many
What does this mean''Que hace el fotografo? Why is it the above dictionary cant tranlate a word or a sentence most of the time''!!!
11 Answers
dan said:
Please forgive me if this doesn't go to all that responded to my spanish problem...I'm new with the computer as well as spanish.....But thank you all for responding, I really appreciate it!!
You brought it on yourself (je je). You asked two questions. Quentin answered the first one (What does "Qué hace le fotógrafo'" mean') Once Quentin had replied, there wasn't anything left to say on that topic. Your second question ("Why is it the above dictionary cant tranlate [sic] a word or a sentence most of the time''!!!"), however, is one for which there are a wide variety of opinions (often, heated).
Please forgive me if this doesn't go to all that responded to my spanish problem...I'm new with the computer as well as spanish.....But thank you all for responding, I really appreciate it!!
I'll try to stay marginally on topic here (translation IS related to this forum)...
Computers do not, of course, actually understand anything.
The above statement is rather anthropocentric, in my opinion. What is understanding, after all? A parakeet doesn't "understand" the human words it speaks, at least from a human perspective, but it understands them from its own perspective. A dog that responds to dozens of words is higher up the scale of understanding, and a gorilla capable of signing is higher yet. Language understanding is therefore not either black or white, but a gray scale. So I would say that, rather, a computer's understanding of language is much shallower than our own.
But I agree with the second part of your post wholeheartedly. That is, humans won't design a computer that can translate well, but we will design a computer that can design a computer that can translate well. We humans can communicate by language perfectly well without any formal input of rules whatsoever (my grandfather never went to school a day in his life), but we intuit the rules through exposure and use. An adaptive program with access to a huge corpus, as you mention, will be what finally makes the big break-through.
And while it is unlikely that that will happen in the next couple of decades, I believe that the big change will be not incremental, but quantum in nature, so it could conceivably happen at any time.
samdie said:
Interestingly, the real progress that was made with chess-playing programs happened when they stopped trying to supply the rules and and wrote adaptive programs (which were then pitted against one another) and allowed them to develop their "own" strategies/rules. It is my understanding that the best chess-playing programs don't play like humans (and master chess-players can recognize the "computer-style" of playing) but they do win.
Samdie, I've done a machine program that can beat 99% of the players in the world, using most of the latest development in computer chess, and I can tell you: it counts numbers. I know the programmers of 3 out of the 5 strongest programs in the world (all of which could beat that Deep Blue), and they regard their machines as blind and stupid. Very fast and very effective in searching, but completely clueless about chess playing, and chess is as simple as 2+2 compared to the simplest sentence you can think of in any language. A computer chess program is still a big calculator, not an intelligent program capable of coming up with new ideas. I insist: artificial intelligence is nowhere near close to even analyse a simple sentence the way we do, and chess is a simple problem highly suited for machines to surpass humans in performance just by using brute force.
lazarus1907 said:
For a machine to be able to translate, it must be able to understand emotions, social impact, connotations,... In other words, it must be able to think like a human person.
Not really. Computers do not, of course, actually understand anything. They apply a collection of rules to a set of data. The problem is how to construct a comprehensive set of rules and, for languages, we're a long way from that kind of understanding. Interestingly, the real progress that was made with chess-playing programs happened when they stopped trying to supply the rules and and wrote adaptive programs (which were then pitted against one another) and allowed them to develop their "own" strategies/rules. It is my understanding that the best chess-playing programs don't play like humans (and master chess-players can recognize the "computer-style" of playing) but they do win.
It may be that the same approach will turn out to be necessary for successful translation, i.e. present the programs with a massive corpus of sentences and their translations and have the programs infer the rules. On the other hand,
the "rules" of natural language are not those of chess (which, by comparison, are rigidly defined and few in number). I am inclined to think that there may, eventually, be machines that can compete with (most) human translators but, at the same time, I suspect that the goal of perfect translation is "undecidable" for the same reasons that apply to Turing's "halting problem".
MT could be feasible one day, maybe, for certain types of translations. Give it a text with irony, puns, funny sounds, metaphors, colloquialisms, exaggerations, strong cultural connotations, etc... and it will produce a rubbish translation, as usual. For a machine to be able to translate, it must be able to understand emotions, social impact, connotations,... In other words, it must be able to think like a human person. I won't say it is impossible, but so far, progress is close to zero in this sense. One thing is to play chess, where a computer simply has to analyse numbers, and another one is to model all the complexities and peculiarities of the human mind, and all different cultures, which change constantly.
I know, but if you have a knowledge of the language, you can generally work out a decent translation. Just don´t take what the translator says as gospel. More to the point, why did the poster give this thread the title "many"'
Just for fun, I plugged this into the MACHINE translator (please, folks, don't call these translators; I'm a translator, and I'm human), and got this:
Lo sé, pero si tiene un conocimiento de la lengua, en general puede elaborar una traducción decente. Simplemente no tienen lo que dice el traductor como evangelio. Más al punto, ¿por qué dar el cartel de este hilo el título de "muchos"?
The first sentence isn't bad. The second sentence is puzzling, because I would think that the MT would have recognized the command form, but it didn't. The third sentence is typical MT, because it interpreted poster to mean cartel, but it also messed up the tense and the subject.
I'm confident that MT will one day be feasible, but I am also confident that it will not put me out of work within the remainder of my working life.
Ian Francis Hill said:
I am a computer specialist . THEY can still not deal with language 100% and probably never wiil.
I know, but if you have a knowledge of the language, you can generally work out a decent translation. Just don´t take what the translator says as gospel. More to the point, why did the poster give this thread the title "many"'
I am a computer specialist . THEY can still not deal with language 100% and probably never wiil.
¿Qué hace el fotógrafo? What is the photographer doing?
For more than a single word definition try the Translation tab rather than the Dictionary tab. Scroll down the page that pops up and you will see another two machine translators in addition to Google's. Type your sentence into Google at page top and at first, at least, all three machines will translate it. Putting all three translations together, so to speak, you can usually (not always) come up with what your sentence really means.
Anyone can contribute a better translation to Google, so maybe someday fairly soon that translator will be very good. Meanwhile what any machine translator can do is limited but using all three can be a big help.
One reason the machines have trouble translating well is that even in a simple question or statement, things absent or less important in English like tildes and word order, e.g., can be so important in Spanish. And it is too often too hard for most of us beginners including me to type Spanish into them entirely accurately. Type into the dictionary the word "que" and then try the word "qué", e.g., and you will get a better idea of what the machines are dealing with.
Not to mention idiomatic usages.......................
One of the reasons is that people throw things into the translator using misspellings and improperly punctuated sentences.
¿Qué hace el fotógrafo?
What is the photographer doing or making'
