Re: Why grammar is so complex a subject

--- Shreyas Sampat <ssampat@...> wrote:

> Gary Shannon wrote:
>
> >I think I've finally figured out why grammar is so
> >complex. It's because it's an artificial attempt to
> >discover "rules" in what is really a monsterous
> >collection of exceptions. There ARE no rules; only
> >exceptions! Tens of thousands of unique patterns of
> >words learned by rote which, in reality, have no
> >underlying theoretical reason for existing other
> than
> >generations of acquired habits passed down with a
> bit
> >of alteration and streamlining from one generation
> to
> >the next.
> >
> >
> I think this theory predicts a much larger set of
> probable languages,
> with fewer common qualities, than are in evidence.
>
Counter argument: For emerging proto-humans in a
rudamentary hunter-gatherer society there are a very
limited number of things that need to be discussed,
and the ways of putting those things together into a
single utterance are mathematically very limited. "Me
goat see", "Goat me see", "Me see goat", "Goat see
me.", "See goat me.", "See me goat." Which, for
reasons of survival, would have to be differentiated
in meaning from "Tiger see me.", "See tiger me.", etc.

Of those six one or two could, through usage, become
the habitual utterance, and thus create the "rule" of
SVO or SOV or whichever applies. That's a pretty small
set of possibilities to choose from, so all languages
that might have been invented in isolation would tend
to fall into a very small number of word order
systems.

Then there's the matter of how many ways case might be
marked to clarify who is being seen by whom. "Tiger
see at me.", "See tiger to me.", "Me me tiger see.",
"See tiger me me."

Then, as sophistication grows, tense might become an
issue. But again, there are a very limited number of
possible ways to mark the tense of a verb, and so
tense marking would again fall into a very imited
number of categories. "See before tiger me me.", etc.

I think that biological constraints, universe of
discourse limitations imposed by the necessities of
hunter-gatherer survival, and limitations imposed by
early proto-human brain capacity would tend to mold
languages in such a way that variety would be
relatively limited, except in details of lexicon which
would be subject to infinite diversity.

Once these limited ways of expressing the simplest
ideas became well established as speech habits, and
passed down from generation to generation, new layers
of sophistication would tend to adhere to the earlier
usage habits and the "rules" of grammar would begin to
emerge, not because of some inborn "grammar gene", but
because simple repetitve usage turned certain
formalized sentence templates into cultural habits.

What starts out as a simple set of sentence templates
becomes elaborated over time to the point where modern
children have to internalize thousands of templates in
order to go from "Me no see puppy. Puppy go bye-bye?"
to "I was wondering what ever became of that cute
little puppy you used to have."

The earliest childhood templates are those that the
earliest proto-humans would most likely have created
for themselves, and so ontogeny once again mirrors
phylogeny.

The interesting implication of all this is that if
true, it should be possible to do complete parsing of
any arbitrary natural language sentence with a process
not much more sophisticated than the
search-and-replace function of an ordinary word
processor program, using a set of template matching
patterns. In fact, that sounds like an interesting
project to attempt!

--gary

Replies