Translation tools

ABBYY Lingvo.Pro

2010-11-01T14:31:00.002+06:00

The portal beta has already been available (through invitations) for quite a long time, but I didn't have time to write about it. Actually there is not too much to write. At the moment it is a convenient place to search on-line dictionaries and parallel texts — at least for English-Russian language pair.

Basically what guys from ABBYY did was collect all popular concepts and needs and create a portal that would embrace and address them all. The idea is likely to turn successful, but the implementation also matters.

So, currently it is an on-line interface to the engine that allows you to search ABBYY Lingvo dictionaries, phrases and translation memories that ABBYY created from publicly available parallel texts. For English-Russian pair, there are about 5 million words in the memories which mean you can see a lot of use cases for your word or phrase. I've always wished to do concordance search without opening Trados and loading memories — and my wish came true. Moreover, users can upload their own translation memories (deciding if they want to use them privately or share with other users) and search them as well.

ABBYY also promised to add some "artificial intelligence" to let it combine different existing translation memory units into a translation that never existed in the memory. The idea is not new, but to my knowledge it is still not as widespread as the concept of fuzzy matching, for instance. However, I believe in ABBYY and hope their implementation of this functionality will be effective and useful. Eager to see it.

The portal also lets you add your translation of words and phrases to the dictionaries — something that we love about Multitran, something that makes any portal live and dynamic and thus more efficient. The question here is if ABBYY is going to check what users enter or not and which dictionaries such input will go to. At the moment, what I add goes directly to the dictionaries, and the portal provides my e-mail address as a reference with the entry (not sure if it shows my e-mail to all the users, but if yes, I definitely don't like it).

Another nice little thing our translators loved is that Wikipedia page for the search term is just one click away from the portal, so you can quickly get an idea of what it is if no translation is found.

As we touched Multitran a few lines above, I must admit I like its interface more. It's very basic and text-based; however, when you work with dictionaries you hardly need graphics. At least this is true for me. All you need is fast response and the more results fit the screen the better, preferably no additional mouse clicks. Unfortunately for me, this is different with Lingvo.Pro. Although they have just a few graphics, loading time compared to Multitran is longer, there is less text in the screen and more clicks are needed to see everything you want. And what I personally hate about Lingvo.Pro is losing the focus. When I work and return to Multitran window by Alt-Tab or by a mouse click, the focus is normally in the field where I enter my words. With Lingvo.Pro, I always need to first click in the text box (and manually delete what is already there from my previous searches). Please, guys, do something about it, it will speed up my work at least twice, I think! And waiting for more exciting features!

XBench adds spell-checking plug-in

2010-10-26T14:02:00.002+07:00

A couple of weeks ago ApSIC released a new plug-in for their XBench. Based on HunSpell dictionaries, it checks spelling for a number of languages, including Russian. I briefly tested it and certainly found both advantages and drawbacks which I describe below.

The plug-in works in a separate window and allows you to confirm which spelling errors are real errors (in my test there were only two of them out of a couple of dozens of suspects), and those errors are then included to the list of all the errors which you can easily export and send, for instance, to a translator or to an editor.

It is apparently great that you can spell-check your project while doing other quality checks; it basically means that users can now collect really comprehensive reports on errors found in their bilingual documents.

To my mind, it is also the most convenient way to spell-check TagEditor files. I personally don't like TagEditor's built-in spell-check feature. What I normally do is copy the text from TagEditor, paste it to MS Word, delete hidden text, Trados tags and source text and then spell-check the rest. Because of those deletions, some words glue resulting to an increased number of spelling errors. XBench handles TagEditor files correctly, thus eliminating this problem.

However, there are a few quite shortcomings as well.

The most important one is that HunSpell dictionaries are not rich enough - at least for Russian. Therefore, the plug-in reports too many words that in fact do exist in Russian.

The other quite serious thing is that there's no suggestion feature and no hints where the error is. Sometimes it is not easy to find the error at a glimpse, especially if you have to handle a long list of different errors, and your eyes are tired. Sometimes there are even more complicated problems. In my practice, some clients correct Russian translations using non-Cyrillic letters that look like Cyrillic ones (i.e. Russian А and Latin A, Russian Р and English P etc.) It is a spelling error, but you can never recognize it visually, and this will result into keeping this misspelled word and possibly adding it to the dictionary (as soon as the plug-in supports this feature). To my knowledge, HunSpell project does support suggestions, so this is the implementation drawback. Hopefully the next version of the plug-in will support suggestions or at least hints.

All in all, the plug-in is already helpful despite that some functionality is really missing.

Important! There was a response from the developers: I really didn't notice that there is the suggestion feature. In order to use it, just put the mouse cursor over the misspelled word.

Translating Idioms

2010-08-26T14:03:00.000+07:00

It was just not long ago when I needed to translate quite an idiomatic text from Russian into English. My old friend Multitran is normally helpful in such cases, but it was nice to discover WikIdioms project. I learned about it from the recent issue of Multilingual magazine (fortunately or not, it was the most interesting post in this issue for me).

As far as I can guess, the project was started not long ago and uses Drupal site engine which is very sofisticated and powerful, but sometimes really strange. I tried to use search to find a Russian idiom, and the first thing I found out was that Enter key does not work in Search field. You necessarily need to click Search button to start search - at least in my poor IE8. Then I found out that its search algorithm is not easy to understand. I tried different popular Russian words, and it found nothing though I knew for sure that there are idioms containing these words in the database. I ended up searching for a simple word "а". Try to guess what was found?

Search results
◦ние за волкот, а волкот на врата
◦Сидеть с помытой шеей
◦Сидеть с помытой шеей

I have no idea why the last entry is listed twice, and what is much more important, I have no idea where is а in this last entry.

Anyway, I do hope the project is going to eveolve and become a nice and useful dictionary. At least I'm going to use it both for work and for learning other languages idioms, and I do hope to contribute, too.

Can I Ever Note?

2010-08-17T18:31:00.002+07:00

I was reading a blog post on a localization topic and was going to share some thoughts about it, but Evernote, the utility mentioned in the post, was so attractive that I first decided to check it out. By the way, I must admit that Russian localization of the site is pretty good. The only problem for me was that it is sort of haunting: I wanted the site to appear in English and I switched language wherever I could, but as soon as there was no choice it appeared in Russian again.

So, the idea is brilliant, I think. I have been looking for a single place to make notes for ages, and I do believe the implementation of the idea is also good. But 'I still haven't found what I'm looking for'. I just could not see how it works. Either I am unlucky or the site creators didn't account for poor users of Internet Explorer (yes, I know that it is a mauvais ton to use it, but believe me I have good reasons for that) - so, when tried to sign up, all I got was that:

I couldn't force it to provide the picture in English. The red text says: Please enter the characters you see on the picture above. Does anybody see any picture?

