Daniel W. Harris, A Griceian

By J. L. Speranza
for the Grice Club

In his talk on Grice, D. W. Harris, said he would "propose a Grice-inspired alternative" to stuff -- and he did!

The point is to see how Grice "avoids [some awful] commitments [to things] "by taking," instead, "the meaning of a linguistic type in a community to be constituted by the overlapping communicative dispositions of its members.

Harris stars by tackling "Grice’s Definition of Timeless Meaning"

As club members know, H. P. Grice, sometime Oxford philosopher, "offered" "his fullest constitutive account of timeless meaning in the penultimate Logic and Conversation lecture, and eventually settles on the following definition for the timeless meaning of a complete utterance-type (e.g., a sentence) for a group G.1, 2

(1 Grice (1989: 127). All references used by Harris to Grice are to (1989)).
(note 2. For Grice, linguistic meaning will be just one kind of timeless meaning, and there may be no principled basis for separating it from other varieties. For the purposes of Harris's study, it's best to be loose with the distinction).

---- GRICE WRITES:

“For group G,
utterance-type
X
means ‘*ψp’ ”
=df.
“At least some (many) members of
group G have in their repertoires the
procedure of uttering a token of X if,
for some A,
they want A to ψ† that p,

the retention of this procedure being for them
conditional on the assumption that at least
some (other) members of G have, or have had,
this procedure in their repertoires.”

---

Harris comments:

"For all its obscurity,", it's clear than most I've read on the topic. "I think," Harris thinsk, "this definition holds the solution to a variety of problems that plague mainstream theories of linguistic meaning."

Harris thus tries "to make good on this potential by interrogating

(a) Grice’s peculiar way of
formulating his definiendum.

(b) Grice's notion of someone’s having a
procedure in their repertoire, and

(c) Grice's notion of a group
of speakers or speech community.

Although Harris thinks that "Grice’s account generalizes nicely to what he calls incomplete utterance-types (including words and sub-sentential phrases), Harris says he won’t focus on how this generalization works, for the time being.

Instead, Harris will stick mostly to examples of sentence meaning in order to demonstrate the general character of Grice’s account.

Harris then goes to tackle "Attributions of Timeless Meaning."

Take note of the form of Grice’s definiendum—his schematization of an attribution
of timeless meaning.

X means ‘*ψp’

This formulation is painstakingly deliberate.

Earlier in the William James lectures (as repr. in "Utterer's meaning and intentions" (Philosophical Review) and "Utterer's meaning, sentence-meaning, and word meaning", Grice had argued that,

"whereas attributions of utterer’s occasion meaning should be articulated with a
that-clause (i.e., ‘In uttering X, U meant that P’), attributions of timeless meaning
to utterance-types should be given in terms of a quoted expression matching
the grammatical category of the expression to which meaning is being attributed."

Grice indeed explicitly claims that “[t]his difference is semantically important”
(118).

This claim about timeless meaning attributions reads naturally as a descriptive
claim about ordinary usage of one sense of the English verb, ‘to mean’.

Although Harris is not trying to defend this reading, he thinks the claim is interesting from the standpoint of understanding Grice’s views.

Even read as a stipulation, it sheds light on Grice’s understanding of the nature of timeless meaning.

That-clauses, after all, are commonly taken to refer to intentional contents.

In holding that the same wording is inappropriate for attributions of timeless meaning, Grice implies that a sentence’s possession of timeless meaning does not amount to its bearing any straightforward relation to a proposition.

And since Grice does not shy away from talk of propositions in general, the fact that he explicitly avoids such talk in this instance cannot reasonably be regarded as motivated by naturalistic concerns.3

So although Grice is committed to the idea that the timeless meaning of an utterance-type plays a role in constraining what a speaker can mean in producing an utterance of that type, he also seems to take for granted that timeless meaning is not a kind of content.4

Consider now the form of the second relatum in Grice’s definiendum. To wit:

*ψp

Grice calls the schematic operator variable ‘*ψ’ a “dummy mood indicator” (118)
or “mood marker” (110).

It

“corresponds to the propositional attitude ψ-ing
(whatever that may be), as, for example,

⊢p

corresponds to

believing (thinking)

and

!p

corresponds to

intending” (123).

Grice’s decision to separate out this element from the schematic sentence variable ‘p’ inside the quotes is puzzling, and this notation does not reappear in particular instances of timeless meaning attribution.

Moreover, the result is an unexplained technical use of quotation.

Harris finds it tempting to think that the result is unnecessarily confusing.

Grice’s motivation here is simpler than his technical machinery suggests.

Grice wishes to point out that the timeless meaning of an utterance-type constrains
more than just the contents of speech acts performed by utterances of that
type—it also constrains their illocutionary force.

Moreover, this constraining is tied to particular features of the utterance-type—features that Grice schematically represents as

*ψ

but which typically take the form of grammatical mood, intonation (in speech), and punctuation (in writing).

5

Harris then tackles "Having a Procedure in One’s Repertoire"

(3 For Grice’s liberal views on proposition talk, see (1989: 137). e point remains significant even if one suffers from a naturalistically-minded squeamishness about intensionality.

In that case, one would interpret that-clauses as reporting the truth conditions of speech acts and mental states.

And so to deliberately avoid such usage, and to point out that this avoidance is semantically significant, would be to suggest that timeless meanings lack truth conditions.

As far as Harris can see, Harris's use of ‘content’ and ‘propositional content’ are neutral between these interpretations.

