Mobile menu

Off topic: Why human translators can never be replaced
Thread poster: Todd Field

Todd Field  Identity Verified
United States
Local time: 05:03
Member (2003)
Portuguese to English
Oct 29, 2003

A colleague of ours brought up an interesting analogy in favor of the continued existence of human translators:

Think of all the centuries of musical composition. In standard Western music, there are only twelve half tones which are combined in various sequences of pitches, harmonies (polyphony), time durations and rhythms. With only twelve little notes, millions of songs have been made, from the Cantigas de Amigo to AC/DC, and no two are the same.

Now, take human language. Imagine the number of words alone (much more than twelve, of course!). Multiply them by all their possible combinations and sequences. Add the fact that regional dialects, the creation of new words every day, multiple meanings in different contexts, idioms, technical jargon and so on make the true number of words in a language absolutely immense. Multiply this by the total number of possible world language pairs. The math is astounding.

If anyone out there has ever wondered if their career would ever be made obsolete by a machine, I think this is some pretty hard evidence that this can never happen.

Just an interesting analogy to pass along to other Proz.com colleagues!


Direct link Reply with quote
 

Narasimhan Raghavan  Identity Verified
Local time: 16:33
English to Tamil
+ ...
Can a computer match this? Oct 29, 2003

This happened a few years back. In those days I was translating without the aid of computers or online dictionaries. Translation was done in long hand. I was stumped by a term "ver. Spiegel". It was not a standard abbreviation. I left the term for the moment and proceeded with the work. The next day I was reading a thriller by Dick Francis. The hero was crossing a closely guarded frontier post. The guards were checking the underside of the car with a sliding mirror. I had a flash. I mentally translated sliding mirror into German. Veschiebespiegel! I took up the translation. The term fitted the context to a T. The client confirmed the correctness of the term. Problem solved. Can a computer match this? I think not.
N.Raghavan


Direct link Reply with quote
 
Tobi
Local time: 13:03
English to German
+ ...
You're right! Oct 29, 2003

You're absolutely right!

Todd and Monica Field wrote:


If anyone out there has ever wondered if their career would ever be made obsolete by a machine, I think this is some pretty hard evidence that this can never happen.


But if anyone out there is afraid of loosing his job to a machine, he does not know anything about his business. Not in our lifetime - as long as it shall be!

I rather fear the dumping prices which get common more and more ...

Cheers,

Tobi


Direct link Reply with quote
 

Parrot  Identity Verified
Spain
Local time: 13:03
Member (2002)
Spanish to English
+ ...
A debt of centuries Oct 29, 2003

Between the "House of Knowledge" (Bait al-Hikma) and the School of Toledo run some 400 years, which accounted for the passage - through translation - of archaic and classical learning into modern thought. The debates that took place there around the terminology of algebra, medicine, philosophy and a plethora of other subjects are hardly within the scope of machine translation. The real question we may be dealing with here is the value of original texts themselves: how many are truly necessary to human learning, as against "disposables". And even then, every age requires a re-interpretation of the key texts in order to advance this knowledge.

Direct link Reply with quote
 

Bart B. Van Bockstaele  Identity Verified
Canada
Local time: 07:03
Dutch
+ ...
Sliding mirrors Oct 29, 2003

Actually, if there is an area were computers excell even now, it would definitely be this type of situation.

The bad news is that "Never say never" still applies.

It is now more than 20 years ago that I have written a tiny program (on a computer with a memory of 8 kB *not* MB!) that was able to read and solve the small problems children who are just learning to calculate get in school. (like: Bobby buys two apples for 2 CAD, and three more apples for 0.5 CAD. What is the price of one apple?).

Granted, this is a very limited application field, but it was also a long time ago and with a computer with 8 kB of memory that had very limited abilities.

Writing a program that correctly translated the text, was a snap. I added that later on.

Will computers one day be able to work just as well or even better than human translators? Absolutely, beyond a shadow of a doubt.

However, while computers have been invented a couple of centuries ago (officially around 1820 but it is debatable), they have only really be around for about 40 years, an extremely short time for the most versatile machine ever invented.

People who are eager and happy to point out that computers cannot do this, are unable to do that, and will never do something else, totally miss the point. Computers have *no* limitations whatsoever.

My first 20 or so kg PET 2001 of 1978 and my 1.5 kg Lifebook P2110 are essentially identical: they both process vast amounts of extremely simple instructions and they both rely on memory to hold those instructions, their data and the intermediate products.

On top of that, not many people seem to realize that our brains are far closer to digital machines, than they are to analog ones. Make no mistake, it is perfectly possible to create a complete human superbrain in a computer, a brain that will outthink any human. And that is only the beginning.

The good news, for us, is that technology evolves rather slowly. It seems somewhat unlikely that our generation of translators will be replaced by computers.

Make no mistake however, our demise is closer than you think. I suspect that some of us will actually live long enough to see the first human replacing computer translators. And computer authors are not far behind that.