When I opened it in Opera, it showed not only the picture, but a control to hear the characters in case I don't see them. But I'm afraid I'd like to use a single Web browser like I'm looking for a single note solution. And why cannot we make sites that at least work (let them work differently, but just work, i.e. provide all functionality that is necessary to use the site) in all popular browsers?

PROMTMan

2010-08-16T14:48:00.004+07:00

He SUDDENLY appears next to you and quietly translates everything! ;) Try to recognize PROMT translation of popular songs!

This is about me.

If you want to check who you are, just follow the link. You will be given 18 PROMT-translated pieces from popular songs in English (and to increase your chances, you first select four preferred bands and the timeframe :)) and you will have to choose the right song out of three suggested options. If you succeed, the game will be reversed: you'll get three pieces of text and will need to guess which one belongs to the particular song, and 18 songs again. Then you'll get this prize picture like I did.

I guessed all 36 songs and thanks to people who got the idea of this wonderful contest: I had a lot of fun!

ABBYY FotoTranslate

2010-07-13T16:29:00.001+07:00

The application was recently mentioned somewhere in the news, and I really liked the idea to mix FineReader OCR system with Lingvo dictionaries and make it mobile, so that you can take pictures on our way and check what the words on the pictures mean. If you’re travelling and not speaking languages well enough, it may help you get your bearings in some situations. I recalled that a year ago I had to park a car in the streets of Vienna and as my German was quite poor I really could not understand what all different parking signs say. What I did was sending SMS with the sign text to a friend of mine who spoke German and she translated. What I did when I came back home was taking a course of German :)

However, FotoTranslate also has quite serious limitations:

it works only on Symbian (i.e. Nokia smartphones),
translates only into Russian and
only from English, German, French, Spanish and Italian.

First off, those languages are relatively common, quite many people speak them to some extent and many people are learning them. Secondly, you can easily get more flexibility by having Lingvo installed on your mobile phone (which is, by the way, not limited to only Symbian) and typing what you see on the sign or in a restaurant menu. Many mobile phones would allow you typing in those languages, and even if you don’t have some language-specific characters, Lingvo is “smart” enough, so you will be able to get the translation. And last, OCR technology works not good enough in some circumstances which may result into getting a wrong translation. So, translating via taking pictures seems to be fun – but it is neither faster nor more accurate. Apparently it would make much more sense with more “exotic” languages, especially those Russian people normally cannot type – e.g. Chinese, Japanese, Thai, Arabic, Georgian, Armenian or Hebrew.

These were my thoughts after reading the application description. And as I have Nokia E75, I immediately downloaded and installed it and then played with it for a few minutes. My value judgments are as follows:

I think the installation was sort of long, but this indeed may depend on the phone, its characteristics and its condition.
It took me some time to get the camera work. While it was making quite good pictures using the built-in camera application, FotoTranslate kept saying that the camera lens may be obstructed or that the object is out of focus. Sometime later, something happened, and the camera began working normally.
The recognition technology is pretty good, though I tried to challenge it by taking pictures of a text on the computer screen, colored glossy magazine cover and a text printed on a non-flat surface (namely on a mug).
Dictionaries are also good enough and provide a lot of meanings of all recognized words – pretty much like Lingvo.
I will delete the application as soon as I finish this article for the reasons mentioned above – I just don’t need it. But it may be really good and useful for certain people in certain circumstances. It can also be a real fun for children, but I don’t know any children who would have Nokia smarthpones.

Additionally, as far as I read in one of ABBYY forums, a version for iPhone is expected in August 2010.

Escape from salespeople

2010-06-24T16:38:00.001+07:00

I’m not sure if I’m unique… I‘m even quite sure I belong to the majority. I don’t like salespeople. Sometimes I even hate them. My attitude towards them is (not surprisingly) directly proportional to their technical knowledge (or at least the visibility of that knowledge) and inversely proportional to their persistence.

Professionally, I often deal with people selling software, and I would place almost all of them into the following three categories:

1. “Not technical guys”. Their presentations are voluble and facile; they apparently describe the best software that ever existed on Earth. But… “I’m not a technical guy, I’m just a salesperson, I can put you in contact with the right people…” – it’s their answer to any in-depth questions.

I really believe you have to be a good salesperson to sell you-don’t-know-what. I don’t see they’d have a goal. Even if they seem to have the goal, it’s talking, not selling. And I‘m certainly afraid that a company that did not invested into training salespeople properly may also have not invested enough into software development and/or support.

Whenever I start talking to salespeople, I do know what kind of software they sell – I simply come to them because I’m interested in their particular application, not because “I liked the name of your company and decided to find out what are you guys doing”. Moreover, I normally know at least 2 of their competitors and would like to compare. I don’t need this generic info; I need the facts, benchmarks, differentiators, advantages, plans and their vision of the future. Am I really so unique in that? Is general information really what the rest of people seek? No, I don’t think so.

2. “Generally speakers”. They are often not so brilliant speakers, and this is mainly because they don’t need to learn by heart what to say. They know what they sell, but again they describe the main functionality for the broad audience. However, I don’t care of broad audience, I know what their software is about and I want to know what the difference between it and XXX’s system is. But if I ask this question, I get a generic reply – additional highlight of what they already said. I think this is because they are really committed to their software – so much committed that they just didn’t care to benchmark it against the competitors. They do like many little things about their application and will tell you everything about them – but they don’t have an idea if their competitor’s software has the same or more interesting things. They may suggest sending you more information via e-mail, but the outcome is that YOU have to be insistent to get the info you need. Even if they sell wonderful software, they have a good chance to lose lots of potential buyers.

3. “Ubiquitous”. They are ready to answer all your questions. They’re in fact eagerly waiting for any opportunity to talk to you. You make them happy showing your interest, and they are everywhere ready to serve you. They naturally follow you, so you eventually get tired and feel uncomfortable. They give you lots of arguments in favor of their software, but rarely listen to your “contras”. They have a classification of possible "contras" and ways to use them as a start to talk about "pros" which you're not interested in. They have answers to all your questions, but these are standard “yes, but…” answers you don’t need. And when you discover your only wish is to get rid of them they already know your schedule and your favorite trails and they are waiting for you on the right time at the right place.

And luckily there are exceptions. There are salespeople who know what they sell, try to find out what you need and have the most accurate answers to your questions, who are exactly where you need them and when you need them. They listen carefully to what you say and make sure they understood you correctly. They also make sure you understood what they wanted you to know. They know how to guide you step by step to the point where you fall in love with their software. They don’t know if you’re going to buy it eventually, but they’re pretty sure you’d remember it and them as salespeople. And if they eventually realize they sell not what you need, they won’t insist, but the software after all will be step by step updated towards your needs. This is so simple, isn’t it? Why is it so rare that I meet salespeople of that sort?

And another little thing that I like about salespeople is arguments like “You’d rather buy today or tomorrow it’ll get…”

Come on, do you really sell a one-day thing? Is it really important for you that they buy today or is it important that they just buy? Do you really think they will buy if you push them? Some of them may, but do you really think they’re going to be a good referral in the future? Or is today sale more important for you than a dozen in a few months? Do you really mean to tell them they’re not going to get what they need if they don’t buy something they don’t need first?

