Presumptive Griceians

By JLS
--- for the GC

ACADEMIA being what it is, in France, they are calling people 'neo-neo-Griceans'. Give me a break! I want good old, or paleo-, if you must, Grice!

Then there's Levinson's thick book with MIT: _Presumptive Meanings: A Theory of Generalised Conversational Implicature_. Does he count?

Why does

"She bit her nails"

seem to suggest "she bit her OWN FINGER
nails"? Why can't "him" and "Ben" co-refer in "Rosa locked him in Ben's
room"? Why can't "Tuesday" mean "today" if it happens to be Tuesday and you
say, "Let's meet on Tuesday"? Cases like this are the stuff of Levinson's
new book with MIT Press!

I reproduce below the table of contents of Levinson's book. Since I believe
it's the first sort of all-covering statement of what _implicature_ is or
amounts to, it may help to review the territory it covers (see appendix
below for the table of contents of Levinson's book).

I. GRICEAN PRAGMATICS: THE STATE-OF-THE-ART. Levinson writes:

I grant that [the notion of Generalised Conversational
Implicature] is _not- easy to defend
(p xiii).

The idea is that _particularised_ implicature may be _easier_ to defend...
His defense, unlike Grice's, which is philosophy-motivated, is
linguistics-motivated. He relies on a number of disciplines. Notably, on
p.xix, Levinson writes of the "fresh interest in implicature in AI circles"
-- which we discussed in this FORUM -- and writes, "a good idea of what is
going on can be got from glancing at the proceedings of the
1996 AAAI symposium on implicature". Re the philosophical tradition proper,
besides mentioning Grice all over the place, Levinson makes on p.4 an
interesting claim about Wittgenstein:

An utterance is not, as it were, a veridical model
or 'snapshot' of the scene it describes, although
much talk of truth-conditions might lead one to suppose
the contrary (the work of the early Wittgenstein explored
this initially attractive idea, which has never
been quite washed out of our thinking since).

I guess I fail to see what idea he is referring to, and why would it be
washed out if it's attractive?

My favourite way of modelling implicature is via practical reasoning, and
so I was warmed to see Levinson's reference to Schelling. Levinson makes
the claim more that Grice's essay on 'Meaning' will fit the rationality-model:

I assume a kind of coincidence of interests, treating
linguistic communication as a "game of pure co-ordination"
in the game-theoretic sense of Schelling's _The Theory of
Conflict_, a PICTURE I THINK PRESUPPOSED by Grice's 'Meaning':
the utterer is trying to find an ECONOMICAL MEANS of
invoking specific ideas in the addresee, KNOWING THAT the
addressee has exactly this expectation.

I'm sceptical about this, though. I think Grice's motivations are other,
and, though I see Levinson's pint, can't see how a picture to be shaped in
1960 (Schelling's book) can be _presupposed_ by an obscure essay dated in
1948! I think the picture that Grice's essay of 1948 presupposes is
Stevenson's brand of emotivism (The only reference cited by Grice there
being Stevenson's _Ethics & Language_).

There is one notion that I think is central to Grice: viz. the idea of
_reason(ing). Levinson deals with it here and there, not so much in the
abstract, but as it gets shaped in particular _reasoning patterns_. He
shares with Grice, though, an unashamed form of rationalism. Thus, on p.14, he writes:

what is conversationally implicated (unlike what
is _conventionally_ implicated) is _not_ coded, but
rather INFERRED on the basis of some basic assumptions
ABOUT THE RATIONAL NATURE of conversational activity,
as sated in the cooperative principle and its constituent
maxims of conversation.

It would be unnecessary to point this rationalism (for which I also refer
to HINTIKKA in the biblio below) were it not for Wilson/Sperber's attempt
to _deny_ that Grice's constraints are _rational_...

II. GRICE'S PHILOSOPHICAL MOTIVATION IN 'LOGIC & CONVERSATION'. On p.18,
Levinson points to a rather important issue in exegesis: the comparison of
editions! (a task for which Levinson relies on Arundale). E.g. in
'Presupposition & Conversational Implicature', Grice discusses the
_philosophical relevance_ of "generalised conversational implicature" in a
passage that he edited out when reprinting the essay in _Studies_. That
passage by Grice, which Levinson quotes, and which has relevance to our
discussion of "voluntarily" reads:

generalised conversational implicatures are the ones
that seem to be more controversial & AT THE SAME TIME
MORE VALUABLE FOR PHILOSOPHICAL PURPOSES.

Echoes of Austin on "voluntarily" come to mind, and indeed the whole
discussion in 'Prolegomena', and the A-philosopher. Grice continues:

In my 1967 lectures, I put forward the case that the notion of
GENERALISED conversational implicature might be used to deal with a
variety of problems, particularly in Philosophical Logic, BUT
ALSO IN OTHER AREAS.

As I say, Grice's consideration of Oxford-types of philosophising (Austin,
Hart, Ryle 1949, etc) comes to mind. All in all, though Levinson seems to
favour the interpretation that Grice was _only_ concerned with the
divergence between _natural-language_ connectives and their formal
counterparts, he notes that Grice's motivation is indeed to clear out the
way for a clean account of _logical form_ and truth-conditions, as it were.
In this, his interpretation of Grice contrasts with the following
_mis-interpretation_ we find in the work of Dummett when he writes that

the notion of conversational implicature was invented
in place of the general semantic concepts which had
been expelled in the original determination to pay
attention to nothing but the actual use of particular
utterances. 1978:445.


III. IMPLICATURE-THEORY FOUND TOO "WEAK". On p.26, Levinson points to a
problem for the Particularised Implicature theorist, with which, as a
linguist he is particularly concerned:

general patterns, in which, for example, MARKED
forms [e.g. "He murdered her voluntarily"]
suggest marked extensions ["within a fishy context"]
CAN'T COME OUT of a theory of contextual biases alone.

