MT means Mad Translation, not Machine Translation (or Automatic Translation) ! SPEECH TO SPEECH is a BSTL-Technology !


About Machine Translation (MT)

Machine Translation or Mad Translation ?

The purpose of this site is to re-establish the truth about the real (im)possibilities of Machine Translation systems (also called automatic translators). This is a quite complete synthesis with automatic translation examples, analyses from specialists and even expectations about the future of this technology. All participants to the realization of this site are linguists or people working in the field of foreign languages.

This site will also make you discover what is the BSTL-Technology, see bottom of this page, the Last Minute paragraph about SPEECH TO SPEECH Translation...

The company L&H (Lernout & Hauspie, developers of Power Translator) is presently KNOCKOUT (September 2000)... Among all the reasons for this failure, you will find one here below. They are one of the BSTL-Technology designers to the detriment of their shareholders!

For example, here follows what has been said during an international conference (Translators and Computers Session, ATA November 4-8, 1998) by Mr. Amar Almasude (coordinator of instructional technology, Francis Marion University, Florence, South Carolina) : 'Machine translations are still considered inadequate, even useless.' To corroborate this, read first some funny examples... Or, if you wish to translate automatically English into Japanese and vice versa, read this product comparison and the examples provided... You will be amazed! If you understand Spanish, read this product overview containing also an interesting example of machine translation...

Do you want to read a benchmark between 4 automatic translators available on the market ?

This should warn you... But we still advise you to read what follows:

We have found the following advertisements in magazines...

High-quality automatic translation of texts of all types will not be available for some time to come, if ever! An ideal world would be one where no pre-editing or post-editing was necessary!

Maybe in twenty or thirty years from now there will be super-powerful computers with much smarter Operating Systems (neuronal, bionic or whatever) that can produce high-quality automatic translations. The trouble is, you don't want to have to wait until then. What can you do in the meantime? Globalization is knocking on your door... Can you afford to wait for these (hypothetical) tools to become available? We suggest you accept help from less ambitious but more effective tools. Please note that translation tools do not mean you can replace human translators, but they do facilitate communication with foreign addressees. Human translators will always be necessary to deal with cultural or legal aspects (laws can change, business practices vary from one country to another, etc).

On illusions of today:

What we are concerned about are so-called self-contained full translation systems (or automatic translation system).
Many automatic translation projects have had to be abandoned after years of development because of insurmountable problems. For example, the 'METAL' project (Siemens) was stopped in 1995 after more than 30 years of development (beginning in the early sixties)... R&D in the field of automatic translation has been going for more than 50 years now, and no satisfactory solution has been found yet! NONE of the methods that have been looked at have been able to provide the right answer. Some machine translation developers talk about 'mainstream linguistics', 'universal grammar', 'statistical modelling', 'derivation trees', 'parsing algorithms', etc. All these methods are dead-end approaches because they CANNOT cope with meaning! Maybe the answer would be to be able to understand how the human brain works and to develop a program that simulates all its ultra-complex functions... When that is possible, part of the problem will have been solved! But when will that be?

Those interested should read 'La Traductique' (P.Bouillon, A. Clas - Editions AUPELF-UREF, ISBN 2-7606-1616-9) to learn about the feasibility of automatic translation. This book's introduction is very readable and the following paragraph sums up its conclusion:


Various people claim that automatic translation programs work very well in all contexts (and especially yours), producing high-quality translations that can be used directly by anyone! This is not true! If you don't believe us, test these programs yourself with sentences matching your needs. The results speak for themselves! They are unusable directly and require thorough human verification, because they contain many mistranslations and grammatical mistakes. These programs try to define the context of the words arbitrarily, using statistics, and most probably, the meaning chosen by the program will not be the one you had in mind.

As explained in Bouillon's book (§ 10.8), 'the most reliable and effective translator will always be a human being'. We agree absolutely, and urge you to adopt this point of view in order to safeguard yourself from unpleasant surprises...

For the unconvinced:
We advise you to read the book 'The Possibility of Language' by Melby and Warner, John Benjamins Publishing Company (1996) or else read the following page: 'Why Can't a Computer Translate More Like a Person?' by Alan K. Melby.

