Does Saussure's Law Apply Synchronically to Lithuanian?

--- Was: Lith. Acc.pl.
Richard:
My understanding is that acute and circumflex are only distinguished
in stressed long vowels and diphthongs.

Miguel:
Certainly not only in _stressed_ position, otherwise de Saussure's
law (which states that an acute syllable attracts the stress from a
preceding circumflex or short syllable) would be an absurdity.

Richard NEW:
Am I to take it from this that syllables with a diphthong (including,
e.g. /am/) or a long vowel are _phonetically_ acute (falling) or
circumflex (rising), rather than merely _underlyingly_ acute or
circumflex?

Richard:
I have notes that say that Saussure's law applies synchronically. Is
this wrong? I certainly can't reconcile it with the dative dual and
plural endings (o-stems -ám, áms and a:-stems óm, óms) where rãtas
forms rãtam, rãtams, not *ratám, *ratáms, and similarly rankà forms
ran~kom, ran~koms.

Sergei:
I'm not sure about dual (dual forms has disappeared from Lithuanian
toward the middle of this century except some pronouns, so didn't
bother to find out :), but I'll do that in some hours if nobody
outruns me), but -áms and -óms are historical -ãmus and -õmus (the
forms are registered in Old Lithuanian texts), with normal
contractional metatony.

I'm sorry, I missed the point of your question. I must admit I'm not
sure I completely understand what the statement "Saussure's law
applies synchronically" would mean, since even in most cases when it
really applies the morphemes triggering retraction have _historical_
acute only, and synchronically have a syllable with a short nucleus,
not bearing any pitch accent at all.

Richard NEW:
What I had read long ago was that there was a general rule that short
vowels were underlyingly circumflex (LH), and that long vowels and
diphthongs were underlyingly acute (H). Exceptions and basic stress
were supplied by the lexicon.

This was then modified in the following steps (I quote from memory):

1. Surplus stresses are removed.

2. Words that need a stress and lack one are supplied with one.

3. Saussure's law is applied.

4. Final acutes are shortened.

5. Non-final, stressed, a and e (generally) lengthen.

6. Single mora circumflex became single mora acute.

This process does imply that unstressed syllables contrast acute and
circumflex. However, the formulation also implied that syllables
that lose the accent by Saussure's Law differ from those that never
had it. The only example of the latter I can think of is the
nominative and instrumental singular of a-stems with mobile stress
and circumflex accent on the root. Do they actually contrast?

These rules clearly need to be supplemented with

7. Late contractions, such as -ãmus > áms.

From Sergei's response, Steps 3, 4 and 7 seem to be too complicated
to be the synchronic process. It appears that they are now replaced
(at least for him) by:

3'. Certain morphemes retract the accent from a preceding circumflex
syllable.

I would then say that 'Saussure's law applies synchronically' if
Rules 1 to 7 were a more accurate description of Lithuanian than
Rules 1, 2, 3', 5 and 6.

Richard.