This reminds me of Grice's reply to Davidson in the symposium on Intending
(cited by Grandy/Warner below) where Grice delivered his 'Reply to Davidson
on intending', "a thorough criticism", in the words of Davidson. In that
symposium, Davidson, trying to be more Gricean than Grice, was saying that
an utterance of the form

1. Jones intends to go to Rome.

IMPLICATES

2. Jones _thinks_ that he will go to Rome.

Grice, as reported in Pears, noted that the theory did sound good, indeed
perhaps too good to be true. I.e. Grice hoped that the connection between
intention and belief was _stronger_ than the Theory of Implicature could
allow! (for his own brand of theory, where 2 comes out as a standard
_entailment_ of (1) see Grice 1971 below.

IV. GRICE'S TYPE OF TRASCENDENTAL ARGUMENT. One thing that Levinson
discusses if not from a philosophical perspective, but from a new welcomed
linguistc-cum-AI one, is Grice's reliance on trascendental arguments.
The idea (suggested by Grice in Studies, p.30) being that conversational
maxims are not just Rational: they are REASONABLE. If we don't rely on
them, we are not speaking rational. The maxims are like a "precondition for
rational communication", as Strawson puts it (Strawson's essay in PGRICE).
Levinson deals with this in with reference to Grice's recourse to the
"pirots" -- a form of "metaphysical transubstantiation"! Interestingly,
while Grice says that he is using "pirots" after Russell, and Carnap (_The
Conception of Value_, p.140),

My creatures I call _pirots_ which Russell and
Carnap have told us, are things which
karulise elatically.

Levinson points out that the word is used by Chomsky in _Syntactical
Structures_ so I'm slightly confused! It would seem as tho' Chomksy got the
idea from the Philosophers? In any case, Levinson relies here, rightly, on
Grice's _unpublications_ over the years:

Grice routinely adopted a design perspective in
order to tackle fundamental problems in philosophy.
He would imagine artificial life-forms, which he dubbed
_pirots_ and ask how they would have to be minimally
endowed in order to display the particular aspect of human
behaviour he was interested in. p.29

Following Grice's analogy, one can (as I have) dub the set of
conversational maxims so construed, and expanding on Grice's joke on Kant,
as the Conversational Immanuel! For a discussion of this, I found J.
Baker's essay in the Grice Symposium rather interesting.

V. METHODOLOGICAL PROBLEMS WITH THE NOTION "IMPLICATURE". GRICE ON
INDETERMINACY. Levinson refers to J. D. Atlas's discussion of some
methodological problems involved in the "recovery" of Implicature. The
discussion reminds me of Biro's essay for _The Monist_. The idea is that
the recovery of a conversational implicature can only be achieved _ex post
facto_ (or "post hoc", Levinson, p.367), since the number of intentions
which are in principle retrievable is in theory infinite. Levinson thinks
that viewing Grice's maxims as "addresse-oriented heuristics" helps solve
the problem though. E.g. with things like:

3. ?Some if not all of the students split off the mob

Levinson writes:

Some theorists have assumed that the recognition
of utterer's meaning intention is simply a matter
of running the planning-reasoning backwards: we
observe the behaviour and figure out the underlying
intention. The problem is that this cannot work,
for the simple reason that for all inference systems
one cannot work backwards from a conclusion to the
premises from which it was deduced -- there is always
an infinite set of premises which might yield the same
conclusion. Consider the conclusion

4. q

from the sets

5. p, p -> q

6. p & q

7. p v q, -p


etc."

Levinson does not mention at this point that one of the defining features
of implicature, for Grice, is, indeed, INDETERMINACY! (Studies, p.40)
(which Levinson fails to list in his overview of implicature's properties
on p.15) Note that Levinson is otherwise happy with Grice's notion. Thus,
when criticising "default logics" (see below "CETERIS PARIBUS") he writes

These logics, in their present form, provide a
RESTRICTED SET of limited inference rules,
INCAPABLE OF MODELLING the OPEN-ENDED, CREATIVE,
and INDEFINITE SET of inferences tha come under
Grice's theory. p.48.

VI. LEVINSON CONTRA HORN ON LEXICALISATION. The case of the Squarish
Psychoid. In discussing Carnap's and Bar-Hillel's notion of informativeness
(which Levinson notes received some good criticism by Cohen -- ref. below)
Levinson mentions that if in a closed universe of discourse we have a cube,
a pyramid, and a cone, and utter

8. The squarish block

the default implicature will be:

9. The block is _not_ a square.

This, however, Levinson notes, is merely an implicature due to the simple
fact that

10. A square (or a cube) is (surely at least) squarish.
Levinson, p.34.

This reminds me of discussion I've had with Horn, and may explain why Horn
may _not_ be altogether happy with _all_ the sections in Levinson's book?
Horn would say that "squarISH" _lexicalises_ (9). I just don't know, but
I'm more incined for the radical pragmatic approach a la Grice and Levinson here. He writes:

Anyone who rejects the notion of a GENERALISED
conversational implicature or a preferred interpretation
for that matter will have to find SOME way OR OTHER
to explain lexicalisation patterns in English, if he
thinks there are any! The nonce-implicature theorist can't
just explain them. According to any such theory, there are
_no_ general tendencies to be found, or if there are,
they have the mere status of _mere behavioural tendencies_
playing no role in the systematic generation of implicature.
On such a theory, there is no more reason for

11. not all

(the O-corner of the Square of Opposition, see below) to
resist lexicalisation than for

12. not some.

to undergo it (as

13. none.

Only a theory of default or preferred interpretation can
explain how something AS DEFEASIBLE & PROTEAN as a pragmatic
suggestion can IMPOSE SUCH FUNDAMENTAL CONSTRAINTS ON
THE ENGLISH LEXICON. p.71

I am reminded here of Austin's interesting remarks on "inadvertently" (in
'A Plea for Excuses')-- which Tapper may check if they are quoted in Horn's
book. I am _also_ reminded of one delightful outburst by Grice, when he had
found that English had made such fine distinctions for him, "How clever
language is!", (reported by Warnock). But this is rather tricky. Levinson,
for example, considers one Horn scale which he sees as having special
_philosophical relevance_:

14.

i.e. "I believe" +> "I don't know", and notes, "Stating

15. Jones believes that p.

implicates that

16. Jones doesn't know that p.

and notes:

It is likely that our analysis of the LEXICAL
meaning of such verbs has SYSTEMATICALLY FAILED TO
TAKE IMPLICATURES into account.

