--- In
cybalist@yahoogroups.com, "Brian M. Scott" <BMScott@...> wrote:
>
> At 1:04:13 AM on Saturday, July 5, 2008, stlatos wrote:
>
> > --- In cybalist@yahoogroups.com, "david_russell_watson"
> > <liberty@> wrote:
>
> [...]
>
> >> 'Aptya'/'Athwiya' is indeed an irregular correspondence,
> >> but one on obviously much more solid ground than
> >> 'Thraetaona'/ 'Targitaus'.
>
> > Yes, of course, but the principle that changes might occur
> > in only one word in a language and still be valid and
> > identifiable must be used for both.
>
> It is an empirical fact that irregular changes occur, not a
> matter of principle,
Are you objecting to my use of the word "principle"?
> and there is no guarantee that isolated
> instances can even be identified, let alone demonstrated.
> Even when an irregular relationship is very likely, the
> details of the change are generally undemonstrable in the
> absence of intermediate forms.
It's the same for all historical linguistics. I'm not going to
treat irregular rules as rendering all attempts at reconstruction
impossible.
> There is obviously some
> value in finding a plausible pathway, but in the case of an
> isolated change it isn't subject to confirmation; it's a
> Just-So story, and the proposed 'rules' have no real
> evidentiary support.
I disagree. Consider:
comfortable, comftrable
What caused this?
comfortable
comfrtable
comftrable
First came V2 > 0, a common though not necessarily completely
regular change, then metathesis to correct an impossible cluster.
This intermediate form is undemonstrable because it occurs within the
minds of speakers, not in speech or writing, but the stage with frt
before ftr is needed to explain the whole.
Similarly, a proposal like:
axptyo
axptya
axwtya
axtwya
axtwiya
a:twiya
a:Twiya
is not and likely never will be proven, but it does follow rules of
ordering (pt > wt > tw first as tw > Tw is reg.), met. to correct
impossible clusters (wt > tw), and an explanation of apparent
irregularity (only one word with xpt so only one underwent it), etc.
Most importantly, it is capable of being (dis)proven based on newly
discovered evidence.
The Laryngeal Theory came about very similarly: no apparent
possibility of certain demonstration, but proposed as a rational
explanation, then later new discoveries validated it. It's an
unlikely chain of events that would never have been bet on beforehand,
but I'd say it was still right to look for an explanation even when
proof might never have materialized.
> > You can object to the particulars of my reconstruction,
> > but not the theory behind it.
>
> I do in fact object to one of your fundamental
> methodological principles: crudely stated, that if two words
> might have a common source, they do, at least if you can
> come up with derivations that you find plausible.
I split up those considered of common origin if they don't fit my
rules as well: *pr- 'far, around' and *pYr- '(up) to, through' are
different, based on such changes as e>o by P, p>f but pY>p, etc., in
dif. languages. It's all based on evidence; if you think it's too
thin then say so, but don't act as if I'm following a ridiculous
principle of my own making.
> As I
> think Piotr once commented, it leads to multiplication of
> phonemes and derivations full of metatheses, assimilations,
> and dissimilations that often seem rather ad hoc.
PIE is already said to have p/t/k and W/0/Y. It is an added
specification that W/Y only are fitted to the k-group. If a
reconstruction of Slavic took place far in the future, why would
someone be obliged to give overwhelming proof for each of many
phonemes ALSO having a pal. variant, with correspondences dismissed
because SOME involved irregular changes?