[Edited at 2003-10-29 12:20]


Direct link Reply with quote
 

two2tango  Identity Verified
Argentina
Local time: 08:03
Member
English to Spanish
+ ...
Not afraid of computers Oct 29, 2003

Bart B. Van Bockstaele wrote:

People who are eager and happy to point out that computers cannot do this, are unable to do that, and will never do something else, totally miss the point. Computers have *no* limitations whatsoever.



It is funny to see how different our perceptions can be. I have been working with computers for a long time and I believe they are full of limitations.

In my opinion, a computer able to think real thought ot to understand a joke belongs to science-fiction, with a strong stress on the latter.

This topic, so important for translators, appears every now and then in one forum or another. I transcribe below what I wrote in the thread http://www.proz.com/?sp=bb/new&ViewTopic&post=98415#98415

Not afraid of computers

I have first-hand experience with computers and software, and I like them a lot, but I am not at all worried about machines taking over our jobs.

Computers are very good and fast at performing without distraction and with little or no error simple jobs or decisions with very clear rules.

When a computer performs a very complicated task, it is because the programmers have previously cut it into pieces or "elemental jobs" small enough for the very limited "intelligence" of the computer and its software.

We could say that a task, no matter how big or complicated, could be done by a computer if it can be reduced to a finite number of simple decisions and operations. For instance Deep Blue's chess victory on Gary Kasparov was a show of brute force, not of a defeat of human race at the hands of a new silicon god.

Now when we talk about translation, this "elemental task" would be the translation of a segment. And in average translating a segment is not simpler that translating the whole text, and therefore not into computer's field of advantage.

And of course there is the argument that computer science moves so fast that what looks a dream today becomes trivial five years later. But computers are only tools, and in this case the limitation resides in the linguistic side. We should start worrying the day translation can be reduced to a linguistic algorithm (including register, context and idioms).

Instead of competition, I foresee a better integration of human translators and their computers.

Enrique Cavalitto


Direct link Reply with quote
 

Bart B. Van Bockstaele  Identity Verified
Canada
Local time: 07:03
Dutch
+ ...
Different perceptions Oct 29, 2003

Our different perceptions are most likely caused by different experience(s), or by our different perception of time.

For many people, "never" means, "not in the near future" or "not in my lifetime" or sometimes even, "not in the next couple of hours". For me, "never" means "never": not before our universe existed, not during its existence, and also not afterwards. Never. Something may happen after I die, therefore, it will not have happened during my lifetime, but it will still have happened, and therefore the word "never", in my opinion, does not apply.

When I started programming in Belgium, there was no such thing as a university "education" to learn it. As a matter of fact, by the time the first university program was initiated I had already twice defeated the so-called experts of the department of education by developing programs that could not be written (according to them). Not that this was hard to do: they didn't have the benefit of having seen a computer other than in a picture.

The Deep Blue story is an old and very interesting one. I remember the time when it was said that computers were not intelligent and could never replace humans because they were easily defeated by even an absolute novice. When that was no longer true the story was changed to allow for the increased abilities of the program. I then said that there would be a day that chess would be considered an inadequate problem because one needs no intelligence to win. I was laughed at, at the time, but here we are.

Actually, the brute force of Deep Blue shows my point. Whether we like it or not, computers are ultimately far superior. We are limited machines, evolving, but at an extremely slow rate. Our advantage lies in the fact that we simplify a lot, and "instinctively" (actually, physically) see patterns. However, that property is also the direct cause of a lot of our problems. In traffic, we often don't perceive danger when it is present, and often do perceive danger when there is none. With accident-causing behaviour as a result in both cases. It is the cause of the well known optical illusions and the cause of countless other illusions.

In a natural world, that has obviously given us an advantage over other species. But our world is not so natural anymore and humanity as a whole is having problems adapting to a world developed by no more than a handful of very abnormal (as in far to the left or the right of the Gauss curve) individuals. We are evolving to a world where a more or less objective assessment of reality is more advantageous than our pattern based over-simplified perception. That'll give computers and their programs even more of an edge. Not that it matters, ultimately, computers will have the advantage in the natural world as well, thanks to their program's ability of seeing reality more clearly.

Most of that is in the not-so-near future. But that future is a lot nearer than "never".

Deep Blue is not at that stage yet, but the time is not far off that its brute force will give it the ultimate advantage over a human player. Not that it matters per se. Personally, I've never accepted the game. It is trivial and utterly useless, a waste of time, contributing absolutely nothing to humanity, just as other entertainment such as television, soccer or getting drunk.

Ah, yes, also note that there is basically no difference between Deep Blue and a PET 2001. They are both perfectly able to defeat a human chess player. Of course, Deep Blue with its vastly larger memory and far from marginally higher speed, would solve the problems a couple of centuries faster than the PET, but that is a quantitative difference, not a qualitative difference. The PET was actually probably more reliable than Deep Blue, and that is a qualitative difference.

In computers as we know them, the architecture is ultimately unimportant. Even the program efficiency is ultimately unimportant. It is relatively easy to design self-modifying evolving programs. Our human brain has those abilities as well, but is far less flexible and a lot slower in doing so.