But, why would it? I mean, it would not be part of the _lexical_ meaning,
(by definition) would it! In furthering Grice's linguistic botanising, as
it were, Levinson proposes, on p.111, the following table for what Grice
(Studies, p.291), in joining the linguists, call "factives":

Verbal Doublets Giving Rise to Quantity Implicatures:
====================================================
weak verbs strong verbs
NOT entailing entailing
their complements their complements
=====================================================
say, claim, hold disclose, divulge, reveal
deny, reject disprove, refute
predict foresee
believe know
==


VII. PROTOTYPES. In his comment on L. Tapper, S R Brown (THIS FORUM)
mentions Wittgenstein's idea on family resemblances. Levinson discusses in
passing this (e.g. p.37) vis a vis E. Rosch's prototype-theory. This he
does when attempting to re-formulate Grice's second maxim of quantity
("don't be more informative than required") as the addressee-oriented
heuristic, "Minimal specifications get maximally informative or
STEREOTYPICAL INTERPRETATIONS" (p.37). E.g. if I say,

17. Jones brushed his teeth voluntarily.

given the presumption that one always does so voluntarily, implicates that
there's something fishy about Jones's brushing his teeth. The issue,
though, seems to be why _assume_ that Jones brushes his teeth voluntarily!?


VIII. A TYPOLOGY OF THE EFFECTS ON THE EXPRESSION OF VARIOUS TYPES OF
IMPLICATURE. Towards a Systematic Diagnosis. On p.41, Levinson offers a
table where some of the linguistic effects of various types of implicature
are analysed on a yes/no (and "no-answer") basis. It is, it seems to me, a
good case of how the linguist's work can complement with the philosopher's
craving for generalisation!


Gricean Heuristics Scrutinised


. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . Q1 . Q2 . M1&M3 .
. .what isn't . what is . what's said .
. .said, isn't. expressed simply,.in an abnormal way.
. . . is stereotypically.isn't normal .
. . . exemplified . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. I. negative . . . .
. inference . Yes . No . Yes .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. II. meta- . . . .
. linguistic . Yes . No . Yes .
. basis . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. III. contrast . . . .
. between . Yes . N/A . No .
. strong/weak . . . .
. forms . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. IV. contrast . . . .
. bewteen . No . N/A . Yes .
. synonymous . . . .
. forms . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. V. meta- . . . .
. linguistic . Yes . No . Yes .
. negation . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. VI. inference . No . Yes . No .
. to stereotype . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. VII. overriding. . . .
. particularised . None . Q, M . Q .
. implicature . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .



It would be a good thing to add to the table both Grice's Quality-
(springing from the idea that the utterer is offering trustworthy
information) and Relevance- implicatures as well.


IX. HOW MUCH "CETERIS PARIBUS" IS IMPLICATURE?

We have discussed in this FORUM the idea of Ceteris Paribus -- vis a vis e.g. Schiffer's criticism in _Mind_.

As things are, Levinson discusses _ceteris paribus_ quite a bit,
since after all, he is defending the idea that what's crucial for
understanding implicature is that it is a certain inference which will be
drawn by default in the absence of indications to the contrary. I must
admit that I was kind of wedded to this idea, but now that I see the
dangers of relying too much on ceteris-paribus too much (resulting in
viewing conversation as routinisation) I'm no longer so sure! Levinson writes:

Default logics aim to capture a peculiar
mode of reasoning -- viz. the notion of
a REASONABLE assumption, a ceteris
paribus assumption. The inference from

18. Tweety is a (cartoon of a) bird.

to

19. Tweety can fly -- if scared by a cartoon of a cat.

is an inference that we seem to make _UNLESS_
there is some inconsistent assumption already made.
e.g. to the effect that

20. Tweety is a (cartoon of a) penguin.


p.45


This is why Levinson (relying on Atlas/Levinson) thinks that


21. The dog scared the birds.


implicate


22. The birds _flew_ away.


Since what dog would reasonably scare such delightful critters as
_penguins_ anyway?


X. FORMALISING IMPLICATURE: IS THERE A NEED?

Although not very happy with formalising per se (he says formal notations date so soon), Levinson notes that there's something to be said about this newsish attempt, not so much to formalise implicature, but changing the perspective, to see IMPLICATURE as a model of practical inference. He finds a system of practical inference as the most plausible way to at least retain the Gricean motivation:

plan generation & plan recognition are what
MUST BE INVOLVED if meaning in general has
the character of Grice's non-natural meaning.
p.53

In this context, Levinson quotes work within the AI paradigm which are
mainly PhDs (Hirschberg, Mccafferty, Wainer, and Welker -- and discussion
of Mccafferty by Grice's literary executor, Grandy).

XI. PLAGUING SCHOOLDCHILDREN FOR TWO MILLENIA: THE SQUARE OF OPPOSITION.
I'm happy to learn from Tapper that Horn has posted his essay on John's
children and the King of France (being bald) in his site. Levinson deals
quite a bit with various squares of opposition, which he notes, have been
plaguing us since Boethius. One such is as below, which I'm reprinting
because it interestingly places the pleonetetic quantifiers identified by
Altham.

1.0 A . . . . . . . . . . .E ~1.0
all none
. contraries .
. .
. .
0.5 many ????? few ~0.5
. .
. .
. subcontraries .
some all
0.0 I . . . . . . . . . . .O O.O


Necessity

23. []p

and Possibility

24. <>p

fall square in the square if with Kripke we see them as, respectively,
"true in ALL possible worlds" and
"true in AT LEAST ONE possible world". Also "/\" (and) and "\/" (or) -- to
Grice's notation -- as they parallel "all" and "some" (all: "/\x." some:
"\/.x"). I'm glad Tapper discusses Mill in his "Horn
Reprints". I note that Levinson does, too, and mentions yet _another_
important name in the history of Gricean logic: Morgan. I'm adding
Levinson's probability factors as per his p.85 table, too, since Tapper is
a probablist & he will probably like this. And anyone, Levinson borrows the
idea from Horn 1989:237!

