Re: Natural Language and Natural Selection

From: Stevan Harnad (harnad@cogsci.soton.ac.uk)
Date: Sat Mar 02 1996 - 20:11:01 GMT


> From: "DONNA CRUMLEY" <DHC195@psy.soton.ac.uk>
> Date: Wed, 28 Feb 1996 16:39:26 GMT
>
> Chomsky and Gould, for example suggest that language is a side effect
> of other evolutionary forces such as an increase in over all brain size
> and ' constraints of as yet unknown laws of structure and growth '.

This suggestion is based on an analogy with structural side-effects,
based largely on the constraints of 3-dimensional geometry on growth and
form. But how could this spatial analogy apply to UG, which is purely
abstract and symbolic?

> An example given in the article is that humans have two teats because
> they usually bear single children and often have twins, the number of
> teats corresponds to the maximum number of children born at any one
> time. There wasn't even a suggestion made that the human body wouldn't
> be likely to have one breast, or more than two for that matter. The
> above example is the Darwinian method of looking at and adapting things
> into a theory.

But it's hard to see how this structural property (the symmetry of the
body) applies to UG...

> Gould's criticism seems very sensible at first, but then
> an example is made of the eye. No physical process other than natural
> selection can explain the evolution of an organ such as the eye, this
> is because it is the only physical process in which the criterion of
> being good at seeing can play a causal role.

Yes, but UG just HAPPENS to be universal to all natural language peaking
humans. It is not the only possible basis for a language with the same
expressive power as natural language. Nor is it the simplest or optimal
one. So why UG, rather than something with the same power, but much
simpler, and perhaps even learnable?

And how would the structure of UG have been shaped by evolution? Unlike
the eye (and its various homologues in other organisms), which is the
only kind of solution to the problem of transducing light, UG is far
from the only solution to the problem of communicating by language, and,
in particular, communicating in a shared, mutually understandable
grammar.

> Gould believes that most evolutionary change
> is confined to bursts of change which are relatively brief on the
> geological time change.

Fine, but are we to believe that one of there bursts just hiccupped out
UG, with all its complexity?

> The general conclusions and arguments coming from Gould's ideas seem
> to be that Natural Selection could be used to explain language and
> where it comes from.

It could explain language. But can it explain grammatical capacity?
And what about the poverty of the stimulus? If Skinnerian trial and error
reinforcement could not shape grammatical capacity, how could evolution?
Is there not an even more impoverished "stimulus" for evolution?

> 1. Speakers need a learning mechanism for labels for
> cultural innovations. The example of a screwdriver is given in the
> article.

Yes, but much simpler conventions than UG would have done just as well,
and might have been learnable, to boot.

> 2. It may be difficult to develop a huge innate code. ( I don't quite
> understand where this point comes from, even after reading it over a
> couple of times).

Not sure what you mean, but this probably refers to the complexity of
UG.

> The process of language evolution is also looked at in detail.
> There must have been a time when there was no language and then
> language 'happened'. As this happened there must have been genetic
> variation among individuals in their grammatical competence!!

Hard to imagine...

> Society and it's effect on language development was also looked at,
> the lifestyle of many humans led to an increased need for cooperation
> with others in order to survive.

All true, and relevant to "language" in general, but to UG capacity in
particular, which is subject to the poverty of the stimulus? What can
culture do with or for that?

> A comparison is made between humans and apes, the conclusion drawn was
> that there is very little, if any association between the ' language'
> of humans and apes. Even when sign language was compared in deaf
> children and apes there was still no association. This does NOT go
> against the argument that human language evolved by natural selection
> though... the debate on this one goes on.

There are surely some brain structures that evolved by natural selection
and are specific to language and that are not shared by other species.
But it is not clear how UG figures among them; it is an oddball, in
being a very complex structure, yet not a geometric one. Moreover, it is
universal, but not functionally necessary; simpler alternatives would
have been just as adaptive.

> The article concludes with the result it set out to show which was that
> language has developed through a process of natural selection. I would
> have liked to have seen some other ideas in the article which disagree
> with the natural selection hypothesis, I wonder what Pinker would have
> to say about the work of Skinner and others in the same area.

He would rightly echo Chomsky's point about the poverty of the stimulus.
But if grammatical capacity was not shaped by Skinnerian learning, how
could it have been shaped by evolution? Is it a better alternative to
suppose that UG all happened at once, by a sudden mutation?



This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:24:15 GMT