So, either Mandarin or Spanish would be the obvious choices then, isn't it?
Te he pillado, tio.
Or, for the minority out there that speaks English: Got you there, buddy.
Perhaps I should have been more explicit, but the point was: there is nothing intrinsically wrong with search results in Dutch, as long as you've got the option to look for pages in the language you would like. And that is not English by default.
Well, there are a few contrasts that correlate with earlier findings, but for the rest you are completely right. This isn't exactly great research, at least not the way it's presented here.
Influential though Chomky may have been, the general form of UG (as you so appropriately put it: "Basically they claim that there is a langauge module in the brain.") is not really his (Broca, Wernicke, anyone?). His and his followers' are the claims about pretty specific parameters and principles, which are a lot harder to prove and can obviously not be determined from such general data (and which I don't expect anyone to find either, but that's another story).
The Growth-Point Theory is a slightly outlandish idea and its believers are trying to get it more widely accepted. After this article, they will probably try to link their theory to it, something like: if deaf children without exposure to language can develop language in more than one modality, so their theory must be true, although they'll present it the other way around, starting with the design of this experiment and claim this result as a prediction of their theory. And of course, this experiment shows nothing of the kind, but doesn't contradict it either. The rest is politics. That's what I meant by "agenda".
This study doesn't prove anything of the kind. As reported, it only shows that people can learn language. Of course that includes the capability of developing language constructs. How else did we ever start speaking? It also shows that you don't need to be able to talk or hear in order to develop language skills, and that's not really new either.
You're overlooking a few major problems: writing good rules is hard, adding efficient fault tolerance to a parser is hard (NLTK doesn't have it), and coding information is a dictionary is a lot of work.
If you want to settle for a few very simple errors in English only, then it's not that much work, but any sensible error in English and particular other languages requires a lot of work. Believe me, I've done my PhD on this topic, a European project, and now work for a company that actually builds the Dutch grammar checker for Microsoft.
If you want, I can answer some questions, or give you some pointers, in case you are interested in the topic...
Although it is nice to see mentioning of my trade a/., this paper has about the status of a student's essay. It doesn't even mention literature after 1998!
Yes, all of Word's grammar checkers are on the dumb side, and yes, there are better checkers out there, but they're all academic. And nobody is so interested in improving Word's grammar checker that they'll spend a huge amount of money turning those into real products.
What is would take to make an OS grammar checker? A simple one? One year of work, but you'll only annoy your users. A good one? Ten years. Your ideas about grammatical rules is slightly primitive: it's *hard* to check grammar. Without a good dictionary, it's nearly impossible (and that's several years worth of hard work).
In Dutch (my native language) the same happens quite frequently, and yes, it's annoying. However, it's not the grammar checker that does this, it's the spell checker...
Way overpriced. What's the traditional rule again? 20 to 30 times the annual profit (after taxes)? How are they ever going to reach, let alone sustain a 1 billion dollar profit per year?
Rule number one of experimental research: get a representative sample. Four is not enough even for objects within one group and these items are distributed over 2 groups (tonal vs. a-tonal), with one group containing only one item! That's *BAD* research.
Furthermore, it is highly obvious that tonal music produces a steeper curve according to the measure chosen, since the number of possible notes that follow each other is more limited than in a-tonal music, where in some pieces the requirements strictly demand that the same note shall not be repeated before any of the 11 other notes has been played. That accounts for quite a lot.
Furthermore, the article *completely* fails to explain why a steeper curve would make something more understandable. Zipf's law can be found in nearly all natural processes (check Mandelbrot's work on it), but that doesn't mean that e.g. interruptions in the telephone system will look like a language to us.
I have been thinking about that too, and perhaps it's Outlook and the (**!*!*!!) Exchange server that make it so slow. When I'm working on a dual processor machine, I don't notice any problem, but on a single processor machine with a few Word documents or Visual Studio (.NET 2003) open, things really slow down. But the one constant factor is Outlook (Office 2003). Perhaps people should be less addicted to e-mail...
We all have 2GHz Intel machines with 256Mb RAM here, and XP definitely doesn't run comfortable, unless you have the patience of Buddha.
A student approaches his master and asks him: Master, how come my 3GHz Hyperthreaded four processor system with 2Gb of RAM feels so slow, yet I never hear you complain about your old 386? Doesn't it run slower? And the Master responds: A hare will think the grass is dead, while a turtle might see it grow. A penguin on the other hand, doesn't even know what grass is. The student was immediately enlightened, went home and programmed a web server on his Commodore 64.
That's, errr, kind pathetic. You were fantasizing with your mates over an unknown woman and saying things that must have been rather sexual in nature? Well, that's pretty inmature. But then to be disgusted because Andrea turns out to be a man, is homophobic as well. Let me guess, you're from Texas?
That still doesn't make it literature. Although you can differ about the definition of "literature", take a look at this definition from WordNet. I think people here are confusing meanings 1 and 3. There is no Nobel prize for any other meaning than #1.
Slashdot Mirror
So, either Mandarin or Spanish would be the obvious choices then, isn't it?
Te he pillado, tio.
Or, for the minority out there that speaks English: Got you there, buddy.
Perhaps I should have been more explicit, but the point was: there is nothing intrinsically wrong with search results in Dutch, as long as you've got the option to look for pages in the language you would like. And that is not English by default.
What's wrong with Dutch? Can't read it? Most people in this world can't read English either...
Is this the one where you simply present all pages you've ever encountered in one huge list?
You'd better not, since that has been patented by the British Museum!
