Jan Dejnozka Discusses Grice

I HAVE BEEN VERY FORTUNATE to have Jan Deknozka discuss some fine points of Grice on meaning -- elsewhere.

Here is, for example, Grice's "fifth" (and kind of final) re-definition as far as
his essay "Logic & Conversation: Utterer's Meaning & Intentions" goes (in
Studies in the Way of Words, p.114).

To consider some of the observations by Dejnozka, elsewhere, I shall make use of what Grice has a the "ranges of variables":

Range of Variables
Asterisk-
sub-psi: mode (indicative/imperative)
associated with psychological attitude
Psi (believing/desiring)
E: inference-element
f: features of the utterance x
c: modes of correlation
(such as iconic, associative, conventional)
U: Uttere
Phi, Phi': properties of persons (possible addressees)
Psi: believing/desiring.
Psi+: Psi if clause (II) is operative, and
"think that U Psi-s" if clause (II) is
non-operative.

(GM) "U meant by uttering x that (asterisk sub-psi) p iff
(EPhi)(Ef)(Ec):

I. U uttered x intending x *to be such that* anyone
who has Phi would think that
(1) x has f.
(2) f is correlated in way c with Psi-ing that p.
(3) (EPhi'): U intends x to be such that anyone who
has Phi' would think, via believing (1) and (2), that
U Psi-s that p.
(4) in view of (3), U Psi-s that p;
and
II.(operative only for certain substituends for "asterisk-sub-psi")
U uttered x intending that, *should there actually be*
anyone who has Phi, he would via thinking (4), himself Psi that p;
and
III.It is not the case that, for some inference element E,
U intends x to be such that anyone who has Phi will both
(1') rely on E in coming to Psi+ that p.
(2') think that (EPhi'): U intends x to be such that anyone
who has Phi' will come to Psi+ that p *without* relying
on E.

I hope the symbols are clear enough. They can be rephrased in terms of ordinary English such as "believing", "desiring", etc. Grice was too fond of generalities!

Jan:

"[It would also help if you] explain "PROTREPTIC vs.
EXHIBITIVE MEANING" in ordinary language. [...] Isn't
PROTREPTIC vs. EXHIBITIVE MEANING another Gricean
"ambiguity" of different "kinds" of meaning?"

You are right, it WOULD. Fortunately, it was my sloppy language. Grice
actually claims it's "utterances" which are exhibitive (all of them). Some
happen to be "protreptic" as well. The idea being that by an exhibitive
utterance you just exhibit the psychological attitude you intend to mean
(the belief that the cat is on the mat, the desire that the cat is on the
mat). On the other hand, by a "protreptic" utterance, you also intend your
addresse to adopt that psychological attitude (This he discusses on pp.111ff).

"Would this be a correct paraphrase of the analysis
we've been discussing?

I'm rephrasing it a bit. My liberties and apologies. I'm also correcting
your version as per Grice's definition above.

(GM) For any utterer U, any utterance x, and
any propositional complex p,
by uttering x U means that p iff
I. There exists a property Phi, there exists a
feature f, there exists a correlation c, such that,
by uttering x U utters x explicitly or implicitly
intending x *to be such that* if anyone who has
property Phi would think that
(1) x has f.
(2) f is correlated in way c with Psi-ing that p
(3) (EPhi'): U explicitly or implicitly intends x to
be such that anyone who has Phi' would think, via
believing (1) & (2), that U Psi-s that p.
(4) in view of (3) U Psi-s that p;
and
II. (operative only for certain substituends for "asterisk-
sub-psi) By uttering x U explicitly/implicitly intends
that *should there actually be* anyone who has Phi, he
would via thinking (4) himself Psi that p.
III. It is not the case that, for some inference element E, U
explicitly/implicitly intends x to be such that anyone
who has Phi will both
(1') rely on E in coming to Psi+ that p
(2') think that (EPhi'): U explicitly or implicitly
intends x to be such that anyone who has Phi' will
come to Psi+ that p *without* relying on E.