Segments in – segments out, or can translations sound natural?

2010-06-11T13:40:00.001+07:00

What are essential attributes of a successful translation project?

Anyone would mention client instructions, TM and glossary. The quality and success of a project is often measured as the degree to which those three were adhered. And indeed translators like when the TM is clean and the glossary is unambiguous, and they totally love when client instructions say: your translation should read as if it was originally written in your language. We are all artists, and we do appreciate when customers want us to show how good we are in our native language.

So, we enthusiastically launch our TM tool and start translating as if we’re copywriting the text anew in our native language. Nice picture, everyone’s happy.

I recently was asked to review a short translation. The client style guide was neat and short (does anyone like long style guides?) and said that the readers should not notice that the text is a translation. The instructions from the project manager were also pretty clear: please correct grammar and syntax errors, please don’t do any stylistic or preferential changes and make sure the translator followed the style guide and the glossary etc.

And the translation was good. It didn’t have any grammar or syntax errors, it followed the glossary, it was really a good piece of work. Except for one small thing. It didn’t follow the style guide. Anyone could tell a mile off that it was a translation. No Russian speaker would write it like this if only she or he would write in Russian from the very beginning.

Needless to say, the client wanted some particular examples of this “non-natural flow” to discuss them with the translator, and in the attempt to collect some I realized there is nothing to pass to the translator, and it is not an issue of this particular project at all. It’s just that a translation as it is being done now can never sound natural.

Look at what we use every day: translation memory, translation tools, machine translation… What we’re doing is always a translation. Period. All we’re left to do now is forget the sweet dreams about natural flow. As long as translators are in a segmented view of a translation tool they think in segments, too, and the text will always be a translation. That’s it. No matter how good we are as translators, as long as we translate segments we get segments. Original text was first written and then segmented, that’s the difference. If we take a picture, cut it into pieces, copy each piece and then glue them together we'll never get a picture as if it was originally painted as a whole. Lines will be cut.

Even if our client is extremely loyal and lets us change anything from 100% percent matches to text segmenting, we’re still in the translation environment and all we can produce is a translation. And it is not that bad in many cases. But whether this is good or bad is another story, and this one was just about unachievable “natural flow”.

TweetDeck's Unicode Support

2010-06-09T15:42:00.001+07:00

Nothing special from me today, just very quick notes.

I noticed yesterday that TweedDeck (isn't it the most often downloadable piece of software at the moment, at least in our industry?) was not showing Cyrillic, and therefore some posts from my friends made in Russian through the sites looked like " - - @ !"

Well, it could be easy to believe sort of just 5-10 years ago, but not now when we have Unicode everywhere. So, I searched TweetDeck's web site and indeed found an answer. Remarkably, the question was asked and then answered by the same person.

By the way, someone in the support thread complained about the inability to search by Cyrillic words - I think it was back in 2009 when the question was asked. My version of TweetDecks easily searches Russian words - at least after switching to international font.

So, if anyone still needs characters other than Latin in TweetDeck and still doesn't know how to turn them on, here's the tricky thing: just select Settings -> Colors/Fonts and select International Font/Twitter Key under TweetDeck Font - that's it.

Alchemy Catalyst 8.0 - Activation

2010-06-09T15:20:00.002+07:00

Today I fought against Alchemy Catalyst. As a TILP CLP, I've got a lifetime license from Alchemy, one of the most interesting takeaways from Level 1, I think.

I installed it on my home computer because I upgrade my office one quite often and always forget to release licenses. So, I wanted to activate Catalyst today and used remote desktop connection to my home computer as usual. Though Catalyst wouldn't activate via terminal service.

I presume this is a sort of "protection" against sharing licenses or whatever, though I didn't like the idea. The only thing that I could not do via remote desktop was activating. Accessing and using the tool is still possible (and thanks for that at least!), but as I needed to use the software immediately I had to drive home to activate it and then back to the office to use it. I love to drive, but not when I'm under the deadline pressure...

Besides that, Catalyst 8.0 seems to be as nice as it was in the age of being Corel Catalyst - one of the first localization tools I tried. Hopefully I'll find something more to write about as I work with it.

Plug-ins for XBench

2010-06-07T15:51:00.003+07:00

A couple of weeks ago our engineers found that DataBridge, a Russian translation agency, developed a plug-in for XBench to check Russian translations. I certainly wanted to try it (and by the way, it is also free).

Additionally, it's been 7 months ago when plug-in functionality was added to XBench, and I've heard from a few people that they are developing for XBench. I decided to search the Internet to see if there are any other plug-ins on the market, but Google showed nothing.

So, DataBridge seems to be the first to provide publicly available extension to this quite popular tool. Unfortunately it generates as much "noise" (or false positives) as simple regular expression search. My understanding of a plug-in, however, is that it may (and shall) combine different regexes and other commands to implement complex search algorithms.

I'm not sure if XBench supports other approaches, but hard-coded criteria for Partial Translation and Incomplete Translation make those checks mostly useless. At least in my case what they generated were 100% false positives. For instance, I got an error every time I had a 10-character heading translated to a 7-character heading (or shorter).

Despite those issues, the initiative itself is great, and I guess the developers will focus on refining the search results and will translate their efforts into a powerful tool for Russian linguists.

As I put it in November, the next (and quite trendy indeed!) move for ApSIC, to my mind, would be to start XBench user's community to give a real boost to plug-in development and certainly to maintain a centralized list of such plug-ins.

And if anyone knows of other XBnech plug-ins and wants to share the info, please let me know!

[Simplified?] Russian Language Day

2010-06-06T13:46:00.003+07:00

UNO has recently assigned June 6 (the birthday of Russian poet Alexander Pushkin) to be the Day of Russian Language. And I really liked yesterday's Russian BBC post (in Russian) about it and about current language trends.

I think many people who love Russian language (and I'm sure translators are a part of them) recently have been concerned that the language is degrading. Objectively, the boost of unprofessional book publishing (which means both low-quality books and illiteral typesetting) in early 90th became one of the reasons why young people nowadays write and speak poorer Russian than was taught in ex-USSR; then the Internet arrived, and people learned a lot of americanisms and geeky slang (without learning English though), and developed lots of new "russified" words from English ones, and started writing "Olbanian" (garbled Russian, but while it's garbled for some people, others are growing up with it and it's quite natural to them) etc.

It's almost everyday I'm talking to my almost 6-year-old daughter about what she hears and says and what she should write - luckily she feels the difference, and being able to speak regular day-to-day Russian she is at the same time able to spell words correctly and build really complex descriptive sentences using quite a rich lexicon.

