On Sun, 25 May 2003, Glen Gordon wrote:
> [...]
> The theorized assimilation of **to-s to **so-s before arriving at *so
> (with an unexplained loss of *s to boot) is a long list of devout
> beliefs while valiantly ignoring any pertinent facts like *te-syo where
> assimilation has evidently not occured. There is so much wrong with
> this theory that its credibility will never be salvaged.
Excuse me for barging in, but being the scoundrel behind this heresy I
should perhaps be the one to defend it. It is wrong to demand that
assimilations recur elsewhere, many events of assimilation and
dissimilation being "spontaneous" and not governed by rules. The
assimilation seen in Sanskrit mama 'of me' vis-a-vis Avestan mana and OCS
mene does not force a word like manas 'thought' to follow suit; nor does
the dissimilation-triggered /l/ of German Himmel exclude a form like
Rahmen where the same sequence survived unchanged. These are seemingly
schizophrenic facts of the languages we have to accept.
I admit *to-s > *so-s > *so is merely a desperate attemt to salvage
underlying morphological regularity and at least phonological
"naturalness" (finding phonological regularity unattainable). I do not
admit that two changes constitute "a long list". We often have to show
more stamina than this requires.
Jens