Extract from this page : "...It is interesting to observe how various persons who have not worked on machine translation react to the title question of this paper. Some believe that there is no fundamental difference between humans and machines. They assume that the quality of machine translation will someday rival the quality of human translation in all respects. They point out that computers can do arithmetic much faster and more accurately than people. Then they remind us that math is harder than language for many students. Furthermore, they take it as obvious that the human brain is ultimately a type of computer. From this basis, they conclude that it is just a matter of time until we have a new kind of computer that will function like the brain, only faster and better, and will surpass the capabilities of humans in the area of language processing. Others take a contrary position. They believe that humans and computers are so entirely different in the way they work inside that computers will never approach the capabilities of human translators. Still others are puzzled by the question. They were under the impression that the problem of machine translation was solved years ago..."
"...The fact of the matter is that machine translation is a problem that is far from solved. Experts in the field agree that computers do not yet translate like people. On some texts, particularly highly technical texts treating a very narrow topic in a rather dry and monotonous style, computers sometimes do quite well. But with other texts, particularly with texts that are more general and more interesting to humans, computers are very likely to produce atrocious results. Professional human translators, on the other hand, can produce good translations of many kinds of text. People can handle a range of text types; computers cannot. Where the experts disagree is on the question of why computers are so limited in their ability to translate..."

Read the following page: 'Machine Translation: The Disappointing Past and Present' from Martin Kay, Xerox Palo Alto Research Center. You can also read this extract from the HEISOFT AG WEB Site.

These two pages will help you understand why automatic translation will never work!

Do you want to some examples of actual automatic translation results? Below is one example produced by a program available on the market:

Source sentence in English:
This avant-garde movie the press has spoken about has been defamed by the critics in spite of an original advertising campaign.
Translated into French by a commercial automatic translator:
Ce film d'avant-garde que la pression a parlé environ a été diffamé par les critiques malgré une campagne publicitaire originale.

Comment: 'the press' should have been translated as 'la presse', meaning the newspapers and the journalists who work for them. The translation chosen by the automatic translator was 'la pression', meaning the act of pushing something firmly... a very bad choice! The next part of the sentence 'the press has spoken about' should have been translated into French as 'dont la presse a parlé'. In this case, the program did not identify the verb 'to talk about' ('parler de' in French), and translated 'about' with the adverb 'environ' in French, which means 'approximately' instead of the preposition 'de' in French, meaning 'on the subject of'.

Another example...

Source sentence in English:
Our program will allow everyone to understand one another better. It will also give you a little push to help you start or increase your interactions with other countries.
Translated into French by a commercial automatic translator:
Notre programme permettra à chacun de comprendre un un autre meilleur. Il vous donnera également poussée pour vous aider commencent ou augmentent vos interactions avec d'autres pays. This translation does not mean anything! Have it read by a French speaker to see what he/she thinks about it...

These mistakes show how automatic translators make serious grammatical mistakes and that the results can be incomprehensible... Moreover, these software packages are expensive!!! Would you buy a car without an engine or a printer that couldn't print your documents?

Do you want to send garbage to your correspondents? Of course NOT!

Perhaps you can now see why you shouldn't use an automatic translator in your everyday work. You will jeopardize your public image if you send or publish automatically translated texts!

If you want to see for yourself what automatic translation really looks like, go to the AltaVista Search Engine, type in any keyword as a search and submit your request. When you receive the answer from this search engine, you will notice that the pages cited can be automatically translated for free into different languages (if you are an English speaker, try this option from a French site for example). Use this 'Translate' option and have fun reading the results (you will notice that these results cannot be used for serious purposes and IN NO CASE in a business environment). Remember, this free automatic translation service is the outcome of more than 25 years of research and development. If 25 years of development give such results, do you think the solution will ever be found ?

Here follows a Spanish article written by Mr. Francis Pisani in EL PAIS Digital (Spain) concerning this kind of solution:
'Desde hace varios meses, AltaVista (http://www.altavista.com) ofrece traducciones simultáneas sin coste alguno en su sitio. Instalado en Nevada, Sancho (http://www.sancho.com) propone la traducción automática al inglés de los sitios en español. En ambos casos, los resultados son con frecuencia hilarantes.'
In one sentence, here is the summary of this paragraph: 'The results of automatic translation are frequently hilarious...' So, automatic translation is a good joke!

If you want to know more about difficulties and ambiguities of natural languages, refer to these pages : Linguistic Problems and Complexities.

Do you want to read another MT test (in French; very funny page) ?

Conclusion

Translation is a very difficult thing requiring much feeling and understanding of cultural aspects, which is not available in a computer. When will a computer be able to think and to interpret the environment (social and cultural) as a human being ? Don't forget that some grammatical structures in a given language do not exist in another language: in this case, how to translate without interpretation ?