Anyway, the linguists BBC talked to are pretty sure that what we're seeing now is not really fast-changing language, and changes in Peter the Great times, for instance, were much more extensive. On the other hand, experts note that Russian is way too far from being popular in the computer world, and suppose that a new writer of a Leo Tolstoy's level would add to the language popularity. Don't know if one day we'll be looking for Russian-to-English literature translators for a "War and Peace"-like book, but at least I'm glad experts think my language is not under that serious mutation as I was afraid it is. However, I still worry as my 6-year-old seems to really speak the most "great and powerful" language among her friends, and I'm sort of afraid she's going to be a black sheep in the world of "simplified Russian"...

PROMT 9.0 and MT into Russian - a really short notice

2010-06-04T12:58:00.001+07:00

I was too late yesterday at GALA's PROMT 9.0 Demo webinar. I only could listen to Alex Yanyshevsky telling about last couple of features he was going to show.

I totally forgot about PROMT since their version 8. I do believe MT into Russian is unreal at the moment - at least if you don't have large high-quality and garbage-free TMs, and who has them? So, I use MT to get the basic idea of texts in other languages that I cannot understand. And I never translate into Russian, but into English.

Yesterday I could see a couple of examples of PROMT to translate the same text into Russian and into Portuguese. I'm not good at Portuguese, but the example was simple enough, so I could easily see that Portuguese translation at least makes sense. It really provided the basic idea of the source. Needless to guess: Russian did not. And "PROMT is a Russian company" - that's something customers say when they plan to use PROMT to do MT into Russian. Despite that, its out-of-the-box quality seem to not change compared to version 8.0, and is pretty much the same as Google's. Then why should I pay for PROMT? (Indeed, what I say is valid only for Russian, I do believe it provides good results for Spanish or French etc.)

I checked PROMT's web site briefly, and what made me interested was that their next version is going to be hybrid. Eagerly awaiting!

XBench again...

2010-06-02T19:40:00.001+07:00

This tool may seem to be my favoite, but that's not true. Unfortunately my favorite tool is not developed yet. But XBench is free and really very helpful in some aspects, so if I find something interesting in it I just cannot help sharing this.

I didn't use it for quite a long time, so ApSIC released build 396 back in February, and I only found this out last week. This build is claimed to load big files much faster and to have more enhancements.

Key findings:

1. It took XBench like 5 to 10 seconds to load my at least 100 MB of TagEditor files. I think it would take something like a minute in previous versions. So it's really much faster now.

2. I did not notice other enhancements - I believe this is because I don't use the involved features.

3. Where XBench is really great for me is at finding little yet important inconsistencies. If you translate similar texts on a daily basis you may just not notice a slight change in a model name or in a list of operation systems etc. and just apply a Trados match without changing it. If you load your files along with your TM to XBench and set both "Inconsistency in Target" and "Inconsistency in Source" checkboxes, it will find what you didn't notice.

4. There is another free utility named XBench which is a benchmarking tool for Macintosh, note this when googling for ApSIC XBench.

More about plug-ins for XBench in the next post.

Russian Rules

2010-06-01T11:32:00.007+07:00

If you ever hesitated about Russian spelling, grammar, punctuation etc. and spent dozens of so important minutes looking through hard copies of grammar references and other books, you will really appreciate what Ilya Birman from Chelyabinsk did. He prepared a web release of Russian grammar rules with quite powerful and flexible search capabilities.

If you speak Russian (and I believe you do if you’re interested in Russian spelling), I’d rather let Ilya himself to introduce his project:

Of course, I played with the site myself – unfortunately for a very limited period of time, – and found just a couple of smallish things that I was not happy with. First off, the site is apparently not optimized for Internet Explorer which I still use for some strange reasons (or, more exactly, without any particular reasons). It looks differently and not so nicely in IE, but what is much more important, I didn't notice any functionality problems. The second thing is that the search is "too intelligent". It works as you type and provides the results before you typed everything you wanted. This seemed slightly confusing for me, but I think it's something you can easily get adjusted to.

This is certainly not the first effort to make an interactive Russian grammar reference, but compared with probably the most popular gramota.ru, Ilya’s project is the first comprehensive attempt to index and codify them, and I'm really happy we have it now. Thanks, Ilya!

P.S. Additionally, Ilya's blog on Russian language is also quite interesting for those who love Russian.

The dogs bark, the caravan goes on

2010-05-24T16:34:00.001+07:00

Note. As far as I know, the closest English equivalent of the title is "the moon does not heed the barking of dogs", but I really didn't like it. I think something like "the moon shines while the dogs bark" would be more correct because the idea is that all parts continue doing what they were doing: the dogs continue barking, the caravan continues moving, and they actually have no relation to each other.

The idea to make service providers pay for the services they provide is indeed great. This is something that fascinated everyone in the translation community, and this is probably one of the most lively discussed topics at the moment.

What happened? L10NBridge announced they are going to work with their vendors only through their Translation Workspace. It's provided on software-as-a-service basis, and you have to pay a monthly fee for using it.

What is it for the vendors? Basically this is that you have to pay for the possibility to provide your services to LionBridge - something that Tom Sawyer invented loooooong ago, and I don't know if he was the first to do that.

What is it for LionBridge? Honestly, I don't think they just want to make extra money, though "you never can tell about bees", especially if you're not a bee. I think their main idea is... to streamline, to increase efficiency, to differentiate - and all sorts of those concinnous words which mean doing things better than you normally did.

What is it for onlookers? All the vendors use lots of other software in order to provide their services - and they don't mind paying for all those pieces of software. All the vendors agree to pay for portals like ProZ etc. LionBridge offered a sort of combination, and the problem is that LionBridge is not a third party in this game. Will we one day see vendors force customers to pay for purchasing services exclusively from them? Why not?

What is going to happen? I'm certainly not the smartest person in the world, but neither the most stupid one. And I do believe LionBridge management is at least more or less the same. Doing what they did, they expected the vendors' reaction to be like this. They expected to make vendors unhappy, they expected many vendors to say good-bye to them, they expected "much ado about nothing". And they think they can handle it. They think they will be going on - and what is more important, they will be going on in the direction they chose. On the other hand, their vendors will go on as well, and their directions will be not bad, too. So, no matter how loud the dogs bark, both LionBridge's and vendors' caravans will be going on.

XBench is alive and kicking!

2009-11-18T14:42:00.007+06:00

After some time of silence, XBench development has become visibly active. On October 14, 2009 they released new version 2.8, and it took ApSIC less than a month to release its new build with hot fixes.

The tool, which was powerful enough for a tool available for free, is now even more helpful and useful. The most important things, to my mind, are regular expressions support and faster search engine, but this is not the whole list. XBench has always been the tool that supported the widest variety of file formats, but it expanded here as well having added a few more formats with the new release. Additionally, its checklists were enhanced and allow for better organization.

Last but not least, as of October 7, 2009, XBench allows you to develop plug-ins to it. Now you can implement your own types of checks as a DLL and make XBench run those checks together with ones that are built into it. All you need to do that, is to be familiar with software development and read this document carefully.

