Re: [tied] Re: Indo-Iranian Vowel Collapse (was: IIr 2nd Palatalisa

Richard wrote:

> My feeling is that there was something about PIE that made the merger
> of /a/ and /o/ highly likely, but not certain. Different dialects
> therefore show different outcomes and different degrees of merger.
> It reminds me of NW Germanic umlaut, which is by no means uniform
> (not even Old English umlaut is!) and the varying development of
> short stressed Latin /e/ to /ie/ in daughter languages.

Or Celtic lenitions, or the treatment of nasal vowels in Slavic...

> Is there any respectable way of describing such tendencies? Torsten
> would probably suggest class differences (as in his Nordwestblock
> monologues), or high v. low registers, but I'm not sure that they
> would work.

Some initial states of a language system may constitute a favourable (or
"natural") environment for some types of processes, partly
predetermining their later direction and dynamics. For example, a
language with strong expiratory stress on initial syllables is likely to
reduce and drop inflectional and derivational endings. The information
originally contained in the lost syllables is then likely to be
re-encoded as the phonologisation of various assimilatory modifications
of the root syllable (i-umlaut, a-umlaut, u-umlaut). This will result in
parallel but non-identical developments in the daughter languages. The
palatal umlaut of /a/ occurs in all the NW Germanic languages, but note
its almost complementary environments in ON (not in light syllables:
gestr, marr) and OHG (not in heavy syllables: gast, meri), while OE has
it across the board (giest, mere < *gasti-, mari-).

Piotr