CAT Boondoggle

Once more I made a foray into computer assisted translation, and once more I was appalled. I keep looking off and on for better tools. They probably exist, but the right ones have not come along for me yet.

Computer-based or computer-assisted translation is generally caught in a bind, because they operate with a set of assumptions that are faulty.

True enough Champollion decoded hieroglyphics by means of comparing three sets of script to one another and basing his work on the notion that the three text on the Rosetta stone were relating the same thing in different languages, which gave him the keys to the Egyptian language. He was able to decipher hieroglyphics.

But ever since linguistic scholarship is getting increasingly bogged down in mistaking the symbols for the thing they represent, and generally confusing form and content. This leads to the whole absurdity of computer translation, and even computer assisted translation falls into the same trap.

With a very Platonic twist, A Course in Miracles says: "Let us not forget, however, that words are but symbols of symbols. They are thus twice removed from reality." (ACIM:M-21.1:9-10) In other words, the words give form to a symbol in the mind, behind which is a reality which is being approximated by those symbols. The same meaning could be symbolized differently in a different medium, but it should be evident that there is no direct transformation possible of one set of symbols into another, unless we first step back to the meaning behind those symbols.

My favorite example of criticism along these lines is the Dutch author Jan Willem Kaiser in his commentary on the Gospel of Mark, where he describes the failings of most all Bible translations exactly because they talk about only the manifest story, without ever grasping the content. What the Bible translators don't get is Jesus' word that it all comes to us in symbols, and the business of biblical translation continually attempts to transpose the symbols without recourse to the latent meaning. Kaiser makes the obvious point that unless the translators are actively serious students of Jesus' teachings, there is little chance of a good outcome.

Unfortunately, most translators labored under Paul's misinterpretation of Jesus, and his fundamentalist, literal distortions in which he almost consistently confuses content and form, perhaps indicated best by his belabored conclusion that Jesus was talking about the resurrection of the body, not the mind.
Those of us who study ACIM today often end up re-discovering the deeper meaning behind the symbols, and suddenly seeing through the parables of the Bible, which would lead to very different translations. Examples are plentiful, such as Jesus' term metanoia in Greek, which actually means change of mind, and it is rendered most often under the influence of Pauline theology as repentance, which was not what Jesus was talking about. He was teaching in parables of the possibility of changing our mind. An example such as the Samaritan woman at the well in John 4 becomes much clearer when you connect it with the teaching of special relationships in the Course. Indeed she was not 'married' and even her current 'husband' was not her 'husband' -- the whole episode was an account in parables of Jesus' teaching of special relationships, which our mind/soul engages in on the level of the body, and which can never be relationships because they are premised on separation and reinforce separation, the one and only real relationship is what the Course calls the Holy Relationship, which is the healing of the illusory separation that keeps our mind engaged seeking outside itself, and the restoration of the true communication through our true self. Pity the poor translators who thought Jesus was talking about 'husbands,' when he was really talking in parables of relationships, just like he was not speaking of 'breads' in Mark, but of spiritual nourishment. Nor did he speak of daily bread in the Lord's prayer. On and on.

The same is true about translation, and the two lines preceding the above quote actually clarify this, and I'll quote the para:

Strictly speaking, words play no part at all in healing. 2 The motivating factor is prayer, or asking. 3 What you ask for you receive. 4 But this refers to the prayer of the heart, not to the words you use in praying. 5 Sometimes the words and the prayer are contradictory; sometimes they agree. 6 It does not matter. 7 God does not understand words, for they were made by separated minds to keep them in the illusion of separation. 8 Words can be helpful, particularly for the beginner, in helping concentration and facilitating the exclusion, or at least the control, of extraneous thoughts. 9 Let us not forget, however, that words are but symbols of symbols. 10 They are thus twice removed from reality. (ACIM:M-21.1)

The only reason why Ken Wapnick, Ph.D., as a clinical psychologist and not a linguist, has been effective in overseeing the translation of ACIM in all the languages in which it now appears, is because he has focused on content with the translators, not on form. And of course the translators need to master their respective languages, but the effectiveness of their translations hinges only on their comprehension of the content. Once that is in place they can express that understanding in their respective tongues. The accounts of Chiao Lin Cabanne with her Chinese translations of the Course speak volumes. She did the first translation into classical Chinese based on her linguistic knowledge, and then started learning the Course, and tranlated it into today's simplified Chinese, only to discover that as she learned the teachings of the Course, she had to do her first translation over, so she did, and her new translation in classical Chinese now reflects her own growth with the material.  The Course is an extreme example in some ways, but the experience with its translation is a demonstration of why today's notions of translation are completely delusional, because they remain stuck in the notion of transforming sets of symbols into one another, and exclude the intervening step of comprehension in the process.