Well done, I believe. The market of QA tools has been quite live recently, and the fact that we have a powerful and evolving free tool is really pleasing.

And if ApSIC takes the next step and provide support for a community of plug-in developers, I really believe we may expect XBench to be the most comprehensive QA tool on the market in a very short time.

Nanotechnologies in action!

2009-11-13T11:23:00.005+06:00

This post in fact is going to deal neither with nanotechnologies nor even with translation directly. However, I'm sooo surprised that I just feel I have to tell the entire world about it immediately.

I need to apply for a new foreign passport as my current one "ran out of space". I traveled so much that there are no free pages for visas any longer though it is valid for two more years. I felt very reluctant about collecting all the necessary papers, calling the officials to make sure what and when and how... and I naturally pushed myself to at least search for the necessary phone number in online yellow pages. And there was a web link under all the phone numbers, and I decided to check it.

What a surprise! They have all necessary information there! It's quite nicely categorized, to my mind, and includes everything you need to deal with passports, migration etc. Open hours, lists of necessary documents, rates, document templates - everything is there. And what really killed me - you can make an appointment with the head of Federal Migration Service in Tomsk via the site, and the appointment will be using Skype!

I can bet almost 100% of Russian citizens (at least outside Moscow) could not even think a provincial government agency may be so "electronically advanced". What they still miss is accepting the documents via e-mail or FTP, sending the passports back via DHL and... site translation into at least English :)

ABBYY Lingvo х3: Kazakh version

2009-07-02T12:45:00.005+07:00

In May 2009, ABBYY released two Lingvo dictionaries with Kazakh: three- and multilingual versions. Three-language version supports Kazakh, Russian and English. Kazakh office of Palex, my company, tested the demo version of the three-language version, and here is what they found.

The demo version allowed to evaluate term search in the Internet, in Wikipedia and online dictionaries. Translation samples, synonyms and explanations are also a pleasing feature.

Explanations are what make the most dramatical difference compared to the electronic dictionaries we used before. At the same time, the dictionary includes new terms in economics and law domains, which is very relevant to the situation in Kazakhstan where it is a requirement to have all documentation translated into national language.

The main shortcoming of this dictionary is the lack of direct links between English and Kazakh. Despite of that, the dictionary is still a very useful resource for English-Kazakh translators.

As for the actual translations in the dictionary, some of them are erroneous. However, we didn't notice any typos like in other dictionaries. the translation is shipshape.

We believe that of the existing dictionaries with Kazakh support, ABBYY Lingvo х3 provides the most comprehensive range of features, and the quality of its translation is quite high.

ABBYY Lingvo vs Multitran: crowdsourcing in dictionaries

2009-06-29T08:42:00.006+07:00

Just another subjective opinion :)

I've always liked Lingvo. I'v been using this dictionary for ages, I like people at ABBYY, and Tatyana Danielyan, their Director of Project Management, was so kind to award me with the full version of their dictionary a year or more ago, and it was extremely helpful in my trips to non-English-speaking countries etc.

And I was occasionally using Multitran. As we still experience some network slowdowns, it was inconvenient to look up words on the Web; it took much longer than searching Lingvo.

However, this weekend I was translating a text from a completely new domain for me. Therefore, I had to check each and every term against a dictionary, and to make sure this is the term that is actually used in Russian. All of a sudden, I discovered that Multitran is much more useful from this point of view.

Multitran became interactive many years ago. Registered users may add their terms to it and expand to different domains, and due to this crowdsourcing it's really much more "live" and "up-to-date" than Lingvo is. Additionally, their style to present information proved to be more convenient for me, and because of that, even in times of network slowdown it took me less time to come up with a correct term in Multitran than in local Lingvo dictionary.

Both dictionaries lack Swedish, though, which I personally would like to have.

Google Translate

2009-06-28T12:25:00.005+07:00

Recently I started learning Swedish. And I immediately started interacting in Swedish. And to not waste my time for boring word-for-word translation with regular check of grammar rules, I tried Google Translate. It provided quite sensible translations from Swedish into English. Sometimes when I didn't have time to recheck what it translated for me, I sent out some stupid Swedish text, but in most cases I ran through the whole translation to learn from it and make sure it makes sense, and it was quite good. It really inspired me. Indeed, everyone is saying machine translation is already here doing things for us, but it's just another pair of shoes when you see it in practice and can easily use it.

So, I certainly wanted to test it with Russian. I didn't hope to get too much, but I've always believed that statistical machine translation is better for Russian than rule-based one, and with Google's huge text corpus, I was pretty sure that it should make at least some sense.

No, it didn't.

(to be continued, however, after some more tests with Russian...)

Tricky clean up in Trados

2009-06-25T11:11:00.004+07:00

Tring to solve our yesterday's problems, we spent quite a lot of time testing different things, and eventually came up with an interesting finding.

Cleanup settings do not depend on the language combination set for the TTX files to clean up. What it depends on, is the language combination of a TM you clean the files up to (even if you select not to update it).

So, if you just need to clean up some files translated into German and don't care which TM you have open, and it happens to be a Chinese or Japanese TM, you will receive your translation without any spaces between sentences. And just the other way round, if you clean up Chinese files to a dummy Italian TM you won't be able to get rid of the spaces.

So, just be careful when cleaning the files up!

SDL and its support again

2009-06-24T16:52:00.004+07:00

Yesterday we ran into a problem. Our main platform for translation at the moment is SDL Trados 2007. And we don't have premium support agreement because when we started using it there were no such agreements at all, and their support was very responsive and helpful (one of the main reasons why we opted for SDL). Then we haven't had any serious problems for many years, and even if we had some, we always had some opportunity to ask a question or to find an answer in their knowledge base (still open to registered customers, wow!).

So, eventually we ran into a problem. We apparently checked everything we could, ended up with that it is not any setting problem, and decided to contact SDL support, to show them where the problem is and to ask if it can be solved.

So, we went to http://www.sdl.com/en/support. It said: go to My Account, and everything is going to be fine. OK. My Account apparently directs me to My Support. Happy to do so. My Support offers me to review my Primary Services Agreement (which I apparently don't have) or go to SDL Trados Knowledge Base. OK again. Knowledge Base is good. But I don't find answers to my question there. I remember there was a link to ask a question. Not any longer. Or it's just so well hidden that I failed to find it.

It's not too often I disturb their support, but if I do it I really need it. Now there's no way. OK, we'll develop a solution; at least they use XML for the bilingual files, so we can easily parse them without Trados.

But I really like the trick. SDL lost their main advantage (for us, of course), but they translated it into another one, into becoming so deeply integrated into our and our clients' processes during the Good Support Era that it's really impossible just to switch to another tool. And on the other hand, we already developed a lot of proprietary utilities to do this and that in Trados files... maybe we can easily translate it into our own TM system? :)

Vison of a QA tool

2009-06-22T15:39:00.003+07:00

It's for about five years I have a vision of an ideal formal translation QA tool - at least for our needs. A tool that would spot errors correctly, determine different languages (which is so easy!) in a multilingual project and not report inconsistent translations when the same phrase is translated into different languages.

