I'm left without understanding how de Saussure's law works in the Lithuanian
nominal paradigms. According to my Lith. grammar, it works in the ins.sg. and
the acc.pl. of most declensions. Further in the nom.sg. of a:-stems (but not
the e:-stems!) and in the nom.pl. of adjectival o-stems.
As far as the singular goes, a laryngeal explanation suits the facts best.
Saussure's law affects the instrumental in *-h1, and the a:-stem nom.sg. in
*-h2. It does not affect the oblique forms of the a:-stems, where *h2 occurred
intervocallically (G.sg. -a:s < *-eh2-os) or indeed the a:-stem nom.pl. -os <
*-a:s < *-eh2-es.
In the plural, we have a different picture. The acuteness of the acc.pl. (and
the nom.pl. of adjectival o-stems), apparently reflects the PIE short prosody of
e.g. o-stems nom.pl. *-oy and acc.pl. *-ons, as opposed to long prosody in the
oblique (e.g. G.pl. *-o~m). If this principle had been applied in the singular,
we would expect the nom. and acc. sg to have been subject to Saussure's law as
well. In fact, where mobility shifts the accent to the last syllable in the i-
and u-stems, we see that the accent on at least the nom.sg. [but not the
acc.sg.] is acute (-ìs, -ùs). And yet, an accent class 2 noun like tur~gus, in
violation of Saussure's law, does not change to *turgùs in the nom.sg.
=======================
Miguel Carrasquer Vidal
mcv@...