4 See §6.

5 Harris develops this point in §6.

The other piece of novel machinery in Grice’s definition is

ψ†

which is a device used to abbreviate one of two complex constructions involving a variable

ψ

that ranges over propositional attitude verbs.

Grice’s definition of the operator can be found immediately preceding his definitions of timeless meaning (124), and his justification for positing its
ambiguity can be found in his revisionary discussion of the “M-intended effects” involved in meaning intentions (123).

Much of the explanatory burden of Grice’s proposal is carried by his idea of
utterers having a procedure in their repertoires to produce an utterance of some
type if they want to perform a certain type of speech act.

At first glance, this notion may look unexplained and vague.

But Harris thinks (and I agree) that this formulation carries several advantages.

One thing Grice gets in exchange for some vagueness is agnosticism (or scepticism) about the the psychological states that constitute semantic competence.

Aside from avoiding the practical pitfall of having his theory hang on future science, the chief reasons to avoid psychological speculation is that there are good reasons for thinking that semantic competence is multiply realizable.

That is, the relation that holds between an expression and an utterer just in case that expression has a certain meaning for the utterer could obtain in virtue of different sorts of facts about the utterer’s psychology.

Grice's MARTIANS.

Consider an alien being who is very similar to us in terms of its verbal behaviour,
but whose psychological makeup is very different from our own.

We might discover these differences by studying cross linguistic evidence, or evidence about how aliens of this species acquire language, or by some other method not directly tied to the alien’s present verbal behavior.

Harris thinks we should be very tempted to say that, since the alien uses linguistic expressions in the same way we do, the expressions have the same meaning for it.6

We can make a similar point without appealing to extraterrestrials.

Consider an utterer, call him U, who uses and understands the complex noun phrase ‘gender gap’ (or, as I prefer, 'Weltanschauung') in the familiar way, but who has learned the phrase as an idiomatic unit.

Utterer U does not know the words ‘gender’ or ‘gap’, (or "Welt" and "anschauung") at least in the sense that Utterer is undisposed to utter either word separately, and would fail to understand an utterance of either word outside the context of the complex phrase.

Although, in effect, Utterer treats ‘gender gap’ (and "Weltanschauung") as a single expression, Harris thinks it is reasonable to say that _sentences_ containing it
mean the same thing for him that they do for the rest of us (or German speakers, if you are one of them).

Nonetheless, it is very plausible that different psychological facts constitute Utterer’s competence with the phrase, 'gender gap' or "Weltanschauung".

Despite this advantage of Grice’s psychological neutrality, it is clear from his
struggles with the concept that even he found his ‘procedure’/‘repertoire’ formulation to be unfortunately obscure (126–128).

Harris thinks we can do better than Grice by talking about what Harris calls communicative dispositions instead.

"He is a runt".

Grice briefy considers replacing his talk of having a procedure to utter X in one’s repertoire with talk of being disposed to utter X, and then quickly rejects this formulation, after considering the example of his prim and proper Aunt Matilda, who might not

(6 If you’re not convinced about this for externalist reasons, see footnote 12.)

be disposed to utter a vulgarity ("He is a runt" -- for an undersized person) in any situation despite its having meaning for her (127).

In the end, Grice “abandon[s] the attempt to provide a de"nition” of having a procedure in one’s repertoire, “and content[s] himself with a few informal
remarks” (127).

Harris thinks "Grice stopped too soon." And we agree.

One needn’t look further than Grice’s “informal remark” about what’s going on with Aunt Matilda to see how he might’ve pursued a dispositional account further.

Grice considers the situation in which some utterance-type X
is current for some group G.

That is to say, to utter X in such-and-such circumstances
is part of the practice of many members of G.

In any case Grice's prim and proper Aunt Matilda, a Catholic convert, incidentally (a member of G) may be said to have a procedure for X, even though, to use Grice's phrase,

"she herself would rather be seen dead than utter ['He is a runt']"

For, yet, she knows that some OTHER members of her G, unfortunately, do have a readiness to utter X in such-and-such circumstances. [127–8]

In other words: although Grice's prim and proper Aunt Matilda is not disposed to utter a vulgarity such as "He is a runt" to mean an undersized person, under any circumstances, she is still disposed to interpret an utterance of a vulgar expression in the usual way, and 'runt' literally, too, and this is because she belongs to a speech community, unfortunately, in which, as things happen, _others_ are disposed to use it in that way.

Although Grice's Aunt Matilda lacks the performative disposition -- the disposition to utter the vulgar expression, 'runt' to mean, 'under-sized person' in order
to perform a certain type of speech act (to "insult", gratuitously) — she possesses the corresponding interpretive disposition —- the disposition to interpret utterances of the vulgar expression ("runt") as speech acts of the customary type.

With this case in mind, Harris submits that what it takes for an utterer U to belong to a speech community G for whom an expression X has a certain meaning is,
roughly, that at least one of the following two conditions to hold.

First,


(a) Utterer U is disposed to utter
X in order to perform speech acts
of certain types
when addressing members of G, and members of G are disposed to interpret
S’s utterances of X addressed to members of G as speech acts of those
types.

(including "insulting" other members of the group).

and

(b) Utterer U is disposed to interpret utterances of X addressed to utterer U by members of G as speech acts of certain types, and members of G are disposed to utter X in order to perform speech acts of those same types when addressing utterer U.

