Re: [tied] Mother of all IE languages

From: Richard Wordingham
Message: 27906
Date: 2003-12-02

--- In, "S.Kalyanaraman" <kalyan97@...>
> The basic flaws in the combined methods of lexicostatistics and
> glottochronology are:
> 1. The limited number of words chosen (2449 lexemes from 87
> languages; thus about 29 lexical items per language) as cognate
> pairs may not be a true representative sample of 'genetic'
> relationships among languages.

If you assume that 'lexemes' means pairs, then the calculation is 2
* 2449 / 87, approximately 56. However some lexemes, such as those
for 'two', 'five', 'I', and 'mother', will have occurred in far more
than 2 languages.

> The choice of the sample is not
> random and hence, application of statistical methods of analyses
> result in spurious correlations.

The only bias I can see is a risk that meaningss have been chosen
because their words tend to resist replacement in Indo-European
languages. I doubt that this has actually biased the results. Do
you have some other effect in mind?

> 2. The concept of 'genetic' links among languages is itself a
> questionable assumption, since there could be a linguistic area
> where interchanges occur with a rapidity and intensity,
> in relation to technological advances such as farming
> tools/techniques or alloying of minerals. The 'genetic' links
> that a particular language is learnt only from a 'genetically'
> related language speakers.

I don't think you mean this. You may like to note that several
creoles have been included (Sranan, a couple of French creoles). I
am impressed by how close they come to their lexical parents.

> This is a false assumption. For example,
> in the Sarasvati Civilization, we have indications of
> metallurgists's lexical repertoire which is based on new
> such as bronze or brass alloys (or pancaloha, five-metal alloys).
> such words related to new technologies, the words are likely to
> entered the parole of, say, Munda, Dravidian and Indo-Aryan
> almost simultaneously. In such cases, the assumption of
a 'genetic'
> relationship gets immediately falsified.

The word lists were designed to be lists of meanings that all
languages have. A few meanings actually change their words very
frequently - 'road' comes to mind as one of them. Words of the
sort that you are discussing above are excluded from the lists by
design. That is not to say that some words are not vulnerable to
replacement as a group - native number systems have been swept aside
in much of East Asia, and we've recently had much discussion about
the plausible Semitic origin of PIE *septm 'seven'.

> 3. If indeed agriculture spread from Anatolia and if IE language
> traces intruded into Bharat

Can't find the place in my atlas. The language of this list is
English, though we don't normally make a fuss about it. (I'm
speaking purely for myself, in a personal capacity.)

, it should be possible to find IE
> agricultural terms in indic. But, this is not the case.

Does anyone sensible believe that agriculture was brought to India
by IE speakers? The conventional theory is that IE was brought to
India (sensu lato) by pastoralists.

> The next exercise of Russel D. Gray and Quentin D. Atkinson should
> include some lexical items from my Indian Lexicon; there is a
> possibility that the Anatolia hypothesis will be disproved for
> agricultural and metallurgical cognate pairs.

As a matter of interest, has anyone done similar studies with
technological (or craft) terms?

If it's not too far off-topic, what is the current view on the
origin of North West Indian agriculture?