like the Hotel California of translation… arrival isn’t a problem, but checking out can be an issue” (see People who rock the industry). It is pretty much the same way I feel about Across, although my issue is more about – well, let us call a spade a spade – honesty, quality and freedom. Touting for a “free” Personal Edition is like advertising free cheese in the mousetrap, to paraphrase a Russian saying. “Accelerating the workflow” out of the mouth of a project manager is an unscrupulous excuse for delivering “Masse statt Klasse” (quantity, not quality), like the Germans say, and an euphemism for manipulating and enslaving those poor buggers who deserve it for falling for such an obvious mousetrap cheese.

My editing job on an Across Language Server took me many times longer than with or without any other tool. The “unfortunate Russian translator” turned out to be a translation agency (!) from Belarus with an outspoken spammy website. I am not sure about the geography of other translation providers for my agency client, but for me, editing a Russian translation from a Belarusian language service provider on the German company’s Across server felt like an online approximation of a trip to North Korea. Considering the established workflow when working with Across tools and a drain of experienced translators for other languages, I presume my experience is by far not unique. Needless to say, I decided it was the last job, translating or editing, I would do with Across. Soon afterwards I also severed my business relationship with this company. I never came to regret this decision. If translation resellers build their relationship with individual translators on Across, it is not an insignificant technical change. It is a grave, no-nonsense development prioritizing on the short-term comfort of the reseller’s project managers to the detriment of translation quality and the relationship with the suppliers. It is a short-sighted, unsustainable strategy because the reseller will inevitably lose its base.

Now, a year later, I have no idea how this company fares. Considering the personnel turnover, another generation of project managers is probably underway (knowing this company I can imagine how they speak about “dynamic expansion”, “more personnel”, “dozens of highly qualified project managers” and “thousand of experienced translators” – for all possible languages). I don’t know about their figures, but the emails which I still receive from this company with a certain regularity are automatically generated messages for anonymous “dear translators” to inform them of the “temporary unavailability of our Across Server”. Apparently, once you left your footprint at Across Systems, you are not easily forgotten.

Speaking out of experience of working with Across CAT tool, I think I understand Jerzy very well. Start to consider a job with Across (“Free CAT Tool for Freelance Tranlsators”) and you arrive at a crossroads. It is not a technical matter or a question of personal preference and taste. Across, in my opinion, is a true qualifier. If it is a translation agency which insists on you using or trying Across “for free”, chances are, it is this sort of an agency I had the pleasure to deal with. If you are a serious, quality-driven translator, just don’t waste your time: a polite decline would probably work best. If you are a budding translator who never worked with any CAT tools and want to try some free version, there are MemSource Editor or OmegaT which I heard good things about. If you are determined to persevere, consider MemoQ or SDL. But advertising your skills and experience as well as mentioning Across among your available tools is like sending an ambiguous signal. Inadvertently or not, you are making yourself available exclusively for translation resellers which, with a little help from Across, are going to use and dispose of you as they please.

But let’s not over-dramatize things. After all, I started this post with talking about harmless consumer reports in the holiday season. We have been told again and again that tools are tools and it all depends on the way they are used. Tools”R”Us and I would be the last to cry over steaming boats which destroy our handicraft treasures. But, speaking elliptically, tools are tools, however, there are tools and tools. This post is about a tool intended for a purpose which – if we are honest and serious translators – is definitely incompatible with our business. Everything is a two-way street and, whether you are a freelance translator, or a translation agency dependent on other translators, building your business on Across language tools is like building your business on a dangerous terrain. Across gives a straightforward sign for it. Ignore at your peril.

The picture above is by my daughter. I had little motivation to rummage through free stock photography sites and look for pictures with tags like “no entry”, “wrong direction”, etc. A tiny, quiet street where I live does have a one-way street sign, but I didn’t feel like making a photo. I finished this post soon after Jerzy Czopik published his “No Across” statement, but my daughter needed more time. Considering the tight school schedule, piano lessons, dancing classes and all the distractions of the Christmas season, it is okay. It is certainly not the best picture my daughter has ever drawn (plus my photoshopping the colors a bit), but the subject of this review isn’t pretty either. If you ask me, this CAT tool is simply the worst translation software I ever got to know. Thank you, Jerzy, and all for mincing no words. It was time to raise your voice.

Zweisprachiger Text auf dokumenta in Kassel - Aus der Sicht eines Übersetzers

Some of the buttons I really enjoy clicking on when working with MemoQ are Split Segment (Ctrl+T) and Join Segments (Ctrl+J):

Segmentierung in MemoQ.

In fact, it was a revelation to know that you can have a – Divide et impera! – break from a rigid segmentation structure of the source text instead of blindly following it in your output (as in older Trados versions). An often overlooked drawback of CAT tools is, to quote my colleague Gabriele Zöttl: “Übersetzungsprogramme zwingen den Übersetzer, ein Segment nach dem anderen genau in der Reihenfolge des Ausgangstextes zu übersetzen. Bei den meisten Texten, die mir auf den Schreibtisch kommen, führt diese Vorgehensweise zu einem höchst unbefriedigenden Ergebnis.” Since each language has its own rules and habits, chopping up the text into the original segments in the usual TagEditor manner (“Dateien im proprietären, in Segmente zerhackstückten Trados-Format”) is a mark of a seriously flawed translation approach.

It is also great to know the source text can be altered, too, from within the programme, with Edit Source (F2):

Taste zum Editieren des Quelltexts in MemoQ - Übersetzungssoftware (CAT).

As somebody who translates many but reads very few, if at all, manuals, I owe this knowledge to Kevin Lossner with his tips on fixing source segmentation and editing source text. I wonder if he actually plans to compile his brilliant MemoQuickies into a sort of hands-on tutorial, as there seems to be a niche still unfilled (no “CATs for Dummies” so far).

As for segmentation, I wonder even more if this Mut zur Lücke (German: courage for the gap) in the bilingual text on the picture above could be defined as courage to LEAVE gaps or CLOSE gaps:

Segmentierung des bilingualen Texts mit einer fehlenden englischen Zeile

It certainly depends on the current state of political correctness (or your views on it), but much more on the translation DIRECTION. As a casual visitor to this year’s documenta (13) in Kassel, I have no idea whether it was a translator from German into English who had to break the segmentation in this otherwise perfectly aligned text or a Übersetzer(in) following the rules of German political correctness. I wish it was the former, but suspect it is the latter. Also, makes the translation longer, if you charge by the TARGET words…

« Older entries § Newer entries »