XII. IS GRICE'S RAMSHACKLE WORTH REVERING?

One thing I hate about
Grice-bashing of the sort Levinson occasionally engages in (:)) is when he
writes of Grice's scheme as being rickety! He is discussing the idea of the
conversational categories having to be four (Quality: Trustworthiness,
Quantity: Informativeness, Relation: Relevance, Manner: Perspicuity) etc.
based as they were on Aristotle and Kant. Levinson writes:

it may be objected that my scheme shows TOO
MUCH REVERENCE for Grice's RAMSHACKLE
SCHEME. p.75

Anyway, I've seen worse ramshackles!

XIII. GOLFING NUMERALS & AND OTHER TRIGGERS OF PARTICULARISED IMPLICATURE.

Murphy has posed an interesting case for Gricean analysis concerning not
only the quantifiers (how do you explain '?Some of the students if not all
split off the mob') but also the cardinal number of cows in a herd:

25. ?Three if not all of the ten cows
were culled from the herd.

Levinson, in discussing the Gricean account of absolute cardinality, draws
on an example by Horn which I found of interest (having discussed with Horn
the alleged ungrammaticality of

26. He's nine years young. Horn notes (in his PhD for UCLA) that if you are
playing golf, you are bound to use NUMERALS in rather tricky ways.

Horn has drawn our attention to the fact that in certain
SPECIFIC CIRCUMSTANCES, ENTAILMENT SCALES CAN BE INVERTED.
For example, in tallying golf scores, where the HIGHER score
is the LESSER number one might feliciously say,


27. John can do the round in 71, IF NOT 70.


This scale-reversal does not appear to arise with other quantifiers (Sadock
1984:143) and indeed, Levinson notes, "otherwise seems restricted to
pragmatically defined scales". Can not the same be said of culling cows and
splitting activists?


XIV. VARIOUS "CHRONIC" IMPLICATURES. If we say


28. Rodrigo Vanegas lived in Colombia.


the implicature seems to be


29. Rodrigo Vanegas no longer lives there.


Levinson notes of course this is all pragmatic, as Harnish had long shown
(Harnish reprinted in Kasher). Levinson makes the following comment about
time which I liked

most formal treatments of TENSE in English reflect
the ASYMMETRY in informational specificity by stating
in effect that the PRESENT tense of an event holds
if the time of an event coincides with the DESIGNATED
UTTERANCE time, whereas the PAST tense holds if there
is at least SOME time preceding utterance-time at which
the event occurred (Cann 1993:242). GIVEN THE LINEAR
NATURE OF TIME, IT IS ALSO THE CASE THAT ANY PRESENT
EVENT BECOMES A PAST EVENT: THE PAST IS MORE EVENT-
POPULATED THAN THE PRESENT! p.95

True! But then cfr. Augustine. "The past does not exist, therefore, how can
it be _more-populated_ than anything?"


XV. A PHILOSOPHICAL LOOK AT PROPOSITIONS.

The meaning of "at". When
discussing " "Modified Occam Razor" (do not multiply senses beyond
necessity"), Grice cast doubts as to whether it is proper to ask about the
_meaning_ of things like "or" or "to", which are so _brief_. Propositions
have never been a hot topic with philosophers. In this book, Levinson
attempts an analysis of the scale

30.

which I did found of interest. Levinson considers that a sentence of the form:

31. Larry Tapper is near the university.

entails, for Levinson,

32. Larry Tapper is _at_ the university.


Or: to be "at" some place is to be "near" some place. Rather odd, but then
... true!


XVI. CONTEXT, CONTEXT, and MORE CONTEXT.

The polemic between the generalised conversational theorist (such as Levinson and Grice in his more philosophical moments) and the particularised conversational theorist (like Wilson and Grice when trying to be a reductionist) seems to be one about the relevance of context. Levinson is concerned with Hirchberg here. Julia Hirschberg is a nonce-particularised implicature
theorist. Most of her examples (in her PhD for the U. of Pennsylvania --
now a Garland book) involve the idea of a "rank" (as opposed to a "scale").
A rank can be _very_ contextual and specific. The example cited by Levinson
goes:

33. A: Did you get Paul Newman's autograph?
B: I got Joanne Woodward's.


The implicature is


34. No, I did not get Paul Newman's autograph.


The reasoning relies on a scale, viz.


35.


It would seem as Levinson finds this all too much!


XVII. MODAL IMPLICATURES: THE POLEMIC.

I enjoyed R. B. Jones's comments on the implicatures et al of "may". Levinson deals with those, in part, in terms of Loebner's degrees of tolerance for certain pleonetetic quantifiers. E.g.

36. Most students split off the mob,
but most students did not.


is nonsense ("most" is intolerant with itself), but the following is not:


37. Few students split off the mob,
and few did not.


Levinson notes that "LIKELY" and "not likely" are INTOLERANT. Thus the
oddness (indicated by "?") of:

38. ?It's likely she'll go & likely she won't

XVIII. MORE LEAKAGES IN GRICE'S RAMSHACKLE, and how to start learning to
feel rather grateful about them.

In discussing Horn's Negative Raising (i.e. the implicature of

39. I don't think you've refuted Grice.

as being

39. I think you have NOT refuted Grice.

Levinson notes:

the very occurrence of this negative raising
as examined by Horn 1978 seems like a LEAKAGE in the
blocking mechanism whereby the first quantity
submaxim bounds the second quantity submaxim.


And then, when one would expect some resolution, he notes,

We can be GRATEFUL for that LEAKAGE though,
because it makes clearly visible the ANTINOMIC CHARACTER
of the two conversational submaxims, even if it _is_ a
bit of a theoretical embarrassment!


Echoes of splitting students and culling cows...

