NOMINALIZATION AND AGENT CAUSATION: VENDLER/CHISHOLM
1. This metal is light
and
2. This light metal
do not differ in "meaning" but
rather in some other feature or features.
Such features have remained controversial
for some time. I am not going to examine
the details of Meinong's doctrine of nominalization
(nor of meaning), because what I'm interested
in is the employment of nominalization in
characterizing causal sentences.
Russell made use of nominalization in portraying
the difference between 'implies' and 'therefore'
when he tried to capture the difference between
"formal implication" and mere "implication,"
at one point saying:
In grammar, the distinction is that between
a verb and a verbal noun, between, say, "A
is greater than B" and "A's being
greater than B." In the first of these,
a proposition is actually asserted, whereas
in the second it is merely considered. (Principles
of Mathematics. Norton. p. 35. Discussion
of nominalization would, after Russell, be
neglected for the most part until Cocchiarella
and more recently Landini took up the concept
in examining fundamental issues in Russell's
philosophy of logic. (e. g., N. Cocchiarella
"The Development of the Theory of Types
and the Notion and the Logical Subject in
Russell's Early Philosophy." _Synthese_.
45. 1980. pp. 75-115. and Gregory Landini:
"On Denoting" Against Denoting."
in _Russell_. Summer 1998. pp. 43-80). In
1970, Chomsky wrote a very influential paper
on nominalization which I have discussed
in connection with Russell elsewhere (this
can be accessed under the title "Chomsky
and Russell on Nominalization" at
http://www. channel1. com/users/srbayne/histanalytic2.
htm).
What we find in the above quotation from
Russell is the idea that if we nominalize
a sentence in the manner in which he describes,
there is a feature, assertion, that gets
"cancelled" in the process. Another
way of looking at it is to say that nominalization
takes away from a sentence any implication
of truth or falsity, which for Russell were
logical notions, leaving us with conceptual
content, something which for Russell is mainly
psychological. Now what I am going to propose
is this:
Just as nominalization cancels the assertion
property of a sentence, nominalization cancels
the "event feature" of a sentence
- should that sentence express an action.
"But what," the reader may ask,"is
this going to accomplish?" The answer
is this: once it is understood that the event
feature associated with the verb, and not
just its assertion feature, is cancelled,
there will be implications for any assessment
based on the semantics of natural language
of the doctrine that the agent who acts is
a
*cause* of his or her actions. Two philosophers
figure prominently in the debate over the
reality of "agent causation": Rodrick
Chisholm and Zeno Vendler. How, then, might
nominalization figure in causative constructions?
Consider a schema for representing the form
of such sentences
3. e1 caused e2.
Here "e1" represents some event;
however, there are clearly locutions such
as
4. NP caused e2
where the causal antecedent appears to be
the referent of a noun phrase ("NP")
which need not be an event. Supposing we
substitute the name of a person who is an
agent for "NP":
5. Tom caused e2.
In a very interesting essay by Zeno Vendler,
one which will figure prominently in our
discussion, Vendler contends that in all
such cases there is available to us a construction
such as
6. Tom caused e2 by Xing.
Suppose we restrict ourselves to "basic
actions," such as arm raisings, fist
clenchings, wrist turnings etc. which require
no further action in order to be performed.
The relevant question is: "When I raise
my arm do I cause my arm to go up?"
Here is what Vendler says
A. Hence, since the sentence 'He caused his
arm to rise by... is incompletable, the predicate
cause... is disallowed, and, a fortiori,
the predicate 'is the cause of'... John,
in simply raising his arm, did not cause
the raising his arm, and he is not the cause
of the event. ("Agency and Causation"
Midwest Studies in Philosophy IX. 1984. _Causation
and Causal Theories_. p. 372. Shortly I shall
return to this argument, but first a word
on the effects of nominalization on sentences
of the form (3) expressing a causal relation.
Consider one such sentence.
7. John's kicking Bill caused him to fall
down the stairs.
