COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	%searle[s89,jmc]		Notes on Searle's Speech Acts
C00009 ENDMK
C⊗;
%searle[s89,jmc]		Notes on Searle's Speech Acts

5 - S. asks how are assertions to be verified.  AI has an
additional method.  Ask whether the mechanism will actually
work in a program.  However, it should be admitted that little
of what I propose has actually been incorporated in programs.
Therefore, I am appealing to my intuition and that of my readers
about what might work.  An important form of criticism would be:
``I tried that and it didn't work.  Here's what I did.  How
does McCarthy propose to get around this.''  Less work and less
convincing.  ``Here's why it won't work.''  Still less convincing.
``I just don't believe it will work.''  Even less convincing.
It may work and it may not, but that's not what X really means.

8 - The ability to use words like ``analytic'' or ``meaning'' correctly
is not the same as the ability to define them.  Let's see if S.
says that.

Good idea.  Include in understanding the ability to recognize which
cases are borderline.  Maybe his example about whether green things
are necessarily extended can be fodder for discussion of ambiguity
tolerance.


12 - ``My only point at present is that when certain preferred models
of explication fail to account for certain concepts it is the models
which must go, not the concepts.''
Most often, yes.

14 - Converting {\it knowing how} into {\it knowing that}.  This is
a narrow form of knowing how, i.e. the ability to characterize
particular utterances.  Are other forms of knowing how just the
ability to do in each case what one can't necessarily describe
in general.

The mistakes we make and the mistakes I shall make in linguistic
characterizations in the course of this work will be due to such
things as not considering enough exampees or misdescribing the
examples cossdiered, not to mention carelessness, insensitity
and obtuseness; but, to repeat, they wil ntt be due to over-hasty
generalization from insufficient empirical data concerning the
verbal behavior of groups, for there will be no such generalization
nor such data.

OK, if ``obtuseness'' is to be taken to include failure to
solve difficult scientific problems.  Thus, Searle should say
that Lorentz didn't invent the theory of relativity because he
was obtuse.  If he won't say that, then there's another cause
of mistakes.

I have no substantial quarrel with the methodoloy Searle proposes
to adopt in 1.3.  Of course, I propose to supplement it by thinking
about the role the use of language plays in the interaction between
entities with beliefs and desires.  However, my methodology for
that is the same as Searle's; I propose to use my intuition about
what the goals are and what methods are capable of achieving them.

16   The unit of linguistic communication is not, as has
     generally been supposed, the symbol, word or sentence,
     or even the token of the symbol, word or sentence, but
     rather the production or issuance of the symbol or word
     or sentence in the performance of the speech act.

This agrees with my idea that speech acts are to be considered
as elements of interactive procedures, i.e. as executions of
program statements.

17 - a theory of language is part of a theory of action

25 - perlocutionary act - get him to realize, frighten them
	get his master to take him for a walk

28 - Expressions don't refer.  Only speakers refer. - hmm
unnecessary dogma

referring

What does it mean to refer?  ``Fido'' refers to a particular dog
relative to a context-dependent system of reference---call it
$C7$.  ``Fido'' may refer to something quite different, e.g.
Napoleon, in some other context.  $C7$ may be preferable to
this other system for reasons related to the whole complex
of ways in which $C7$ refers.  It may well be that ``Fido''
refers to the same dog in all useful contexts.  Then we
can say that ``Fido'' refers to that dog {\it simpliciter}.

39 - His dog can perform some simple illocutionary acts.

43 - reference to Grice, philos. review, july 1957, Meaning

57 - "axioms" for promising.  Nonmonotonic.

It will be particularly illuminating if Elephant promises are
designed to have only some of Searle's properties.  Maybe we
can better design Elephant so that the promises of Elephant
programs are always sincere.  An elepant is faithful one hundred
percent.