In theory I agree. Experimentally, trying to teach her that other toys are connected to different sounds, e.g. that the black-and-white cow is “moo”, didn’t produce any reaction so far. And I believe she doesn’t understand the meaning of the word “not” yet, so I can’t explain that some things are “not meow”.
I guess this problem will fix itself later… that some day she will start also repeating the sounds for other animals. (But I am not sure what is the official sound for turtle.)
I am sure this isn’t necessary, but, you do realize that she’s going to learn language flawlessly without you actively doing anything?
Instead of saying “my ” my daughter used to exclusively say “the that I use.” My son used to append a “t” sound to the end of almost every word. These quirks sorted themselves out without us mentioning them. =)
The phrase “actively doing anything” is too slippery. What one person does passively another may do actively. People who post on Less Wrong tend to do things consciously more often than the general public. The theories which say that children acquire language without anyone doing anything special are no doubt studying the behavior of normal people.
The conclusion is that Viliam is probably simply thinking out loud about things that most people consider only subconsciously and implement in some way but don’t know how to articulate. If you try to acquire a foreign language by merely listening to native speakers converse, you will learn very little. Children learn language when adults adapt their speech to their level and attempt to bridge the inferential distance. Most people do this by accident of having the impulses of a human parent.
There’s overwhelming data that parenting can prevent language acquisition. But that requires extreme degenerate cases—essentially child abuse on the level of locking the child in the closet and not talking to them at all.
For typical parenting, I agree that it is unlikely that variance in parenting style has measurable effect on language acquisition.
I don’t see why you would expect that to be affected much either.
Vocabulary is a good measure of intelligence because words have a very long tail (Pareto, IIRC) distribution of usage; most words are hardly ever used. I pride myself on my vocabulary and I know it’s vastly larger than most English speakers as evidenced by things like a perfect SAT verbal score, but nevertheless if I read through one page of my compact OED, I will run into scores or hundreds of words I’ve never seen before (and even more meanings of words!). This doesn’t bother me since when you reach the point where you’re reading the OED to learn new words, the words are now useless for any sort of actual communication… Anyway, since most words are hardly ever used, people will be exposed to them very few times, and so they are a sensitive measure of how quickly a person can learn the meaning of a word. If people are exposed to the word ‘perspicacious’ only 3 or 4 times in a lifetime on average, then intelligence will heavily affect whether that was enough for them to learn it; and by testing a few dozen words drawn from the critical region of rarity where they’re rare enough that most people would not have been exposed enough times but not so rare that no one learns them normally (as determined empirically—think item-response theory curves here), you can get a surprisingly good proxy for intelligence despite the obvious fragility to cheating. (Of course, you can get around it, and the SAT does, by simply having lots of semi-rare words to draw upon. Have you ever looked at comprehensive SAT vocab lists? There’s just no way most people could memorize more than a few hundred without, well, being very smart and verbally adept.)
Since there are so many potential rare vocabulary words to learn (and which could be sampled on an IQ test) and since parents speak only a small fraction of the number of words a person is exposed to over a lifetime… Even a parent deliberately trying to build vocabulary can’t have that much of an effect. Similarly, cramming vocab in school won’t have much of an effect either because schools don’t do anything to exploit the spacing effect which might relatively equalize everyone’s retention; all that cramming a list of 10 vocab words a week won’t do much or might even exacerbate vocab differences as the smarter kids remember the vocab while everyone else just forgets them like usual. I would expect heritability of vocabulary size to increase over lifetimes as any childhood effect of craming fades following the forgetting curve.
So to sum it up:
parents can offer only brief exposure to a small subset of the semi-rare ‘vocab’ words that one can genuinely use and which will be measured by vocab sub-tests
the forgetting curve ensures that brief exposures will typically be forgotten due to lack of reinforcement
intelligence differences lead to large differences in how many vocab words will stick in addition to any sort of gene-environment effect like smarter kids seeking out books which will give them more exposure to semi-rare words, which swamps any effect from #1 which survives #3
Vocab subtests are a part of many IQ tests, so there are probably existing family or twin studies which have the relevant estimates you could dig out to see how much shared-environment (which will cover parenting style) affects vocabulary subtest performance (which can probably be reversed to get a total vocabulary size) and whether it shrinks with age like shared-environment on intelligence does.
It’s also a good measure of how much do you read (or how much have you read as a child). People who read books—lots and LOTS of books—have a very good vocabulary. People who don’t, don’t. There is certainly a correlation, but vocabulary is still just a proxy for intelligence, maybe not a good proxy for math people or in the ’net age.
I just skim http://lesswrong.com/r/discussion/comments/ and occasionally C-f for my name. (My PMs/red-box are so backed up that I haven’t dared check them in many months, so this is how I see most replies to my comments...)
I’m not convinced that using different names is a really helpful idea. It requires an extra transition later on. Well no real harm done. But I wonder about the principle behind that: Dropping complexity because it is hard? I agree that child directed speech is different. It is simpler. But it isn’t wrong. Couldn’t you have said “the cat meows” or “this toy meows” or even “it meows”? That would have placed the verb in the right place in a simple sentence. The baby can now validly repeat the sound/word.
The necessity of negative examples is well-known when training classifiers.
In theory I agree. Experimentally, trying to teach her that other toys are connected to different sounds, e.g. that the black-and-white cow is “moo”, didn’t produce any reaction so far. And I believe she doesn’t understand the meaning of the word “not” yet, so I can’t explain that some things are “not meow”.
I guess this problem will fix itself later… that some day she will start also repeating the sounds for other animals. (But I am not sure what is the official sound for turtle.)
