noam chomsky on universal grammar and the genetics of language with captioning

Uploaded by kimberlymurza on 11.05.2012

>> The first question, Mr. Chomsky, comes from Kris --
Kristof Gudrow [phonetic].
How have your ideas on universal grammar changed over the years?
Are you more or less convinced of the theory now
than you were initially?
>> Well, there's -- there's a lot of confusion
about the notion universal grammar.
Universal grammar had a traditional meaning.
But, in modern linguistics -- the last 50 years or so --
it's had a technical meaning which is not unrelated
to the traditional meaning but it's not identical either.
Universal grammar is just the name for the theory
of the genetic component of the language faculty.
I mean, transparently, there's some genetic component.
Right? There's a reason, say,
why my granddaughter reflexively identified some part
of her environment as language related which is no small trick.
Nobody knows how to duplicate that.
And then more or less reflexively picked
up the capacity that we're all now using.
Whereas her pet, say, kitten or a chimpanzee or song bird
or whatever it maybe
with exactly the same inputs couldn't even take the first
step, can't even identify part
of the environment is language related.
Obviously, not the later steps.
Well, there are two possible answers to how that happens.
One is it's a miracle.
The other is there's -- she has some specific genetic capacity
that's like the capacity that had her grow arms
and not wings, let's say.
Just some fixed -- or had a mammalian visual system
but not an insect visual system.
Now this is not controversial
for anything except human higher mental faculties.
For some reason when people investigate human higher mental
faculties, they have to be insane, you know.
You can't accept the approach that we take
to everything else in the world.
They're kind of a mythological dualism.
Every else in the world we study
by the standard methods of science.
But, when we talk about human -- how our mental faculties,
we have to become mystics.
So, therefore, there's a controversy about the existence
of universal grammar which is like --
which means a controversy
about whether there is some genetic property
that distinguishes humans from everybody else which leads
to these -- to the ability to do -- doing what we're now doing.
But there shouldn't be any controversy about that.
The only question is: What is it?
Well, there have been theories about it from the 1950's
when these studies began up till the present.
And it's a living field so they keep changing.
So, in that sense, yes, my views
about universal grammar keep changing.
I'll say, when Anne walked into my office as a graduate student
and told me I was wrong about everything, I said, okay.
My new view has changed, you know.
But, in that sense, sure, there's gonna be constant change
until the field is -- disappears or is dead or something.
And it's a -- there's a long way to go.
These are not trivial questions at the --
the sort of general tendency of change --
not every linguist would agree by any means.
It's a personal opinion.
In the early stages when the first question was asked
seriously about 50 years ago as to how we are capable
of doing what we do all the time, how are we capable
of understanding, of producing expressions which have --
we've never heard which may have never been uttered
in the history of the language and do it over infinite range
with very strange properties that they have as soon
as you look at them, how can we do it.
The only answers seem to be that each
of us has a highly intricate computational system
in the brain which yields these very specific results.
But that then poses a paradox because it must be the case
that we all -- all humans have the same genetic capacity
with marginal variation.
The reason is, if you take a child from say a --
a hunter-gatherer tribe in the Amazon and the child is raised
in Cambridge, Mass., it'll --
may just become a graduate student studying quantum physics
at MIT with no difference from anyone else, you know.
And conversely.
So we all have the same capacity.
And it's more or less understood why.
The capacity developed very recently in evolutionary time
and probably in some window
between 100,000 and 50,000 years ago.
Something like that.
And that's just the flick of an eye.
So whatever happened never changed except
extremely marginally.
So we're all fundamentally identical
for all practical purposes.
Human genetic variation is very slight anyway
and superficial differences but not very profound.
For an outside, say,
an extraterrestrial observer looking at us the way we look
at frogs would say there's only one human and one language
with minor variations.
So, on the one hand, it -- it's gotta be uniform.
On the other hand, the --
it seemed to be the case
that each particular language had a highly intricate
and complex system of rules -- computational system --
and they're very different from one another.
And that is a paradox.
In fact, you know, a serious paradox.
Well, over the years, there have been efforts to deal with it
to try to overcome the paradox.
A major step was taken --
and here are views on universal grammar, at lease for many
of us, did change radically was around 1980.
>> I was still there.
>> You were there, yeah.
[inaudible speaker] it's her fault.
[Laughter] When a different view of the matter sort
of crystalized what's called --
sometimes called the "principles and parameters view."
The picture that -- the prince --
that there are fixed principles which are really invariant.
Nobody has to acquire them.
They're part of universal grammar.
And then there's a number of options that can be taken called
"parameters" that the child has to pick up from experience.
And they have to be pretty simple.
You have to be able to pick them up from limited evidence
because that's all there is.