The resulting foundational theory of timeless meaning holds that the meaning of
an utterance-type in a community of utterers is constituted by the overlapping performative and interpretive dispositions (together, communicative dispositions) of
the members of the community with respect to utterances of that type.

Harris thinks this is what Grice should have said, not only because it seems to integrate the insight of his “informal remarks” back into the definition while maintaining the spirit of his overall approach, but also because it opens the door to a powerful account of the nature of speech communities.

-- Note that, to complicate things, -- it is a sophisticated type of vulgarity: 'runt' as interpreted by Aunt Matilda is 'metaphorical', besides being a 'vulgarity'. The metaphorical vulgar extension relies on the non-vulgar use of 'runt' as in 'runt of the litter'. It's only as applied to 'undersized person' that Aunt Matilda finds problem with 'runt', since it's a hyperbolic metaphor of the worst kind. --

It is after the sojourn with Aunt Matilda's vulgarities, that Harris tackles 'Convention'.

Consider the final clause of Grice’s definiens:

the retention of this procedure being for them conditional on the assumption that at
least some (other) members of G have, or have had, this procedure in their repertoires.

This clause represents the only aspect of Grice’s theory that we might be tempted
to describe as involving convention.

Of course, Grice uses the word ‘convention’ and its cognates in a casual way here and there in his writings —- notably in his discussion of the long defunct conventional implicature -- but as Neale and most points out, we should not
read too much into this usage (1992: fn12):

Grice’s use of the word ‘conventional’ in ‘conventional meaning’ should not be
taken too literally, and it is rather 'conventional', if we can be autological, for it is Grice’s view that linguistic meaning is not to be
explicated in terms of what other philosophers might think of as convention.

He says it's OPTIMALITY that explains meaning, never convention. (Grice's example of his "Deutero-Esperanto").

The ‘other philosophers’ Neale has in mind are D. K. Lewis and those who have
followed him in giving a highly precise and intellectualized account of convention.

--- including Schiffer and Loar.

According to the Lewisian account, the fact that

an expression X means that p in a
community of utterers is constituted by the fact that the community
bears an actual language relation to a function from expressions to meanings (a
language) which assigns p to X.7

A community bears the actual language relation
to a given language just in case there prevails
a regularity to use the language
that has become conventionalized in the community.8

And a regularity

R

counts as a convention in a group G, according to Lewis,

just in case the following conditions hold:

(i) Everyone in G believes that
everyone else in G acts in accordance
with R.

(ii) there is a general preference
among utterers and addressees in G for conformity
to R;

(iii) the fact that R exists
gives everyone in G a reason (together

(7 Note that public languages in Lewis’s sense are not folk-theoretic languages like French and English.

On the view Harris favours it may be possible to offer some notion of a community of
English speakers that would do extensional justice to the folk, but Harris's theory of meaning is delationary about such notions insofar as they play no ineliminable explanatory role.

8 Just what sort of regularity we’re talking about is a matter of debate. According to Lewis, the actual language relation is built on a conventionalized regularity of truthfulness and trust -- a convention to utter sentences only if they are true in the language, and to interpret other members
of the community as doing the same. Schiffer (1972, 1993, 2006), Bennett (1976), Dale and others have suggested that the relevant sort of regularity must be spelled out in terms of the Gricean meaning intentions with which utterers utter expressions.
with their other goals—in particular their interest in communication) to continue
conforming to R;

(iv) There is at least one other possible regularity R′ that
would serve the purposes of the members of G just as well if they were to conform
to it instead; and

(v) it is common knowledge among members of G that
(i)–(iv) obtain.
9

This account is, as Harris notes, "plagued with difficulties", and it's no wonder Grice rejected it en bloc.

Some of these arise from Lewis’s highly intellectualized account of convention. The blame is Quine, his tutor.

Lewis’s theory commits him to the view that utterers possess a surprisingly rich variety of propositional knowledge, both about their own beliefs and behaviors, and about those of the other members of the speech communities to which they belong.

According to the late professor Lewis, my ability to verbally communicate with you is explained, in part, by the fact that we are a members of a community that stands in the actual language relation to some language L.

And this entails that we both know (and know that we know, etc.) that each of Lewis’s conditions (i)–(iv) obtains with regards to our
community and L.

But aside from general worries about over-intellectualizing semantic competence, there are more speci"c worries about the theory’s psychological predictions.

Assuming that you and I have begun to converse aer having just met, do I really know (and know that you know, etc.) that you behave according to a specific regularity with respect to some Lewisian language L?

It seems not.

Even aside from problems about whether I could be justified, quite
generally, in holding beliefs about the higher order beliefs about someone I have
just met, it seems quite clear that, at best, I could only have a very rough idea of
the language that you speak.

How could I know that you conform to the relevant regularity with respect to L, and not, say L*, which differs from L only with respect to expressions I have never heard you utter?

In short: theories of convention fail to account for the fact that speakers generally manage to operate on partial and imperfect information about one another’s linguistic competence.

A related problem arises from the fact that any given speaker routinely
switches back and forth between a wide variety of heterogeneous styles, registers,
and dialects in the course of conversing with shiing interlocutors.

Sometimes, these heterogenous ways of using language directly con&ict with one another, as in the auto-antonymous uses of ‘livid’ and ‘nonplussed’ in respectively
more and less traditional or proper dialects of English.

Moreover, these shifts in usage tend to occur quickly, as the result of shifts in social setting or addressee.

The Lewisian’s only way of explaining such shifts in usage is to say that they involve a wholesale change in the language being spoken.

