Re: The role of analogy, alliteration and sandhi in counting

> Look at the threads, especially at
> http://groups.yahoo.com/group/cybalist/message/14923 ! As I
> understand Miguel's idea, *pW was neither stable nor common, and
> tended to become *p or *kW. However, the elimination of *pW
> proceeded at different rates in the different branches, and with
> varying preferences. The branches need not yet have become

separate

> languages. What we see is similar to the Germanic merger of *w and
> *hw. It is complete in most dialects, and from a sample of modern
> dialects one might very easily reconstruct just *w. That would
> leave isolated forms like English <who> /hu:/ (compared to <what> -
> /wOt/ in much of English) to hint that things were once more
> complicated.

A loose idea (as usual): Suppose the substrate language of NW Europe
(Pre-IE Nordwestblock) had dialect variations p/kW. That result in
doublet (p/kW) loanwords in the successor IE languages (example:
http://www.angelfire.com/rant/tgpedersen/pd.html , check the Cushitic
forms, and *panna/*kanna?), perhaps to the point where the
Sprachgefühl demanded "resolution", preferring p-forms over kW-forms
or vice versa, whereupon the same preference spread to th rest of the
language? Is this a good way to explain the Irish 'cothrige'-forms?

Torsten