This week I used one of today's most popular CAT tools, Wordfast Anywhere, which interestingly was originally developed by one Yves Champollion, who is a descendant of the other Champollion. Evidently it is a powerful tool, if it is used right, but the question is if there are any 'right' uses of these tools. To illustrate that question: yesterday I found out that in Holland there is a TV quiz-program which is based on quotes from user manuals of appliances, and the contestants are supposed to guess what the appliance is, and they almost never can. That is probably a testimony to the state of computer-based and computer-assisted translation.

While we see people who can't spell stumble with spellcheckers because of homophones, translation memory, and CAT cause worse problems when they are misused, and it is harder to use them right than to misuse them. I have related my experience with a translation agency a few years ago. In that case it was a very prosaic project - a quote for the electrical installation in the residence of the American ambassador in The Hague. The first translator supposedly had been a professor of linguistics who taught Dutch at some college in NY, and according to the agency she had high proficiency with CAT. My guess was that the translator was Turkish, and had learned Dutch and English from a correspondence course. Almost all the terms of the trade were absurdly wrong, such as a ground fault interrupt circuit, which became something like and 'earth leak switch' in a literal translation from the Dutch. Those are exactly the things where a TM is supposed to help you, but again, absent an understanding of the topic, total nonsense will result. I ended up redoing the translation entirely, and correctly, and in record time, such that the agency asked me how I could do it so fast, and I told them that it was because I was not using the CAT tools which they had initially demanded I use.

Still, I continue to hold out hope for better tool sets for serious translation, the first requirement of which would be the complete elimination of any attempt at computer translation, and an understanding that this is not how the process works. Any meaningful assistance must come from better access to more and better dictionaries, grammars and so on, to support translators who are actually capable of thinking in both languages. And then if they also have some mastery of their subject matter, perhaps there is hope. In the meantime I raised my translation rates to 2-3 times higher with CAT tools than without.

Copyright © 2012 Rogier F. van Vlissingen. All rights reserved.

Babylon V8

As reported here earlier, I've become enamored with Babylon translation support software, and the new version V8.0 is definitely a worthwhile improvement, if you think of it as a support tool for the translator.

For better or for worse, the company continues to include machine translation also, which is patently ridiculous and pointless.

Here is what I wrote to them:

quote (slightly edited from the original letter to the company).
I really don't understand why your company would be still keeping that stupid "translate" function in there. It is so mind-numbingly stupid, that it ceases to be funny after three tries.

Anyone who would understand what language even is, should understand that this is a categorical and structural impossibility to ever achieve. EVER. Language processing does not work in the way computer scientists seem to imagine - and I say this having read more computer science than the average Ph.D. The fundamental thinking error is to think that the mind and comprehension proceeds from the concrete to the abstract, when clearly the reverse is the case. This same issue is the underlying error behind all the nonsense about artificial intelligence. Just because a computer needs to synthesize the abstract from the concrete, and is therefore essentially structurally incompetent in higher level operations, so also it can never produce language, never mind how intricate the languages for programming it, because again of the unavoidable, fundamental flaw of having to deduce the abstract from the concrete, which is a process in which there is no conformal mapping, if the missing component of human experience is missing, which is the only conceivable guarantee that a translation is a translation and not gibberish.

Seen from this viewpoint, words don't make meaning, meaning makes words, or rather finds expression in words. Words evoke that meaning. Translation is to a) understand, and b) render into another language. This means the mind has to go up to the abstract level, and then descend again into the concreteness of another language, in proper idiom.
There is not even the remotest possiblity of a mathematical transform, or a logical transformative process on the phenomenological level of language that could ever produce a meaningful translation, except of something so trivial that you would not need a translation anyway. Like I said above, this whole notion is categorically absurd. And to persist in it, makes your company look stupid.

It sounded pretty smart when you talked about tools for translators. The translate function does not belong.
unquote


Copyright © 2009 Rogier F. van Vlissingen. All rights reserved.

Progress with Translation...

