| Pages in topic: < [1 2 3 4 5 6] > | How long will human translation last? Thread poster: LucyPatterso (X)
|
|---|
Roy OConnor (X)
Local time: 00:19
German to English
| Increased non-employment | Sep 18, 2012 |
Phil Hand wrote: LucyPatterson wrote:
As machine translations improve, there will be less work for translators. No, there won't. Are there fewer people making cars these days? Are there fewer chefs? Is there less work for accountants, now that we have calculators (and sophisticated accounts programs)?
How soon will this happen? Is it already happening now?
I don't think many translators will actually become unemployed, but in the long term there will be fewer jobs available for translators. Does anybody remember switchboard operators, typing pools? And yes, before calculators and computers people had to do the work of number crunching. What happened to sprayers and welders working on cars when robots took their jobs?
There will be less work for translators in the future - it is only a matter of time, the writing is on wall....
| | | | Kay Barbara
United Kingdom
Local time: 00:19
Member (2008)
English to German
+ ...
| MT is overrated | Sep 18, 2012 |
Roy OConnor wrote: Phil Hand wrote: LucyPatterson wrote:
As machine translations improve, there will be less work for translators. No, there won't. Are there fewer people making cars these days? Are there fewer chefs? Is there less work for accountants, now that we have calculators (and sophisticated accounts programs)?
How soon will this happen? Is it already happening now?
I don't think many translators will actually become unemployed, but in the long term there will be fewer jobs available for translators. Does anybody remember switchboard operators, typing pools? And yes, before calculators and computers people had to do the work of number crunching. What happened to sprayers and welders working on cars when robots took their jobs? There will be less work for translators in the future - it is only a matter of time, the writing is on wall....
Roy, I think your examples do not fit the MT debate: you only mentioned dull, non-creative, repetitive tasks which were predestined to be performed by machines which are faster more accurate, reliable and cheap etc. I can't see that you have a point there.
Translation is much too complex a process to be performed by machines. As Jeff rightly pointed out, it's (mostly) based on statistics and probability and that's a very poor approximation of the translation process.
When there is an ambiguous word/phrase in the source text, you will have to look 2 sentences back or maybe 3 ahead or use common sense to know what the intended meaning is - you cannot arrive at a solution using maths or rule-based MT.
For expressive texts, the translator needs to reproduce the function (effect) of the source text in the target text - now I'd love to learn how machines are supposed to to that.
Human translation will last as long as human intelligence cannot be replicated - brute force number crunching will not help.
To be fair, I should add that MT'ed weather forecasts are as good as they can be ...
| | | | | Various replies | Sep 18, 2012 |
Tomás Cano Binder, CT wrote:
I completely disagree with this. If you are minimally conscious about the main aspects of a quality translation (meaning, style, spelling, localisation aspects, prevention of common pitfalls), MT definitely does not speed up work, but makes it not only more cumbersome and slow. It makes work also a lot less rewarding since a machine dictates the general guidelines of your style and wording. It sounds a bit sad to become a freelancer thinking that you will be your own boss... to become the servant of a machine translation tool.
I have improved the speed of my work with MT without having to compromise any of the above. I also find integrating MT into my work procedures to be an interesting challenge, which makes the rather routine translation work I do intellectually more demanding and rewarding. The texts I translate are technical and require expertise, but provided you have that expertise, the work is not cognitively taxing at all. Being a servant of the tool makes it more fun.
Shai Nave wrote:
When I wrote that professionals should charge a rate the reflects the work that they are doing I meant that the professional should not earn less per hour or per day of work just for using a certain tool (which is a process that already started by some stakeholders looking for MT-post editing for ridiculous rates although the work usually takes as much as time as translating the content from scratch, if not more), and also take into account one's investment in hardware, software, the time for maintaining that environment, and every other technology related investment (in time and money).
I mostly agree, I think the hourly rate of translating should remain comparable to that of other skilled trades. And I definitely don't encourage anyone to fall for badly paid post-editing jobs, but it's a good idea for translators to start trying out the numerous free or inexpensive MT tools that are available for their existing CAT systems (Studio for example has five or so MT plugins). The extra investment compared to normal translation capable systems is insignificant, but you are right that the time that has to be spent learning the tools and procedures is expensive. The obsession with the per unit rate is not going to disappear, although I hope the tools to estimate the units will improve (not every new word is equal, as we all know). Also, the prices will keep dropping in most fields, whether we like it or not. Remember two years ago when the crisis hit? Lots of people were intimidated into dropping prices then, even though it lowered their real earnings. Price drops due to improved efficiency will be accepted more easily, since the real earnings will only return to the old standards.
