|
part of the meaning of these sentences in our semantics. Nouns that are understood
as parts in this way are generally relational (cf. English the steering wheel of the
car, the West Wing of Meston, the goosh-injecting tyroid of the space rocket ; see
Barker 1995, Barker and Dowty 1993).
(253)
Jsteering wheelK = λy.λx.λs.steering-wheel(x)(s) & part-of(x)(y)(s)
I encode the ‘part-hood’ of a part-denoting noun like steering wheel as a sepa-
rate property (‘part-of(x)(y)(s)’, to be read as ‘x is a part of y in s’), rather than
writing steering-wheel(x)(y)(s)), to make this aspect of the meaning of such nouns
217
completely explicit in the notation. I leave it open here whether the part-relation
between individuals ‘part-of’ is to be identified with the relation ≤ in our situation
semantics. I will assume, however, that the former entails the latter, i.e., that the
following holds:
(254)
∀x∀y∀s [part-of(x)(y)(s) → ∀s
0
[s
0
≤ s & y ≤ s
0
→ x ≤ s
0
]]
Determining what it takes to be a part of something else in the relevant sense is a
difficult (and ultimately philosophical) question, and I will not attempt to spell this
out in any detail here. One relevant and intuitively plausible aspect of part-hood is
that it is not necessarily a permanent property of an object to be part of something
else (though this may be the case for certain parts that are in some sense essential).
24
I will appeal to our intuitive understanding of whether the discourse participants in
a given scenario would see (or be able to see) something as part of something else
or not. What matters for our purposes is whether it is common ground whether
an individual is part of another one. For example, in Hawkins’ example above, the
second scenario, where the goosh-injecting tyroid is found in a tool box as an object
of its own, provides no clear evidence for the hearer to see it as a part of something
else, even if it used to be part of a space rocket. But in the first scenario, it is clearly
a physical part of a larger machine and the hearer has no difficulty in seeing it as
such in the given situation.
5.4.2
Type-Shifters for Relational Nouns
One general question that any compositional semantic account of relational nouns
has to address is what happens to the relatum argument when the noun combines
with another expression, e.g., a determiner (or an adjective, via Predicate Modifica-
tion, for that matter), that is normally assumed to combine with a property, rather
24
But this is a complicated matter. Bach (1986) considers the sentence We found part of a Roman
aqueduct, which does not require the aqueduct to still exist (nor to have ever existed!).
218
than a relation. One common tool for dealing with such problems is to introduce a
type-shifter (in the spirit of Partee (1986)), which reduces the arity of the noun by
existentially quantifying over its first argument. In a situation semantics, there are
several ways of formulating such type-shifters.
(255)
Some Type-shifters for relational nouns in a situation semantics
25
a.
JΠK = λR.λx.λs.∃y. [R(y)(x)(s)] & x ≤ s & y ≤ s]
b.
JΠ
w
K = λR.λx.λs.∃y [R(y)(x)(w
s
) & x ≤ s]
Π in (255a), for example, requires the relation to hold in s and both of the argu-
ments of the relation to be part of s; Π
w
in (255b), on the other, requires the relation
to hold in the world of s and only the ‘external’ argument to be part of s. For cases
of part-whole bridging, the first option yields the desired result:
(256)
JΠ(steering wheel)K =
λx.λs.∃y. [steering-wheel(x)(s) & part-of(x)(y)(s)& y ≤ s]
The full weak-article definite then is analyzed as follows:
(257)
J[[the
weak
s]Π(steering wheel)]
K
g
=
ιx.∃y. [steering-wheel(x)(s) & part-of(x)(y)(s)& y ≤ s]
This will pick out the unique individual that is a steering wheel in s and which
is part of something else in s, where s is the situation introduced by the situation
pronoun on the weak article. This interpretation is exactly what we want for cases
of part-whole bridging, as the whole is part of the relevant situation in them. (A use
for the type-shifter in (255b) will be seen towards the end of the chapter.)
The other examples also are nicely captured by having the type-shifter Π present.
A DP like the West Wing is standardly understood to denote a part of a building.
25
As Chris Potts (p.c.) has pointed out to me, we may ultimately want to distinguish the status
of the part-condition(s) from the status of R.
219
What is causing difficulty in the scenarios cited above is that the obvious choice of
building for the hearer does not render a felicitous interpretation, as the building
that the dialog is taking place in does not have a West Wing. But without any prior
context providing evidence to the contrary, the location that the discourse takes place
in seems to provide what is at least a strong default for the choice of the situation
relative to which the definite is interpreted. If the other building (Taylor) that does
have a West Wing had been explicitly mentioned before, or if the QUD had made it
Do'stlaringiz bilan baham: |