Some time ago I wrote here about Babylon Software, which I'm starting to like a lot, particularly because it is a Swiss-army style tool for translators. Extremely flexible. And the other thing I like is that this company does understand that machine translation is a stupid idea, which only a total geek could think of, who thinks the mind was modeled on a computer chip in lieu of the other way around. Once you get that point, it's obvious how true it is that a computer is a stupid instrument that just under certain conditions can execute certain instructions more efficiently than we can, but only a total moron would mistake this for thinking. So expressions like machine translation and artificial intelligence are a permanent oxymoron, just like smart bombs, or military intelligence.

So, as a company, Babylon is now addressing themselves to making tools to address the needs of translators. For some reason they did retain their machine translation as an option, and I can recommend it anytime you need comic relief. It will give you the same sorts of assinine results as the translation function on Yahoo - another group of nerds who don't get it.

You can configure it anyway you wish, and I like to have both dictionaries, (meaning, synonyms, etc.) in the language of origin, as well as a translation dictionary, and this gives you the tools you need at your fingertips.

The free dictionaries that come with the system are worth what you paid for them, but the actual dictionaries are not too expensive. I am beginning to find this tool indispensable. But, more important than the good dictionaries that are available for the system, and the flexibility of configuring it so they come up to your liking, the ability to customize this system is indispensable. For example I have one or two authors I'm working on whose vocabulary is so unique, that I'm developing my own concordance to their work, as a backbone to a multi-year translation project, using the Babylon Glossary Builder. Considering all of this, this product is indeed remarkable, and if I use it long enough it will become indispensable, especially for these very specialized projects.

Copyright © 2009 Rogier F. van Vlissingen. All rights reserved.

Translation Software Revisited - Babylon

Some time ago I had some funny experiences with a translation bureau, who would not hire me as a translator, because I did not use all kinds of translation software tools, which I despise. But then they did hire me to revise a translation of one of their "qualified translators" which was a complete disaster. In fact I laughed so hard while doing it, that I suggested to them that the translator in question probably was a Turkish national who had taken some correspondence courses in English and Dutch, and who now figured they were a translator. Evidently not so. According to the company, said "qualified translator" was very qualified indeed, the person was a language professor of Dutch extraction at an American college, who also would use all the software tools the company required its translators to use. It ended up that I did it faster without the tools, than the original translator did in with the tools, and I made a lot fewer errors, not to mention that the errors could have been costly, since the project was a specification for an electrical installation.
Thankfully for their customers, this particular company went out of business.

Most of the issues related to a problem we all see regularly day to day, when people who can't spell use a spelling checker, and they end up accepting homophone words in the wrong places, such as saying "then" for "than," etc.

At long last I've decided to test some of these tools anyway, in particular because I bumped into one company which seems to get it that they should focus on helping translators to do their job better, not replacing them. From a description their program sounded impressive, and they are offering me a chance to use their software as part of an ongoing evaluation.

Babylon.com initiated a Translator Outreach Program emphasizing that no machine translation can be a substitute for the experienced human translator. It is Babylon’s aim to reposition its translation software among professional translators. Despite their slogan “translation @ a click”, aiming at Babylon’s general users and describing the ease of use of its software by simply clicking on a word to receive a translation, it is important for Babylon to recall the essence of its software: the fast and efficient look up in many dictionaries, which – as Babylon is convinced – also is of great benefit to professional translators. Babylon Ltd. Is inter alia provider of English Dutch translation and Dutch English translation solutions.

Translators that would like to join the Babylon Outreach Program for Translators can do so at: http://www.babylon-blog.com/translator-outreach/ and will receive a free annual license of the Babylon software.

So, I'll start on this evaluation with all the skepticism I can muster, and then I will report back in this spot after I have some experience. Right now I'm at the end (ca. 7/8ths) of one 300,000 word project, and halfway through another 80,000 word project, during all of which I've probably consulted a dictionary 25 times, and asked a translation group on-line about 5 times for help with a difficult phrase. The rest has been subtleties requiring human interaction and careful weighing of very contextual stuff, as well as simply stupid mistakes in terms of employing English turns of phrase in Dutch, the kind of a mistake that software makes more likely, not less so.

One thing I've also learned the hard way from this project was that correcting a really bad translation, as I did on one book, is far worse than doing a fresh translation from scratch. This is why machine translations will cost you time, not save you time. However, if this tool can add convenience, great. Given the stats above it won't be about time savings, but the convenience could still be worth it.


Copyright © 2009 Rogier F. van Vlissingen. All rights reserved.

The Thomas Edison Effect

For anyone who has watched the development of computing, it is evident that the experience of vision and invention constantly must remind us of Thomas Edison, and his winning attitude that with every failed design the working light bulb came closer, which eventually proved to be correct.