Shai Nave wrote:
Besides, when I wrote that I couldn't disagree more I mostly opposed your statement that now is the ideal time to switch to MT. First, not everyone wants to lose our unique, knowledgeable and professional voice and replace it with a generic solution. So MT, which is a tool, is not something everyone would choose to use. Secondly, I disagreed with your suggestion (to use a tool that increases one's efficiency while charging the same and riding this wave as long as one can), I think that it is unethical.
I think this is an admirable attitude, it's really good to see someone have loads of pride in their profession and a strong sense of fair play. I wouldn't worry about translators losing their voice, the profession is just going to become more and more technical as time goes by (this might even lead to the status of professional translators improving). We are going to acquire new competencies, which I don't think is a bad thing. I also don't think riding the wave is unethical, that's the way innovations are adopted in our society.
Jeff Whittaker wrote:
If there were no more human translations, machine translation could not improve.
Even if you totally disregard rule-based MT systems (which were the only systems until about 20 years ago), this statement isn't true. The current popular systems like Google Translate may be glorified sub-segmental translation memories, but more sophisticated systems are being developed. Automatic or manual annotation of existing corpuses (with morphological, syntactic and semantic information) will also be more useful than simply adding more text mass to the corpuses. Rule-based MT isn't dead either, for instance the MOLTO system seems promising.
Roy OConnor wrote:
There will be less work for translators in the future - it is only a matter of time, the writing is on wall...
This is true. There are two kinds of translators that will thrive in the future: the hardcore professionals with specialist expertise (they will always be needed), and technologically oriented translators who can produce massive volumes (compared to current standards) of bulk translations (but not so bulk as to be entirely machine translated).
| | | | Roy OConnor (X)
Local time: 00:19
German to English
| Don't ignore the power of science & technology | Sep 18, 2012 |
I think Kay describes the current situation very well. My prognosis though is for the long-term say in 10 to 20 years.
For a long time speech synthesis and particularly speech recognition were regarded as the Holy Grail, but these are now very common techniques, although by no means perfect. The same applies to MT in my opinion. The problem with speech processing in those days was that analogue filters were inadequate and digitisation was too slow. Today's problem with translating i... See more I think Kay describes the current situation very well. My prognosis though is for the long-term say in 10 to 20 years.
For a long time speech synthesis and particularly speech recognition were regarded as the Holy Grail, but these are now very common techniques, although by no means perfect. The same applies to MT in my opinion. The problem with speech processing in those days was that analogue filters were inadequate and digitisation was too slow. Today's problem with translating is probably more a question of algorithms and data bases or maybe even things like neural networks.
It's difficult to predict exactly what developments are around the corner, but they will come eventually. ▲ Collapse
| | |
|
|
|
Kay Barbara
United Kingdom
Local time: 00:19
Member (2008)
English to German
+ ...
| Controlled language | Sep 18, 2012 |
In order to give a bit more balanced view, I felt I should mention the use of controlled language to achieve workable MT results. So in ideal settings MT may be a viable choice of translation. However, controlled languages have been around for quite a while I think but their applicability seems to be not without problems.
Even if those ideal environments were to be adopted by more (big) companies, I still don't see a significant impact on the translation industry in the foreseeable future... See more In order to give a bit more balanced view, I felt I should mention the use of controlled language to achieve workable MT results. So in ideal settings MT may be a viable choice of translation. However, controlled languages have been around for quite a while I think but their applicability seems to be not without problems.
Even if those ideal environments were to be adopted by more (big) companies, I still don't see a significant impact on the translation industry in the foreseeable future. ▲ Collapse
| | | | | Just enjoyed the giggle of the day | Sep 18, 2012 |
I just had to share this!
I occasionally collaborate with a Danish colleague on marketing.
We had a Danish text about underwear. Not the flimsy, glamorous sort: this was designed to control a mature, buxom figure to make the rest of the fashion house's designs more flattering.
Today my colleague forwarded a mail from the client:
Someone in the back office had run the Danish through Google Translate and found a very different result from ours. It was not ev... See more I just had to share this!
I occasionally collaborate with a Danish colleague on marketing.
We had a Danish text about underwear. Not the flimsy, glamorous sort: this was designed to control a mature, buxom figure to make the rest of the fashion house's designs more flattering.
Today my colleague forwarded a mail from the client:
Someone in the back office had run the Danish through Google Translate and found a very different result from ours. It was not even close to the original, quite subtle Danish...
I had composed a piece with expressions like 'firm up the curves' and 'garments sit snugly', and I used the word 'figure'. (As on the bottom line, so it was marked WRONG WORD.)
'We have designed' (my suggestion) was changed to 'The clothes were designed in order to...' (Introducing a passive - you can't win!)
No more firming up curves, folks.
"The garments are touching your body and tighten up as needed..." 
Edited to add that the client is sending the spring catalogue to us as usual - and not just putting it through Google Translate...
