Sergei has sent this to me off list, but I assume it was actually meant as a public response, so I take the liberty to forward this very interesting description to phoNet.

Piotr

----- Original Message -----
From: "tarasovass" <S.Tarasovas@...>
To: "Piotr Gasiorowski" <gpiotr@...>
Sent: Friday, May 31, 2002 12:17 AM
Subject: Re: Pitch Accent


--- In phoNet@y..., "Piotr Gasiorowski" <gpiotr@i...> wrote:

> > Are there any known instances of qualitative changes in vowel
inventory, in which pitch accent is a decisive factor?
>
> > Such as
> > [á] > [ó] but [â] > [â]
>
> As far as I know, pitch-accent contrasts are accompanied by clear
qualitative differences in Modern Lithuanian (if Sergei Tarasovas is
listening to us, he can give you some first-hand information about
that).

With pleasure, but I'll have to limit myself to Standard Lithuanian,
as within the limits of a posting it's hardly possible to provide a
description which would involve the dialects.

In Standard Lithuanian (or, rather, those South-West Auks^taitian
dialects that underly it, because, actually, those speakers of
today's Standard Lithuanian who haven't yet neutralized [or
transformed to non-prosodical form] the pitch-accent opposition are
oriented to their native dialects as to the pitch accents) the
picture is as follows:

-- in the syllables where (phonologically long = phonetically tense)
monophthongs (i.e., /e.:/,/i:/,/u:/,/o:/,/a:/ and /e:/) serve as a
nucleus, the acute and circumflex pitch accents are actually not
accompanied by any substantial qualitative differences: acuted vowels
(prosodicaly signalled mostly by an abrupt change of the F0 and/or
loudness -- not necessarily by the classical 'sharply up -- smoothly
down' scheme, by the way) are more tense and sharply articulated (and
thus marked), circumflexed (prosodicaly signalled mostly by a stable
or slightly rasing F0/loudness) are less tense and (more laxed, thus
unmarked), but one definitely has to have a good ear to catch the
difference clearly.

-- in the falling diphtongs- and diphthongoids-based syllables
(/ei/,/ai/,/ui/,/au/,/el/,/al/,/il/,/ul/,/er/,/ar/,/ir/,/ur/,/em/,/am/
,/im/,/um/,/en/,/an/,/in/,/un/), by constrast, the pitch accents are
accompanied by drastic qualitative differences: in acuted diphthong
(oid)s, the nucleus (which bears prosodical prominence) is more
tense, articulated rather sharply and retains it's "default" quality,
while in circumflexed ones it is never tensed and often loses it's
"default" quality, the glide being more tense and sharply articulated.
In diphthongs, the nucleus is assimilated to the glide and a bit
reduced; thus:
acuted /e'i/ [æ.i] vs. circumflexed /ei~/ [ei.] (here [e] is just
that -- IPA's close-mid -- or just 'mid' -- front vowel, [æ.],[i.]
for semi-long (tense) [æ],[i]), /a'i/ [a.i] vs. /ai~/ [&i.] (here and
below [&] stays for various flavors of schwa), /u`i/ [U.i] vs. /ui~/
[Ui.] (here [U.] is a semi-long IPA's [U] slightly shifted to [u] but
still not tense, and [U] is an IPA's [U] shifted to [&]), /a'u/ [a.u]
vs. /au~/ [åu.] (here [å] for IPA's open-mid back slightly labialized
vowel).
In diphthongoids, it's mostly reduced:
/e'l/ [æ.l] vs. /el~/ [el.], /a'l/ vs. /al~/ [&l], /i`l/ [I.l] vs.
/il~/ [Il.] (here [I.] is a semi-long IPA's [I] slightly shifted to
[i] but still not tense, and [I] is an IPA's [I] shifted to [&]),
/u`l/ [U.l] vs. /ul~/ [Ul.], the same with /r/, /m/, and /n/-based
diphthongoids.
Such a sharp qualitative contrast makes merely prosodical distiction
a bit superfluous, and many Lithuanian dialects, as well as idiolects
of Standard Lithuanian actually use qualitative (plus, to some
extent, quantitative) features only to oppose historically acuted and
circumflexed diphthongs and diphthongoids.

Sergei