XIX. THE PHILOSOPHICAL RELEVANCE OF ONTOGENESIS, or what we can learn from
children learning English -- from Speranza.

Murphy has noted that the Grice models poses yet another problem for anyone who is sensitive about langauge acquisition, since, it would seem, a person may be born, live, and even die, without ever learning the truth-conditional table for the conditional. On the other hand, Levinson notes that child language
often PROVES or CONFIRMS Grice. He discusses the lexicalisation "braker".
Given that there _is_ a noun, "brake", it would be otiose to have _another_
noun, BRAKER. However, since a child may be unaware of the noun BRAKER, it
will be predicted, by Gricean maxims? that he will say things like

40. ?the braker
(Levinson, p.139).

and he will! Another pair of items he discusses is:

41. Jones went to the club.
42. ?John goed to the club.

A child is bound to say things like (42) at some stage of her Gricean
development. The interesting Gricean idea about this, which I have felt in
my own ontogenetical behaviour is that the child (and I) will perceive that
"went" can not MEAN the same thing as "go" in the past, since it sounds so
different! ("went" being, literally, the past of "wend". The child is a
Gricean though. Levinson notes:

Children produce regular inflections, like (42)
at the same age (or even AFTER) they are correctly
saying (41), because they APPEAR TO ASSUME
that "went" is a DIFFERENT verb from "go" with a VERY
SIMILAR, but CONTRASTIVE meaning, and this
hypothesis may be maintained for years. p.140

In all these ontogenetical observtions, Levinson relies on work by CLARK.
Interestingly, Clark writes in the dust cover of Levinson's book: "A
remarkable achievement", even though, as Levinson notes, Clarke is a
nonce-theorist!

XX. YOUR WEAKEST MEANING: "Some", "Voluntarilty", etc.

We have discussed here the Grice/Austin (and Mates/Cavell) polemic on 'voluntarily'. Levinson notes something which may relate:

On the original Gricean or standard radical pragmatic
view, the LEXICAL CONTENT of an expression is always
taken to be its WEAKEST EXHIBITED MEANING with the
assumption that the pragmatics supplements the expressions
to give their stronger interpretations. p.185

For Austin, though, this ain't necessarily so...


XXI. METALINGUISTIC NEGATION: its philosophical relevance. In 'Logic &
Conversation', Grice considers (p.53) the utterance

43. He doesn't merely believe it.
He knows it.

Always looking for speaker's contradicitons (as when pointing that it's
always ilogical, for any predicate Phi, to say, "A is Ph-er than Himself")
Levinson notes that

44. He doesn't merely believe it.

- if understood as

45. He doesn't believe it.

would turn Grice's 43 into a contradiction. Therefore, Levinson suggests
(p.211) it's better to regard the negation there as merely META-LINGUISTIC,
as in

46. It's not a toMAYto. It's a toMAHto.

(hard to formalise in logic this, he notes), i.e. as applying to the _calling_ (or dubbing) the referent "he" a believer. Lot of discussion (by Horn and Carston) on this. Must say I liked Levinson's idea of the contradiction being involved here. As he notes in a different context, "what could be a better trigger to a Gricean inference than a falsehood, categorial or other?"

XXII. IMPLICATURE-INTRUSIVE CONSTRUCTIONS.

In arguing for intrustive constructions (i.e. where the implicature builds in the truth-condition) Levinson notes an example which combines some of Murphy's puzzles. Levinson is concerned with how a definite descriptor

47. The king of France

may contain an implicaturish phrase

48. The King who owns some of the properties here.

Two interpreations of (48) are possible. Levinson's example concerns what
he sees as a perfectly well-formed and clear utterance:

49. The student who cheated on some of the exams
should be pardoned, but the one who cheated
on ALL of the exames should be expelled.
(Levinson, p.219).

Levinson wonders how, if Grice is right that "some" implicates "not all",
are we to make sense of (49)? Surely, via Grice,

50. a student who cheated on ALL of the exams

is _also_

51. a student who cheated on SOME of the exams.

Yet, we seem to obviate these Gricean considerations when using such
phrases -- sloppy languagers as we is! Actually what is at work is a
principle like this, it seems. If we symbolise a Horn entailment scale as

52.

-- where "S" is the strong expression and "W" the weak -- we'd have, "Given
the scale

53. THE W

CAN NOT (by virtue of "W" alone) be distinct from

54. the S.

since anything that satisfies S will sastisfy W. But in fact "the W"
_happens_ to be felicitous in these cases, because "the W" is
_pragmatically strengthened_ to _exclude_ "The S"" (p.220).

XIII. ANAPHORA STRIKES BACK (with a Gricean vengeance).

We have discussed with Murphy the philosophical relevance of anaphora in this forum -- and http://www.analytic.org (vis a vis R. Helzerman's liar paradox, and in my
case, vis a vis SR Bayne's criticism of Gricean nonce implicature). I was
thus warmed to have a thorough Gricean treatment of anaphora by Levisnon.
He dedicates the whole chapter 4 of his book to that, and though I note
that his bibliography does not contain many works by philosophers on
_anaphora_, that's just as would be expected (one great exception being
Good Ol' Donnellan in Cole). Examples Levinson is concerned with include:

55. John-1 thinks that I admire the idiot-1
(Levinson, p.299)


-- where the subscript "1" indicate co-reference. Levinson also considers
cases like

56. Clinton-1 said that the president-1 had not been informed.

where co-reference is understood via an implicated contrast vis a vis
"presidential capacity". Levinson notes that "pronominal reference tracking
can be very difficult for children." I'm relieved! Another of his examples
-- merging on ungrammaticality and '?' status is:

57. Rosa locked him in Ben's room.

Levinson suggests that "him" and "Ben" are not likely to be corferential
there. (p.307). If people just knew how we spend our weekends!

XXIV. POSTSCRIPT. UTTERER'S MEANING AS INCLUSIVE. A reductio ad absurdum of
Cavell.

Levinson notes that, for Grice, utterer's meaning must be
understood INCLUSIVELY (p. 381). In this connection, Levinson has the
following philosophical note that may please L. Tapper, who brought Cavell
to our pragmatic discussions:

Bach 1994:143 notes that a slightly consequence
of some defintions of "what is said" is that

58. Jones said that p.

ENTAILS

59. Jones meant that p.

in Grice's sense of meaning-nn. To avoid this, Grice is
forced to say that a speaker who intends a NON-LITERAL
interpretation of, for example, a trope, has _not_ _said_
that p, but merely "made it as if to say that p". Indeed,
ONE CANNOT, IN PURE GRICEAN TERMINOLOGY, TALK ABOUT
"SAYING ONE THING BUT MEANING ANOTHER INSTEAD! p.398

Wonder where Cavell fits in! And may the debate ensue...


