Re: Ah, look at all the lonely languages

--- In cybalist@yahoogroups.com, "Davius Sanctex" <gruposdavius@...>
wrote:

> [ ... ] How much evidence needs to be compiled to show a

relationship, one language to another.

> [peter] It is regular patterns of relationship, rather than

quantity of similar words, that linguists look for. But there are a
number of actual cases where opinions differ from linguist to
linguist.

> ---
> Is there some statistical proof which give us the significance of a

relationship? Say, if for two languages A and B 1000 cognates are
identifidied, this is more significant if for the two languages A and
B only, say, 250 cogantes are identified.

>
> I think a test based on a predetermined list (for example Swadesh

list) can provide such a test, by comparing the hazard expected names
of cognates, with the number of true cognates, what do you think
about?

>
> Davius S.

Roughly speaking, a high number of cognates ( > 10%) on the Swadesh
100-word is sufficent prima facie evidence of relatedness, but a low
number is not necessarily disproof. Some pairs of Austronesian
languages get very low scores. With dead languages, you may find
that too much of the vocabulary is missing to use such a test. Some
of the ancient languages at the Rosetta Project have very patchy 100-
word lists.

Richard.