From: Torsten
Message: 68171
Date: 2011-11-02
>The difference between your method of explaining those cases with an optional sound law and one of explaining it by eg. dialect mixture is that the latter approach tries to explain the case by actual occurrences in the real world, which makes it more falsifiable, whereas your method precludes that in principle. Thus that approach should be preferred over the one by optional sound laws because of 1) its greater expressivity (it explains more), and 2) its consequent greater falsifiability.
> Attempts to find a "real but highly non-obvious sound law" often
> lead to linguists attempting to fit a law that must be tailored to
> fit every case, with a strange set of env. and env. exceptions,
> making them unlikely. This includes such changes as e-a > a-a in
> Celtic, cases of KY>K not >C and the reverse in Armenian, etc.
>