In my own modest way I was able to see some of my visions realized when I was leading the design strategic IT systems for a shipping company, and even later I found out that some other designs of mine which were prototyped in the late 80's ended up being implemented nearly thirteen years after my departure from the scene.

Since the end of 2000 it seems I've been practicing bad timing, or so it seems for I got into the Internet game just when it was ending, working for what then seemed to be a promising ISP, Verio, just having been acquired by NTT of Japan, to form NTT/Verio. The most I got out of that was that they paid for a year of Japanese studies at the Japan institute in Manhattan, for the rest, I got to watch from the inside the collapse of the dot.com boom, which was then happening all around us and to us, and projects were going nowhere. In 2001 I got involved in the launch of a managed security service, with Riptech, later to be acquired by Symantec, to form Symantec Managed Security. The events of 9/11 squashed the fun. By late summer I had built up an interesting order book, but after 9/11 the market disappeared for about two years, and NTT/Verio was going through spasmodic RIFs every three months or so for all the time I spent with them, and by March 2002 my turn had come.

Regardless of all the frustrations, it was a fruitful period of exploring a lot of ideas which had been on my mind for a long time. It was during this time that my thinking about serious on-line collaboration began to take form, along with the fact that personal computers were such an obvious security disaster, that it should be possible to actually organize greater security on-line than a "personal computer," which from a security standpoint is leakier than a sieve. One of the first companies that I found really inspiring in that regard was www.safe-mail.net, which is still in operation today. The magic of their infrastructure, which is pretty flexible, is that they have a built in, fully automated deployment of PKI, guaranteeing the integrity of communications within the domain. On the other hand the fallacy then is that in their standard consumer offering they provide access just based on username and password, which is inadequate identification for any type of secure transactions, but the potential exists to integrate secure identification, which today is available in many flavors.

At a later time, I began this blog simply to vent some of my ideas which I had not been able to realize, and that in turn led to a number of exchanges which some day may become fruitful.

Fundamentally I think that the whole thing about web 2.0 now makes it even more critically important to develop serious solutions for on-line personal workspace, which should be designed to provide better security than the physical world does, your PC in particular. The on-line world cannot offer us serious solutions as long as it exposes us to needless security risks, such as the avalanche of identity theft which is now going on. For the time being web2.0 is mostly driven by the ad-supported business paradigm, which seems to have become gospel, because of the evident success of Google, to such a degree that even Microsoft in its desperation is now working hard to compete with Google. Competition is good, if nothing else as a gauge to measure one's own progress, but any time businesses become obsessed with their competitors it usually spells trouble, for it indicates that they are in doubt about their own identity or mission. The effort is all about getting better returns from advertising, and the user experience is only the means to that end, which does not bode well for the user experience in the long term.

What is needed as a vision is an understanding that security, privacy are an asset, not a liability, and that the mission of online services should be to solve a customers problems, without giving them additional liabilities they did not have before. Consumer resistance to on-line payments is substantial as a result, because many consumers walk away when they feel their security and privacy is being threatened by the all around negligence of the on-line culture of the moment. In our future therefore the real solutions that will arise, which will form durable on-line businesses will need to be worth paying for. The ad-supported model leads companies to chase fads, and to permanently sacrifice the customers security and privacy, for ease of use, convenience, not to mention data mining, which remains an invasion of privacy no matter how you slice it. And just because the world seems to be in denial about it, does not mean the customer has lost their senses, and is not aware of it. There just don't seem to be many alternatives right now, although solutions like Safe-Mail play into this sentiment.

I believe that a fundamental analytical insight is that communication is not complete without a financial transaction capability, which is what remains one of the weak links on-line, and getting weaker by the minute with every theft of credit card numbers that are stolen. It seems to be a miracle that there are any left that have not been stolen. The mission is an integrated work environment which makes my on-line life a viable solution to the practical restrictions of the physical world, but as long as it increases my risk, with new and unacceptable exposures, it condemns itself to being a faddish and unstable business.

For any naive reader who thinks I'm too pessimistic about the current situation, think again, just now in the June 2008 issue of PC World, a columnist seriously recommends doing your on-line banking on a cell phone, since as of yet they have fewer security problems, and in another column I'm reading, the author suggest not entrusting ones medical data to either Google or Microsoft, unless and until there are laws to protect us. So, on the whole companies are their own worst enemies by taking the user for granted.


