Re: [tied] searching for common words for all today's languages

From: Richard Wordingham
Message: 43267
Date: 2006-02-05

--- In, "Brian M. Scott" <BMScott@...> wrote:
> At 4:18:22 PM on Sunday, February 5, 2006, Richard
> Wordingham wrote:

> > --- In, "Brian M. Scott"
> > <BMScott@> wrote:
> > Some radar systems regularly search for and find signals
> > that are way below the level of noise.
> But under what circumstances? I'm no engineer, but I
> suspect that this requires knowing a good deal about the
> nature of at least the noise and quite possibly the signal
> as well, and probably not just qualitatively, either.

That much is true; this was largely meant as a warning about glibly
saying something is lost in the noise.

> It's
> not clear that it's possible to have that kind of knowledge
> about language, and we certainly don't have it now.

It's probably not being applied numerically, but I'm not so sure we
don't have the knowledge. I suspect we need to apply information from
the daunting field of semantics, but I also fear that that will only
allow us to go a little way below the level of the noise. A crude
example is looking for word families to eliminate recent loans.

On the noise side, we may well have the problem that, statistically
speaking, the association between sound and meaning is not arbitrary.

We are straying well off-topic, but I'm not sure where to divert this