I guess there are minor, unessential, variants of Grice's fifth
redefinition

"I count as presupposed by [GM] 5 explicit/implicit
propositional intendings, and 10 propositional believings/
desirings. They need not involve utterances, and thus
circularity need not be involved. But some propositions are
so "abstract," surely we could not intend or believe them
in the absence of language. Eg

2. The meaning of a statement is the method of its verification.

surely requires at least an internal utterance in order to be
intended/believed, or (to the extent possible) desired. All
beliefs appear to be "actual" as opposed to dispositional,
and similarly any "desires".

We should recall that, for Grice Phi and Phi' range "over properties of persons
[...] appropriate substituends for Phi and Phi' will include such diverse
expressions as "is a passerby", "is a passerby who sees this notice", "is a
native English speaker", and "is identical with Jones". So, for 2, we can think that Phi stands for "is identical with Carnap"!

In connection with "psychological reality" and in reply to MJ Murphy, Dejnozka
comments
"I agree with D. Wiggins that it is logically
NECESSARY that we cannot refer to anything if
we cannot single anything out. Thus I agree with
MJ Murphy, broadly speaking, that is, insofar as
"singling out" is a psychological process. But
just look at [GM]: intentioons, beliefs, and
desires are running all over the place in [Grice's]
conditions for (1) (By uttering x U means that p).
Why, intentions, beliefs, and desires are practically
the only conditions he requires! Are you saying that
these are not "psychological" for him? Is he a
"materialist" or a "neutral monist"?

My idea that he was a "realist" did not seem to have convinced Jan]. Well,
His specific essay in philosophical psychology is his "Method in
philosophical psychology: from the banal to the bizarre", in THE CONCEPTION
OF VALUE. I guess he can defined as a "functionalist". He certainly was no
materialist, or "neutral monist". He was a "constructive functionalist", of
the Aristotelian type. He thought of concepts like "believing" (and
especially "desiring", since he defines "believing" in terms of desiring in
that paper) as theoretical concepts brought about to explain the link
between perceptual input and behavioural output.

"[The way [Grice] organises his essay against
counterexamples addressed against the necessity
and sufficiency of his analysans] does not answer
my criticism [which concerns what he] succeeds in
doing. When I say "concentrating on the periphery,"
[...] I mean that's what he did."

I see, because you think he should have based his analysis on the idea of an
addressee or audience or recipient. I take your point (Doesn't mean I share
it!).

"Could you please explain Grice's distinction between
"theory" and "analysis"?

Well, not really. But a comparison between his approach to "meaning" and to
"desiring" may help. "Meaning" he analysis in terms of "desiring" (since
"intending" is a conjunction of "desiring" and "believing", and the latter
is analysable as "desiring"). But he does not postulate "meaning" as a
theoretical concept to bridge the gap between perceptual input and
behavioural output. He merely states necessary and sufficient conditions.
That's an "analysis". On the other hand, a "theory" is were concepts such as
"desiring" belong. He does suggest that it is possible to construct a
"theory" rather than an "analysis" of meaning:

It remains to be inquire whether there is any reasonable
alternative programme for the problems about meaning
other than the provision of a REDUCTIVE ANALYSIS of the
concept of meaning. The only alternative which I can
think of would be that of treating "meaning" as a
THEORETICAL CONCEPT which, together perhaps with OTHER
THEORETICAL CONCEPTS, would provide for the primitive
predicates involved in a semantic system, an array whose
job it would be to provide the LAWS and HYPOTHESES in
terms of which the phenomena of meaning are to be
EXPLAINED. If this direction is taken, the MEANING of
a PARTICULAR EXPRESSION will be a matter of HYPOTHESIS
and CONJECTURE rather than of INTUITION, since the
application of THEORETICAL CONCEPTS is not generally
thought of as reachable by INTUITION or observation.
(Grice, Retrospective Epilogue, "The primacy
of utterer's meaning", in Studies in
the way of words, p.358)

"Does this mean, when uttering

3. I'm tired.

that c can concern only U believing that U utterer is
tired? "As your [addressee]" seems to imply [addressee]
meaning. It seems to me that if [utterer's] meaning is truly
primitive, there should be no need to mention an [addressee]
audience at all - not even the utterer as utterer's own
[addressee]."

Indeed, revising Grice's fifth re-definition above, one sees that while he
allows for "U" to range over "utterer", he actually doesn't use "A" for
"addressee", and the analysans is expressed in terms of "anyone who has
property Phi, or Phi'". I just used "addressee" as a matter of economical
convenience, because I think that that is what he has in mind. But it's just
anyone who happens to recognise that U means.

