bedroom, an extremely neat bookshelf, a special smell, or almost anything
thing else that would make Max salient to the speaker.)
Recall next the minimality condition proposed by Stainton to rein in over-
generation, discussed on p. 15 above. The difficulty with this absolutely nec-
essary minimality condition is that it brings Stainton’s account dangerously
close to mine, making it look almost indistinguishable. If the minimal propo-
sition that can be gotten with a property P is that P is instantiated in some
salient object, then it’s awfully close to saying that we end up with the equiv-
alent to P as in (40).
One possible difference is with respect to individuals. When ‘Barbara Par-
tee’ is uttered as in (14a) above, I claimed that the sentence
syntactic
that was
actually produced was That is Barbara Partee, but with That is elided, as
in (41a). There are good reasons not to think this is a predicative use of the
name, but rather an identity statement (or ‘specificational’, in the common
terminology; see Mikkelsen 2005 for an extensive recent treatment). On my
account, as long as the reference for that can be recovered, ellipsis is possible.
For Stainton’s account, the question is whether this is the minimal proposition
developable. One serious competitor for minimal status would be an existen-
tial predicate applied to the individual: Barbara Partee exists. But this doesn’t
seem to be a good parse of (14a). So in this case, the conventional ellipsis of
That is
does a better job of accounting for our intuitions about this example
than the grasping of the minimal manifest property and application of it to
Barbara Partee.
The major apparent syntactic advantage of this line of analysis is that it
straightforwardly accounts for the presence of the nominative case on nomi-
nal expressions in case-marking languages. In Greek or German equivalents
to (41), we would find only the nominative. If case assignment is effected
by an asymmetric agreement relation between the NP (which by hypothesis
lacks a case value inherently) and some case assigner (for nominative, typi-
cally taken to be the head of the clause itself, T), then T must be present to
assign nominative.
The difficulty with taking this line of reasoning at face value is that nom-
inative
7
case is also the case used for all kinds of labeling, from street signs,
7
While I am concentrating on the nominative as found in typical analytic case system lan-
guages like those in western Europe, I mean all these remarks about the ‘nominative’ to apply
to the least marked case in a given language. In analytic nominative/accusative languages such
as Japanese or Korean, an entirely caseless form is the least marked, and is used for labels. In
ergative/absolutive languages like Basque, it is the absolutive that appears. See Merchant to
traffic signs, book and movie titles, store names, product names, etc. In all
such cases, the nominative must be available extra-sententially (presumably
by virtue of being in such a ‘label’ environment or construction) and there
is therefore little deep reason to insist on T as the only source of nomina-
tive. (Note that I am not claiming that nominative is a ‘default’ case in these
languages: allowing for ‘default’ case assignment would void all Case Fil-
ter violations, and make incorrect predictions in structures with resumptive
pronouns as well; see Merchant 2004b, to appear for discussion.)
In any event, the above approach, whatever its merits, is difficult to extend
to the following examples.
(43)
Quantifiers as arguments of a manifest property
a. An editor of Linguistics and Philosophy
b. Three pints of lager.
c. quite a lot of children.
While the ‘limited ellipsis’ analysis gives a reasonable paraphrase for
(43c), it is less felicitous with (43a,b): an editor of Linguistics and
Philosophy
and three pints of lager. The latter example is a special
subcase I return to in detail in section 5. The problem with the first example
is that there is nothing in the context (besides the chair) that could easily be
construed as providing a referent for a deictic or demonstrative. The intended
assertion, according to Stainton, is something like THIS CHAIR IS FOR [an
editor of Linguistics and Philosophy]. The problem is fully general: if I am
instructing you where to put namecards at a wedding dinner beforehand, I
can look at the seating chart, walk around the table and point at successive
empty chairs, saying ‘Sam’s mom’, ‘The bride’s best friend’, ‘Laura Skotte-
gaard’, ‘Some random guy Susan is bringing as her date’, etc. and thereby
assert that each of the indicated chairs is intended for the named or described
person. Such a content is hardly available to equivalent uses of ‘That’s Sam’s
mom’, ‘That’s the bride’s best friend’, etc. Instead, I would appeal here to
the labeling function: there is some construction in which a situational deixis
can be linked to a nominative DP ‘label’, where the label can be the name
of the object so labeled or stand in some pragmatic relation to it (as in ‘I-
94 Minneapolis/St. Paul’ on an interstate highway sign—this sign appearing
appear for discussion of split languages: I expect that in split languages, the absolutive will be
used in the labeling function as well (as it is in Hindi/Urdu, for example, and Georgian).