From: Brian M. Scott
Message: 60650
Date: 2008-10-07
> From: "Brian M. Scott" <BMScott@...>I'm sorry, but the fact that you think that widely
>>>>> Sometimes, wide-spread words are LWs when isolate
>>>>> words are in fact cognates, so I'm not sure this
>>>>> reasoning is really adequate.
>>>> You're missing the point. Richard isn't claiming that
>>>> using the proto-language instead of choosing from its
>>>> daughters *guarantees* that you won't find false
>>>> cognates; he's merely claiming -- correctly -- that you
>>>> are much likelier to get them if you allow yourself to
>>>> choose from the daughters. The probability that a word
>>>> reconstructible for the proto-language is actually native
>>>> to the family is clearly higher than the probability that
>>>> a word chosen at random from one of the daughters is
>>>> native to the family.
>>> I understand the argument but I'm not convinced. There
>>> are counter examples.
>> The fact that you can talk about counterexamples shows that
>> you do *not* understand the argument. The argument doesn't
>> deny the existence of widespread loanwords.
> I understand the argument,
>>> What are Ringe and April McMahon investigating ?You can easily find them on the web.
>> I told you: quantitative techniques for doing historical
>> linguistics.
> Do you references for their work ?