🚀 go-pugleaf

In-Reply-To: <1993Jan9.005144.10836@midway.uchicago.edu> deb5@ellis.uchicago.edu (Daniel von Brighoff)

>         As a linguistics undergrad at the University of Chicago, I learned
> that the principle balancing out "majority rules" is "phonetic plausibility."  
                        .
                        .
>         Now, the downside:  if this argument seems circular, it's because
> (as faras this example goes), it is.

The phonetic plausibility argument is circular, but there are two
other sorts of plausibility argument which are not phonetic and are
therefore independent support for a hypothesis.
Firstly, a statistical argument. In fact the IE p > Gmc f is on one
of a whole series of differences between Gmc consonants and IE one.
It seems statistically unlikely that all the other consonant system
would change in more or less the same way leaving Gmc with the
original system. (Unless it could be shown that Gmc had broken away
and had missed out on innovations in all the other dialects.
Secondly, there's the question of the geographical distribution of
innovation. Say you have a range of dialects like this
(geographically)

                    X X X X X X A X X X
                     X X X X X X X X X X
                    X B X X X X X X X X
                     X X X X X X X C X X

What are the chances that a feature common to A,B & C but not found in
all the X dialects is an innovation? Well, it's possible, but clearly
much less likely than that they have simply not taken on an
innovation in the X dialects. In IE, the k is found in all the
Western dialects *and* the single most easterly, Tocharian. This
actually supports the k>s hypothesis without relying on majority vote
or phonetic plausibility and thus improves the credibility of the
whole hypothesis.

Peter

----------------------------------------------------------------------------
Peter Christian                         
  Dept of European Languages              peter@gold.ac.uk
  Goldsmiths' College, London.            petex@cix.compulink.co.uk
----------------------------------------------------------------------------