Re: Natural Language and Natural Selection

From: Stevan Harnad (
Date: Sun Mar 03 1996 - 18:05:11 GMT

> From: "E.J.Fletcher" <>
> Date: Sun, 3 Mar 1996 15:48:19 +0000 (GMT)
> All human societies have language. Therefore did language evolve
> because it made us better at surviving and reproducing? It is traits
> which enable an advantage to be held over others, which achieves
> dominance over time. After all "nothing succeeds like success"; success
> which is "founded on a process of blind variation and selection. Blind
> because no one steers the variation in any direction; it is random"
> (Dawkins). So language "appeared at some point, over some form of time
> scale, installing in mankind some form of advantage. It obviously was
> instrumental in self survival, otherwise such a trait would not be
> evident now.

Mathematics has also helped us survive, but did that "evolve" too?

It is very likely that capacity for language is based in our brains and
our genes, and was selected for because it helped our ancestors survive
and reproduce. But it is not logically impossible that it was merely
invented by some individual or group, and then passed on by "word of
mouth," so to speak, without any specific basis in our brains or genes.
I mention this only by way of suggesting that we should not jump too
quickly into a genetic, evolutionary explanation of everything that is
advantageous; there are other ways to pass advantages on besides genes.
Indeed, language itself is the MEANS of passing things on from
generation to generation without any genetic basis or genetic change.

And remember that the big question for this seminar is not about the
evolution of "language" in general, but of our specific, and apparently
species-universal grammatical capacity: UG (Universal Grammar). Did that
evolve? How? What was the variation? What were the selective

> However, was language determined by evolution initially? Surely
> language development is a consequence of the evolution of
> consciousness? Language would not exist if cognition
> (thought/experience) was not present , after all language
> allows expression of thought or experience; if only a somewhat poor
> one. Perhaps it is not fair to say the way in which we formulate
> language in our minds, and the factor that prompted us to do so ,
> originates from conscious reasoning: but through a process of
> unconscious reasoning. Nevertheless language us to express and create
> some form of labeling for our experiences, to which we are, of course,
> conscious. After all to experience is to be conscious.

A bit garbled here, for a kid-sib to sort out: You seem to be thinking
we had to make up our minds to say something before we could invent
language. Maybe so. Maybe it was shaped gradually out of other forms of
communication that we still share with animals (alarm reactions, mating
calls, etc.).

The relation between consciousness and language is no doubt a
complicated one -- but then so is the relation be consciousness and
anything we do, including both communication (deliberate and
involuntary) and just plain old behaviour.

> Chomsky points to a universal grammar (U.G.) which is innate,
> suggesting that grammatical rules are inborn. this is in a sense
> logical, as grammatical rules are centred around what it is possible to
> say and what it is not possible to say. What you cannot say is never
> produced, so no correction occurs. Therefore the process of learning
> not what to say has been made redundant.

Why is it "logical" that you should know what's grammatically right
without having to learn the rules? Can you think of any other area in
which this is true (mathematics? chess?)? And if you didn't learn them,
yet you "know" them, how did they get into your brains and your genes?

> Pinker and Bloom suggest that "all languages are complex computational
> systems with basic rules and representations". Surely it is more than
> this however. Basic rules and representations of what? Cognitive
> processes? Experiences?

Whatever it takes to make you, for a lifetime, produce and recognise
all and only the grammatically correct sentences of any human language.
First, it's a mystery how that can be done at all. Then, linguists
gradually piece together what rules would be needed to produce and
recognise all and only the grammatically correct sentences. They
discover UG. Every 4-year-old child already "knows" UG. How?

> Chomsky, Fodor, Lenneberg and Liberman etc.
> suggest that the mind is a "series of freely activating computational
> modules of which language is a product". Where did the input originate
> from? should this not be the question which neoDarwinists should be
> searching for, as it appears that language is a byproduct of such
> "processing". Is experience, or unconscious reasoning the cause here?
> After all, is language not a product of cognition?

No doubt, but the hard question is about grammar. Language is a bigger
question, but it is obvious that some of its properties can be learnt
(e.g., a vocabulary, pronunciation) and some things could have evolved
(the tendency to vocalise, the specific speaking and hearing capacity of
human beings). I agree that there is a question of whether thought drove
language or language drove thought. Trouble is, we do not yet know what
thought is, whereas we are getting a clearer idea of what language is.
Some theorists (like Fodor) are even speaking about a "language of
thought," and even an inborn one!

