On 2008-04-08 04:00, Patrick Ryan wrote:
> Piotr, if you do not know this is sophistry, then I will be surprised.
>
> I said nothing about analyzable within PIE. I pointed only to a pattern of
> final elements in *CVC roots that were reminiscent of very common root
> extensions added to *CVC.
>
> In order to 'relate' them, you must divide them. If you cannot divide them,
> they are not relatable.
>
> It is really that simple, and I believe you know it.
The problem is only that you need some good independent evidence to
justify your divisions. Otherwise you will remain the only person who
sees them. It's all to easy to fool oneself into "seeing" a pattern
among random data. It's just how our brain works -- it's a powerful
pattern-recogniser, so it often recognises a pattern that isn't there.
Just look at the sky at night -- you will see Orion, Cassiopeia, the
Plough (or Big Dipper, if you like) etc. -- you can almost see the lines
linking the stars up into familiar shapes.
> My Goodness! Shades of the very poorest reconstruction methodology! That is
> what has made Proto-Caucasian such a pathetic enterprise.
>
> What would you imagine for the original: *gWeuiH2m?
Hullo, Patrick, I'm here. That straw man isn't me.
> You can pick and choose what will be "retained', and what will suffer
> execution at another Picaud's Law.
>
> Frankly, I had not expected this as an argument from you; I missed it in
> the
> article of I would have called you on it.
Because there is no such argument there, or in my previous posting.
> There is _absolutely_ no evidence for a variably resolved cluster that
> violates every rule we know of PIE structure.
>
> The constituent parts of *pla(:)H-, as you well know, would be likelier to
> be *pele- and -*H(2).
>
> But in this particular case, 'set into motion', my belief would be that the
> pre-PIE form was *pAlá:-, naturally long. But let us not get into that
> question yet.
What root are you talking about? I cited *pleh1- 'fill' just as an
example of a triconsonantal root that cannot be further analysed given
our present knowledge. I didn't want to suggest that it was related to
*pleu- 'swim' or anything else.
Comparative analysis is like a microscope. It allows you to see
historically underlying fine structure in synchronically indivisible
words. For example, you can blow up Eng. young until you can see old
morphological divisions: *[h2ju-h3n.]-k^ó-. It shows that the modern
sequence of three phonemes [jVN] reflects no fewer than three PIE
morphemes. To complicate matters further, the internal divisions have
become so blurred that the modern vowel straddles two morphemes and the
final velar nasal reflects *n and *k^ at the same time. Needless to say,
in order to achieve such precision and to convince other people that the
etymology is correct you have to process an enormous amount of data --
actually, the reconstruction above summarises the work of several
generations of IE linguists.
Like other types of microscope, the comparative method has its
limitations. We cannot reconstruct ad infinitum because the longer the
range of comparison, the scarcer the data and the worse its quality. You
still see shadows of patterns, like constellations in the sky, but you
can't be sure they are not optical illusions.
Piotr