Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: Fw: Intro to natural language processing



Rob,

RF> There are so many books to read. If you could
> give me a hint that Deacon even begins to explain
> syntax (other than to say it is "rudimentary")
> perhaps I would read his book.

Deacon never said that syntax is "rudimentary",
but what he did say is that the rich syntax of
modern humans must be the result of a long period
of evolution.  Furthermore, the early stages of
that evolution must have started from something
simpler.  The major questions are what and why.

Deacon's primary area of specialization is
neurophysiology.  He branched out from that
basis to anthropology and the analysis of the
related evolutionary evidence.  He is not a
linguist, but he has some understanding of the
issues, and unlike Chomsky and Jackendoff, he
did not have the inherited baggage of Universal
Grammar and a Language Acquisition Device.

RF> If you define NL to be fundamentally symbolic
> I imagine symbols do become very important to your
> model. But I don't think symbols explain syntax.

No.  Deacon does not start with a definition of
language, instead he starts with an analysis of the
vocalizations of mammals and birds, with an emphasis
on primates.  Many people have done things like that
over the past 60 years, but Deacon focuses on various
ways of producing sounds, and on the neural connections
of the vocal mechanisms.  He relates that evidence to
the kinds of vocalizations and to the disturbances
caused by various neural lesions and handicaps.

For example, Deacon relates autism and William's syndrome
(WS), which are almost exactly complementary kinds of
neural disorders.  Autism is characterized by a disorder
or deficiency in the frontal lobes with essentially normal
development of the rest of the cortex, but WS leaves the
frontal lobes mostly intact, but with significant reduction
in the rest of the cortex.  (There are other factors involved,
but read the book.)  One interesting point is that autistic
and WS individuals have almost complementary strengths and
deficiencies:

  1. Autistic people have severe problems in learning and using
     langugae, and they have serious difficulties in social
     interactions and in understanding other people's emotional
     states and inclinations.  Despite their language handicaps,
     many of them are highly or even extremely talented in
     areas that require combinatorial fluency, such as music,
     mathematics, and many kinds of problem solving.

  2. WS people are almost the exact opposite.  They are very
     fluent in the use of language to the extent of being
     garrulous, and they are extremely social to the extent
     of being a nuisance.  They often have large vocabularies,
     but many of their words are not "connected" with any
     nonlinguistic knowledge or experience.  Despite their
     verbal fluency, their IQ scores tend to be far below
     normal, and their problem-solving abilities are minimal.

When I was reading that I couldn't help thinking that the
robot R2D2 was autistic and C3PO had William's syndrome.

Again, many other people have related autism and WS, but
Deacon goes further and relates them to a wide range of
other studies.  That is why the book takes over 500 pages
and why I can't summarize it effectively.

When talking about symbols, it is essential to note that
Deacon (and I) talk about those issues in Peirce's terms.
One fundamental distinction (which Peirce stated early in
his career and developed much further over the years) is
the tripartite division of icon, index, and symbol:

  1. An icon determines its referent by some similarity.
     Examples are a picture or diagram that resembles the
     referent in whole, in part, or in some abstraction
     of some aspect.

  2. An index determines its referent by some physical or
     causal connection.  Examples include a finger pointing
     toward somthing, a weather vane indicating the direction
     of the wind, or smoke indicating fire.

  3. A symbol determines its referent by some social habit
     or convention, which is independent of any similarity
     or causal connection.  An example is the picture of
     a black telephone with a rotary dial, which began as
     an icon and has evolved into a symbol that has little
     or no similarity to today's cellular phones or even
     desktop phones.

Deacon analyzes a wide range of signs to show that nonhuman
mammals and birds are very good at learning icons and
indexes, but only with great difficulty and in rare cases
have any of the experiments with chimps and bonobos crossed
the boundary from index to symbol.

RF> ...It is the opposite of what we would expect if language
> learning started by combining words according to general
> concepts. It is more what we would expect language started
> by analogizing between examples, and concepts only began
> to emerge later from collections of many such analogies
> between examples of usage.

Deacon doesn't discuss that particular example, but your
second sentence comes closer to his interpretation.

John