syncretism w/o paradigms

[Martha McGinnis]

>One standard type of argument for underspecification is that
>features are not morphologically signalled on (some class of)
>controllers, but enter into agreement nevertheless.

Agreed.  But this general point doesn't necessarily bear on
impoverishment, if, with Heidi (and Halle & Marantz), we assume that
syncretism also arises through underspecification of Vocab items.
Vocab underspecification would definitely NOT affect what features
are available for agreement.

>There appears to be a markedness generalization, say a filter of the
>Noyer kind: *[person, gender] = gender distinctions restricted to
>the third person. This filter constrains possible vocabulary items
>in these languages, both pronouns and agreement morphemes (no word
>class marks both person and gender). But--on the assumption that
>agreement is copying/matching--the controller must be fully
>specified for features. The subject must be [1 sg f] when Martha is
>speaking, this is matched on the targets, even though no single
>vocabulary item can spell out all of the features.

OK... I see the point.  The absence of gender marking in 1/2 person
is a "metasyncretism" in that it arises in sg/pl pronouns and in
finite verb agreement, as well as cross-linguistically.  So it's
probably not the result of underspecified Vocab items being inserted
into specified syntactic nodes -- it's probably the result of
Impoverishment.  On the other hand, 1/2 pronouns trigger gender
agreement on adjectives and participles.  So it seems features *can*
be Impoverished after triggering agreement (unless the agreement
arises from some other source, as you noted).

>Is this more or less what you were looking for?

Yes, I think it is.  Well, never a dull moment!  Thanks.

So now we return to Heidi's question: is there any reason to maintain
BOTH Vocab underspecification (i.e. Vocab items having a subset of
the features of the nodes they're inserted into) AND Impoverishment?
The possibility of triggering agreement might have been a way to
distinguish them, but if it isn't, then how can they be
distinguished?  There is a distinction in principle: if
impoverishment is an operation (or a filter), it has to apply in
specific environments, while underspecified vocabulary items tend to
appear in heterogeneous environments ("elsewhere" environments).
This might rule out an impoverishment analysis in some cases (I need
to think about this more before I attempt any examples).

If Vocab underspecification is possible in principle, minimalist
aesthetics might tempt us to prefer it where possible -- because it
involves minimizing the features of a Vocab item, rather than adding
a potentially highly specific Impoverishment rule.

What do you think?

-Martha
--
mcginnis at ucalgary.ca