-----

APPENDIX: Table of Contents of Levinson's book:


0. Formal conventions
0.1. Typographical.
0.2. Symbols & abbreviations.
0.3. Preface
1.Defining Grice's "generalised conversational implicature"
1.0. An argument for the very existence of "generalised convesational
implicatures"
1.1. H. P. Grice's original programme in _Logic & Conversation.
1.2. A three-layer versus a two-layer approach to communication theory.
(two layer: utterer's meaning vs. expression meaning; three-layer:
utterer's meaning; utterance _token_ meaning and utterance _type_ meaning).
1.3. Grice's nine conversational maxims as addressee-oriented "heuristics",
rather than cognitive principles.
1.4. Towards a motivated typology of generalised conversational implicature
1.4.1. Defining Grice's conversational submaxim Q1 ("be as informative as
is required") as "What isn't said, isn't"
1.4.2. Defining Grice's conversational submaxim Q2 ("Don't be more
informative than required") as "What is expressed simply is stereotypically
exemplified"
1.4.3. Defining Grice's conversational submaxims M1/M3 ("avoid obscurity of
expression" & "be orderly") as
"what is said in an abnormal way isn't normal"
1.4.4. Interactions of implicatures and clash of heuristics.
1.5. Non-monotonicity & reasoning by default.
1.5.1. Formal non-monotonic reasoning systems in general.
1.5.2. Conversational Implicature Seen as Non-Monotonic Inference
1.5.3. The defeasibility of Horn's SCALAR implicature
1.6. Some failed attempts to reduce generalised conversational implicature
to nonce utterer's meaning
1.6.1. Two problems with Sperber/Wilson's cognitive principle of relevance
1.6.1.1. Their model employs deduction, but conversational implicature is
notably non-monotonic.
1.6.1.2. Their model, if it makes any clear predictions, makes the wrong ones.
1.6.2. Other attempts: implicature as co-operative accomodation
(Mccafferty, Thomason -- based on DK Lewis)
1.7. Conversational implicature & its bearing on patterns of lexicalisation
2.1. The phenomena "to be saved": "the continuation of progress depends to
a large extent on the possibility of 'saving the phenomena'" (Grice,
Studies, p.379).
2.2. An examination of Grice's maxim Q1 ("What isn't said, isn't")
2.2.1. Its Associated implicatures
2.2.2. Entailment scales ("Horn Scales")
2.2.2.1. Quantifiers & modal operators: Horn's arithmetic Square of
Opposition & Altham's Pleonetetics.
2.2.2.2. Entailment scales over NON-logical predicates
2.2.2.2.1. The General Lexicon
2.2.2.2.2. Absolute Cardinality: A reply to Kempson.
2.2.2.2.3. Closed-class morphemes & function lexemes.
2.2.2.3. Contrasts based on Lexical Opposition
2.2.2.4. Scalar implicature: generalised or particularised?
2.2.2.5. An examination of CLAUSAL implicature
2.3. An examination of Grice's Q2 maxim ("What is simply described is
stereotypically exemplified"
2.3.1. Formulae
2.3.2. Some illustrations.
2.4. An examination of Grice's M1/M4 conversational maxims ("What is
abnormal isn't normal").
2.4.1. Larry Horn's division of Labour.
2.5. Joint Implicature: the Projection problem for Implicature.
3. Some progress in the semantics/pragmatics interface
3.1. The problem: do pragmatic factors intrude into truth-conditions?
3.2. The received Negative Reply.
3.2.1. H. P. Grice's view. Accounting for the "implicatural" side to
truth-conditions
3.2.2. Grice's First Diagnosis for "What is Said": Disambiguation
3.2.3. Grice's Second Diagnosis for "What is Said": Indexical Resolution
3.2.4. Grice's Third Diagnosis for "What is Said": Reference Assignment
3.2.5. Further diagnosis for "What is Said": Ellipsis Unpacking
A: Who came?
B: John (+> came).
3.2.6. Further Diagnosis for What is Said": Generality-Narrowing
3.2.7. A Way Out of "The Grice Circle": how can truth conditions be the
input for implicature _and_ viceversa?
3.3. A Review of INTRUSIVE "pragmantic" constructions.
3.3.1. Comparatives
3.3.2. Counterfactual Conditionals
3.3.3. Negation
3.4. The Ultimate Case for Pragmatic Intrusion: Referring
3.4.1. Implicatures that Determine Definite Reference
3.4.2. Implicaturally-determined reference & the Referential/Attributive
distinction
3.4.3. Assessing the Obstinate Gricean Theorist's Retreat.
3.4.4. Two Types of Gricean Pragmatics: pre-semantic & post-semantic.
3.5.1. Formalising the distinction.
3.5.2. Gricean pragmatics & Fodorean Modularity
3.5.3. Sag's AI-based proposal
3.5.4. Kadmon's Discourse Representation Theory
3.5.5. A synthesis
3.5.6. Finally Getting at _propositions_ & truth-conditions.
4. Intra-sentential anaphora & implicature: the case against Chomsky.
4.1. The Grammatical Phenomena.
4.2. Inferring Co-reference vs disjoint reference.
4.2.1. Local anaphora
4.2.2. Inferring co-reference
4.2.3. Inferring disjoint reference
4.3. Chomsky's Binding Theory
4.4. Anaphora in Language History & Language Learning
4.5. Towards a Generative Pragmatics
5. Retrospect & Prospect
5.1. Predictions
5.2. Presumptive meaning & general aspect of reasoning
5.3. Prospect.

References.