On this view, I may speak several different languages over the course of a single conversation, and a

9 Lewis (1969, 1975).

staggering number over the course of my whole life.

This consequence of the Lewisian view not only illustrates the seriousness of the epistemic problems raised in the last paragraph, it also exposes the Lewisian theory’s inability to recognize the intuitive connections between different ways of speaking the same (folk-theoretic) language, such as English, or Greek.

Even the squarest Midwestern housewife can at least get the gist of Lil’ Wayne’s latest hit.

But clearly, there is no single function from expressions to meanings such that she and and Wayne stand in the actual language relation to it.

These epistemic problems obscure the metaphysical problem of whether there
is even a fact of the matter about which language is spoken by a given community.
Hawthorne (1990, 1993), Lewis (1992), Schiffer (1993, 2006) and R. E. Dale, have
dubbed this the meaning-without-use problem.

Due to the productivity of natural
languages and the "nitude of human discourse, there are many sentences that
have meaning for a group of speakers despite there being no speaker in that
group who has ever used the sentences.

This poses a problem for the Lewisian account, since it follows that speakers of natural languages stand in the actual language relation to an infinite number of languages, all of which agree in their assignments of meanings to those expressions actually in use, but which diverge with regard to the indefinitely many remaining expressions.

And this conclusion seems incompatible with the fact that these remaining expressions have possess meanings.

The late Prof. Lewis attempts to deal with this problem by arguing that the language determined by any particular spoken fragment is speci"ed by the unique extrapolation
of whatever grammar applies to the fragment.

The idea of an extrapolation of a grammar, according to Lewis, “presupposes the distinction between straight and bent” grammars (1992: 151).

Lewis doesn’t offer, alas, a full definition of
what counts as a straight grammar, but he suggests that “any grammar that any
linguist would actually propose” would be included (150).

This distinction between
straight and bent grammars would rule out grammars specifying languages
following straightforward compositional principles that break down only
for sentences above one thousand words in length, for example.

Clearly, Lewis’s straight/bent account is worryingly impressionistic, and
Schiffer (2006) persuasively argues that these worries are justified.

Schiffer proposes that we replace it with a psycholinguistic account (286):

What makes L a language that is used by x even though the used fragment of
L is also a fragment of infinitely many languages other than L is that it’s the
language determined by the internally represented generative grammar implicated
in x’s language-understanding processes.

This theory has its own fatal defect.

Since speech communities are heterogenous, in the sense that no two members of a community share an idiolect, Schiffer’s solution to the meaning-without-use problem renders his specification of the actual language relation unsuitable for a theory of meaning for public, folk-theoretical languages, such as English or "Griceish".

Any appeal to facts about an utterer’s individual psychology would require arbitrarily focusing on a single community member’s idiolect while ignoring others.

Even if we could cook up an account of the actual language relation capable
of dealing with the meaning-without-use problem, the gross epistemic problems
with Lewisian convention discussed above would be waiting to ruin the party.

Many philosophers have recognized these problems, but few have attempted to
replace the Lewisian account with anything falsifiably precise.

----- [vide Speranza, tackling it in terms of Grice's optimality guise of value, vol. 5,]

Some philosophers have taken
the notion of convention and (public) language to be primitive concepts in their
discussions of linguistic meaning, or have shrugged off the difficult conceptual
problems in the hopes that a satisfactory theory will some day come along.10

Others have forcefully argued that no such account could ever emerge.

The
most prominent of these skeptics is Noam Chomsky (1980, 1986), who holds
that a concept as imprecise and politicized as public language should be given
no place in serious theorizing about language, and that although linguistic conventions may exist, they fail to explain anything interesting about the nature of
language.11

Although Harris is tempted to tentatively agree with Chomsky, it is be-

(
10 Unexplained references to public languages and linguistic convention abound in the philosophy
of language.

Harris has Michael Devitt in (Harris's) mind as someone who sometimes urges that we may take convention as a primitive.

Devitt's discussion of meaning in public languages (for example, in
§10.5 of Ignorance of Language) winds up looking a lot like the account Harris gave in his study, albeit with
more emphasis on a notion of convention that is loosely explained to the point of nearprimitiveness.

Moreover, although Devitt insists on the intelligibility of a relatively folk-friendly notion of public languages, it is not clear (at least to Harris and to me) whether that notion plays any explanatory role in his foundational theory of public meaning. E.g.:

“Suppose that speakers in the community share a literal meaning. If this sharing is partly explained by the appropriate causal relations between speakers dispositions, then that literal meaning will be the conventional one in the community”
(2006: 179).

This suggests a view according to which the conventions that are constitutive of an
expression’s having the meaning that it does are merely those that govern that very expression, i.e. without the holistic detour through public languages.

This might get around some of the problems Harris raises, though it is worth noting that since the conventions governing complex linguistic types consist of both the lexical conventions that apply to individual words contained within them as well as the recursive conventions for applying them, it may turn out that an average
sentence exemplifies a very rich body of linguistic conventions, and there would still be analogous worries about the ways in which these conventions would apply to hitherto unfamiliar expressions.

(11 Also Donald Davidson (1984, 1986, in PGRICE) offered some sad arguments that linguistic meaning is not essentially conventional).

yond my current ambitions to argue his ambitious point.

Suffice it to say that any theory of linguistic meaning that deals in public languages and linguistic conventions gambles on the hope that these deeply problematic and as-yetunexplained concepts will one day be illuminated.