It would intelligently detect where a capital letter is required and where it is a false positive to report it etc. It would easily and correctly handle flexing languages when checking terminology. It would automatically not report untranslated segments like "Windows Vista". It would easily open any of numerous bilingual file formats, allow to make changes on the fly and once changed, automatically recheck if the error is in fact corrected. It would provide you with clean and intuitive reports where a click of a button allows to correct all the errors of the same type, make a note for the future that this error is highly possible again and advice on adding it to a checklist. It would accumulate statistics and provide recommendations to translators and QA specialists based on different criteria. It would automatically "pre-correct" errors in new projects based on its experience with previous ones.

To be continued or never published :)

SDL Trados Studio 2009, too much for 1 GB RAM?

2009-06-17T10:12:00.008+07:00

At Localization World Berlin a week ago, I non-intentionally stopped by SDL's booth. I don't really like to make an opinion on a product based on any demonstration, but only on my own experience in trying and testing it.

So, they've been showing there that new appearance and how it works in general (which is to my mind not interesting again because if you've been using translation environments tools for ages, you can easily imagine the basics of any new tool in this category).

When I was briefly beta-testing Trados Studio (and the tool is in fact really nice and convenient), I didn't see any preview. I didn't have time to really try to find access to it, but it was interesting to take a look at how this feature works. So, eventually I asked at the booth if the Studio supports preview and how. The guy certainly started demonstrating. He pressed a few buttons, and the preview started being generated. It took some noticeable time to generate it, and I asked if the computer running this demonstration was slow.

'Yes, it's quite an outdated one', the guy responded, 'It only has 1 GB of RAM'. Working currently with 1 GB of RAM and feeling not bad about it most of the time, I didn't say anything. However, I thought that not many potential users of this tool are likely to have faster computers.

Besides that, the preview works great, it updates every time you confirm a segment (which is in fact what MemoQ's preview has been doing at least for 1 year I've been testing it), and is in general a very good reference for a translator. What may be a bit inconvenient is that it is in a separate window, so you have to switch. But for those who use small screens, it's even a benefit, and I assume there is a way to include it to the single Trados Studio window if you have a larger screen (this is something I have to check though).

But what I'm eager to do now is to check the real speed of the tool running on a real average workstation which most of the translators use.

Regular expressions

2009-05-06T13:59:00.004+07:00

I believe many translators already know how useful regular expressions are. They're not easy, and I'm pretty sure not that many translators use them in practice, but they (regexes) do really great job.

But they're a pain as well. It considerable depends on the regex implementation in a CAT tool (or TEnT, which stands for Translator's Environment Tools and is pretty fashionable word now), and I so far dealt with two such implementations. One was in MemoQ, and what I really liked about it was the possibility to specify exceptions for each particular regex. I consider such implementations to be very flexible, but as far as I can see, they didn't built regexes into QA process where I mostly use them.

The other implementation is in Trados QA Checker 2.0 where I often use them. What I really liked about this implementation was that I asked the tool developer to support variables pass between source and target segments, and he did it. At the moment I asked about it, it was extremely important for me, and at that moment I got the most flexible regex-enabled QA tool I tried.

However, when we put regexes in "mass production" at the office and tried to describe all common errors with regexes, we ran into a big trouble. To check that a word is not occasionally capitalized after a comma, what a good idea. Translators often change punctuation marks, often decide to join two sentences into one and may easily forget to change the case of the next letter. But if you consider how many occurrences of capitalizaed letter after a comma really are allowed (all proper names at least), you'll understand this regex is gonna be a headache, especially in a technical manual.

In fact, our engineers created a lot of regexes which all had to easy the life of QA people. And each of them had its wrong side and eventually generated a lot of noise. It certainly doesn't mean that regexes are bad. It doesn't even mean that our engineers didn't think well (although sometimes they did). It simply means that the implementation needs to be more flexible.

For me, minimum requirements at the moment are:

- variables support (including the possibility to pass variables from source regex to target and vice versa)
- exceptions to regexes
- forks support (if-then-else).

And the next step, I guess, is gonna be a scripting language that will allow to create short script using regexes, maybe even a lite version of Perl. Oh, I wish... :)

MemoQ 3.5.15 - just some smallish notes

2009-04-30T10:37:00.005+07:00

I'm not quite experienced with MemoQ. I used to test it a year ago, but didn't really perform too many jobs in it.

Anyway, after attending MemoQFest and hearing so many words about how great it is (and I did like it when I tested), I decided to eventually seriously test it. I was a kind of disappointed to get MemoQ4Free trial license which is apparently because I have this serial number for more than a year. And I didn't have a chance to contact MemoQ people asking for a more capable trial version. Will do.

So, I decided to test by translating MemoQ help files. But all I could do with this license was to translate onl one file because I could only have one file per project. There was no point to create several single-file projects as they are not allowed tro share a translation memory anyway. So, before I really start testing I'll have to eventually send my thanks and appreciations for the event to Kilgray.

However, I noticed a couple of interesting facts (drawbacks indeed as I always notice drawbacks). They may be of no importance for many users, but are sometimes annoying for me. And I don't know if they can be corrected by setting some options or not, I couldn't find ways to do that in 5 minutes, so I postponed searching.

So, those two facts are:

1. If a segment that is already in your TM is a part of another segment (in my case it was the name of a dialog box: the first segment contained the name itself, and some other segments referred to this name then) it automatically suggests you the translation. This is indeed a ver nice feature, not drawback. What is to my mind a drawback is that if ou select to insert this translation, and if it was included into tags in the segment being transalted, the tags disappear. So, you have to paste them from source text or cope and paste the translation manually.

2. This one also relates to tags. Differnet people have different work habits, and I prefer to have source text copied into the target, then select some number of words and replace them with the translation (just type it instead of the selected text). I select the text using Ctrl-Shift-arrows on the keyboard. And again, if there is a tag in the the text, and if I try to select the text before the tag, it includes the tag into the selection as well. So if I just retype, the tag is deleted. And if I want to preserve the tag, I need to press Shift-Left arrow twice to move to the end of the word preceding the text. This is something that really poisoned my life with MemoQ.

Is Web 2.0 really here?

2009-04-24T18:39:00.004+07:00

I'm currently sitting at MemoQFest, Kilgray's first user conference, but this questions interests me for much longer, and especially at each conference I attend.

People are speaking about Web collaborations - that's OK and possible now even in Russia where some freelancers still use analog modems to connect to the Internet. But people are also speaking about submitting online requests for quotations etc. Nothing new indeed, translation portals exist for ages, and what people are talking about now is just tools that work behind the portal to automate workflows. But are our clients really that mature?