Amen to that. And please mod the parent "insightful". Or, if you want to insist on "Funny", give it the attribute "bitter irony"...
Actually, the moderators went all over the place. I got around 20 mods up and down or so (didn't count them though).
But did you ever take a look of the specs for Longhorn?
I guess that'll be enough to run Longhorn then.
Well, there are a few contrasts that correlate with earlier findings, but for the rest you are completely right. This isn't exactly great research, at least not the way it's presented here.
Was I the only one that read "ACME" instead of Amec?
IAACPL (I Am A Computational Psycho Linguist)
Influential though Chomky may have been, the general form of UG (as you so appropriately put it: "Basically they claim that there is a langauge module in the brain.") is not really his (Broca, Wernicke, anyone?). His and his followers' are the claims about pretty specific parameters and principles, which are a lot harder to prove and can obviously not be determined from such general data (and which I don't expect anyone to find either, but that's another story).
The Growth-Point Theory is a slightly outlandish idea and its believers are trying to get it more widely accepted. After this article, they will probably try to link their theory to it, something like: if deaf children without exposure to language can develop language in more than one modality, so their theory must be true, although they'll present it the other way around, starting with the design of this experiment and claim this result as a prediction of their theory. And of course, this experiment shows nothing of the kind, but doesn't contradict it either. The rest is politics. That's what I meant by "agenda".
This study doesn't prove anything of the kind. As reported, it only shows that people can learn language. Of course that includes the capability of developing language constructs. How else did we ever start speaking? It also shows that you don't need to be able to talk or hear in order to develop language skills, and that's not really new either.
9 96411 had more details. But notice that some of the people in the study have other agendas and hope that acceptance of this study can help them further their own views http://mcneilllab.uchicago.edu/topics/gp.html.
Anyway, the New Scientist article http://www.newscientist.com/news/news.jsp?id=ns99
As in 100% - 12%?
You're overlooking a few major problems: writing good rules is hard, adding efficient fault tolerance to a parser is hard (NLTK doesn't have it), and coding information is a dictionary is a lot of work.
If you want to settle for a few very simple errors in English only, then it's not that much work, but any sensible error in English and particular other languages requires a lot of work. Believe me, I've done my PhD on this topic, a European project, and now work for a company that actually builds the Dutch grammar checker for Microsoft.
If you want, I can answer some questions, or give you some pointers, in case you are interested in the topic...
Although it is nice to see mentioning of my trade a /., this paper has about the status of a student's essay. It doesn't even mention literature after 1998!
Yes, all of Word's grammar checkers are on the dumb side, and yes, there are better checkers out there, but they're all academic. And nobody is so interested in improving Word's grammar checker that they'll spend a huge amount of money turning those into real products.
What is would take to make an OS grammar checker? A simple one? One year of work, but you'll only annoy your users. A good one? Ten years. Your ideas about grammatical rules is slightly primitive: it's *hard* to check grammar. Without a good dictionary, it's nearly impossible (and that's several years worth of hard work).
In Dutch (my native language) the same happens quite frequently, and yes, it's annoying. However, it's not the grammar checker that does this, it's the spell checker...
You can turn it off (did you ever see the zillion options?) and the word doubling check is done by the spell checker, just for your information.
Well put. I wish all psychologists would see it that way...
Way overpriced. What's the traditional rule again? 20 to 30 times the annual profit (after taxes)? How are they ever going to reach, let alone sustain a 1 billion dollar profit per year?
For Christ's sake, computers are mostly used as tools. And who keeps their old tools around for so long? Only neanderthals:...
Rule number one of experimental research: get a representative sample. Four is not enough even for objects within one group and these items are distributed over 2 groups (tonal vs. a-tonal), with one group containing only one item! That's *BAD* research.
Furthermore, it is highly obvious that tonal music produces a steeper curve according to the measure chosen, since the number of possible notes that follow each other is more limited than in a-tonal music, where in some pieces the requirements strictly demand that the same note shall not be repeated before any of the 11 other notes has been played. That accounts for quite a lot.
Furthermore, the article *completely* fails to explain why a steeper curve would make something more understandable. Zipf's law can be found in nearly all natural processes (check Mandelbrot's work on it), but that doesn't mean that e.g. interruptions in the telephone system will look like a language to us.
Resuming? Utter, utter non-sense.
I have been thinking about that too, and perhaps it's Outlook and the (**!*!*!!) Exchange server that make it so slow. When I'm working on a dual processor machine, I don't notice any problem, but on a single processor machine with a few Word documents or Visual Studio (.NET 2003) open, things really slow down. But the one constant factor is Outlook (Office 2003). Perhaps people should be less addicted to e-mail...
We all have 2GHz Intel machines with 256Mb RAM here, and XP definitely doesn't run comfortable, unless you have the patience of Buddha.
A student approaches his master and asks him: Master, how come my 3GHz Hyperthreaded four processor system with 2Gb of RAM feels so slow, yet I never hear you complain about your old 386? Doesn't it run slower?
And the Master responds: A hare will think the grass is dead, while a turtle might see it grow. A penguin on the other hand, doesn't even know what grass is.
The student was immediately enlightened, went home and programmed a web server on his Commodore 64.
That's, errr, kind pathetic. You were fantasizing with your mates over an unknown woman and saying things that must have been rather sexual in nature? Well, that's pretty inmature. But then to be disgusted because Andrea turns out to be a man, is homophobic as well. Let me guess, you're from Texas?
That still doesn't make it literature. Although you can differ about the definition of "literature", take a look at this definition from WordNet. I think people here are confusing meanings 1 and 3. There is no Nobel prize for any other meaning than #1.