What, then, should we make of the last clause of Grice’s definition of timeless
meaning?

It does seem clear that, with this clause, Grice intends to gesture at the
idea that linguistic meaning comes about and is sustained by something like
convention.

But, as Neale points out, Grice’s version of convention is much weaker than Lewis’s.

It does not require everyone in a speech community to participate in the relevant regularity; it requires only that most members of the community to be aware of some of the other members’ participation.

And it makes no mention of common knowledge, which Grice (and I) know (mutually) nothing about.

Most importantly, Grice’s account does not deal in public languages.

And so, insofar as Grice can be said to posit linguistic convention at all (an inane verbal issue of the type I rejoice in), his account is not encumbered by the problems associated with Lewisian theories of linguistic meaning.

Nonetheless, Harris still thinks that it is worth asking whether Grice’s last clause should be there at all.

Or in some other place.

One problem is that, in stating that the retention of a meaning procedure
should be

“conditional on the assumption [emphasis ours] that at least some other members of G have, or have had, this procedure in their repertoires,”

Grice’s definition contains a glimmer of over-intellectualism that infects Lewis’s
theory of convention.

But then he WAS an intellectual.

To say that an expression’s having meaning in my speech community depends on my having an assumption about the procedures and repertoires (or communicative dispositions) of other speakers still seems to place an unlikely cognitive burden in the heads of language users.

Harris, for example, thinks we could allay this worry by replacing ‘assumption’ with the less cognitive-sounding ‘appreciation’.

The idea being that it may make sense to say that I appreciate your communicative dispositions without entertaining (or even being disposed to entertain) any thoughts about them.

Devitt seems to be on the right track when he says that a regularity of language use should count as convention if the participants’ shared use “is partly explained by the appropriate causal relations between the utterers and addressees’ dispositions.”

Devitt continues:

"it is hard to say precisely what causal relations are appropriate for the shared literal meaning to be conventional, but the center of what has to be said is that any speaker has her disposition because other speakers have theirs and hence regularly use the expression with that meaning. [2006: 179–180]

Defined in this very loose and minimal sense, it is difficult to see how Devitt
could be wrong that many linguistic conventions exist, and that some fleshed-out
theory of them would go a long way toward explaining why most speakers
possess the communicative dispositions they do.

Nonetheless, Harris is still not THAT convinced that linguistic convention, even in this minimal sense, should have a role to play in constitutive accounts of linguistic
meaning.

The appeal to convention purports to explain how speakers come to be communicatively disposed as they are, and why they stay that way.

But so long as an utterer is disposed in the right way, should these details make any difference to an account of the metaphysics of meaning?

In other words: are the causal origins of linguistic meaning essential to its nature?

Harris thinks there are "at least some reasons to DOUBT such an [incredible] thesis."

Take a freshly-programmed English-speaking android, or an alien born with fully-formed linguistic competence much like our own, or a popper of Latin pills, or any other hypothetical individual capable of garden-variety linguistic communication, but whose communicative dispositions have unconventional origins.

The idea of excluding these individuals from speech communities consisting of individuals with whom they could carry on normal conversations is ad hoc.12

So even if most or all actual humans are communicatively disposed as they are due to their participation in conventions in at least Devitt’s minimal sense, and even if this minimal version of linguistic conventions don’t raise any of the problems faced by Lewis, the possibility of non-conventional meaning pushes Harris to think that linguistic conventions should be kept out of the constitutive picture.

It is only THEN that Harris goes on to tackle "Speech Communities".

After all, the Gricean account of meaning Harris has reconstructed and extended appeals to speech communities, and so some explanation of their nature is needed in order to shore up the foundations of the theory.

A natural criterion for the individuation of speech communities falls out of the idea that the meaning of an expression in a community depends on the way in which the communicative dispositions of its members overlap.

Utterers and addressees conversationally overlap to a minimal ex-

(12 ere is, of course, a literature on this sort of thing, springing from Davidson’s (1987) Swampman thought experiment.

Externalists will be tempted to deny that a Latin pill could allow a person to really speak Latin. Harris is not sure where he stands on this kind of issue when it comes to intentional content, but he thinks the line of argument is plainly mistaken once the different explanatory roles of intentional content and linguistic meaning are appreciated.

Brie&y: if the explanatory project of the theory of meaning is ultimately grounded in providing an answer to Neale’s Master Question — “How are we able to accomplish so much…by making various noises or marks” (2011: 4)?—then swampman-type worries about linguistic meaning seem to be ruled out on methodological grounds.

It is easy to imagine that if there were a latin pill, it would quickly become the only way anybody came to speak Latin (why bother reciting all that
Ovid?).

The issue of whether or not they would be speaking “real” Latin is silly.

Clearly it would be just as worth while to answer the Master Question about them as about us, and clearly, if the concept of linguistic meaning is worth positing in our case, it is also worth positing in theirs)

tent if they have corresponding communicative dispositions.

U’s performative disposition PD corresponds to A’s interpretive disposition ID (and vice versa) just in case PD is a disposition to address utterances of some expression X to A in order to perform speech acts of type Y, and ID is a disposition to interpret utterances of X by S as Y-type speech acts.

Given this setup, we can say that there are very many speech communities—as many as there ways in which speakers conversationally overlap.

This account conveniently falls out of the theory Harris lays down without
commitment to any metaphysical building blocks other than the communicative
dispositions already in play.

Plus, the resulting theory of speech communities has significant explanatory power.

