Rules vs. Lists

[Rob Freeman]

My point is partially that "the linguistic pie can be sliced many
ways". Thanks for acknowledging that.

But there is more. Something happens when there can be more ways of
slicing than there are things to slice.

There is another aspect. The idea of more rules than examples seems
only surprising at first, but it is important. The thing is, if there
can be more rules than examples, you can be never done slicing. That's
because for every list of  "slices" you make there can be another,
longer, list to be made. Each list of rules you make either
constitutes, or produces, an even longer list, which implies a longer
list etc.

Such a system operates on itself to constantly produce complexity.

It's just a quirk of the system, but very nice, because it predicts
change, drift, etc, and also gives us considerable scope for
complexity, "new ideas", even "free will" if you like. (The system is
less specified than one which can be abstracted with a smaller number
of rules, it is unstable, even random at one level, liable to go off
at tangents and develop in completely different ways, produce
different languages etc.)

So it is not quite that "there is no difference between one method of
slicing it or another." Because no set of slices is complete. Each
set, list, of "slices" always implies another, larger, set. More than
one larger set actually, should we choose to look for them.

I think this is right. It seems to be the case. Worth investigating, anyway.

Whether there is no end to the list of grammars which linguists can
derive may be questioned by some, but it seems sure there is no end to
things that can be said. I'm not sure if the premise of a set without
end is open to scientific investigation. It is hard to falsify.
Fortunately it is easy to go to the other end of the problem and
explore whether you can find more and more rules from a given set of
examples.

It seems quite a small thing, that there might be rules than examples,
but it has consequences that imply a qualitative difference in how
language works which go beyond what any given speaker does.

I think we should look at the possibility carefully.

-Rob

P.S. We can look at the significance of tone for the analycity of
phonemes if you like. It may be relevant to the idea of more rules
than examples. But I would like to hear other people's opinions first.
In particular I'd like to hear how this relates in standard theory to
that other problem of phonemes being modified in context, voicing
assimilation in Russian obstruent clusters was it, the classic example
of this?

On Mon, Jul 7, 2008 at 9:05 PM, A. Katz <amnfn at well.com> wrote:
> Okay. Your point is that the linguistic pie can be sliced many, many
> different ways. I don't disagree, but I have another point that I have
> been trying to make: there is no difference between one method of
> slicing it or another, when we are studying how a language works. If it
> all adds up correctly, all the different ways are equivalent, and there's
> not any reason to prefer one method over another, unless we have adopted a
> particular constraint, such as economy of rules or mathematical elegance.
>
>
> Now, a particular speaker may adopt one way, and another speaker may adopt
> a second. A third speaker may adopt a third. There may be as many
> different ways of parsing a language as speakers, although that is
> doubtful and perfectly open to scientific investigation.
>
> It's okay to study the details of how speakers process language. It is
> also okay to find ways to describe language apart from speakers. What is
> not okay is to confuse what any given speaker does with how the language
> works.
>
>
> Best,
>
>      --Aya