We have two events asserted to be in a causal
relation, viz., 'John's kicking Bill' and
'his falling down the stairs'. Notice that
both expressions are nominalizations (of
'John kicked Bill' and 'He fell down the
stairs' respectively). But now I want to
raise a question which is central to my position:
Supposing that John's kicking Bill is an
event, is 'John's causing Bill to fall down
the stairs' descriptive of yet another event?
That is, one which is identifiable with neither
e1 nor e2 in our fundamental schema, (3),
which is exemplified by (7)? I want to say
that the answer is no, but other philosophers
appear to want to say just the opposite.
Here is what Chisholm says,
B. Suppose, then, that on a certain occasion
a man does cause an event p to happen. What
now of that event - that is his thus causing
p to happen? (Person and Object.
1976 Open Court. Careful examination of the
claim being made reveals that not only are
we, if Vendler is right, dealing with a sentence
ultimately of the form
8. A man caused an event p to happen by doing
Z
we are committed to an event not identical
to Z or p which is the *causing* of p by
Zing. So we have three events, even though
(3) makes mention of only two, e1, and e2.
Clearly something has gone wrong. If I am
right, what has gone wrong is that a third
event has been slipped in surrepticiously.
What I deny is that whenever we have a sentence
of the form 'e1 caused e2' there is another
of the form 'e1's causing e2' which is an
event, call it e3, in addition to e1 or e2.
If this is the case, then Chisholm's doctrine
of agent causation is in jeopardy. But are
we prepared to hand Vendler a so easily won
victory? Perhaps not. Vendler believes he
is arguing that basic actions are not caused
by an agent. He believes he has shown this
by arguing that a statement of the causal
relation fails in the criterion he has established
for such relations, (A). There is some reason
for doubt. What Vendler claims of basic actions,
alone, one can claim of any event which stands
in the relation of immediate precedence to
another event which it causes. But are there
causal episodes where the cause is immediately
precedent to its effect? My inclination is
to think not. Whenever we have two events,
e1 and e2, such that one can be said to be
the cause of the other it would appear that
there is always some event, e3, between them
of which it can be said that it is by e3-ing
that e1 cause e2. If this is to say that
there are no causally initiating events in
nature - an idea going back to Aristotle
- then we do have a distinction between basic
actions and the events of nature. If so Chisholm
is wrong. So we conclude this part of our
examination with a question: "Are the
events of nature continuous with one another
in the mathematical sense? Dedekind, and
Russell's treatment of causation, come together
here in ways I cannot take the time to specify.
But now a concluding word on nominalization.
(5), above, is merely an instantiation of
(4). But suppose that instead of substituting
the name of a purported agent we substitute
some nominalization of a causal sentence.
If we were to admit some event e3 in addition
to e1 and e2, that is, if we were to say
that e3 was the event of e1's causing e2,
would we now be faced with asserting additional
causes of e3? Are we committed to some sentence
9. e1's causing e2 caused e4
where e4 is caused by neither e1 nor e2?
We are clearly faced with a multiplication
of events offending our respect for economy.
My proposal that there is no such event as
e3 carries with it the idea that when I nominalize
a cause sentence of the form (3) not only
is the assertive feature cancelled but the
event feature is cancelled as well. So on
my view a sentence such as 'e1's causing
e2' is neither the assertion of a causal
relation nor a term referring to some event.
Whether it "refers to" or in some
sense "signifies" a proposition
is another matter. The claim is merely that
cancellation of the assertion component concommitantly
cancels the event feature. This may have
implications for very interesting work done
by Davidson, James Higginbotham, Parsons,
and others (perhaps the finest in depth treatment
that takes standard linguistic theory into
account is "Events, States and Times"
in Richard Larson's and Gabriel Segal's Knowledge
and Meaning: An Introduction to Semantic
Theory. We may want to explore event structure
as part of the inflectional or agreement
structure with a more attentive eye on philosophical
problems.
|