[Edited at 2012-09-18 14:49 GMT] ▲ Collapse
| | | | Jeff Whittaker 
United States
Local time: 18:19
Spanish to English
+ ...
| John Fossey
Canada
Local time: 18:19
Member (2008)
French to English
+ ...
Roy OConnor wrote:
What happened to sprayers and welders working on cars when robots took their jobs?
There will be less work for translators in the future - it is only a matter of time, the writing is on wall....
Having been in this industry, I totally disagree with this view, and I think it is a good argument for why human translators will never disappear.
Today, if you are a plant that needs to hire painters and welders, you will be hard pressed to find them. It's not that they are not out there, far from it, but the demand far exceeds the supply. Those jobs that the myth says were taken over by robots were the jobs that were not suited to humans anyway - repetitive, mind-numbing, dangerous jobs that took skill to do perfectly but were hard for a human to maintain a quality level all day long. Those welders went on to learn how to program and supervise the robots, using the skill in welding technology that lay behind their trade but becoming welding technicians and engineers. Productivity went through the roof and costs came down for the finished product, but the income of those welders is now far higher than in the pre-robot age.
The same will happen in the translation industry. Modern technological tools will make humans more productive, require more skill, dramatically improve throughput but at the same time raise translator's income, as they become worth more due to their higher skill level. Unit prices for translation as a whole will drop, but I believe we're going to see an increasing range of prices - lower prices for easy work that requires a translator to oversee the machine, such as tables of part's lists and such, and much higher prices for things that really matter and need translator's skills, such as the legal documents and contracts that come about from the business opportunities that arise when the language divide is reduced.
[Edited at 2012-09-18 14:20 GMT]
| | |
|
|
|
John Fossey
Canada
Local time: 18:19
Member (2008)
French to English
+ ...
| MT becoming GIGO | Sep 18, 2012 |
Jeff Whittaker wrote:
Machine translation is based on matching existing translations. If there were no more human translations, machine translation could not improve. In fact, the more bilingual machine translations that appear on the net and the more error-filled bilingual documents submitted by incompetent human "translators", the more the system collapses in on itself. Who will be left to program the machine?
A very important point, and one that I think will be a limiting factor to the development of machine translation.
It can already be seen - MT results in a flawed translation - that erroneous translation is posted on the net as a bilingual document - the MT web crawlers pick it up and use it as input for more MT - resulting in compound flaws. After a while the "statistical" MT model starts reporting that this translation has a certain statistical likelihood of occuring and "institutionalizes" the bad translation.
If you use a web sourced resource such as linguee.com, you can already see this happening.
As is said in all computer related things: GIGO - Garbage In, Garbage Out
[Edited at 2012-09-18 14:34 GMT]
| | | | Roy OConnor (X)
Local time: 00:19
German to English
| The demise has already started | Sep 18, 2012 |
In reply to John, the demise has already started. How many young upcoming translators are finding it difficult to find work? Proz's pages are full of such people desperately seeking advice.
Also, those of us who have been translating for many years can compare how efficiently we work compared to those early days when computers were primitive and there was no web or internet. Ten of us now maybe do the work of 15 or 20 translators in those days.
True there will always be... See more In reply to John, the demise has already started. How many young upcoming translators are finding it difficult to find work? Proz's pages are full of such people desperately seeking advice.
Also, those of us who have been translating for many years can compare how efficiently we work compared to those early days when computers were primitive and there was no web or internet. Ten of us now maybe do the work of 15 or 20 translators in those days.
True there will always be a need for a certain amount of human translation. But there just won't be room for the same number of workers in the profession. ▲ Collapse
| | | | Phil Hand
China
Local time: 06:19
Chinese to English
It's possible that there's been a drop in the amount of work available for German translators, though frankly I doubt it. (Do the Common Sense Advisory have any figures?)
But my pair has gone from about zero 30 years ago to being probably the second largest pair in the world today. The amount of trade between China and the world has driven this first wave of demand; greater cultural and social integration will drive a much larger second wave.
I do agree that we're more... See more It's possible that there's been a drop in the amount of work available for German translators, though frankly I doubt it. (Do the Common Sense Advisory have any figures?)
But my pair has gone from about zero 30 years ago to being probably the second largest pair in the world today. The amount of trade between China and the world has driven this first wave of demand; greater cultural and social integration will drive a much larger second wave.
I do agree that we're more efficient than we used to be, thanks to computers. And quality has improved. But from where I'm standing, the quantity of work is rising exponentially. And whatever MT might be achieving in French and German, it's nowhere with Chinese. Like John said, there isn't a base of existing quality translations to build MT systems on yet. ▲ Collapse
| | | | Yaotl Altan
Mexico
Local time: 16:19
Member (2006)
English to Spanish
+ ...
| The cloud and TMs | Sep 18, 2012 |
Phil Hand wrote:
...