"Also, I disagree that I must be interpreted by others as
meaning

4. Jan Dejnozka.

or that others must have heard of me. Think of a message
in a floating bottle, or a fortune in a fortune cookie

5. Help! I am being held prisoner in a Chinese laundry!

I think we can understand [U's] meaning in this case without
knowing or having heard of the prisoner. [...] Trivially, of
course, any time I write

6. I.

in object-language ordinary [...] discourse I'm referring
to myself. We know at least that the writer of the fortune
was writing about him/herself. But it seems an abuse of
language to say we must have heard of the writer. We have
never heard of the writer, and worse, we have no idea who
s/he is. We do have at least one conceptual way of referring
to the writer - as the writer of a certain "fortune", but
that's based on the uttering, not on the meaning. I could
just as easily refer to the author of a piece of nonsense who
means nothing. [...] I often communicate with people who
have never heard of me. If I couldn't do that, I could never
introduce myself!

9. I am Jan Dejnoska.

The points by Dejnoska are well taken. But note that while Grice is interested in providing different properties identifying possible addresses (anyone having Phi or Phi'), he does not provide corresponding identifying properties for "U". I guess he
just means that if you happen to utter (9), you'll be introducing yourself,
or that if you happen to READ (5), you (qua addressee) will happen to
identify the U as someone who meant this or that. Grice seems to be
concerned with the identification (by the utterer) of his potential
addressee, but not of the identification of the utterer by the addressee.
This goes to show that he is, unlike you, into an utterer's centred
analsysis meaning rather than the one you propose around the notion of an
"addressee".

"Thank you for agreeing that c is underdefined by Grice.
I assume you mean by "underdefined," "not defined narrowly
enough." I'm saying the analysis or definition is too broad,
that the analysans applies to more cases than the analysandum does."

Well, Grice says that c stands for iconic, associative, and conventional.
That's not THAT broad. I guess here the issue is that claim
x has f.
If x is a hand wave, and f specifies the type of handwave it is, the
correlation will be iconic. Associate correlation may be that you pretend to
cry (shed tears) so I associate that with you meaning that something's
saddening you. Finally, a conventional correlation would be "You eat like a
pig" when I mean that you eat like a pig (The fact that "pig" refers to pigs
in English is totally conventional and arbitrary).

"Why (can't I lie when I say (3)? Goldbricks say it to get
out of work. When I say "intend to lie," I don't mean
(the utterer's) intention by "intend." I mean having the
ULTERIOR purpose of lying.

Oh, I see, so it's an intention that it's not captured by (GM) (Grice's
analysis of meaning) above. It's something "ulterior", and so beyond
"meaning". Lying does not depend of course on "c" being a conventional
correlation. You can lie "non-verbally" when the correlation is just iconic,
or associative, and then, even not every conventional correlation is of
course linguistic. If you put a sign of a "bear" in a wood, it means that
there are bears in the wood. You can lie by putting that sign, but that's
beyond what the sign means. It reminds me of a cartoon by Bailey. He sees a
sign that reads

10. Bear to the right.

And the stoopid soldier says to the sargeant. "I see no bear to the right"!
:) So, he thinks the sign is misleading, and probably deceptive (while it
means "keep driving on the right of the road).

"It makes perfect syntactic sense in ordinary English
(to say one believes a sentence).

10. Whenever 'This is a bank' is believed.
is syntactically correct even if "bank" can mean either
a financial institution or the side of a river. In fact,
many cases of beliefs apparently but not really in
conflict are due precisely to ambiguity. But even aside from
that, we could make "c" the correlation that if x has f (as I
specified f) and x is true, then p is believed whenever p is
believed. Grice's condition would be perfectly satisfied, but
it might as well be the condition that

11. 1 + 1 = 2.

Point taken. I was merely displaying my prejudice that it's not
sentences that we mean, but "propositional complexes". I mean, if we are
both together in Brazil, and it starts to rain. You'll think that it is
raining, and so will I. My claim is that we'll think the same propositional
complex. I don't need to claim that you think an English sentence or that I
think a Spanish sentence. That seems to presuppose that language is too much
involved in our thinking processes, as I think it isn't. I can think lots of
things that I cannot possible put into words...