An utterer’s communicative dispositions can vary greatly from one conversation to the next, depending on who is involved and what the speaker believes about them.

As a result, speech communities and linguistic meaning both have a distinctly protean character.

This character represents a very difficult problem for Lewisian theories of meaning, but I think that the Gricean model can explain it.

The communicative dispositions of a speaker are presumably determined by a
variety of psychological factors that Harris divides into two kinds.

I.

On one hand, utterers’ communicative dispositions depend on what Harris calls hard
determinants.

Hard determinants are factors that typically don’t change from one conversation
to the next -- paradigmatically, an utterer’s standing linguistic competence.

If I am a monolingual English speaker and you are a monolingual German speaker,


(13 this places severe limitations on the degree of conversational overlap we can
hope to attain, short of taking language classes. A speaker’s conversational dispositions in any particular situation also depends on so determinants—paradigmatically, speakers’ beliefs about each other’s linguistic dispositions, including their beliefs about what they take to be the hard and so determinants
thereof.

If I am a monolingual English speaker and I believe you to be a monolingual
German speaker, my performative dispositions toward you will be
severely constrained, and I may be less likely to correctly identify your English
utterances, should my beliefs about you turn out to be wrong.

All of this follows from Grice’s theory of speaker meaning.

According to Grice, what an utterer means by an utterance is constituted by the communicative intentions with which he produces it.

For an utterer to mean something in producing an utterance, U must have both an informative intention to bring about an effect (e.g. a belief) in an addressee, and a communicative intention to 12

13 Since the theory Harris presents doesn’t deal in public languages, the idea of a monolingual speaker of some public language is meant to be taken in a pre-theoretic way.)

bring about recognition of her informative intention.14

But, as Grice (1971) and others have pointed out,15 one cannot intend to do what one believes to be impossible.

If I believe you to be a monolingual German speaker, or even perhaps if I believe
you to have inaccurate beliefs about my own linguistic competence, I will
be likely to believe that you would incapable of working out my communicative
intentions were I to speak in English, in which case I will be incapable of forming
those intentions.

Needless to say, this will curtail my performative dispositions with respect to you.

This situation gives rise to a fascinating dynamic. Or not.

Among the most interesting determinants of an individual’s communicative dispositions are his beliefs, tacit or otherwise, about what he takes to be the hard determinants of his interlocutors’ communicative, and ethical, dispositions.

Such beliefs may be absent entirely, as in the case of an isolated and linguistically uniform society.

That is, they may be rough-grained, as in the case of an individual striking up a conversation with a strangerr.

But they may be fine-grained, as in the case of one secret agent speaking
to another in code.

In any case, it seems clear that our ability to intelligently adjust our communicative dispositions from one conversation to the next entails
that we have at least some tacit appreciation of our interlocutors’ linguistic competencies.

At the same time, it is worth noting that most speakers’ beliefs about
each other’s hard determinants are not genuinely explanatory of their communicative
abilities.

When I strike up a conversation with a stranger, my belief that he
speaks English may license my performative dispositions toward him.

If I believed he did not speak English, I would approach the interaction differently, or not at all.

But for the reasons I’ve outlined in §4, relations borne by speakers to
public languages contribute nothing to explanations of their abilities to communicate
linguistically.

So although the contents of our beliefs about each other’s linguistic competences are generally unexplanatory, they still play a crucial role
in communication by giving rise to roughly the right sorts of communicative
dispositions.

This dynamic should be particularly fascinating for theorists because these
same unexplanatory beliefs also play a role in the intuitions and inferences that
guide our theorizing.

Given how crucial these beliefs actually are to our communicative practices, it is only natural that we would tend to treat them as obvious in our philosophical reasoning about the nature of language.

But although

For reasons
which have resulted in the vain deaths of many trees, it probably should not be taken as a definition of speaker meaning, but rather as a pair of necessary conditions.

Influential alternatives are found in Grice (1989: chs. 5 and 14), Schiffer (1972), Bennett (1976), and Neale (1992).

15 See Grice (1989: 125) and Neale (2004: 77; 2005: 181; 2011: fn.62).

folk-theoretic concepts of language and linguistic convention may play a significant
role in guiding our communicative dispositions, and may therefore be unavoidable
day-to-day for practical purposes, we should not therefore conclude that the folk theories must be on to something.

Indeed, if Harris's arguments are considered, such theories, as well as their more technical Lewisian cousins, are deeply mistaken.

In the final section of his talk, Harris referred to "The Blueprint Theory of Meaning:.

That is, a brief discussion of the relationship of semantic theory to
its foundations.

Although a given theory of the facts constituting linguistic meaning needn’t fully determine a theory of the nature of meaning itself, it should be clear that some kinds of foundational theories are much more amenable to particular theories of meaning than others.

The Lewisian foundational theory, for example, in taking public languages to
be functions from linguistic types to meanings, tends to comport with the view
that for an expression to mean such-and-such is for it to stand in some relation
to such-and-such a meaning entity.

For the late prof. Lewis, who conceived of the meaning of a sentence as a proposition, this was natural enough.

But, since the 1960’s, a mountain of work on the context-sensitivity of natural language expressions has made it crystal clear that there is no such clean relationship between linguistic types and meaning entities.16

Within a Gricean theory of communication, the properties of a communicative act that an addressee must identify in order to successfully interpret it are determined by the utterer’s addressee-directed intentions.17

Since the forming of these intentions is impossible if the utterer believes they won’t be realized, a reasonable utterer will be capable of felicitously performing a communicative act only if he has provided his addressee with adequate information with which to infer his intentions.

