Language of the Future

[Kennigit]

So, we all know that in the future, translation software will only continue to get better. There are various goals that can be set--or at least imagined--and they may or may not happen sequentially, in parallel, or even at all.

Terminology​

​

​

Spoiler :

translation vs interpretation

Generally, "translations" means 1:1 accuracy or very close to it, whereas "interpretation" generally means "getting the jist/approximate meaning of it". Some people like to be very detail oriented and point out that most real-time or live translations basically need to be interpretations, e.g. live translating an idiom or phrase unique in one language, rather either translating literally or having the time to really translate the idiom or phrase into a new phrase that entirely captures the meaning properly.

Constructed language

A constructed language would be something like esperanto, which has no historial basis in terms of culture and is constructed solely for the purpose of trying to get a "universal language" or "universal secondary language"

Rosetta stone

One text translated to many other languages.

Kennigit generated postulates (KGP)

• A "universal translator", I would say, would be expected to occur before the ability to implant devices into the brain that can "install" a language to the human.
  
• A "universal translator working through text" would be expected before a cochlear implant or other hearing aid is able to recognize a human's speech and translate after a short real-time delay.
  
• A "unversal interpretor working by hearing a third party's voice" (i.e. the listener should hear the translation dominate the volume received--such as having a headset at the UN where you listen to your interpretor) would be expected before some voice modification where you hear the speaker talk to you in your own language, rather than hear a third party voice (in the future, will be entirely robotic) generate the spoken words. Or, of course, before an implant that allows the speaker to talk in your language.

The list goes on, in both scope (what are future goals) and timeframe (what will be achieved and when, and what first?). But, for instance, it may be pointless to try to develop a translator that uses the original speaker's voice and transmits it in some translated way by adjusting all the proper frequencies, etc, to reconstruct the same tone with other words (otherwise you have to have a prerecorded bank of the speaker listing out vocab words) if everyone is satisfied hearing a third party voice.

Current discussion and technologies:​

I have spent 2 minutes and 23 seconds learning about this subject (coincidentally, the length of this video

Link to video.​

The process of google translate is to look at translated documents and try statistical tests to have pattern recognition. That is, the translation works by sampling larger and larger data sets, and is a monte carlo method (not deterministic, i.e. the computer does not know the rules of every language but guesses it). The video should explain better.

Spoiler additional discussion :

Thus, perhaps translating Swahili to Japanese may not be very effective, as there are probably not many translated texts between swahili and japanese. I'm sure google's pattern recognition may make use of secondary translations, e.g. perhaps there are many more english to japanese translations than spanish to japanese, so a step of spanish -> japanese may flow as spanish -> english -> japanese, or to make use of a "rosetta stone", let's say a chinese document all translated into english, spanish, and japanese, and use that document to compare the three together to construct rules for spanish -> japanese (rather than look for samples of spanish documents translated directly into japanese)

Pictures perhaps work better:

I assume (though will not google it) that most professional translators are supposedly very knowledgeable in 2-3 languages, and work by knowing the "rules" of those languages entirely.

This thread and the future:
​

I'll ask a few questions to prompt discussion (my answers in spoilers):

1) timeframes? 10 years? 20? 50?

Spoiler :

I certainly expect in 25 years the equivalent of google translate to work very fluently and not be the semi-clunky mess it is now (similar to how subtitles on videos are clunky--e.g. youtube scripts, though I don't know how closed captioned seemed to have done things well and was from a long time ago).

2) Is universal translation the final goal, or is it to make everyone a native speaker in a language (e.g. esperanto, a constructed language).

Spoiler :

"Computationally" speaking, if we ever have the ability to "install" a language to a person, there is no reason to ever need to install chinese, korean, spanish, english, etc: just install everybody with 1 language. "Translation" would be a thing of the past. However, due to the cultural importances associated with language & the hope to preserve languages, I doubt this will ever be achieved

I also think that is a huge neurobiological task that I don't think will happen. There can be tools to help children or adults learn languages better, but I don't think we'll be able to just make a 6 year old polyglot

And why would we, since I think a hearing-aid like device would work better than making the user "learn" the language.

3) Will a secondary, constructed language, like esperanto, ever exist in a widespread sense once universal translation becomes more realistic?



3b) Will a secondary, constructed language, like esperanto, be human derived (such as esperanto) or exclusively in computer pattern recognition?

Spoiler :

3: I don't think so. Humans are lazy.