Focus, though, on grammatical capacity, as it is there that the
cognitive theory (of what there needs to be inside a head, to give it
the ability to produce and recognise all and only grammatically
correct utterances) is most fully developed. It is surprising that that
what it take needs to be in the head turns out not to be learnable,
hence needs to be built in. How could it have gotten there? Is this
general explanation of how "language" got have evolved a good enough
explanation of how UG in particular could have evolved?

> What if language development was purely developed though purely random
> selection. How does this explain how language emerged? Could an innate
> generative grammar have evolved by natural selection. Evolution appears
> to explain development, not how the qualities emerged. e.g. universal
> grammar appears to be inherent ( and thus seems to be a quality
> selected by evolution due to it's successful nature). However this does
> not explain how, or why language originated, it just defines what can
> and cannot be said. Perhaps language was a bi-product of cognitive
> functioning. If Pinker and Bloom suggest "human knowledge and reasoning
> is expressed in a language of thought", then surely cognition
> superseded the development of language.

Kid-sib's head is spinning! What are you actually saying here? First of
all, "development" seems an extra, confusing word. We are talking about
(1) where language and language ability in general came from, and how,
and about (2) where UG and UG ability in particular came from, and how.
(2) is clearly a part of (1). The answer to (2) will be part of the
answer to (1), but a very delicate part.

There are really only two candidate answers in each case: Either they
were shaped by learning (through behavioural variation and feedback
from its consequences: Skinnerian reinforcement), or they were shaped
by evolution (by genetic variation and feedback from its consequences:
survival, reproduction; "Darwinian reinforcement").

There IS one more possibility, besides Skinnerian learning and Darwinian
evolution, and that is INSTRUCTION, but for that one would already have
had to have language, so that can't be the answer. (Neither is the idea of
"spandrels," the kind of previous geometric constraint that sometimes
gives rise to complex structures without any specific selective
pressures shaping them.)

Consciousness, and the evolution of consciousness, is another problem
(and a bigger one).

But the fact that we HAVE UG ability by age 4, and cannot have learnt
it, requires explanation. Is evolution. either sudden or gradual, a
satisfactory explanation?

> It appears that language "assumes" a listener, as a form of equilibrium
> is maintained between the size, and use, of vocabulary needed to
> express experiences, and the minimal language need for the listener to
> fully comprehend the speakers words, without ambiguity. As such Pinker
> and Bloom suggest that there is pressure for evolution of neural
> mechanisms which decode information.

True, but what about UG in particular?

> However when reduced to basics, is it not consciousness which allows us to do
> this? If so the question should not be whether evolution put pressure on the
> formation of neural mechanisms which decode information, but why should
> evolution have led to our consciousness, and how that, in turn, lead to
> language development.

We'll try to deal with the evolution of consciousness later, but for
now, the question is UG as well as language. Other species are conscious
but have neither language nor UG, so the questions must be separable.

Suppose we were born knowing the rules of chess without needing to
learn them: would "consciousness" be an explanation of how we could do

> According to Pinker and Bloom "syntax is acquired by general purpose
> learning abilities". However as suggested earlier universal grammar
> appears to be inherent. It dictates what can and cannot be said. On the
> other hand, meanings may be learnt, as we learn through experience what
> makes one particular experience common to others.

Correct. So the problem is how we know UG, which is unlearnable (because
of the poverty of the stimulus), not how we learn word meanings, which
ARE learnable (unless you believe in "vanishing intersections").

> "Natural selection, while feeding on variation uses it up, therefore
> there are not qualitative differences in basic design". Does this mean
> that a process of natural selection led to U.G., as this is now
> inherent to all humans? Or was U.G. always present even at the origins
> of language itself? Could U.G. be only 5% complete? surely it's all or
> nothing?

Now you're closer to the real question: If UG is too complex to be learnt
during the lifetime of one person (yet everyone knows it by age 4),
could that complexity have been shaped gradually by evolution (how?). If
suddenly rather than gradually, how? And if neither suddenly nor
gradually, was it ALWAYS there in the structure of matter, since the Big
Bang that started the universe, just waiting till enough tissue could
gather in a space the size if a basketball in order to be formed, like a
"spandrel"? (Chomsky sometimes writes as if he believes this/)

> In conclusion Pinker and Bloom suggest that language shows signs of
> complex design for the communication of propositional structures" and
> that "the only explanation for the origin of organisms with complex
> design is natural selection". Perhaps this is true, but not when
> concerning language. How does this explain U.G. Surely language should
> be assessed in the context of the development of cognition.

Kid-sib understood all the preceding paragraph except the last sentence:
UG is supposed to be some sort of byproduct of the evolution of
cognition in general? Why? How?

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