12. Whenever 'It is hot here' is believed.

contains the indexical "here," but there is nothing *syntactically*
wrong with it. It is perfectly correct English. Likewise for

13. Whenever 'You are tired' is believed.
And "x" need not even contain an indexical. "x" could be
14. Mount Everest is in Tibet.

What's syntactically wrong with the clause (15)?

15. Whenever 'Mount Everest is in Tibet' is believed.
That sentence is not only widely believed, but true!

My point is merely that for a Gricean, you
don't believe "sentences". You believe propositional complexes. Griceans
think dogs and cats and squirrels think, yet they don't utter meaningful
sentences.

"Grice is analyzing (1) (by uttering x U means that p)
which is "synthetic a posteriori", and he states as one
of his conditions a sentence which I can interpret as an
"analytic a priori" truth. He might as well have stated (11)
as a condition, for all it illuminates (1)"

Well, as far as it's not too harmful! But Deknozka's point is well taken.

"Embedding Phi' within (2) would have made it harder to
find counterexamples. Would you also pull Phi to the left
of III?"

Grice writes,
As will be seen, for U to mean something it will have to be
possible to identify the value of Phi (which may be fairly
indeterminate) which U HAS IN MIND [emphasis mine. JLS]; but
*we* [emphasis Grice's. JLS] do not have to determine the
range from which U makes a selection
Grice, Logic & Conversation, p.114.

"Grice is trying to analyze (1) (By uttering x U means that p)
and that sounds like linguistic speaker's meaning to me. You
need to take out the "utters" to take out the linguistic character.
It should be just

16. A means that p.

where proposition "p" is non-linguistic.

We should recall that in "Meaning" Grice says he uses "utter" in a rather
"artifical" sense, as not having any necessary condition with linguistic
meaning. A handwave is an utterance, so is intentional laughter, or an
itentional yawn. Anything you "express" is an utterance. Etymologically, to
"utter" means "to out", and is not connected with sentences in a given
language. That's a jargon introduced by linguistics, I think. To think of a
sentence in context as an "utterance". Quite conversely, Grice uses x for
utterance token and X (capital X) for utterance type (A sentence is just a
kind of utterance type) (This he analysis in the essay, "Utterer's Meaning,
Sentence Meaning, and Word Meaning, pp.117ff).

"By audience [or addressee] meaning I meant the
meaning the audience [addressee] attributes to
the speaker [utterer]. This does not imply conversely
that the [utterer] has an [utterer] meaning, since
an [addressee] can mistakenly attribute an [utterer]
meaning where there was no [utterer] meaning. But it
does imply that the [addressee] conversely has a concept
of [utterer] meaning. My case that [addressee] meaning
is more fundamental rest[s] on private language
argumentation"

Different programmes, it seems. To Grice and to me, it
seems utterer's meaning is more basic, because we wouldn't be bothering to
analysis what the addressee takes to be an instance of utterer's meaning
only to be later proved that he was mistaken! Also, if the addresse needs an
embedded notion of utterer's meaning, I guess that that is what Grice is
providing.

"If you have to rewrite it in terms of the very analysis,
that is precisely a vicious regress! Or perhaps you are not
expressing your point well"

An illustration. If Phi happens to be "he is a native
Engish speaker", that property may seem to involve a semantic connotation,
but as far as we define "native English speaker" in terms of "believing",
"intending", "desiring", "utterance types", etc. I see no circularity.

"Grice has be able to specify the descriptions in every case.
But I think language is too poor to do the job even in the case
of adults, let alone children. Descriptions are alway in
language. My 3-year old almost never identifies me by a
description [...] She constantly identifies me and singles
me out, but she is only occasionally describing me or
even referring to me. It would be better to speak of
"properties" instead of descriptions, but if everything
must be identified by means of properties, then there will
be a vicious regress of "properties" needed to identify
"properties".

Well, not necessarily vicious. I'm thinking of a given property defined in
terms of a more "primitive" or conceptually prior property. E.g. she may not
say "dad", but single out as "the man before her eyes", which is a property,
which may be reducible to some "perceptual" property, "the man I perceive
before my eyes", or "the sense datum that replies to my calls"...

