Germanic without the Schleifton

A few remarks concerning the development of long-vowelled final
syllables in Germanic.

(1) Inherited long vowels

*-o: < pre-Gmc. *-a:, *-o:
Goth. -a, NWGmc. *-u

*-o:n < pre-Gmc. *-a:m (as in acc.sg. of strong f.), or as in weak f.
Goth. -o:, ON -a, OHG/OS -a, OE -e (< *-æ < *-a).

The effect of the nasal consists in the preservation of length in Gothic
and the lowering of the vowel (raised to *-u when non-nasalised) in
NWGmc. We can posit

PGmc. *-o:n > *-o:~ > Goth. -o: (denasalised), NWGmc. *-a~ (> -a)


(2) Contractions

(2.1) The gen.pl. of a-stems (PIE o-stems)

Leaving aside the special case of Goth. -e:, we have *-o-om > NWGmc.
*-o: > ON/OE -a, OHG/OS -o

This is different from the development of both *-o: and *-o:~. The vowel
was originally long in Proto-NWGmc., which suggests a late contraction
(later than the raising and shortening of inherited *-o:). If instead of
reconstructing an accentual contrast we suppose that PGmc. had
uncontracted (disyllabic) *-a.a < *-a.an (the final nasal simply lost
after a short vowel) < *-o-om, we get the following changes:

(a) merger of *-o: and *-a.a in Gothic (both shortened in auslaut,
yielding /-a/);
(b) contraction of *-a.a > *-o: in NWGmc.

(2.2) The gen.sg. & nom.pl. of o:-stems (PIE *-ah2-Vs)

The PGmc. form was probably *-o:.az (or less likely *-a.az), contracted
into /-o:s/ in Gothic (with length preserved because of the final
consonant) and into NWGmc. *-o:z > ON -ar, OHG -a:, OE/OS -a

(2.3) The nom.sg. of weak masculines (PIE *-o:n)

Goth. -a, ON -e (> -i), OHG/OS -o, OE -a

Here the ending merges with the reflex of PGmc. *-o: in Gothic. However,
the NWGmc forms can't be derived from *-o:. Leaving aside ON -e (which
may reflect the rarer variant *-e:n), the WGmc. reflexes are identical
with those of pre-Gmc. *-o-om (see 2.1).

It's hard to see why the ending should have become disyllabic in PGmc.;
that of the weak feminines and neuters was simply *-o:~ < *-o:n. The
weak feminines in *-o:n- show *-o:- across the board, presumably on the
analogy of the strong feminine *-o:- < *-ah2-. Perhaps PIE nom.sg. *-o:n
was reanalysed as *-o-on (acc. *-o-n-m., gen. *-é-n-os, dat./loc.
*-é-n-i, etc.) already in pre-Gmc., establishing the pattern "short
vowel + nasal extension + inflectional ending" (cf. new feminines in
*-a:-n- and *-i:-n-, with a stable _long_ vowel). If this scenario is
roughly correct, then

PIE *-o:n > pre-Gmc. *-o.on > PGmc. *-a.an > *-a.a

with further development as in (2.1).

Piotr