Eric Christopherson skrev:
 > On Mar 27, 2008, at 2:21 PM, Benct Philip Jonsson wrote:
 >> The books I had referenced seem to have vanished from my
 >> library's catalog, but according to several online
 >> sources final /?/ induced rising tone and final /h/ <
 >> */s/ induced falling tone in Old > Middle Chinese.
 >> In Tibetan initial /?/ and vocieless stops induced high
 >> tone and initial /h\/ and voiced stops induced low
 >> tone. Final
 >> d/t and s induced falling tone.
 > That's good to know; I have a conlang where high tone
 > develops after voiceless sounds, and low tone after voiced
 > ones, but I wasn't sure that actually happened in any
 > natlang.
 > I actually want the tone to develop into pitch accent, but
 > I'm not quite sure how to go about that.

There is more: single voiced stops became voiceless
aspirated while voiced stops preceded by /s/ or a stop bcame
or remained unaspirated. Voiced stops preceded by a nasal or
/h\/ remained voiced and have a phonologically irrelevant
'extra low' tone. In original voiceless stops an /s/ or stop
prefix inhibited aspiration while /h\/ or nasal didn't.
'Prefixed' sonorants became high, and unprefixed ones low.
In each case the 'prefix' disappeared, except that voiced
stops, even those fron /h\/+ are prenasalized when
postvocalic. For some reason /zl/ and /ld/ also became
/nd)/, and /db/ became /w<H>/.

So you get four tone patterns:

* high level

* high falling

* low level

* low falling

Falling tone can only be realized on final syllables or
monosyllables. Bisyllables will have hi/lo on the first
syllable and falling/level on the second, with the height of
the second syllable irrelevant. Thus it can be argued that
Tibetan has word accent rather than syllabic tone.

/BP 8^)>
Benct Philip Jonsson -- melroch atte melroch dotte se
   "C'est en vain que nos Josués littéraires crient
   à la langue de s'arrêter; les langues ni le soleil
   ne s'arrêtent plus. Le jour où elles se *fixent*,
   c'est qu'elles meurent."           (Victor Hugo)