In a Gricean theory, this provision of information associated with a linguistic type, by a speaker to her addressee, thus constraining both the interpretive options available to the interpreter and, thereby, the intention-forming
options of the speaker, exhausts the explanatory role of timeless meaning
in general, and linguistic meaning in particular.

Given the idea that linguistic meanings are just the sorts of things—propositions—
that speakers express in performing speech acts, it is very natural to sup-

(16 Perhaps, as Schiffer (2005: 120) claims, this “had always been obvious”. Nonetheless, many important works in the philosophy of language have revealed the width of the gulf between sentences and the propositions speakers express using them.

(17 Grice is quite clear that these properties go beyond what the speaker says. See §3 and below.

pose that the linguistic meaning of a sentence constrains the content of a speech
act performed by uttering that sentence by determining it—indeed, by being
identical to it.18 It is worth noting right here that for a property A to determine a
property B by being identical to it is the strongest possible way, to the point of
triviality, that A can constrain B.

There are weaker ways that one property can constrain another.

Various features of a recipe constrain the properties of the dishes cooked according to the recipe without being identical to those properties, and without even fully determining them, for example.

Likewise, various properties of a blueprint constrain the properties of a building constructed according to it without being identical to or even fully determining them.

(cfr. Speranza on the optimality guise of value as applied to meaning. What is optimally to utter to optimally mean that p).

The aforementioned literature on context sensitivity shoots down the idea
that the constraints placed by linguistic meaning on speech act content are this
strong.

With this in mind, epicycles on the Lewisian conception of meaning
have emerged.

The opus classicus of this Kaplan’s ‘Demonstratives’, wherein he
argues that the linguistic meaning of a sentence is a function from a few contextual
parameters to the proposition a speaker would express with the sentence
relative to that context.

Kaplan calls these functions characters.

As it turns out, the contextual parameters relative to which linguistic meaning determines speech act content have not been forthcoming, and so some philosophers have replaced the idea of contextual parameters with a liberal use of the utterer’s
intentions.19

The reasons I call this view an epicycle are twofold.

First, it evinces the attempt to maintain as strong a relationship between linguistic meaning and speech act content as possible—constraint by determination, just with the help of some context.

Second, it plays into the Lewisian foundational story by taking the meaningfulness of an expression to consist in its standing in a relation to a meaning entity—a function from context to content.

What it ignores, Harris notes, is the possibility that linguistic meaning is just not the same sort of thing as intentional content—that it is not a thing at all—and that it constrains speech act content, not by determining it (relative to some parameters, or with the addition of fur-
15

(18 Michael Devitt argues that “what it is for a certain expression to have a certain literal meaning in a person’s idiolect” is, in part, for the person to be “disposed to use that expression to express a concept (a part of a thought) with that meaning” and to be “disposed to interpret that expression by assigning a concept with that meaning to it” (2005: 179).

This passage anticipates the views espoused in §4 of this paper. The main difference—a significant one, Harris thinks—is Devitt’s apparent commitment to the idea that the meaning of an expression is the same sort of thing as that which a
speaker means in uttering the expression.

19 A variant of this view conceives of linguistic meaning as a kind of “gappy content” or “proposition schema”. Guiding metaphor aside, the idea is the same: meaning determines content with the aid of something else, usually speaker’s intentions.

ther ingredients), but merely by guiding interpreters as they infer the properties
of speech acts. e guiding metaphor of the view I propose is the blueprint,
which steers or guides the contractor’s reconstruction of the architect’s intentions
without determining, even relative to a context or with something extra added
in, the final structure.

At this stage, the defender of the gappy content view will likely accuse me of
engaging in a dispute over mere metaphors. What is the substantive difference
between a “character” and a “blueprint”?

Harris thinks that there are good methodological reasons for preferring the blueprint metaphor.

And that it is more parsimonious and less confusing, but Harris said he would save these arguments for a longer day.

The most signi"cant problem I see for content-theoretic accounts lies in what
they usually don’t even purport to explain.

What Harris has in mind are the aspects of an expression’s linguistic meaning that
outstrip its role in constraining merely the contents of speech acts.

Imagine that you say to me:

‘But go screw yourself ’.

I won’t have succeeded in interpreting you unless I come to appreciate

(a) that you have issued a command,

(b) that you have expressed a negative attitude toward me,

(c) that you have done so rudely,

and

(d) that you have implied some sort of contrast between this comment and
the previous one.

Each of (a)–(d) involves my recognizing a property of your speech act other than its content.

Moreover, the inferences I must perform in order to accomplish each of (a)–(d) are clearly guided or steered, to some extent, by information I glean from the particular words and syntax you’ve used, in much the way that my inferences about the content of your act are soconstrained.

In short: we have good reason to think that there are a wide variety of ways in which an expression’s meaning constrains properties of the speech acts performed with it, in much the same way that meaning constrains content.

It may be possible to accommodate this observation within the Lewisian foundational framework by conceiving of meanings as multifaceted entities that partially determine each of the constrainable properties of speech acts.

Schiffer (1993, 2003: ch.3) and R. E. Dale toy with an idea like this when they propose -- if we have understood their views alright -- that the meaning of a sentence might be thought of as an ordered pair consisting of a Kaplanian
character (or something close to it) followed by a type of illocutionary act.

