Re: First language acquisition

John Flynn <johnpf@xxxxxxxxxxx> wrote in
news:Xns9841964FB3AJOHNPF@xxxxxxxxxxxxxxx:

The criticism is mainly directed at the idea of Universal Grammar that
assumes an underlying Uniformitarianism (see Frederick Newmeyer's
chapter in _Language Evolution_, edited by Kirby and Christiansen, for
a short and readable debate on this issue and the almost universal
default assumption that it is true). That is, that all languages
now and all languages in the past (at least, after we entered the
period of fully modern language, chronologically after the proposed
"protolanguage" eras of Bickerton, Wray, Arbib, Mithen, et al.) and
all languages for the forseeable future are of equal complexity and
operate on the same principles (a truly "Universal Grammar", in other
words).

If we take this default assumption (and it's one of the basic ideas in
most linguistic theories; otherwise we revert to the idea that some
languages are 'primitive' and others 'complex') then it comes with the
implication that anyone put into the right environment at the right
age can learn any language that exists now, in the past, or in the
forseeable future. To do this, according to the basic P&P proposal,
we must have an innate ability to cope with whatever aspect of the
Universal Grammar is thrown at us in the linguistic data in our
environment. To address your "time period" question indirectly, it's
not the issue of how quickly we can change genetically compared to
linguistically, but that we *already have the ability to switch within
one generation* from one language type to another. The evidence seems
to point toward our being able to do this quite easily. Child with
parents from language type A can easily learn language type A' as its
native language.

So, according to basic P&P theory...

This means we have been carrying around in our genomes the 'hardwire'
code (in the form of DNA sequences) that will allow us to acquire any
of these languages perfectly. Or, at least, to a degree sufficient
enough to consider that acquisition to be native-like. For this to
take place, we would have to have *perfect* transmission of this
genetic code from parent to child to grandchild, and so on, with no
errors, for every part of this proposed genetic program. Even the
slightest mutation or unfortunate recombination would result in a part
of our innate acquisition ability being defective. You can see how
this variation works in other biological features. No child looks
identical to its parents, for example; or consider how one person can
have a congential heart problem in any otherwise healthy family; or
just consider the variation in a population's hair colouring or eye
colouring or... a whole host of things that we know are controlled by
our genetic inheritance, and errors in copying or detrimental
recombinations can occur to produce, at worst, defective traits and,
at best, neutral variations. Do we see any of this in the language
ability in normal children? No. The default assumption of the P&P
approach is that you can take a child from anywhere and place it
halfway round the world and it will pick up that community's language
flawlessly. And by this I don't mean children with Specific Language
Impairment that may be caused by other developmental defects. I mean
that a perfectly healthy child who, if we assume that its language
ability is totally innate and under genetic control, should be just as
likely to have a mutation occur on its (hypothetical) "noun" gene
complex as it is to be an albino or to have green eyes from brown-eyed
parents. But we don't see any of these variations occurring. We
never see anyone with the effects of a point-mutation (a SNP -- a
single nucleotide polymorphism) in the hypothetical genes that control
our innate language acquisition abilities. With 6 billion or so people
alive today, we should be seeing at least some effects of the
ubiquitous genetic mutation rate. Where are they? Where are the
people who lack, say, a fully functioning gene for the Head
Directionality Parameter but have an otherwise perfectly working
language acquisition ability? Or someone who has a SNP on their
"subjacency" gene sequence? Such people would be amazing to find and
examine, but we just don't find them. It appears that this is the
only gene-driven biological feature that can be transmitted through
hundreds of generations without even the tiniest degree of copying
error.


From what I can understand of what you wrote (and forgive me that's this
is only a rough understanding of the issues involved given the
limitation of what can be written here), whoever that says this has only
a very crude understanding of genetics. I don't know why, for example,
someone would say something about single nucleotide polymorphism when we
don't even know what gene or genes are involved. A number of the
assertions just seem false. The fact there is no observable phenotype
differences doesn't mean that there is no genotype differences, and do
we actually know what we are looking for in the phenotype, so we can
actually collect any data on the any possible variation or the lack of
it? I'm not arguing whether the Head Directionality Parameter or any
other hypothesis is valid or not, just that all these seems to be a lot
of handwaving and no solid data on which to prove or disprove any idea.



