Re: [tied] Re: celtic

03-10-03 15:22, guto rhys wrote:

> Correct me if I am wrong - which I often am.
>
> The initial consonant of PIE *penkWe was analogically remodelled on the
> preceeding numeral *kWetWor (??) in early Celtic. Giving thus *KwenkWe,
> latter *kW- > p- in P-Celtic. Attested by Welsh 'pump', rather thatn **ump.

Not quite. There seems to be a common tendency in Italic and Celtic to
change *p..kW.. into *kW..kW.. by distant assimilation. This is what
happened to *penkWe '5' and *pekW- 'cook' (> *kWenkWe, *kWekW-).

> An identical process happened in Proto-Germanic but it was the form
> for the numeral 'four' that was remodelled on the following numeral,
> thus *kW- > f- (Goth. fidwor).

This is disputable. Miguel has a whole theory about rounded labials (*pW
etc.) in PIE. I think Gmc. *fedwo:r/*fegWo:r/*fedur- is a case of
assimilation in a root containing another labial segment (cf. *wl.kWos >
*wulfaz) and the influence of the next numeral was not necessary.

> Is this the same logic used for the apparent initial p- in the word for
> 4 in some Italic dialects?
>
> How common is analogical remodelling in numerals??

Very common, although it may not have happened in these two cases. In
serial counting, numerals naturally fall into a binary rhythm (as in
"One, two, buckle my shoe; three, four, shut the door ...") and may get
assimilated to each other especially within such pairs.

Piotr