"What is "a recipient"? Can a stone be "a recipient"?
How can one be "a recipient" without being an audience?"

Well, you'll see from (GM) above, that Grice does use indeed "person", as
"anyone having property Phi or Phi'". I guess a certain rationality is
necessary even of the passive kind. The addressee has to be able (for a
successful instance of understanding) what the utterer is intending to
communicate. I can talk to a rock. Reminds me of another cartoon, in today's
Herald (relying on memory)

Snoopy's relative (talking to rock):
It must be very boring having the life of a rock
...
Compared to mine.
Who's here, just talking to a rock...

"How can one be an audience [addressee] without
attributing some speaker meaning to some speaker?"

Exactly, but recall Snowwhite's Stepmother, "Mirror, Mirror, who is the most
beautiful woman in the kingdom?". She probably means something, although her
addressee is just a mirror. Perhaps she is just using a rhetorical figure
and saying the words to convince herself that she's the most beautiful woman
in the kingdom.

"Most importantly, how can U intend the two things (GM)
stipulates U must intend if U has no concept of an
audience [addressee] or of audience [addresssee] meaning
in my sense?

Note that "intending" presupposes that the U must think his
intention is feasible. He cannot just intend that the rock (or the mirror)
will think that the U believes that p. But note that Grice describes the
condition in terms of "thinking", not of "knowing". The U may be wrong. All
that we need is that U THINKS that it is feasible that what he intends may
be realised. I can intend to fly iff I think I can fly. It doesn't mean I do
fly or that I will fly. I should have to elaborate on this... I do think
that feasible intentions are important as far as "understanding" is
concerned. I.e. I cannot intend that the rock or the mirror will understand
what I say if I know they won't! But I may be mistaken, and think they MAY.
In which case, I probably CAN intend the rock to try and understand me. If
understanding doesn't take place is a question of understanding, not of my
having meant THIS or THAT. (Cfr. when you write in your diary, "Dear
Diary,..." You know a BOOK cannot understand what you're writing...

"By "audience meaning," I mean not any meaning the
"audience" "means" (because then the "audience" would be
a "speaker", not an "audience"), but "the meaning the
audience attributes to the speaker"".

I confess this seems to complicate things for me. Why should
I base meaning on its reception on some addressee. Grice mentions "anyone
who has Phi or Phi'" but does not actually speak, in redefinition 5 of
"addressee" as such. His is an analysis of "meaning", not of
"understanding". It was Strawson who first proposed an analysis of "A
understands what U means" as "A knows what U means" ("Intention & Convention
in Speech Acts, in his Logico-Linguistic Papers, London: Methuen).

"Intentions often seem totally out of control! Think of
Freud on unconscious slips. And the more we try to control
them, the less controllable they are. Freud cites an
example (or the translator substitutes one) of a newspaper
in Germany reporting:

17. The battle-scared crow prince had returned to Berlin.

The newspaper printed an apologetic retraction the next day
saying that of course they meant to say that

18. The bottle-scarred clown prince had returned to Berlin.

I cite this as an example only of uncontrollability of unconscious
intent. It might take many years of therapy to control the reporter's
unconscious hostility to Crown Prince Wilhelm, and even then "control"
would "not" be the right word. The aim would be to make the
unconscious conscious - to make it cease to exist as an unconscious
intention, not to control it."

In the Gricean festschrift, ed. R. Grandy & R. Warner, Philosophical Grounds of
Rationality, D. Davidson has an essay precisely on that topic, which he
calls "malaprops". He thinks only a Gricean can account for them. The essay
has by Davidson has been discussed in special Davidsonian collections and
commented inter alia by M. Dummett. The idea being that it's the intended
meaning that matters, not the actual words spoken (Davidson's essay is called

19. A nice derangement of epitaphs.

which is actually the classical malaprop for

20. A nice arrangement of epithets.


"Why couldn't I be drowning alone and cry "Help"?
Well, because we're not rational, and a drowning person
will clutch at useless straws. But that doesn't matter. Grice
is still claiming to be able to analyze every last case of (1)
in terms of the conditions he states. That includes the
drowning example."

Note that Grice does allow U = A, so he can trying to help himself, or he may be
praying to God.