Well, some of them are. They use their own workflow systems which upload project files to their FTP servers and automatically generate e-mail notifications to our project managers. How can we connect it with our Web-based workflow management system? The answer I just heard was to have project managers enter this information into the portal form instead of the client. This indeed will work, but isn't it really ridiculous from the technology point of view to have project managers do something manually in order to have it then processed automatically while both steps can be done automatically?

Clients that are not that mature are often secure. They do care about their information, and they often have a list of software and procedures they can use and a list of those they are not allowed to use at the office. File transfer via a portal is often on such "black lists".

I'm not even going to touch the clients who are just tough and wouldn't want to change the manner of their work just because some technology requires it. Later on they can fall in love with translation portals, but it will take a year or more for them. Do we have just to drop such clients off?

I really wonder what other LSPs think about it. I agree that we can be strict when it comes to payment terms, pricing issues or other things like that because this is what our clients owe us. But as soon as it concerns the way we provide services to our clients, i.e. what we owe them - do we have to push the clients and make them accept our work style? I think, no. That's why I'm still looking for the translation management system that would do everything such systems offer now and will provide integration with customers systems and/or e-mail service.

QA functionality in XBench 2.7

2009-04-19T04:26:00.000+07:00

XBench 2.7 (build 0.183)

Supported checks. This tool officially does not support Unicode, and this is probably the main drawback of the application that eventually resulted in a rather high error level. For this reason, it does not support checking Arabic, Chinese, Farsi as well as Czech and Polish TTX files.
It does not have punctuation checks enabled by default; however, they are easy to enable via XBench rules which may significantly improve its error reporting in real life.
False positives. XBench reported corrupt characters for all non-Latin and non-Cyrillic characters.
Multilingual project support. Like many other QA tools, this one finds inconsistencies between translations into different languages and reports terminology errors in untranslated segments.
Conclusion. This tool is very new to the market, but probably on of the most promising ones to date. Its extensive file format support, additional functionality and extension capabilities together with the fact that the tool is currently free allow to suppose many companies, particularly small ones, may want to select it as their QA solution.

QA functionality in QA Distiller 6.0.0

2009-04-16T04:20:00.001+07:00

QA Distiller™ 6.0.0, build 188

Supported checks. This tool supports the widest number of possible checks which still may be extended using regular expressions. However, a serious drawback is that it does not check tags identity. The test file included a hyperlink which was intentionally changed in all translations; however, QA Distiller ignored it.
Additionally, we couldn’t find a way to check untranslatables in Distiller. There are two places where you can set a list of untranslatable items, and our idea was that the tool should make sure they are identical in source and target text. However, Distiller did not report missing untranslatable which existed in the test file.
Another weak point is number formatting check. Distiller only makes sure the number includes separators specified in parameters of the target language, but does not check the order of the separators. So, for example, it will consider 1,222.33 and 1.222,33 to be the same numbers with regard to number formatting.
Multilingual project support. In addition to reporting inconsistencies between different languages, Distiller also handles multilingual batches together with multilingual dictionaries in a strange way. For the first file it encounters, it tries to match all the glossary files in spite of the language indicated in the translated file and the glossary, which results in numerous “ignored terminology” errors. For the second target language, it tries to match it to all the glossaries until it finds the correct one. Then it perfectly matches the rest of the translated files with correct glossaries and doesn’t generate error messages.
Right-to-left language support. While this is the most comprehensive QA tool so far, it definitely lacks right-to-left languages support. Sentences in those languages are still aligned left-to-right, and if a segment ends with non-Arabic/Farsi/Hebrew words or digits, QA Distiller often handles the end of the segment incorrectly which results in a false error message. Additionally, it reports terminology errors in almost every segment because of incorrect RTL text handling. If you open the same file in MS Word and do a simple search for the glossary term you will be able to locate it easily while Distiller insists the term translation is missing.
It does not support Farsi by default, so we had to define a new language which resulted in reporting too many corrupt characters (480 occurrences).
Additional observations. Inability to change error severity may also be considered as a disadvantage (at least it makes the software less flexible and customisable).
Conclusion. At the moment, this is the most comprehensive, yet rather expensive standalone solution on the market.

QA functionality in ErrorSpy 4.0

2009-04-13T10:28:00.002+07:00

ErrorSpy 4.0, build 001

Supported checks. The total set of supported checks is quite extensive with some specifics listed below. ErrorSpy includes presets for some languages, but they are sometimes incorrect (e.g. incorrect quotation marks for French). It does not support Chinese Traditional as well as right-to-left languages without additional customisation and does not support specifying more than one set of quotation marks in case of nesting. Although it allows to specify decimal and thousand separators to check number formatting, we failed to make it check it. In fact it only reported unmatched figures.
File import. The tool cannot check for skipped segments because it does not import skipped segments at all. This is not convenient if you need to locate any segments that were left untranslated.
False positives. ErrorSpy reported “space required after punctuation mark” errors even if the corresponding checkbox is deselected.
Right-to-left language support. No support by default; however, the tool allows to create new languages. For Arabic, it reports Latin characters to be punctuation marks although they were listed as valid characters in language configuration.
Additional observations. English user interface contains translation errors (for example, one of the checkboxes reads: “spaces that require a space before”).
Another drawback is that ErrorSpy sometimes corrupts the first letter of language names.
This tool does not remember the directory it recently worked with. It is quite inconvenient when you work in a non-default directory.
For some reason, more and more interface elements switch to German with each test run. The interface gets back to English after restart.
The tool crashed on each attempt to check Russian. This may be a problem of this particular installation, but may also be a problem of the whole release. It must be noted, however, that version 3 ran smoothly on the same computer.
Conclusion. Although the tool significantly improved compared to its version 3, the first build seems to be quite unstable. However, with such rapid progress, this tool is rather promising.

QA functionality in Wordfast

2009-04-12T11:49:00.001+07:00

Wordfast version 5.51t3

Supported checks. The amount of checks supported is rather limited, but may be extended using custom macros.
File import. Wordfast determines HTML files by <html> tag at the beginning, but not by the real content, whereas in real life this tag may often be omitted.
Multilingual project support. We failed to make it check terminology against the correct glossary. After checking the Arabic test file, Wordfast continued to apply Arabic glossary to the rest of the languages despite of numerous setup changes, glossary recreation, deletion etc. Even when there was no Arabic glossary existing on the computer, Wordfast still reported Arabic terminology errors. It might have got much better scores if it used correct glossaries.
Right-to-left language support. It doesn’t properly handle right-to-left languages and just like all other tools reports terminology errors in untranslated segments.
Conclusion. As many other plug-in tools, WordFast provides quite a good solution for those who select it as a TM tool and do not want to implement a standalone QA tool.