3b:

Esperanto was/is created trying to blend in mostly western languages afaik (e.g. french, spanish, english). Any human attempt at constructing a language also will be biased. A computer could make up a language today, that would perhaps be the most difficult language ever for a human to try to learn, but it could facilitate creating a "rosetta stone" for the computer to be able to quickly translate any language to any other language, so long as the language X <-> constructed language link is done accurately. This also would easily allow for any new language (aliens, old discovered languages, whatever) to be translated to any other language if the first step can be established.

I don't think this will happen personally. I think the research effort would go towards "patching" things where needed, e.g. say korean <-> chinese works well, the goal will just to get chinese working and then add in korean through that.

tl;dr: A general solution is not expected to me (i.e. no universal secondary language, either in computers or in humans), which brings me to

4) What will be the cutoff for "universal"?

Question 3 would establish true universal language. I, however, think it will just be done for the top 10 used languages, plus maybe 1 or 2 to try to help out w/ ancient texts (e.g. Greek), and call it a day.

5) How widespread will this be implemented? Only lofty UN meetings? The common man?

Spoiler :

obviously text translation is anyone with the internet. But for instance, if we had cochlear implants available, would people bother to get it? Only those that do foreign business? Only the rich? Is there counter-culture (i.e. "Speak American?")

What do you think are the goals of universal translation (scope), and what do you think will be achieved--either technologically or in actual implementation​

 

[Kouvb593kdnuewnd]

In the far future we may not even have any need of any language at all, because in the end language is only a method to transfer information.
What if we could transfer the information to each other in some more effective way like a computer can do.

 

[Kennigit]

In the far future we may not even have any need of any language at all, because in the end language is only a method to transfer information.
What if we could transfer the information to each other in some more effective way like a computer can do.

In the far future there is no reason for humans to exist if robotic AI can entirely capture the human experience

sacks of meat will lose out to technology in the end, and I would expect it to be voluntary

 

[Wolfbeckett]

In the far future we may not even have any need of any language at all, because in the end language is only a method to transfer information.
What if we could transfer the information to each other in some more effective way like a computer can do.

I guess it depends on how you define language. In your hypothetical future I would say that whatever data transfer protocol is being used to send and receive the information would be the de facto "language". If, as you say, language is only a method to transfer information, than a data transfer protocol is just as much a language as any spoken form.

 

[Phrossack]

Language is one of the most powerful elements of cultural identity. A world with one language is like a salad bar that serves only lettuce. It is a clothing store in which everything is size medium. It is an art supply shop that only has red paint. In short, it is singularly drab and awful. I want to learn many languages--Mongolian, French, Russian, Arabic, Georgian, Lithuanian, Hungarian--not necessarily because they're useful, but because they're fun to learn, and give life a different flavor.

As for replacing humans, I will die fighting that concept. I'm quite comfortable being an animal species with both animal wants and needs and the ability to critically think about them.

 

[Hygro]

Whatever gets me the most sex with the hottest babes is what we should strive for.

 

[gay_Aleks]

So.. Swedish? Would it be Swedish?

Or, speaking purely numerically, Sanskrit or Chinese?

 

[Evie]

I'm with Phrossack. While I see translation as progressing in the future - I wouldn't be surprised if your timeframes were a little pessimistic, - I think such translation would be a tool of convenience, and someone truly wanting to learn about a culture, a people would still want to learn the language. For much the same reason I prefer to read books in English rather than the french translation (or in French over the English translation)

And indeed, the world would be much poorer if universal translation became universal language.

 

[cybrxkhan]

I think maybe some sort of universal translator might be useful for mundane things, like translating instruction manuals or something, but it can't replace languages... not for a while, at least. I have to agree with Phrossack and Oda here, even if a universal translator is available I don't quite think it will be able to capture the cultural aspect of language that well. Culture is a very important aspect of language - and a lot of things about the human experience, really - that a lot of people don't realize.

Whatever gets me the most sex with the hottest babes is what we should strive for.

A worthy and noble goal.

 

[Kyriakos]

I don't think it is realistic to ever expect a universal-fast-accurate translator for even as few as 10-12 languages with actual differences in grammar and syntax between them.
Sometimes the inherent differences in the dynamics of such languages renders auto translation impossible, even if no idioms are used.
Maybe if they are all stripped down to some very base level, then it can happen. But surely that would be a horrible idea to begin with

PS: I think it can be said that auto language translating is not outside of the scope of the famous 'Does P equal NP' computer programming question. And the (not proven) answer to that seems to be an echoing No.

 

[Borachio]

The universal language of the future has to be English, imo. Though it might be Mandarin.

Having one language which is universally understood is just so much easier than translating all languages amongst each other, even given a truly effective Google translation tool. Which seems to be a long way off, despite the hype.

Still, even given English as a universal language doesn't mean that all other languages disappear. Indeed, you then get a multiplicity of Englishes. With people's mother tongues still remaining.