"And that's also why "deeming" is a dodge. Either (GM) works
or it doesn't. If "deeming" has to be brought in to rescue
(GM), tnat is a sure sign (GM) failed. But "deeming" is
"audience meaning", not (utterer's) meaning. A deemer is
an audience who attributes meaning to a speaker. So deeming
can never rescue any analysis of utterer's meaning, unless
utterer's meaning is identical with audience meaning,
which makes no sense."

As a matter of record, it is only at a latter stage of his philosophical development that Grice included the notion of "deeming". He discovered that "meaning" had the
features of a "value-oriented" concept (like "person", or "rational", or
"good"), and so he thought that "deeming" was an useful concept to give a
rational to meaning, in terms of the notion of "optimal value". He writes
re: "deeming U as having meant that p":

It is not my business on this occasion to suggest, exactly,
or in any detail, what the deamns for approach or approxima-
tion might be. I will only say that, whatever they are, they
ought to be ones which justify US in DEEMING certain cases
to justify a given ideal even though they do not, in fact,
strictly speaking, justify it; just as in Oxford on one
occasion, there was a difficulty between an incoming provost
and a college rule that dogs were not allowed in college: the
governing body passed a resolution deeming the new provost's
dog to be a cat. I suspect that crucially we do a lot of deeming,
though perhaps not always in such an entertaining fashion.
Grice, Meaning Revisited, in Studies, op. cit, p.302.

"Your claim that Grice is not a trascendentalist is not a
reasoned objection to my argument."

Grice in fact discusses trascendental arguments (that he prefers
to call "metaphysical arguments" in his contribution to the Gricean
festchrift, Richard Grandy & Richard Warner, PGRICE Philosophical Grounds of
Ratinality: Intentions, Categories, Ends. As both Grandy and Warner are
named Richard, Grice called his contribtuion "Reply to Richards". It is
reprinted (in part) in his THE CONCEPTION OF VALUE.

"I had thought from your original exposition that Grice
was analyzing conventional meaning, not what the speaker
wishes to convey. My criticism is postponed to the level of
finding "A utters x meaning p" ambiguous, because the word
"meaning" can be taken to mean either speaker meaning or
conventional meaning. Second intention is meta-intention.
It need not be linguistic."

Note that for Grice what is "conventional" is just one type of
"correlation" but not meaning per se.

"Predictions need not be successful. That's beside the point.
The point is there's an ordinary sense of "means" which is
"is a usual predictive sign of." There's no debating the
English."

There is of course some debating in terms of the nec. and suff. conditions for its
analysis. Grice considers this:

Anyone who says

21. Those black clouds mean rain

or

22. Those black clouds mean that it WOULD rain

would presumably be committing himself to its being the
case that it will rain, or that it did rain.
Meaning Revisited, p.291.

I see Grice as providing an alternate analysis, and actually an alternate
analysans. 22, unlike 21, uses "would", and Grice proposes two possible
analysans as to the "commitments" involved.

"There's an accidental emphatic character to my example,
but there's no debating the existence of this ordinary
sense of "means" in English. Every dictionary is filled with
lexical definitions, and not one of them is emphatic."

It seems that, sometimes, MEAN is used merely an empty emphatic particle, e.g. "Close the door and I mean it". In an ideal circumstance "Close the door" would do!

"If "meaning" is unambiguous, how can there be any special
cases?"

On the other hand, "horse" is unambiguous, but there are many types.

"My edition of the OED lists 24 meanings of "mean" under three
main headings."

We must leave the discussion of those 24
"descriptive" (i.e. non philosophical) usages for a longer day? It still
doesn't prove it's "ambiguous". I take each "meaning" to be actually an
"use" or "usage". In this I'm very Rylean (see his Use, Usage, and Meaning,
in G. Parkinson, ed. Theories of Meaning)! Finally, in another remark to MJ
you write,
"I wonder what Grice would say about intentions and desires as
publicly observable. Talk about them certainly is. We can only
learn how to use the words "intend" and "desire" through public
examples."

Well, I referred to that in trying to draw the line between "analysis" and
"theory". If "desiring" is a theoretical concept is, precisely, "not
generally thought of as reachable by intuition or observation", as I have
quoted Grice as saying above. That's debatable, I know, but the way the
Gricean programme sees it.