QA functionality in SDL Trados QA Checker 2.0

2009-04-11T11:46:00.002+07:00

SDL Trados QA Checker 2.0, plug-in to SDL Trados 2007

Supported checks. Basic set of checks performed by SDL Trados QA Checker is rather extensive compared to other plug-in tools and may be extended using regular expressions. This tool does not allow to specify Chinese full stops as valid punctuation marks. Moreover, Arabic and even Easter European characters cannot be included into forbidden characters list. It also does not check quotation marks and number formatting.
False positives. Unlike all other tools, it generates false positives by counting skipped and empty segments as incomplete ones.
Conclusion. In general, the tool is good enough for translators who work in Trados TagEditor, but may be hard to employ in dedicated quality assurance departments where batch processing of mono- and multilingual projects is normally required.

QA functionality in Star Transit XV Professional

2009-04-10T11:10:00.001+07:00

Star Transit XV Professional, version 3.1 SP 21 Build 617

Supported checks. Star Transit employs the most limited number of checks without any further customisation. Available customisation is provided via fixed value lists and does not allow to add e.g. custom delimiters which in our case was necessary for Farsi .
False positives. Due to the limited number of checks supported by Star Transit it generates one of the lowest numbers of false positives.
File import. Import of TTX files is not correct enough; tags are represented in an unusual manner which hinders work with files.
Right-to-left language support. This tool proved to be surprisingly good at checking terminology in right-to-left languages. It also showed probably the best handling of right-to-left languages in general.
Reportability. The tool does not provide any reports; all errors need to be corrected “on the fly”.
Conclusion. In general, the real functionality of the tool is closest to the claimed one; however, it is too limited.

QA functionality in SDLX 2007 QA Check

2009-04-09T14:03:00.008+07:00

SDLX 2007 QA Check, build 7014
Supported checks. Basic set of checks performed is also rather limited. A user can extend and customise it using regular expressions; however, regular expressions are often beyond the qualification of a QA manager.
This tool does not check number values and does not check number formatting, double punctuation marks and brackets unless you set up a regular expression. It also does not check tags, and though it is hard enough to change tags in SDLX, TTX files converted to SDLX format may contain corrupt tags which won’t be detected.
Skipped translations are not converted from TTX files and therefore are also not found.
SDLX does not allow specifying Chinese full stops as a valid punctuation mark.
False positives. QA Check generates false positives for forgotten translations (counted as partial translations as well). Also many false positives are generated for partial/incomplete translations (they are not differentiated in SDLX) because incompleteness is determined only by translation length, not taking into account the number of sequential source words found in target segments.
Multilingual project support. As many other tools, QA Check Checks translation consistency between different languages.
Right-to-left language support. QA Check displays such texts left-to-right which hinders work with files and leads to reporting non-existing terminology errors.
Conclusion. With additional customisation, this tool is quite a good solution for SDLX users that do not want to involve additional standalone tools into their work processes.

QA functionality in Déjà Vu X Workgroup version 7.5.302

2009-04-08T13:15:00.003+07:00

It's been almost two years since I researched translation quality assurance tools available on the market. The full research was presented at Translating and the Computer 29 conference in November 2008 and is published at my company Web site. Needless to say, I would like to update the results and find something really new in this area.

So far, just to start with some facts, I'll re-publish short benchmark results from the research. And I'll start alphabetically, so Deja Vu is the first.

Déjà Vu
Supported checks. The number of default checks is rather limited; however, the application supports custom SQL queries which most probably allows for extending the amount of possible checks and further customisation.
Multilingual project support. While checking several files translated into different languages, Déjà Vu applies the TM and glossary for the first language to all files no matter what their target language is.
Right-to-left language support. For Arabic and Farsi it reported terminology errors even where the translated term could be easily found using Find feature.
Reportability. The style of error indication probably fits translators who want to check the translation “on the fly”, but is rather inconvenient for a dedicated quality assurance department.
Conclusion. In general, this is one of few tools declared capabilities of which are close to real ones. The tool is only suitable for checking its native files. Although it supports other most common formats including Trados, SDLX and Star Transit, conversion is quite time-consuming and is not in general worth it.

SDLX sickness

2009-04-06T15:02:00.004+07:00

This weekend I thought about what will Trados and SDLX together result into. And I recalled my very first tests that I did in 2002 or 2003. I had to make a comparison and select a TM solution for our new company and our main client. I'll try to find this comparison chart in my archives and publish or quote it here. I do remember there were Trados, SDLX, Deja Vu, Wordfast and some more TM systems.

Based on it, what I suggested to our client was SDLX. Why? Because it was nice, made use of TMs and glossaries in quite a convenient way, supported a lot of file formats and languages and had a free "lite" version for freelance translators. But what was maybe the most important thing was that the company was very responsive. They replied messages immediately, fixed bugs within 24 hours and always were ready to help, even to import our files into a new project if we experienced some troubles doing it locally.

I still remember the names of people who responded my mails. It was so nice to feel they care and you may rely on them. There were a lot of different problems with this software, but if we ever missed a deadline, it was not because of them. They all got resolved promptly.

SDL was growing, SDLX was developing, new features were added and old features were considered to be "obsolete". Free lite version was discontinued. Technical support first became slow, then paid. I don't know how prompt is currently their paid support. Unpaid one is pretty slow and useless, at least if you don't know addresses of real people you need to contact.

Knowing some SDL people almost in person, I'm quite sure they are pretty dedicated and doing their best, just like their predecessors were back in 2002 or 2003. What's changed is the environment they're in. I don't know if it's true for all large corporations, or if SDL just was unlucky on this way, or it it was their intention and they're really lucky to achieve their goal. But I really miss that nice cozy company that I selected 6 or 7 years ago.

A few notes on SDL Trados QA Checker 2.0

2009-04-02T14:35:00.004+07:00

A few days ago I had to perform formal QA on a project translated from English into German. I used SDL Trados QA Checker 2.0 (SP3) for this tasks and have a few interesting findings.

First of all, I received a lot of capitalization error reports. Apparently all nouns are capitalized in German which is not true for other languages, and this result into differences in capitalization. Quite annoying though!

Then, it reports repeated words like "OCR-A OCR-B" or "Task 1, Task 2", which is also annoying.

Third, it reported untranslated segments, and the segments really were untranslated. The problem was that those segments were just untranslatable, and TagEditor didn't even open it, but QA Checker still reported it. It's a good added safety feature, but annoying again. One example of such segment:
©2009

And last but not least, report on inconsistent translation is hard to work with as it doesn't reference the other translation compared to which this one is inconsistent. You have to use search feature and keep in mind what translation was there and what is here. Needless to say, annoying.

I would like to apologize to Patrik Mazanek who developed this plug-in. It in fact is a very useful and very capable tool, it can detect a lot of errors and is quite flexible, but as long as it works good, I apparently don't notice how good it was at detecting this or that error. And as soon as it comes to annoyance, I certainly notice it. Anyway, this post is apprently not to state how bad QA Checker 2.0 is. It's really great. But the purpose of this post is to make people (including Patrik and other developers who will be able to solve the problems) aware of the existing drawbacks.