It's a complex subject, though. So who knows?

But in any case, Jesus spoke Elizabethan English. So that should be good enough for the rest of us. It seemeth.

 

[Winner]

I am afraid machine translation will never achieve true accuracy without an A.I. brain doing some real thinking. So far, it relies on statistical/analytical methods, which can be very useful in many fields (especially in controlled, technical environment settings), but what we get is still an output of an unthinking machine that a 4-year-old child will beat easily.

Example:
The monkey ate the banana because it was ripe.
The monkey ate the banana because it was yellow.
The monkey ate the banana because it was noon.
The monkey ate the banana because it was hungry.

How do you "explain" to an unthinking machine what "it" refers to? What the machine sees is a sentence element that is essentially the same in all cases, even though to human eyes the difference in meaning is quite substantial. So far there are no easy machine translation solutions to this "messiness" of human languages. (Thank the Gods; I am a translator ).

So, to your questions:

1) Unless true A.I. is developed and it co-operates, no machine translation device will ever be sufficient as a tool of everyday human-to-human communication. It can assist you on holidays or when you communicate over the net, but can you imagine dating a woman if you're only able to communicate your meanings to her through Google translate?

2) I think simple economics will force people in the future to be more skilled at foreign languages. Obtaining a working, communicative knowledge of English (there doesn't seem to be any better candidate at the moment, unless there is some serious social engineering project waiting for us in the future) is not that difficult; with proper schooling methods and more cosmopolitan way of life, people will naturally gravitate towards using some universal "lingua franca".

3) I don't think a constructed language will work aside from communities of enthusiasts (but I wish them luck, Qapla'!). That being said, I do believe English as we know it today will become FAR MORE INTERNATIONALISED - i.e., when enough non-native speakers speak it almost daily all around the globe, the language will essentially be seized from the hands of the native speakers. New rules, vocabulary and pronunciation standards will gradually form, perhaps converging in something some scholars have called "World Standard English". The "native Englishes" will then continue on their own trajectory, separating them from the World English.

But this will happen naturally, rather than as a result of some concerted, conscious effort by language engineers

Hope this more or less covers this topic

 

[gay_Aleks]

We already have an universal language.

Unfortunately, some people decided to make several more universal languages, and it all got messy from there.

I am talking about programming languages, of course.

 

[Kyriakos]

We already have an universal language.

Unfortunately, some people decided to make several more universal languages, and it all got messy from there.

I am talking about programming languages, of course.

I c# what you did there.

 

[Tee Kay]

Not gonna change predicting anything in the future beyond about, say, 5-10 years. I'll just say that the world is moving ever faster, the world in 50 years will be nigh unrecognisable in every aspect and kinda it makes me depressed thinking about it.

 

[Kyriakos]

^Apart from the internet itself (and some developments tied to it), what else changed that hugely between the 80s and now? (if the 60s are taken as the start there, there would be also the popular use of computers).

Not that i aim to downplay the internet-induced change, but there is a clear difference there with the 'real world' change.

I think that the mass increase in actual possibility for change, tends to balance out at less pronounced change (less identifiable as the 'same' change for all). By itself this is not negative either. I like the internet, despite that a lot of it surely contains horribly base stuff.

 

[Tee Kay]

^Apart from the internet itself (and some developments tied to it), what else changed that hugely between the 80s and now? (if the 60s are taken as the start there, there would be also the popular use of computers).

No Soviet Union
The Asia-Pacific ex Japan became economically dynamic
Triumph of neoliberalism
Decline of noninternet media
Western deindustrialisation
Increased global movement of trade and people
Militant Islam
Fracturing of traditional identities

The most consequential change though is easily the growth of computing, automation, and the internet. And it's not like these don't have massive real world consequences in economy, culture, politics. Think of all the jobs that have been or will soon be made redundant by automation, for a start.

 

[Kyriakos]

^I agree with all that, but i was thinking of actual changes in the external world in terms of new types of tech. In the 20th century you have the invention of airplanes, computers, in the 19th use of electricity and radio, and in the 18th the commercial steam engine. In the 21st century we are in there doesn't seem to be any obvious new discovery being neared (even the internet was already there in the 90s, but furthermore it has been mentioned, and is part of computers and communication systems). Surely the existent techs are extended, and tie to each other in new manners, but there is no actual new force harnessed or tech practised?

 

[gay_Aleks]

I hope you realise only 15 years have passed since the 21st Century is a thing.

 

[Kyriakos]

i hope you realise only 15 years have passed since the 21st century is a thing.

&#925;&#959;.



(Also: bring back tertiary system computers).