Re: replacement rates

--- In Nostratica@yahoogroups.com,
"H.M. Hubey" <hubeyh@...> wrote:

>
>
> Richard Wordingham wrote:
>
> > ---
> >
> > Weighting is one thing that

calculated replacement rates can be
used

> > for. Mind you, I got

criticised for using IE replacement
rates to

> > assess a comparison of

Austronesian and Austro-Asiatic (or
rather

> > Malayo-Polynesian and Mon-Khmer

as the data turned out).

>
> It would not be too difficult, it

seems to me, to start with the

> differential equation and put a

time

> varying coefficient and try to

fit the data and estimate the

> coeffiecient. Is that not

possible?
Dyen and his collaborators have
already published their maximum
likelihood estimates of replacement
rates for Austronesian and
uncalibrated (i.e. time unit
undetermined) estimates for
Indo-European. Raw data for the
latter is available on-line.

> I think, truthfully, that it is

too early to do that. The IE
situation

> is not too clear. It seems to me
> that you should start with the

aborigine languages of Australia.
They

> are isolated, and there is
> less chance of other factors

botching up the estimation. The
time

> period is reasonably clear
> too :-); anywhere from 40,000 to

60,000 years.
I think they could be a _bad_ group
to work with! They have taboos
which accelerate language change
and encourage inter-dialect
borrowing. (I'm thinking of taboos
on the names of the dead and
'avoidance language' for talking to
mothers-in-law.)
Richard.
---End-of-Message-----

> > I think it would be desirable

to find some way of capturing
cognates

> > when their meaning has drifted

apart. It looks laborious, if only

> > because some way would have to

be found to compensate for allowing

> > more false matches in.
>
>
> I think what linguistics needs is

a "semantic metric". I know how to

> derive at least two different
> ones from data but need time and

money :-) I think all linguists
should

> appreciate the need for
> semantic metrics. It does not

matter how many. That could always
be

> worked out, fixed,
> repaired, improved etc. The trick

is to get going.

>
> >
> >
> > Richard.
> >
> >
> >
>
> --
> M. Hubey
> -o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-

o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o-o

> The only difference between

humans and machines is that humans

> can be created by unskilled

labor. Arthur C. Clarke

>
> /\/\/\/\//\/\/\/\/\/\/

http://www.csam.montclair.edu/~hube
y