Copyright © 2008 Rogier F. van Vlissingen. All rights reserved.

Driving Out Fear

As I have argued for some time now in this blog, the digital divide is not there, it's here. To think that the failure of the digital revolution is in the fact that people in Zimbabwe don't know how to use a computer, or God forbid, don't have access to the Internet is silly.

The failure of the digital revolution lies in the fact that it's just another toy with which we've amused ourselves, which creates more problems than solutions, and which allows us to get side -tracked by problems that are never the real problem. Mainly we get totally obsessed about tools. Companies would be more profitable if the programmers only knew their jobs, and programmers would design better programs if they only had better tools, and after all the finger pointing is over then we outsource the whole shebang to India. And while sometimes they can do it either cheaper or better, more often than not the problems now surface some place else, so then we can bring the whole operation back inside, because new management issues simply make it too expensive to outsource and do it cheaper. Or we try to solve it with a new class of "legitimate" indentured servants with H1B visas.

In the '80's when I at times played an active role in corporate IT development, and became de facto the chief architect of strategic systems for my company, which was an international shipping company, my heros were EWD, WED, and WW, or in full, Edsger W. Dijkstra, W. Edwards Deming, and Walter Wriston.

EWD surely was one of the fathers of structured programming, and of the opinion that programming was a matter of applied mathematics, and that sloppy thinking was the main problem. Undoubtedly, he made a huge and very valuable contribution to the field. He was revered by his students at UT.

I used some amount of structured methodology to good effect.

WED for me hit the nail on the head with a lot of his writing and the methods he thought, and somehow I knew that the essence of it all was his insight that the first obstacle was to "Drive out fear!" And it is also the final obstacle, for fear causes us to design failure in, for failure subconsciously equals job security, since most people view themselves as problem fixers, "fixers," and thus dependent on an unending supply of problems.

Finally for me came Walter Wriston, and his seminal insights in the meaning of information in society and in business, starting from the powerful conclusion that Citibank was in the information business, not the money business. An obituary by Steve Forbes is telling. His writing on The Twilight of Sovereignty was remarkable, given the year, 1992, and it had been prefigured by an essay in Foreign Affairs years before. His insights were equally valuable as applied to management, for he saw that people building positions based on hogging information (always based on fear!), were their own worst enemies in the information economy. Which was not to say they'd stop doing it. Thirty years after Wriston put an internal revolution in motion, Citibank was still teaching clueless exectutives not to do so, but unable to give them reason to stop doing it, as the short term success of information hoggers was as clear as the certainty of their long term failure.

In my own work, I had found that I could sometimes successfully bring to bear the insights and methods of consent management, developed in Holland by Dr. Gerard Endenburg, see www.sociocracy.biz and http://www.gwu.edu/~rpsol/lectures/2006_Lecture_Endenburg.html

For some time I did some management consulting based on Endenburg's methodology, even though I did not like his name for it. His method is definitely part of the new economy solutions bin. But for me it came up short, as it still only addresses the problem of fear circumstantially, by changing conditions, not the cause.

And of course the cause is in the one place where we never want to look in this society: inside. Thus the issue can only be addressed from the top. Only if we do that is there any hope at all to change the games of hide and seek, which perennially surface in software engineering in massive project failures, which probably happen in 2/3rds of the large software projects. So even with the newest technologies, the only hope is in truly addressing fear in the organization, and that means leading by example, from the top down, and allowing solutions to surface. And you cannot do that without also changing the form in which the organization operates, but with todays communications tools there are no limits in what you can do. But without addressing the causes, better tools, or better working conditions, outsourcing or no, we'll just be moving the problem around, and doing nothing else but the proverbial rearranging of the deck chairs on the Titanic.





Copyright © 2008 Rogier F. van Vlissingen. All rights reserved.

Should you wait for Web 3.0?

The whole ad-supported mania is starting to leave a bad taste in people's mouths, and it is resulting in a wave of fads and fallacies, which may be appealing temporarily, but have little long term value in term of providing users what they truly need, because the quick hits always play into whims and wishes, not fundamental or well-understood needs. Plus security and privacy suffer, and people are getting fed up with that also. As always convenience wins out over security, which is still an after thought in most designs. Increasingly services are forgetting to provide convenient ways to interface with them for travellers or others who may either temporarily or permanently have to cope with bandwidth limitations. And social networks become clubs for the bandwidth rich. All those pictures aren't really that relevant!

Time for a change.







Copyright © 2008 Rogier F. van Vlissingen. All rights reserved.