“Time flies like an arrow, and fruit flies like a banana.”
—>
“Die Zeit vergeht wie ein Pfeil, und die Frucht fliegt wie eine Banane.”
“…However, it can’t understand the nuances in the way we convey meaning - that requires abstract thinking. When I say it is raining cats and dogs in English I mean it is pouring down with rain I do not mean cats and dogs are literally falling out of the sky!..”
This is true. However even if a machine could accurately detect idioms, etc, I still don’t believe that this would get to the real root of the problem - which is the quality of the target text. As far as I can see, a computer just isn’t going to be a skilled writer!
Often it’s about artistic judgement calls. There are generally several acceptably accurate and articulate ways of rendering something in translation. The one that we would want to choose has to do with all kinds of things like the required register (is it spoken or written language? What is the intended level of formality or informality?) Beyond that, it also has to do with the “rhythm” or “flow” of a text - the way it reads, if you will. Aesthetically speaking, something might just have a good “ring” to it - especially in a particular context and with respect to the surrounding text.
How in a million years is a computer ever going to achieve any of this!?
There is also the question of the cultural context of the source text - and whether the translator has a reasonably competent knowledge thereof. For example, even the short passage that I posted above from Thomas Mann might be misunderstood in a couple of subtle ways if one didn’t realise that pupils in German schools finish their classes at midday. One might think that the pupils are going home in the twilight, or that they are going home for evening meal instead of lunch, etc.
So, no, I don’t see machine translations ever really cutting the mustard myself.
(Or, as Google translate would probably tell folks in China or Japan: “I always thus fail to notice translations truly cutting down mustard trees with machines”…)
Wonderful!
I think we need to understand how the parts of AI or machine learning we are exposed to via the internet, smart TVs, mobile phones etc. is literally just the tip of a very very big iceberg. There exists even now technologies far more advanced than what we have access too, it’s just either it was developed by the military and is secret or simply that it takes years to bring them to market and satisfy all the regulations.
Part of this advancement is because we are no longer programming computers in a prescriptive, recipe like manner as we use to, although a lot of ‘dumd’ programs still use such algorithmns. Instead, in the field of AI and machine learning, we are programming computers to learn independently of humans, to teach themselves and then teach other programs. The more exposure these autonomous programs receive the more they learn and the more they teach to others connected to the same network. Driverless cars, for example, learn with every journey but they will also share that experience with other driverless cars in its network. I am currently working on a program that monitors a person’s medical behaviour and predicts when they will next require a hospital appointment. Again it is the computer that will learn from a person’s past behaviour and make informed decisions using probability theory and by using thousands of other people who are similar in their behaviours and who have been to hospital as role models to learn from.
It has been estimated that in 2005 we turned the corner on the exponential curve tech development has been following for 40 or 50 years. As such advancements will come more quickly and I can easily see in another five to ten years that Google translate will be very accurate, even understanding subtle nuances in speech and will sound as though spoken by a native speaker. The new Turing test will be when listening to a robot, can we tell the difference between it and a human just by listening. Already developments are taking place in teaching computers “emotion” for want of a better word. In fact I wouldn’t be surprised to hear that a lot of tech we currently say is impossible or that is years away is actually already sitting in some research lab.
Driverless cars is an interesting topic. It’s not at all hard to see how it might work in theory, but I don’t know whether there would be public acceptance? It wouldn’t take too long, probably, before a driverless car would “decide” to avoid hitting a bus by mounting a pavement and killing a mother and young child instead. At that point, I can imagine they would be widely viewed as a public menace - fairly or unfairly.
I would suppose that pilotless airliners are even more feasible? The necessary technology may already exist, pretty much. But who would buy a ticket? (I sure as hell wouldn’t!)
Cars and planes aside, it seems to me that translation and textual composition are fundamentally different to automated systems operation, because they involve aesthetic understanding and judgement. Of course only time will tell, but I’ll believe a computer can translate literature to a remotely acceptable level when I see it.
Dude!
I’m intrigued PrinzCuck, burying your head in the sand is so unlike you - or so I thought 
I have a nasty gut-feeling it’d be my charred corpse that would wind up “buried in the sand” if I booked a trip in a pilotless plane! 
(And without my eagle-eye to keep watch, those dastardly b_tard politicians would soon be nudging back from the outer limits of super-hardcore Brexit, and there’d be nowt to raise the alarm!!)
Can I see one of the perfect translations produced by google translate, please? Thanks
I’m willing to believe that Google translate might do okay with simple plain sentences. One might be able to translate something like: “I just parked my car in the multi-storey carpark” (Or “multi story parking lot” as we would say in the land of the Donald.)
Fine. But how about if the writer was getting all arty-farty, Thomas Mann style, and saying something like: The clock had marked but little time since I entrusted my chariot to the cavernous Wotan’s temple of cars" (!)
The day that a computer could translate THAT without blowing a microchip would be the day it could compose weird and pretentious literature (sorry Herr Dr Mann!) And I just can’t see it happening - at least not in my lifetime.
don`t sweat
I agree, but how much of translation consists of literary translation jobs?
In truth many writers use language artistically to a greater or lesser degree - in things like advertising, etc.
Real high end literature probably doesn’t account for much volume, although the folks who do this work are some of the most well paid.
(We also haven’t talked much about the more specialist areas like legal or scientific translation. In these areas the translator may need some background in law or science, as well as language competence.)
The Russian Wikipedia article on Poe’s “The Raven” has an extensive section about the many translations of the poem from English to Russian and the difficulty the translators experienced in conveying the the content and the subtexts and the atmosphere and the rhythm of the original. (Neither Google nor Yandex translations are mentioned. 