Re: Can relationships between languages be determined after 80,000 y

On 2008-01-29 11:00, fournet.arnaud wrote:

> labials m w b p p?
> dentals n t d t?
> laterals l r tl tl? dl (because of Chinese)
> sibilants s ts z dz s? ts? (because of Chinese)

What has Chinese got to do with Proto-World?

> velars g k k?
> spirants y h H
> glottal stop ?
>
> Any other system is doomed
> because it **fails** to account for all the necessary constrasts.
>
> And more phonemes are necessary to account for constrasts.
> Especially pre-nasalized phonemes.

Using the same logic you will probably add clicks a whole array of
clicks to account for Khoisan and perhaps more places of articulation
(retroflex, palatal) to accommodate Australian and Dravidian, won't you?

> The first book to write on macro-comparison
> is a phonological analysis of existing languages
> containing **no** cognate.
> Looking for cognates is the **last** step
> not the first one.

I most emphatically disagree. Given enough time, anything could have
happened to phonological systems. You wouldn't be able to reconstruct
even the phonological inventory of PIE, by just looking at the daughters
and speculating "what is necessary". Cognates, besides lots of other
important things, allow you to identify conditioned changes, historical
mergers and splits, and without that knowledge no linguistic genius can
guess which features are really old and which represent inovations.

Piotr