I do agree that we're more efficient than we used to be, thanks to computers. And quality has improved. But from where I'm standing, the quantity of work is rising exponentially. And whatever MT might be achieving in French and German, it's nowhere with Chinese. Like John said, there isn't a base of existing quality translations to build MT systems on yet.
But, colleagues:
Don't you consider we are already creating a base of quality translations if we, as a whole, keep using the cloud and creating TMs for others?
Kind regards!
| | |
|
|
|
Sheila Wilson
Spain
Local time: 23:19
Member (2007)
English
+ ...
| Growth market for quality translations | Sep 18, 2012 |
Roy OConnor wrote:
In reply to John, the demise has already started. How many young upcoming translators are finding it difficult to find work? Proz's pages are full of such people desperately seeking advice.
Some of those new translators are undoubtedly having a hard time simply because it is, and always has been, difficult to establish oneself as a freelancer - whatever the field.
But many others never would have been translators in an earlier generation. That's one thing we are seeing - not less work for the same group of pros; but more work for an infinitely larger group of translators calling themselves 'professionals'. It's just so easy: register on one of the tens (hundreds?) of sites where people come to look for translations (dedicated sites like this one or more general 'get work done' sites). Often they don't have to input many details let alone put their hands in their pockets. Five minutes later they are in business (of a sort). Those newbies who pitch their rates at the same level as these 'five-minute-startup' freelancers will never find satisfying work as rates get so close to zero that clients question the value of paying anything at all over and above the few minutes it takes to put the text through an MT.
On the other hand, those who pitch their rates for quality, researched translations will, if they're lucky, have a career ahead of them, if they can maintain that quality. That's what I believe, anyway.
Also, those of us who have been translating for many years can compare how efficiently we work compared to those early days when computers were primitive and there was no web or internet. Ten of us now maybe do the work of 15 or 20 translators in those days.
True there will always be a need for a certain amount of human translation. But there just won't be room for the same number of workers in the profession.
I don't agree. E-commerce is growing exponentially and everyone who trades MUST have a web presence nowadays. That doesn't just mean website translation, it means international clientele and all that entails. I can see the amount of translation needed growing all the time. And while some will be done by MT and the cloud, there will always be lots left for real translators.
For young translators, though, there are some requirements that I personally, at age 56, don't fulfil: Master-level degree in translation studies + other qualifications in languages and linguistics (to prove a real commitment to the profession and a basic competence, and to set themselves apart from those 'five-minute-startup' peers); and very advanced IT and DTP technical skills so they can make full use of every new piece of kit available and can provide a quality result whatever the client's preferred format. I believe there will be less and less work for those capable of producing a fine translation but who don't have the other skills. But us pre-IT dinosaurs won't be in the job market too much longer!
| | | | John Fossey
Canada
Local time: 18:19
Member (2008)
French to English
+ ...
| Growing demand for more skilled translators | Sep 18, 2012 |
Roy OConnor wrote:
Ten of us now maybe do the work of 15 or 20 translators in those days.
That's only 1.5 to 2x the amount of work to be done !
I haven't seen any figures, but I am quite sure there is far more then twice the volume of translation being done now. The world has become global, while in those early days, business stopped at the border, protectionism was the order of the day, and noone knew or cared what happened in other languages and cultures, which were considered "strange". Small organizations certainly had no thoughts of crossing the language divide - now it's routine at all levels.
The difficulty newcomers have breaking into the field just reinforces my point - more experience and skill is needed, making it more difficult for newcomers but it doesn't mean there's less work, far from it.
| | | | brunoccj
Taiwan
Local time: 06:19
English to Chinese
| human translation will last for a long time - at least for my pair. | Oct 24, 2012 |
I'm a Eng>CHT translator, and are doing post-editing for MT.
MT produces CHT target text out of its CHS data, and it's laboring to edit it.
I almost translate from scratch each and every TU.
Even if this is an Eng>CHS job, the MT doesn't suffice because the word order of ENG and CHS are so different.
As for translation of pairs of western languages, human intervention is still needed, until the day AI - which itself requires huge amount of hu... See more I'm a Eng>CHT translator, and are doing post-editing for MT.
MT produces CHT target text out of its CHS data, and it's laboring to edit it.
I almost translate from scratch each and every TU.
Even if this is an Eng>CHS job, the MT doesn't suffice because the word order of ENG and CHS are so different.
As for translation of pairs of western languages, human intervention is still needed, until the day AI - which itself requires huge amount of human input - becomes a reality. ▲ Collapse
| | | | | Pages in topic: < [1 2 3 4 5 6] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » How long will human translation last? | Protemos translation business management system | Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!
The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.
More info » |
| | Trados Business Manager Lite | Create customer quotes and invoices from within Trados Studio
Trados Business Manager Lite helps to simplify and speed up some of the daily tasks, such as invoicing and reporting, associated with running your freelance translation business.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
