Chain shifts & transformed u's, was: Blandness

On Sun, 8 Apr 2001 00:18:33 -0400, Oskar Gudlaugsson <hr_oskar@...>
wrote:

<...>
>But what I did find interesting in this f-r/b-unr discussion, an important
>observation, is what someone pointed out: we never have only rounded front
>sounds, or only unrounded back sounds; they're always a subset of the
>unmarked vowel axis (f-unr or b-r). That makes good sense, and is relevant.
>I wonder if we can make any generalizing statements about how those marked
>sounds generally come about in vowel systems; I see them developing, in
>many cases, through some phonological or morphological mechanisms, such as
>Germanic umlaut and Altaic vowel harmony. Another common origin is probably
>in chain shifts; I suppose French /y/ (from Latin /u/) came about in some
>chain shift, where an older /o/ shifted to /u/, pushing the old /u/ onto
>the front axis; any thoughts on this?
Yep. Chain shifts, of course. This is how I see sound evolution in general,
especially while modelling it ;)

But there's an interesting thing with /u/: it seems that it tends to get
fronted more often than /o/. Also, it seems to me that it allows more
different variants of transformation (unrounding etc.) while /o/ mainly
gets more open / more closed. (I'm making abstraction of diphthongization
and other things that tend to be conditioned by quantity).

Why, I wonder? Cf. the u's of most European langs: they are so often
a bit fronted compared to o's, and historic/orthographic u's have so
many different reflexes/readings (cf. English, French, Welsh, Icelandic,
Swedish/Norwegian, Greek...).

OTOH 'spontaneous' fronting of /o/ seems rather counter-intuitive to me.
In a natlang I'd suppose some intermediary diphthong.

What do you think?

Basilius