ALTHAM J. The Logic of Plurality. Methuen.
APPIAH A. DK Lewis & the conversational implicature defence. In Assertions
& Conditionals. OUP
ARISTOTLE. Categories & De Interpretatione, ed. & trans. by JL Acrkill,
Oxford Clarendon Press. Other citations from OUP edition of complete works.
Cited by Horn in Hall.
ARUNDEL, RB. Studies in the way of words. Grice's new directions in
conceptualising meaning in conversational interaction. International
Communiacation Association. Department of Communication, University of
Alaska, Fairbanks.
ATLAS, JD. Conversation. Oxford U.P.
AUSTIN JL. Philosophical Papers.'A Plea for excuses', cited by Grice in
'Prolegomena'). Ed. GJ Warnock and JO Urmson. OUP.
AVRAMDIDES A. Meaning & Mind: an examination of a Gricean account of
language. MIT (Reprint of her DPhil Oxon under PF Strawson).
. Grice & the social aspect of meaning. Essay presented to the
conference "The H.P.Grice Heritage", San Marino, ed. COSENZA.
BACH K. On the semantics/pragmatics distinction.
http://www.online.sfsu.edu/~kbach/semprag.html
. A synopsis of Conversational Implicature
http://www.userwww.sfsu.edu/~kbach/ci.htm
. The Myth of Conventional Implicature.
http://www.online.sfsu.edu/~kbach/Myth.htm
BARKER J A. Knowledge, ignorance & presupposition. Analysis 35.
BENNETT JF. In the tradition of Kantotle: review of Grandy, PGRICE. TLS 4 360.
. Symposium on the thought of Paul Grice. American Philosophical
Assocation. JPhilosophy 86. Contributors: Grandy, Baker, Stalnaker and Warner.
BIRO JI. Intentionalism & the theory of meaning. Monist 62. Discussed by
Suppes.
. A reply to Suppes. MS.
. Meaning & what is said. In O Neumaier, Mind language & Society.
Cocnetpus Studien, Vienna,
Section one: Grice's reduction.
BLACK M. Meaning & intention: an examination of Grice's views. New Literary
History 4. Repr. in his Collected Essays (Sections include: Grice's theory
of meaningfulness. G's conditions insufficient. G's conditions unnecessary!
G's way out. The circularity of G's approach).
BROCKWAY D. Constraints on interpretation. In The Legacy of Grice, ed HALL.
BURTON-ROBERTS N. Modality & Implicature. Linguistics & Philosophy 7.
CAREY K. The role of conversational implicature in the early
grammaticalisation of the English perfect.
In K Hall, The Legacy of Grice.
CAVELL S. Mus we mean what we say. Inquiry. Repr in _Must we mean what we
say_. Discussed in THIS FORUM by L.M. Tapper.
CARNAP R & Bar-Hillel Y. An outline of a semantic theory of semantic
information. MIT Technical Report 247, repr in Bar Hillel Language &
Information, Addison-Wesley.
CARSTON R. Implicature, explicature & truth-theoretical semantics. In A
Kasher, Conversational Implicature. London, RKP. Originally in R M Kempson,
Meaning and mental representation.
. A reanalysis of some quantity implicatures. MS. London UCL.
. Saying & implicating.
CLARK H. Understanding what is meant from what is said: a study in
conversationally conveyed requests. J Verbal Learn/Verbal Behav 14
COHEN LJ. Grice on the logical particles of natural language. In Y
Bar-Hillel, Pragmatics of natural language. Dordrecht: Reidel.
COSENZA G. H P GRICE'S HERITAGE. University of San Marino International
Centre for Semiotic & Cognitive Studies, Ex-Monastero Santa Chiara.
DAVIS S. Causal theories of mind: action, knowledge, memory, perception &
reference. Gruyter. Includes Grice 1961, The causal theory of perception.
. Pragmatics: a reader. OUP.
DAVIS W A. Implicature: Intention, Convention, and Principle in the Failure
of Grice's Theory. Cambridge Studies in Philosophy.
DONNELLAN KS. Speaker's reference, description & anaphora. In P. Cole,
Pragmatics. Academic Press.
DUMMETT M A E. Truth & Other Enigmas. London: Duckworth.
FAUCONNIER G. Invisible meaning. In K Hall, The Legacy of Grice.
FOGELIN R. Review of Grice, Studies. JPhilosophy
FRETHEIM J. The Effect of Intonation on Implicature. In A. Kasher,
Conversational Implicature,
London, RKP.
GAZDAR GJM. Implicature. Academic Press.
GRANDY R & R Warner, P.G.R.I.C.E.: Philosophical Grounds of Rationality:
Intentions, Categories, Ends. Oxford: Clarendon Press (essays by Grice, P.
Suppes, A. Kemmerling, D. Davidson, S. Schiffer, J Searle, P F Strawson, D
Wilson/D Sperber, J Hintikka, G Baker, R Grandy, J Perry, G Harman, G Myro,
A Code, N Cartwright, J Baker, and R Warner).
. Grice on language. in Bennett, Symposium on the Thought of Paul
Grice.
. On the foundations of conversational implicature In K Hall, The
Legacy of Grice.
GRICE HP. Meaning. PR. 1948
. The causal theory of perception. Aristotelian Society, 1961. In
Warnock, _The Philosophy of Perception_. Oxford readings in Philosophy.
. Logic & Conversation. 1967.
. Vacuous Names. In D Davidson & J Hintikka, Words & Objections.
1968. Discussion of Donnellan.
. Presupposition & Conversational implicature. 1970.
. Intention & Uncertainty. OUP. PBA 1971.
. Reply to Richards. In R. Grandy.
. Studies in the Way of Words.
. The Conception of Value. OUP.
. Aspects of Reason. OUP 2001.
GUMPERZ J J. Conversational cooperation in social perspective. In K Hall,
The Legacy of Grice.
HALL, K. The Legacy of Grice. Berkeley Linguistics Society, Parassession.
Vol. 16 of Proceedings.
HARE R M. Practical Inferences. Macmillan. discussion of Grice and Kenny's