I am sure this isn’t necessary, but, you do realize that she’s going to learn language flawlessly without you actively doing anything?
Instead of saying “my ” my daughter used to exclusively say “the that I use.” My son used to append a “t” sound to the end of almost every word. These quirks sorted themselves out without us mentioning them. =)
The phrase “actively doing anything” is too slippery. What one person does passively another may do actively. People who post on Less Wrong tend to do things consciously more often than the general public. The theories which say that children acquire language without anyone doing anything special are no doubt studying the behavior of normal people.
The conclusion is that Viliam is probably simply thinking out loud about things that most people consider only subconsciously and implement in some way but don’t know how to articulate. If you try to acquire a foreign language by merely listening to native speakers converse, you will learn very little. Children learn language when adults adapt their speech to their level and attempt to bridge the inferential distance. Most people do this by accident of having the impulses of a human parent.
Not actively but maybe subconsciously.
As I already mentioned child directed speech is different.
And also yes: Most children probably can get by without that either.
And also: I’m sure gwern will chime in an cite that parents have no impact on language and concept acquisition at all.
There’s overwhelming data that parenting can prevent language acquisition. But that requires extreme degenerate cases—essentially child abuse on the level of locking the child in the closet and not talking to them at all.
For typical parenting, I agree that it is unlikely that variance in parenting style has measurable effect on language acquisition.
And what about the size of the vocabulary?
I don’t see why you would expect that to be affected much either.
Vocabulary is a good measure of intelligence because words have a very long tail (Pareto, IIRC) distribution of usage; most words are hardly ever used. I pride myself on my vocabulary and I know it’s vastly larger than most English speakers as evidenced by things like a perfect SAT verbal score, but nevertheless if I read through one page of my compact OED, I will run into scores or hundreds of words I’ve never seen before (and even more meanings of words!). This doesn’t bother me since when you reach the point where you’re reading the OED to learn new words, the words are now useless for any sort of actual communication… Anyway, since most words are hardly ever used, people will be exposed to them very few times, and so they are a sensitive measure of how quickly a person can learn the meaning of a word. If people are exposed to the word ‘perspicacious’ only 3 or 4 times in a lifetime on average, then intelligence will heavily affect whether that was enough for them to learn it; and by testing a few dozen words drawn from the critical region of rarity where they’re rare enough that most people would not have been exposed enough times but not so rare that no one learns them normally (as determined empirically—think item-response theory curves here), you can get a surprisingly good proxy for intelligence despite the obvious fragility to cheating. (Of course, you can get around it, and the SAT does, by simply having lots of semi-rare words to draw upon. Have you ever looked at comprehensive SAT vocab lists? There’s just no way most people could memorize more than a few hundred without, well, being very smart and verbally adept.)
Since there are so many potential rare vocabulary words to learn (and which could be sampled on an IQ test) and since parents speak only a small fraction of the number of words a person is exposed to over a lifetime… Even a parent deliberately trying to build vocabulary can’t have that much of an effect. Similarly, cramming vocab in school won’t have much of an effect either because schools don’t do anything to exploit the spacing effect which might relatively equalize everyone’s retention; all that cramming a list of 10 vocab words a week won’t do much or might even exacerbate vocab differences as the smarter kids remember the vocab while everyone else just forgets them like usual. I would expect heritability of vocabulary size to increase over lifetimes as any childhood effect of craming fades following the forgetting curve.
So to sum it up:
parents can offer only brief exposure to a small subset of the semi-rare ‘vocab’ words that one can genuinely use and which will be measured by vocab sub-tests
the forgetting curve ensures that brief exposures will typically be forgotten due to lack of reinforcement
intelligence differences lead to large differences in how many vocab words will stick in addition to any sort of gene-environment effect like smarter kids seeking out books which will give them more exposure to semi-rare words, which swamps any effect from #1 which survives #3
Vocab subtests are a part of many IQ tests, so there are probably existing family or twin studies which have the relevant estimates you could dig out to see how much shared-environment (which will cover parenting style) affects vocabulary subtest performance (which can probably be reversed to get a total vocabulary size) and whether it shrinks with age like shared-environment on intelligence does.
It’s also a good measure of how much do you read (or how much have you read as a child). People who read books—lots and LOTS of books—have a very good vocabulary. People who don’t, don’t. There is certainly a correlation, but vocabulary is still just a proxy for intelligence, maybe not a good proxy for math people or in the ’net age.
Yes, that is about what I expected you to confirm.
And wow, are you omnipresent or how comes you actually noticed the post so quickly (or is there a ‘find my name in posts’ functionality I overlooked)?
I just skim http://lesswrong.com/r/discussion/comments/ and occasionally C-f for my name. (My PMs/red-box are so backed up that I haven’t dared check them in many months, so this is how I see most replies to my comments...)
Then teach her to say turtle. Likewise for other animals; a cat is not a meow.
Yes, as soon as she learns to speak polysyllabic words (which in my language also include “cat”).
My first word was “daddy” (papi in Spanish). It should be possible to start with regular words.
Edited to add: I just looked up “turtle” in Slovak. I can’t believe how much of a jerk I was in my previous comment.
I’m not convinced that using different names is a really helpful idea. It requires an extra transition later on. Well no real harm done. But I wonder about the principle behind that: Dropping complexity because it is hard? I agree that child directed speech is different. It is simpler. But it isn’t wrong. Couldn’t you have said “the cat meows” or “this toy meows” or even “it meows”? That would have placed the verb in the right place in a simple sentence. The baby can now validly repeat the sound/word.
Hard to come by in normal language acquisition, though. So it probably doesn’t quite work like that.