Whether we like it or not, computers (or rather, computer programs running on some device) are nearly certainly the next step of evolution on our planet. From where we are standing, this may seem undesirable, more or less the same way rats, aids-viruses and Siberian tigers probably perceive us as rather unwelcome. From a more general standpoint however, it is neither good or bad. It is only bound to happen.


[Edited at 2003-10-30 02:19]


Direct link Reply with quote
 

two2tango  Identity Verified
Argentina
Local time: 08:03
Member
English to Spanish
+ ...
Still nor afraid of computers Oct 30, 2003

Bart B. Van Bockstaele wrote:

For many people, "never" means, "not in the near future" or "not in my lifetime" or sometimes even, "not in the next couple of hours". For me, "never" means "never": not before our universe existed, not during its existence, and also not afterwards. Never. Something may happen after I die, therefore, it will not have happened during my lifetime, but it will still have happened, and therefore the word "never", in my opinion, does not apply.



You are right about the relativity of the word "never". As an Engineer I tend to concentrate on practical approachs.

Predictions ranging in the thousands or millions of years fall into the realm of faith. I am extremely open-minded in matters of faith.

Now, in practical terms, I maintain my opinion that computers will not replace translator unless language is reduced to the trivial level that the minimal "intelligence" of a computer can aprehend.

Enrique

[Edited at 2003-10-30 13:34]


Direct link Reply with quote
 

Narasimhan Raghavan  Identity Verified
Local time: 16:33
English to Tamil
+ ...
If you talk of eternity, anything is possible Oct 30, 2003

We cannot and need not think beyond the next 50 years. I believe by that time not many of the present Proz.com members will be still members, if at all this portal is still there. These are all philosophical questions left best to philosophers. Within these limitations it is felt that computers cannot duplicate human beings. That's all.
Regards,
N.Raghavan


Direct link Reply with quote
 

Heinrich Pesch  Identity Verified
Finland
Local time: 14:03
Member (2003)
Finnish to German
+ ...
It's not about words Oct 30, 2003

Translation only uses words and sentences, but we do not translate them. We do not even translate languages, but thoughts, concepts and (practically mostly) messages. So there is no way that a symbol processing machine can ever do this, because it does not get the message.
In the virtually boundless continuum of concepts words are only signposts, every language puts its signpost on other locations.
I'm more afraid that sometime everyone would understand English, so most messages would need no translator.

[Edited at 2003-10-30 10:20]


Direct link Reply with quote
 

Narasimhan Raghavan  Identity Verified
Local time: 16:33
English to Tamil
+ ...
You have touched a sensitive point Oct 30, 2003

Already the majority of the Germans, who come to India speak passable English and one source of interpreting job is gone with the wind. Only the French, who insist on speaking their mother tongue, keep us interpreters occupied.
N.Raghavan

Heinrich Pesch wrote:

Translation only uses words and sentences, but we do not translate them. We do not even translate languages, but thoughts, concepts and (practically mostly) messages. So there is no way that a symbol processing machine can ever do this, because it does not get the message.
In the virtually boundless continuum of concepts words are only signposts, every language puts its signpost on other locations.
I'm more afraid that sometime everyone would understand English, so most messages would need no translator.

[Edited at 2003-10-30 10:20]


[Edited at 2003-10-30 15:16]


Direct link Reply with quote
 

PRAKAASH  Identity Verified
India
Local time: 16:33
Member (2007)
English to Hindi
+ ...
Human translators are best, no doubt! Nov 2, 2003

Heinrich Pesch wrote:

Translation only uses words and sentences, but we do not translate them. We do not even translate languages, but thoughts, concepts and (practically mostly) messages. So there is no way that a symbol processing machine can ever do this, because it does not get the message.
In the virtually boundless continuum of concepts words are only signposts, every language puts its signpost on other locations.

[Edited at 2003-10-30 10:20]

I'm totally agreed with my friend as we all know that it's the man himself who created computers and translation softwares but even he can't put his mind or say capability of thinking in a broader and positive frame in the softwares. It is a such a great challenge that computer can never reach up to a man's mind.
Translator uses his mind with full concentration and his lifelong experiences. It may be possible that computers'll provide alternatives, can never be as best as a human translator.
PRAKAASH
prakaasharmaa@rediffmail.com


Direct link Reply with quote
 


To report site rules violations or get help, contact a site moderator:

Moderator(s) of this forum
Fernanda Rocha[Call to this topic]

You can also contact site staff by submitting a support request »

Why human translators can never be replaced

Advanced search






LSP.expert
You’re a freelance translator? LSP.expert helps you manage your daily translation jobs. It’s easy, fast and secure.

How about you start tracking translation jobs and sending invoices in minutes? You can also manage your clients and generate reports about your business activities. So you always keep a clear view on your planning, AND you get a free 30 day trial period!

More info »
Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »



All of ProZ.com
  • All of ProZ.com
  • Term search
  • Jobs