David Fertig: Noun Compounding Question (reply to Carson Schutze)

[Martha McGinnis]

> David Fertig said,
>
>>    It's certainly true that German nouns with -s plurals never take
>>    linking -s, but I think the correct generalization is that nouns with
>> -s plurals never take _ANY_ linking element.
[Carson Schutze replied]
> So, in addition to -s, they also never occur with -n, and is there
> something else you would consider a linking element that's covered by this
> generalization?

    The linking elements listed in traditional grammars include -e, -er,
-(e)n, -(e)ns, and -(e)s. One probably wouldn't want to consider all of
these potential candidates for a noun with an -s plural (because some only
occur on nouns that have the same element as a plural suffix), but that's
not really the point. The point that I was trying to make is that the same
nouns that make use of the default plural formation (-s plural) also have
the default form in compounds (identical to the nominative singular, with
no linking element). I think the Pinker/Clahsen theory predicts this
result. Most of these nouns take the -s plural because as non-canonical
nouns they are not capable of any kind of inflectional irregularity. This
incapacity would presumably apply to the form they take when used as the
first element of compounds as well. (Diachronically, we might predict that
as the -s plural spreads to more normal German nouns, this correlation
might break down.)  If we think of it this way, then Carson Schutze's
question "Why should linking -s care what a noun's plural looks like?"
doesn't really arise. The fact that these nouns take plural -s and the fact
that they don't take linking -s (or any other linking element) are just two
independent consequences of their non-canonical status.

>
>> If this is true, then I think Carson may
>> have answered his own question ("what would prevent the *linking* -s from
>> combining with these nouns?") in his last message when he refered to the
>> Pinker/Clahsen idea that non-canonical nouns (or verbs, etc.) cannot have
>> stem allomorphy (or cannot be "irregular" in Pinker/Clahsen terminology).
>
> I'm not sure I follow my answer to my question :-)
> It seems to presuppose that linking segments would also be analyzed as
> part of a stem allomorph, rather than as a separate chunk of stuff. This
> would be on top of the Noyer/Lieber proposal to treat plural affixes
> (other than -s) as part of stem allomorphs. I see that that could be
> done, but it makes me nervous somehow, and I wonder if it breaks the
> analogy Rolf suggested to Indoeuropean theme vowels. I may be
> misremembering, but I don't recall them being inserted by readjustment
> rules in existing DM analyses. I thought they came in with vocabulary
> insertion--someone correct me! And then someone tell me what according to
> DM would be predicted to be different between the two kinds
> of elements, viz. theme vowels and linking consonants.

    Maybe I got carried away by the parallels between DM and Pinker/Clahsen.
I think Pinker/Clahsen (et al.) would certainly analyze the form with no
linking element as the regular (default) form for the first element of
compounds, and thus all forms with a linking element would be stem
allomorphs. I don't know if DM (or Noyer/Lieber) would do it this way.

    David

David Fertig, Associate Professor of German
Dept. of Modern Languages and Literatures
910 Clemens Hall
University at Buffalo
Buffalo, NY 14260
fertig at acsu.buffalo.edu
Tel: (716)-645-2191 x1202
Fax: (716)-645-5981