Practical Inferences, Analysis Essay.
HINTIKKA K J J. Logic of conversation as a logic of dialogue. In
P.G.R.I.C.E., ed. R Grandy. Repr. in A. Kasher, Conversational Implicature,
London RKP.
HIRCHBERG J. A Theory of Scalar Implicature. New York: Garland. Reprint of
PhD Moore School of Electrical Engeneering University of Pennsylvania
Technical Report MS CIS 85-86.
HORN LR. Structure of logical operators. UCLA PhD
. Greek Grice: protoconversational rules in the history of logic.
. Hamburgers & truth: why Gricean explanation is Gricean. Berkeley
Parassesion, "The Legacy of Grice" in HALL.
. Toward a New Taxonomy for Pragmatic Inference: Quantity-based &
Relevance-based Implicature
In A Kasher, Conversational Implicature, London RKP.
JACK J M. The rights & wrongs of Grice on meaning. MS, Somerville College,
Oxford. Cited by Grice, Studies.
KANT, I. Critique of Pure Reason. Trans N K Smith. Lodnon Macmillan.
KASHER A. Ed. Implicature. RKP.
. Conversational maxims & rationality. Philosophica.
KEMMERLING A M. Utterer's meaning revisited. In P.G.R.I.C.E, ed. Grandy.
. Gricy actions. In H P Grice's Heritage Conference, San
Marino, ed. COSENZA.
KEMPSON RM. Conversational principles. In Kasher.
KENNY AJP. Practical Inferences. Analysis 26.
LEVINSON SC. Presumptive meanings: a theory of generalised conversational
implicature. MIT.
. Minimisation & conversational inference. In Kasher.
. with JD Atlas. What is a conversational implicature?
. 'Implicature'. Chapter III to his _Pragmatics. CUP.
LOEBNER S. Quantification as a major module of natural language semantics.
In J. Groenendijk, et al. Studies in Discourse Representation Theory & the
Theory of Generalised Quantifiers. Dordrecht: Reidel.(A study of "degrees
of tolerance" of quantifiers).
MCCAFFERTY AS. Reasoning about implicature: a plan-based approach. PhD
Universty of Pittsburgh, Pittsburgh, PA. Cited by Grandy.
MCCAWLEY JD. Conversational implicature & the lexicon. In A. Kasher,
Conversational Implicature
MORGAN JD. Formal Logic. 1847: London: Taylor & Walton.
. On the syllogism and on logic in general. _On the Syllogism &
Other Logical Writings_. 1858. London: Routledge & Kegan Paul.
NEALE S. HP Grice & the philosophy of language. Linguistics/Philosophy 15.
. Conventional implicature form the ground floor up. H P Grice's
Heritage conference, San Marin, in COSENZA.
. Obituary for Grice for the University of California, Berkeley.
NOWELL-SMITH PH. Ethics. Penguin
. Contextual implication & ethics. Aristotelian Society.
PATTON T E. The rudiments on meaning: on Ziff on Grice. Foundations of
Language 5.
PEARS DF. Motivated Irrationality. OUP.
PLATTS MDB. Review of R. Grandy, P.G.R.I.C.E. Mind.
RECANATI F. Truth-Conditional Pragmatics and primary Pragmatic Processes.
In A Kasher, Conversational Implicature. London RKP.
RYLE G. The Concept of Mind.
SADOCK JL. On testing for Conversational Implicature. In A. Kasher,
Conversational Implicature. London RKP
SAYWARD C A problem for conversational implicature. Linguistics/Philosophy.
SCHELLING T. The strategy of conflict. MIT.
SCHIFFER SR. Meaning. OUP.
. Compositional semantics & language understanding. In
P.G.R.I.C.E. ed Grandy
. Ceteris paribus laws. Mind.
SEARLE JR. Meaning, communication & representation. In Grandy, PGRICE.
SUPPES P. The primacy of utterer's meaning. In GRANDY. Section IV: 'Biro's
criticism of intentionalism".
SPERANZA JL. "!": Papers for the Grice Circle.
STALNAKER RC. On Grandy on Grice. In J Bennett, Symposium on the Thought of
Grice.
STEVENSON SC. Ethics & Language. New Haven.
STRAWSON PF. If & ->. In R Grandy, PGRICE.
TAYLOR T J. Gricean pragmatics & conversational principles. Oxford: Pergamon.
THOMASON RH. Accomodation & implicature.
TRAVIS C. Review of Grice, Studies in the Way of Words. Mind.
URMSON JO. Parenthetical verbs. Cited by Harnish.
VANDERVEKEN D. Conversational Implicature.
WALKER RCS. Conversational implicature: a reply to LJ Cohen. In S
Blackburn, Meaning, reference & necessity. CUP.
WARNER RO. Reply to Baker and Grandy. In J Bennett, Symposium on the
thought of H P Grice.
. Editor of Grice's _Aspects of Reason_, Oxford, 2001.
WARNOCK GJ. Saturday mornings. In The Philosophy of JL Austin, ed Fann.
Repr in Warnock, Language and Morals. Oxford: Blackwell (ref. to Grice,
"how clever language is!").
WAINER J. Uses of non-monotonic logic in natural language understanding.
PhD. Pennsylvania State Univesrity. Published by UMI Ann Arbor.
. & A Maida, Good & bad news in formalising implicature. In K
Hall, The Legacy of Grice.
WELKER K. Plans in the common ground: towards a generative account of
conversational implicature. PhD Ohio State University. Columbus.
WILSON DSM. On Defining relevance. In R Grandy, PGRICE
. Is there a maxim of truthfulness. In H P Grice's Heritage
conference, San Marino. Ed. Cosenza. . & D. Sperber, On Grice's
Theory of Conversation. In P Werth, Conversation & Discourse. London: Croom
Helm. Repr. in Kasher, Conversational Implicature.
. & D Sperber, Mutual Knowledge & Relevance in Theories of
Comprehension. Repr. in Kahser, A. Conversational Implicature. London, RKP.
WRIGHT RA. Meaning and conversational implicature. In Cole/Morgan, Speech
Acts, Academic Press.
ZIFF P. On Grice's theory of meaning. Analysis 28