Salesmen promoting automatic translation argue that these tools are very useful to understand languages that one knows nothing about... In this case too, this kind of argument makes use of your naïvety ! Let us explain : imagine that you received a German text (or any other language) and had it translated automatically in English (your mother tongue in this example). Don't believe that the result will be accurate, it will contain lots of mistakes of all kinds... Then, your understanding of this text will be such that the meaning of the original document will not reach you, you could even understand the contrary of what was said... Imagine then that you have received an enquiry about your activities and after the reading of the bad automatic translation result, you think that you don't meet the requirements of the sender when in fact you are ready to answer his request... This is a good way to lose your potential customer. Are you ready to take this kind of risk ? With Automatic Translation, you will take it and it is not worth while !

If you choose to purchase an automatic translator in spite of our advice and the examples we have shown you, be prepared to waste your money (read the opinion of human translators about this) ! Believe us, be careful, it is better to hire a translator or to choose translation aid solutions (as dictionaries, conjugation, grammar, etc.) that will really help you learn and translate languages!

Last Minute...

After Machine Translation (sorry... Mad Translation), some ridiculous persons claim that in a near future SPEECH-TO-SPEECH (STS) translation will be available to overcome the language barriers... SPEECH-TO-SPEECH translation consists of 3 steps : Speech Recognition - Automatic Translation - Text-to-Speech... Wow !!! Even between languages like, let's say, Japanese and English !!! This is not Mad Translation anymore, but a technology named : BSTL-T (Be Sure To Laugh - Technology). OK, this was for the fun... Back to reality now : can you imagine what results can be expected from this kind of stupid solution... Voice recognition (it must be, of course, continuous and speaker-independent) is not able to avoid recognition mistakes and is not 100% reliable (because contexts and sentences are often ambiguous, in this case human post-edition is necessary; real-time voice recognition will not be error-free). Automatic translation has been extensively discussed on this page... Linking 2 technologies that are not absolutely error-free will not fail to produce a BSTL-Technology. Text-To-Speech (TTS) is the only good part of this solution. But do you know that marketing people talk about something like 'auto-adaptative Text-To-Speech' that will reproduce in real time the original voice of the speaker (they promise that we will hear the real voice in another language of the person we are speaking with)... Wow again !!! How cool it would be to hear myself talking Chinese or Tagalog...

Hey, I see you dreaming about STS systems... Stop it ! It's only marketing hype ! They have seen this STS product in 'Star Trek' or 'Robocop' and they are trying to sell this silly idea to you... Once again, be ready to lose your money if you want to invest into this kind of technology !

Now, let's face the truth about MT and STS technologies...

What is written here does NOT concern Universities or Academic R&D organisations: their mission is to find interesting research subjects for their students. Students need to be open-minded... And this is a must in the world of education (even if scientists and teachers know that the researches will never lead to marketable products) !

What we want to say is that companies launching such wild hopes are just looking for money... How do they proceed: they issue shares on the stock exchange and this way, they manage to get large sums of money. Even worse, others get public money ! However, they use the credulity of investors... The only purpose of their managers is to get rich quickly thanks to your money... They simply are crooks because they perfectly know that the technology they promise to create will never exist or at least will never be reliable (this kind of product - MT or STS - can only be gadgets) ! Sorry, all this is very sad but it has to be said !

In other words, if you intend to invest in this kind of company, change your mind ! See the Lernout & Haupie (L&H) example, their shares have lost about 85% of their value due to all their lies, and this over a short period (July-September 2000) !

Why has this kind of call for capital to be denounced? Because the money invested in such projects (MT or STS) will be lost and investors will finally lose confidence in all other types of serious technological projects. As a consequence, there won't be any money left to finance really interesting and necessary R&D ! This would be damaging to all of us in the long term and this is the reason why this WEB site exists !

We invite all universities, schools, companies and professionals working in the translation (or any other) fields to link this site into their site. You are authorized to publish these pages 'AS IS' by any other mean (this includes printouts). This way, we can tell the whole world that language learning is the only way to improve globalization! If you have been disappointed by automatic translation don't hesitate to link this site too!

E-Mail
Other link(s) to add, remark(s) ? Then E-Mail us at: madtrans@usa.net

This page has been visited times.

There is no copyright on this home page, except for the BSTL-Technology name. If you want to quote this name, you have to link it with this site as follows : BSTL-Technology or to publish the address of this site as : (c) Mad Translation (http://meltingpot.fortunecity.com/grove/402/index.html)