Al la enhavo

Machine Translation vs Esperanto

de Vestitor, 2013-januaro-06

Mesaĝoj: 22

Lingvo: English

Vilius (Montri la profilon) 2013-januaro-06 23:25:35

sudanglo:Judge it by what will be able to be achieved in a few years time.
I'm afraid it's decades we are talking about, not years.

Vestitor (Montri la profilon) 2013-januaro-06 23:32:17

sudanglo:
For machine translation to be useful like real language interplay it would need to be quicker
Imagine the following Vestitor - Google is working on this already, as I understand it.

You take out your smart phone and speak into it (the phone has been trained on your voice). The phone converts speech to text and invokes translation to the target language (centrally, or in the phone). This is transmitted to the smart phone of the person you are talking to where it is converted to speech and the other party then hears what you have said in his/her language.

This is just like having a normal conversation with the exception that what you say is heard by the other party in his/her language.

Is that convenient enough for you?
Is it? I'm not sure it is. I don't think I'd want to chat up a girl like that, conversation it ain't. Have you listened to the speaking voice on Google Translate, it sounds vaguely naturalistic, but it's a lot worse than even Stephen Hawking's voice synthesiser in its enunciations.

And then there's the old 'what do you do when technology isn't at your fingertips?' question. You pop out for some milk, you see a guy on the street who asks you something (in a foreign language you don't know), but you've left your trusty smartphone - solution to all problems in an age where people have stopped thinking creatively about how to solve their own problems - at home. Chances are high that the day will be saved by some level of English.

Referring to what Oijos pointed out the problem is compounded if the fellow has no English because translation efforts invariably have English as their core.

Now I readily admit that this is not identical to the question of how good actual machine translation could become, but this is because I think it is not even on the horizon. Google, the most popular translator, takes nearly all of its languages through English first to reach another target language -this is the source of many errors. Another cause of errors and confusions is the complete lack of grammar application. Transferring a faulty tool into a shinier, more convenient box is not an improved solution.

I like google translate, I use it, but like everyone else I know it has limitations, some of which I probably can't even judge as erroneous.

erinja (Montri la profilon) 2013-januaro-06 23:38:32

I think this conversation perhaps isn't distinguishing clearly between the various reasons why people use languages. Even a perfect machine translator is suitable for only a certain subset of these situations.

Good situation for a perfect machine translator:
- Tourist who doesn't speak the language needs to order in a restaurant, buy a ticket, or ask to find a toilet.
- Scientist wants to give a talk to speakers of another language
- Internet user wants to watch an online video in another language

Bad situation for a perfect machine translator:
- Making a friend (Do you really want all of your personal conversations going through a machine? What if the battery dies?)
- Forming a relationship and living with someone (so... you're going to have your translation unit between you in bed at night?)
- Working in a foreign country on a long-term basis (You can't just have every conversation go through a machine).

I would summarize it this way - for casual contacts, a machine translator is great. For the long term, you need to learn a language, whatever that language may be.

Totally aside from the point of if/when such a perfect translator can ever come to be, my personal idea is that a translator can't replace Esperanto. Esperanto is already not useful for the tourist, public speaking, and internet situations I described above (except that with a tourist translator, it becomes less important to find a local Esperanto speaker in a foreign city - although a machine can't hang out with you, show you their favorite restaurant, and treat you to lunch).

And a machine translator isn't good for the things that Esperanto excels at (making friends, meeting people, forming relationships). Esperanto is not likely to help you with working abroad. But in a hypothetical office with a language problem, Esperanto is still a better solution than a machine, for long-term versus short-term use.

Vestitor (Montri la profilon) 2013-januaro-07 00:05:11

I think I addressed all of those points, but without ordering them into lists.

erinja (Montri la profilon) 2013-januaro-07 00:37:19

Vestitor:I think I addressed all of those points, but without ordering them into lists.
Yeah, I was writing my message when you posted yours, so I didn't see your message until I posted.

Vestitor (Montri la profilon) 2013-januaro-07 00:44:52

No matter. It probably helps a lot to have put some organisation to the mess of information I tend to provide.

bartlett22183 (Montri la profilon) 2013-januaro-07 18:06:23

In my (not so humble) opinion, there is a subjective element in all this, the difference between human - human interaction, even if at a distance, and human - machine interaction, even if more immediate. Let me clarify.

I have a(n old) desktop computer (not a portable laptop). On it I have many documents (including many in Esperanto). However, rather than sit in one chair and read off a monitor, I frequently print documents (and occasionally even small books) out on paper and sit in a relaxing chair to read with soft classical music in the background. In the same way, I would rather read a book than stare at a monitor.

Still, recently I did acquire a tablet computer which I mostly use for reading (I even have some Interlingua and Esperanto documents on it). Yes, it is itself a (small) computer, but it handles much like an old fashioned book, so I can sit in my easy chair and read with comfort and pleasure (and take it with me to the coffee shop).

Similarly, there might be a time when machine translation has improved over its current state. Nevertheless, just as I prefer a book (even one on a tablet) in the easy chair, I would prefer the "human" touch of dealing more directly with a human language (English, French which I can still read a little of, Interlingua, Esperanto) produced by a real human being rather than a machine translator.

To me, dealing with human - human language interaction, even if I am not fully fluent in a language, is more emotionally satisfying than using a (still imperfect) machine.

Vilius (Montri la profilon) 2013-januaro-07 18:20:53

Yet sudanglo made a good point in the thread about Ido, where this discussion started. The actual quality of machine translation may be completely irrelevant. Even mere fact that such tools exist may provide another argument against Esperanto. Most people don't use them anyway, beyond toying around a bit. But faced with the idea of Esperanto they will be very likely to ask a very natural question "and what about the machine translation?"

Therefore, dear samideanoj, from now on try not only spread the word about Esperanto, but also spend some time to denigrate the machine translation (right after your regular anti-English rant) ridego.gif

acdibble (Montri la profilon) 2013-januaro-07 18:44:36

sudanglo:
Just two days ago, I was in Italy, sitting at a table with nothing but Italians, but non parlo italiano. Using Google Translate was hit and miss
Do not judge the threat to Esperanto by what can be achieved with machine translation at the moment.

Judge it by what will be able to be achieved in a few years time.
I wasn't making any judgement either way, just giving an account of how I was recently in this situation.

RandallBurns (Montri la profilon) 2018-aprilo-21 00:45:33

I think folks may be thinking about this problem incorrectly. Machine translation is not for the immediate future a substitute for human translation or esperanto but a low cost alternative and a supplement.

Google translate is better than nothing for reading street signs, menus and getting the gist of a conversation.

I don't think the choice is human translators vs. 100% automatic translation for documents people care about. The next big step may be products that do say 90% of the work by machine and identify which are the other portions that need human help(which might be either a junior translator or a senior translator or someone with real domain knowledge).

One important point here: Esperanto is far easier to parse than any human language. That may mean that for the next 10-30 years, one will get a much better result writing a document in Esperanto for automatic translation into whatever natural languages require a translation than going from a natural language. I have not seen good documentation on that yet. What I think is that english is just too hard for machines to reliably read.

In the immediate figure there will be a people that do not speak a language well supported by automatic translation. Esperanto provides the lowest overhead way for them to get access to the web via automatic translation if that is their main goal.

Esperanto may also be the most reliable way to communication with machines via a spoken language. Yes, a lot more attention may be put into getting machines to understand natural languages, but there are inherent issues there. I suspect you will have machines doing simple stuff via natural languages, but if you want to instruct them in anything complicated, esperanto will be a lot beter.

Reen al la supro