In order to generalize this proposal, we might think of a sentence meaning as an
ordered n-tuple of constraint properties, each of which delimits some dimension
along which speech acts vary.

But Harris doesn't think that this proposal can work, alas -- (He said "alas" in scare quotes).

Consider those features of a sentence, such as mood and intonation, that guide interpreters in identifying illocutionary force.

In what way do these properties constrain illocutionary force?

They do not determine a speech act’s force, even relative to any stable contextual
parameters—my ability to use indicative sentences to ask questions does not
hinge on some particular contextual parameter(s).

Moreover, if the Kaplanian characters of grammatical moods take as their arguments only the speaker’s intentions -- for example, if the character of the indicative mood amounts to the rule that sentences of that mood should be interpreted however the speaker intends -- the whole notion that mood constrains force is trivialized.

Harris's blueprint model, together with the foundational theory he proposes, can easily assimilate all aspects of meaning to the same basic model.

A given property of a speech act is constrained by the meaning of the sentence
with which it is performed because speakers are communicatively disposed to
associate that type of sentence with speech acts possessing that property,
whether the property has to do with content, force, etiquette, conventional implicature, or emotional expression.

All these constraints are defeasible to varying degrees because the dispositions underlying them are more or less easily overcome.20

The result, I think, is a theory with broader explanatory potential than Lewis’s, or anybody else's for that matter.

(Harris thinks that the constraints that meaning places on content are likewise defeasible to some degree).


REFERENCES

Almog, J., J. Perry, and H. Wettstein

"Themes from Kaplan"

Oxford University Press.


AUSTIN J.L. (1962): How to Do things with Words (Harvard University Press), ed. by Warnock and Urmson.

Bach, Kent (2005): ‘Context ex Machina’ in Zoltán Gendler Szabó (ed.): Semantics
versus Pragmatics (Clarendon Press): 15–44.

Bennett, Jonathan (1976): Linguistic Behavior (Cambridge University Press).

Chomsky, Noam (1980): Rules and Representations (Columbia University Press).
———. (1986): Knowledge of Language: Its Nature, Origin, and Use (Praeger).

Dale, R. E. The theory of meaning. Online.

Davidson, Donald (1984): ‘Communication and Convention’ in Inquiries into
Truth and Interpretation (Oxford University Press).
———. (1986): ‘A Nice Derangement of Epitaphs’ in PGRICE, Philosophical grounds of rationality: intentions, categories, ends.
--- Truth, Language, and History
(Oxford University Press, 2005): 89–107.
———. (1987): ‘Knowing One’s Own Mind’ in Subjective, Intersubjective, Objective
(Oxford University Press, 2001).

Devitt, Michael (1996): Coming to Our Senses (Cambridge University Press).
———. (2006): Ignorance of Language (Oxford University Press).

Grice, H.P. 1938. Privation and negation.
-- 1941. Personal identity. Mind
-- 1948. Meaning.
-- (1989): Studies in the Way of Words (Harvard University Press).
-- 1991. The conception of value.
-- 2001. Aspects of reason.

Harris, D. W. Themes from Grice.

Hawthorne, John. 'A Note on “Languages and Language”’ in Australasian
Journal of Philosophy 68

———. ‘Meaning and Evidence: A Reply to Lewis’ in Australasian Journal
of Philosophy 71

Kaplan, David
Demonstratives’ in (Almog et al. 1989): 481–563.
———. (1989b): ‘Aerthoughts’ in (Almog et al. 1989): 565–614.

Lewis, David. Conventions of language. PhD, Harvard. (Harvard University Press)
———. (1970): ‘General Semantics’ in (Lewis 1983): 189–229.
———. (1975): ‘Languages and Language’ in (Lewis 1983): 163–88.
———. (1983): Philosophical Papers, Volume I (Oxford University Press).
———. (1980): ‘Index, Context, and Content‘ in Stig Kanger and Sven Ohman
(eds.) Philosophy and Grammar (Reidel).
———. (1992): ‘Meaning without Use: Reply to Hawthorne’ in Australasian
Journal of Philosophy 70: 106–10.

Neale, S. Grice. Linguistics
and Philosophy 15: 509–559.
———. (2004): This, that, and the Other’ in
Marga Reimer and Anne Bezuidenhout (eds.): Descriptions and Beyond (Oxford University Press): 68–182.
———. (2005): ‘Pragmatism and Binding’ in Zoltán Gendler Szabó (ed.): Semantics
versus Pragmatics (Clarendon Press): 165–286.
———. (2011): ‘Implicit Reference’ in Gary Ostertag (ed.): The Philosophy of
Stephen Schiffer (MIT Press).

Recanati, F. (2004): Literal Meaning (Cambridge University Press).

Searle, John (1968): Austin on Locutionary and Illocutionary Acts in e Philosophical
Review 77 (4): 405–424.
———. ‘Literal Meaning’ in Erkenntnis 13 (1): 207–24.

Schiffer, S. R. 1970. Meaning. DPhil, Oxon (Under G. Ryle)
-- (1972): Meaning (Oxford University Press).
-- 1986. In PGRICE
— (1993): ‘Actual Language Relations’ in Philosophical Perspectives 7: 231–
58.
———. (2003): e ings We Mean (Oxford University Press).
———. (2006): ‘Two Perspectives on Knowledge of Language’ in Philosophical
Issues 16: 275–287.

Speranza, J. L. Join the Grice Club!

Zermelo, Complete works.