With the appropriate semantic analysis, then, it seems that no extraor-
dinary appeal to pragmatics is necessary beyond what we already assume:
namely that the assignment function is set by the context, not the semantics,
but is used to determine the semantic value of an expression in a context.
Having seen that there is at least one interpretation of ‘slot-filling’ which
does seem to satisfy the requirements for reaching a proposition without the
representational-pragmatic view Stainton proposes, we have slain most of the
dragon, I think. Nevertheless, there still remains a small minority of left-over
cases, which appear somewhat heterogeneous. The question then becomes
whether, on the basis of at least one of these examples, Stainton’s conclusion
can be established. It is to these, then, that I now turn at last.
5.
Scripts, contexts, and syntactic ‘slots’
Every account needs to say something special about the following kind of
data (from Merchant 2004a, and parallel to the Three pints of lager case in
English above). In many languages, objects of certain verbs are marked with
particular cases: in Greek, the object of most transitive verbs is accusative,
and in Russian, the object of certain transitive verbs in certain circumstances
appears in the genitive (in its ‘partitive’ use). This is in particular the case in
ordering food and drinks. In Greece or Russia, to order a coffee or water, one
could say the following:
(65)
a. Ferte
bring.
IMP
mu
me
(enan)
a
kafe
coffee.
ACC
(parakalo)!
please
(Greek)
‘Bring me (a) coffee (please)!’
b. Dajte
give.
IMP
mne
me
vody
water.
GEN
(požalujsta)!
please
(Russian)
where it’s synonymous with (ii) (compare the well-formed (i), and the nonelliptical (ii)):
(i)
Some met Susan, and others did Jessica.
(ii)
Some met a man who knows Susan, and others met a man who knows Jessica.
(iii)
*Some met a man who knows Susan, and others did Jessica.
For
Culicover and Jackendoff, there’s no particular reason why the meaning of the missing VP in
the pseudogapped second clause of (iii) can’t be filled in to mean, on their rule for interpreting
f
, as meet a man who knows. This absence follows from a structural theory of pseudogapping
and VP-ellipsis (see Merchant 2008a for references) on the assumption that Jessica has to
move to a VP-external position, and in (iii), such movement would violate an island. See also
Lasnik 2007 for further arguments against Culicover and Jackendoff 2005.
‘Give me (some) water (please)!’
In exactly the same circumstances, a Greek or Russian speaker could just
as well use the following:
(66)
a. (Enan)
a
kafe
coffee.
ACC
(parakalo)!
please
(Greek)
‘(A) coffee (please)!’
b. Vody
water.
GEN
(požalujsta)!
please
(Russian)
‘(Some) water (please)!’
The very real question in these cases is, where does the case come from?
What determines that the speaker uses the accusative here in German, and
the genitive in Russian? Whatever the vocabulary of Mentalese might look
like, if it is to be plausible as a candidate for the language of thought inde-
pendent of language, it is very unlikely to have anything like the partitive
genitive in it. Instead, a plausible manifest property may be something like
λ x[I WANT x], as Stainton suggests on p. 157. Such properties in Mentalese
do not contain the actual Russian verb xoˇcu ‘I want’ or the like, however. But
it is only such particular verbs that assign (or govern, or determine) this case
in this context—without the actual verb, there’s no obvious reason the speaker
should choose the case she does, especially over the nominative, which is the
case used in all the above instances of nonsentential assertion in these lan-
guages. Therefore, there is no way on Stainton’s analysis to account for the
accusative and partitive genitive in (66). What makes these contexts special
is their formulaic, conventional character, in which particular linguistic ele-
ments are made manifest and license ellipsis. Exactly because they are limited
in number and kind, learned explicitly (a competent foreign learner of Greek
for example, who had never been in a Greek restaurant, may mistakenly use
the nominative here), and seem to reflect syntactic properties of particular
lexical items in the languages, it seems most likely that this is an instance
of syntactic ‘slot-filling’, where the ‘slot’ here is the item being ordered, and
there is a context-dependent linguistic construction that employs the relevant
verb with its case-assigning properties. Whether this verb has been uttered
(though unpronounced) by a speaker is a different question—this is the ques-
tion of whether such syntactic slot-filling is to be analyzed as ellipsis
syntactic
.
Two other such cases can also be mentioned briefly. The first comes from
Dutch: in Dutch, when you answer a ringing phone, you pick it up and say
‘Met’ (lit. ‘with’) followed by your name. So I answer phones in Holland by
saying, ‘Met Jason’. This is short for ‘U spreekt met Jason’ (‘You are.speaking
with Jason’). The preposition met is simply part of the conventionalized means
of answering a phone, but the whole phrase contains a variable (over names)
and is short for something else (which a pedantic or otherwise garrulous per-
son is free to use as well). There’s no way to predict this behavior or inter-
pretation on general principles, linguistic or otherwise. The second such case
comes from Greek, where the names of addressees on envelopes appear in
the accusative, not nominative, case; a letter to my father-in-law is labeled as
in (67a), not (67b):
(67)
a. Dimitri Giannakidi (accusative)
b. #Dimitris Giannakidis (nominative)
This is not because the accusative carries some inherent directional mean-
ing (recipients and the like are marked with the genitive, in fact), but because
there is a conventional ellipsis of the preposition pros ‘to’, which assigns the
accusative. Again, this preposition could be written on the envelope as well,
but need not be.
A final such example is the taxi example, which Stainton gives in (68a)
13
,
and which is similar in its properties to the exophoric sluice in (68b) discussed
by Ginzburg 1992 and Chung, Ladusaw, and McCloskey 1995.
(68)
a. Marco gets into a taxi and says, ‘To Segovia. To the jail.’
b. A passenger gets into a cab and the driver turns and asks, ‘Where
to?’
The main point I wish to make with respect to such examples is that their
form, again, is determined entirely by linguistic elements in what Schank and
Abelson 1977 called the ‘script’ of the situation. In following a script, the
participants know and can anticipate the actions (including the utterances) of
the others following the same script, and can plan accordingly (the notion of
script was developed for artificial intelligence purposes, but is familar from
automated phone booking systems etc.). In such a context, certain particular
linguistic phrases can be expected: they are ‘given’, though not by the imme-
diate actually spoken linguistic precedents, but rather by mutual knowledge
13
Stainton gives this example with the subject ‘Benigno’, but this is clearly a typo for
‘Marco’, since Benigno was in the jail and it was his friend the Argentinian journalist Marco
who was going to visit him, as fans of Almodóvar will recall.
Dostları ilə paylaş: |