And so, for example, in some languages like English, the --
it's called a "head-first language."
So the verb precedes the object
and a preposition precedes the object
of the preposition and so on.
Other languages, like say Japanese,
are almost the mirror image.
The verb follows the object.
The postpositions, not prepositions, and so on.
So the languages are virtually mirror images of each other
and you have to set the parameter.
The child has to set the parameter which says,
am I talking English or am I talking Japanese.
And that can be --
can be determined from very simple data.
So that's a reasonable choice of a parameter.
And the hope was that you could find some finite set
of parameters like a finite switch box
where you set the switch -- the child has set switches one way
or another and can do it on the basis of fairly simple data.
And then, once this enters in through a predetermined system
of principles, you get things
which superficially look very different
but are actually almost identical.
Just differing in superficial choices.
Well, if you'd work that out, you'd have solved the paradox.
It's a long way to work that out.
But that made it possible at least
to confront the issue seriously
without facing an immediate knew yourself contradiction.
And it set off a lot of -- really rich period of research
and inquiry and nothing like it in a --
thousands of years of history of study of language
in the last 25, 30 years of a wide variety
of typologically different languages, new questions
at a depth that could never have been proposed before.
Sometimes the answers leading to new questions and so on.
It's been a very lively period.
And it also raised another question.
What about the principles?
Where do they come from?
And that's, in fact, a choice of parameters.
Where do these things come from?
If they're in universal grammar, if they're part
of the genetic endowment, then they had to evolve somehow.
But not a lot could've evolved because it's too recent.
You know, you go back 100,000 years there's,
as far as we know, nothing.
Humans had the same anatomy.
Anything that's preserved in the fossil records about the same,
you know, hundreds of thousands of years back.
So it's some small change must have taken place in the --
in the brain which somehow allowed all
of this to suddenly blossom.
And pretty soon after that, again, in evolutionary time --
like maybe a couple 10s of thousands of years
which is no time at all humans started leaving east Africa
where we all come from as far as anyone knows.
So some small group developed this system
and then it spread all over the world
and now grow essentially the same.
But what evolved in that short period
of time cannot have been very complex.
You know, you wouldn't expect a series of extensive stages like,
say, development of limbs, you know, millions of years.
Therefore, what you'd predict is
that some other principle external to language --
maybe some principle of nature, the principle
of computational efficiency or something like that,
which is not specific to language,
interacted with a small mutation which just gave rise
to the -- to universal grammar.
Well, that sets forth a new goal of research to ask --
to see if you can determine
that the principles themselves do not have the intricacy
that they appeared to have but are actually the result
of application of nonlinguistic -- in fact, non --
maybe nonhuman principles like general principles
of computational efficiency
to whatever small change took place.
And the small change was probably the capacity to carry
out recursive enumeration basically.
The capacity that gives you the number system, for example.
To take two things -- two objects already constructed
in the mind and make up a new object out of them and then keep
that process up indefinitely so you get an infinite array
of possible expressions.
Each with some semantic interpretation
and some mode of externalization.
Speech or sign.
Whatever it may be.
That would be -- and the goal would be to try to show
that was essentially instantaneous once the small
mutation took place, given the --
this operation or recursive enumeration operation.
It allows you to create a discrete infinite
of expressions -- structured suppress.
Well, that's at least a feasible picture.
The trick is to show that it's true or how close it is to true.
Can you cut away at the apparent complexity of the principles
and show that they can actually be accounted for in terms
of general principles of --
that hold for organisms generally perhaps
and maybe even elsewhere in the physical world.
And that are instantly or almost instantly applied once the
original move is made to whatever small move it was
to produce the capacity for recursive enumeration.
Well, that's a goal, you know.
It's far from being attained.
But, in the last 15, 20 years,
there's been considerable progress towards it.
That is a lot of things that it seemed 20 years ago you had
to assign to the genetic endowment have no --
now have been rather plausibly shown
to be possible consequences of just application.
Particularly of principles of computational efficiency
to a system which had only the ability
to construct an infinite hierarchy of expressions
and that -- we don't know enough about the brain to know how
that might have happened
but that could've been a very small mutation.
Just changing something in somebody's genome
and then spreading through the small breeding group.
So that -- in that respect, it's -- it's a goal, you know,
and steps have been taken towards it.
But you would expect that something like that ought
to be true just from the -- what's known about the history
of evolution of Homo sapiens in very recent times
without much opportunity for selection
to have had any effect.
Maybe a small effect but not much.
So that's, I think -- it's in that --
that's the tendency of thinking --
at least my thinking and some other --
many others on how theories of universal grammar have changed.
But the idea that there is universal grammar that exists
that can't be controversial unless you believe in magic,
you know, for the elementary reasons that I have mentioned.