That's not exactly what is proposed. I used the polysynthetic
language thing as an example. It needn't have been polysynthesis,
but that's a Parameter that is proposed to be fairly high up the
hierarchy of Parameters so it has a lot of dependencies.

If you think about that language-type, there will be features that
are heavily dependent on it and other features that apply only to non-
poly languages. If you have the Poly Parameter set to "yes" because
of the linguistic input of your environment, then it makes no
difference what settings you decide on for features that only crop up
in non-poly languages. In fact, you (or your language acquisition
faculty) won't ever have to decide on some features because you might
never encounter them. This means that the part of the innate genetic
program that allows us to decide on those features will never reach
the surface; they can drift about either being set to an arbitrary
Parameter setting or simply be ignored. Whatever happens to them, the
important thing is that are never exposed at the phenotypic level
where selection keeps them maintained. The genetic component that
controls these unused features can mutate and have copying errors hit
it hundreds of times and it won't matter to the surface product
because the language being learned does not need those features. Just
like eyes in a subterranean species. Natural selection is powerless
to maintain an unexpressed genetic sequence.


I think my question is valid, what makes they think that there are
necessary different genetic traits for each of these hypothetical
features, used or unused? I'm truly puzzled.



And yet... we are also suposed to believe under this approach that we
can take a child from any linguistic background (let's say, a child
whose lineage has spoken only English since it first became a distinct
language in the British Isles... making that, what?, about 1300 years
or so, which equals 50-60 generations, 50-60 occasions that required
perfect heredity and no harmful recombination) and put it into a
typologically-different linguistic environment, and that it will be
able to fire up those unchecked-by-natural-selection gene sequences
necessary for acquiring this new language?


There are probably many possible objections to this kind of argument, it
just doesn't sound right to me. I'm not a geneticist and only took a
course in it a long time ago, so I'm not going to say it is wrong, but I
think it is highly suspect to see language acquisition in the same way
as, say, eye colours. I would prefer to see it more like the processing
of vision in the brain which is a highly complex process and which
rarely goes wrong except in cases of brain-damage. If you can argue
using example like that, then I'd be more convinced.



Something somewhere in
that is not right, and my money is on the proposed theory rather than
the ideas we have about how genetic inheritance works.

Take a look at Mark Baker's _The Atoms of Language_. He offers a
simplified P&P approach for the non-specialist from a true-believer's
point of view. To his credit, he does tackle a couple of biological
objections, but manages to brush them aside with a selectionist Just
So story. Not bad in itself, but he doesn't tackle the really
problematic issues. Still... it's a good read to know what the basic
P&P approach is about.


.

Relevant Pages

  • Re: Ubaidian origins ? Help needed !
    ... majority of HLA types have nothing to do with either types now found ... indian genetics shows in-situ evolution. ... groupings I will consider language information as pertinent. ... I have a paper that links the Basque languages with asian and NW ...
    (sci.archaeology)
  • Re: Evolution Theory Has Scientificly Collapsed
    ... Macroevolution was always a dubious achievement which was never really ... pointed out several years ago that random mutation of complex ... language structures simply cannot be the desperately needed mechanism. ... mistakes the nature of "random" change in genetics and forgets that ...
    (talk.origins)
  • Re: Evolution Theory Has Scientificly Collapsed
    ... Macroevolution was always a dubious achievement which was never really ... language structures simply cannot be the desperately needed mechanism. ... DNA is not a language, and analogies to language inevitably fail at ... mistakes the nature of "random" change in genetics and forgets that ...
    (talk.origins)
  • Re: Is language development evolutionary, or designed by the culture?
    ... I was looking for the consensus view on how KS.0> language changes over time....not the origin of language per se. ... It is the changes in the body that KS.2> are acted upon by the environment. ... KS.2> change the genetics of the population over time. ... TT.3> scientists breed bacteria to do complex calculations. ...
    (sci.lang)
  • Re: First language acquisition
    ... only a very crude understanding of genetics. ... polymorphism when we don't even know what gene or genes are involved. ... If 100% genetically-controlled language acquisition ... features, yet no other biological system that we know of works with the ...
    (sci.lang)