Carson Schutze: Noun Compounding Question (reply to Rolf Noyer)

[Martha McGinnis]

I always figured Rolf's suggestion for the English compounds was the right
one, but I also had the impression that this approach leads to some possibly
odd conclusions if applied to the full range of German cases.
(I don't have Lieber handy, so maybe she addressed this . . . )

According to the summary in Clahsen & friends (Cognition, 1992), the fact is
that any German plural suffix other than -s can occur inside a compound. (You
can also get 'linking' elements, including -s (on stems that *don't* take -s
plurals), and also sometimes -n.) However, we clearly can't treat all the
"extra" sounds as independent linking sounds, because they are conditioned by
stem choice in exactly the same pattern as are genuine plurals: that is, aside
from non-plural linking -s, the choice among allomorphs -e, -er, -en (and
maybe one can argue also -0) is the same choice you get in genuine plurals.
(You also can get umlaut on the stem if the plural takes umlaut.)

To apply Rolf's analysis to German, if these are really the facts, it seems
one would have to say that none of -e, -er, -en are actually affixal
realizations of [+plural], but rather they are all parts of the stem
allomorph. That seems troubling in two respects: first, a German child would
have to figure out that these things, which look like great candidates for
affixes, really aren't, and presumably these compounds aren't common enough
for them to do that; second, we would seem to lose the subregularities that
exist for the choice among 0/e/er/en, if we list all of them as part of all
the stems they attach to.

Thoughts?