in Asa Kasher, ed., Pragmatics:
Critical Assessment (London: Routledge, 1998)
How to delimit semantics is an ongoing problem in linguistics and
philosophy of language. Like syntax, semantics is concerned only with
information that competent speakers can glean from linguistic items
apart from particular contexts of utterance. Anything a hearer infers
from collateral information about the context of a particular utterance
thus counts as nonsemantic information. Even so, it is a semantic fact
about certain linguistic items, notably indexicals (such as 'she',
'here', and 'then'), that contextual facts contribute to determining
what they are used to refer to. Although it is arguable that indexical
reference is not, in general, a strict function of context (Bach 1987,
pp. 175-186), contextual sensitivity is a linguistically marked feature
of indexicals. So the context sensitivity of an expression is not
itself a contextual fact about that expression. On the other hand,
there are certain context-independent facts about expressions that are
not matters of linguistic meaning.
This is where standardization comes in.
What is standardization? A form of words is standardized for a
certain use if this use, though regularized, goes beyond literal
meaning and yet can be explained without special conventions. In each
case, there is a certain core of linguistic meaning attributable on
compositional grounds but a common use that cannot be explained in
terms of linguistic meaning alone. The familiarity of the form of
words, together with a familiar inference route from their literal
meaning to what the speaker could plausibly be taken to mean in using
them, streamlines the process of identifying what the speaker is
conveying. The inference is compressed by precedent. But were there no
such precedent, in which case a more elaborate inference would be
required, there would still be enough contextual information available
to the hearer for figuring out what is being conveyed. That is why
special conventions are not need to explain these cases. So
standardization is different not only from what Paul Grice called
"particularized" conversational implicature but also from conventional
implicature and other sorts of conventionalization, such as dead
metaphor and idiomatization.
Standardization is illustrated by what Grice calls "generalized"
conversational implicature. Whereas "particularized" implicatures
exploit "special features of the context" to enable the audience to
identify what the speaker is conveying (taken at face value the
utterance is not (or is not sufficiently) truthful, plausible,
informative, relevant, or otherwise appropriate), generalized
implicatures are carried by "the use of a certain form of words" (1989,
p. 17). Grice's examples involve certain occurrences of indefinite
descriptions, as in 'He is meeting a woman this evening,' where it is
implicated that the woman is not the man's wife, and 'He broke a finger
yesterday,' where there is a "reverse implicature" that the finger is
the man's own. Just as implicature is a special case of indirect speech
act (Bach and Harnish 1979, pp 165-172), so generalized implicature is
a special case of standardized indirect speech acts, as discussed in
the foregoing chapter.
Standardization is a more widespread phenomenon than one might
suppose, and in this postscript I will identify certain other cases of
it that I have taken up in recent work. These include (1) performative
utterances, (2) standardized nonliterality, (3) what I call
"conversational impliciture," as opposed to implic-a-ture, (4) various
cases of singular reference, including referential uses of definite
descriptions. The breadth of the phenomenon of standardization might
seem to to blur the boundary between semantics and pragmatics or
otherwise complicate the problem of delimiting semantics, but in my
view it actually simplifies the problem. Identifying the diverse areas
where standardization operates demonstrates the virtue of wielding
Grice's "Modified Occam's Razor: Senses are not to be multiplied beyond
necessity" (1989, p. 47). Using the Razor to delimit semantics requires,
as I will explain later, a rather strict notion of what is said, whereby
something is not part of what is said if it is not correlated with anything
in the sentence being uttered.
1. Performatives and Short-circuiting
In making the case that "Performatives are statements too" (Bach 1975),
I argued that the performative practice "short circuits" the steps of
the otherwise more elaborate inference pattern that would be required
on the part of the audience (Jerry Morgan (1978) later used the phrase
"short-circuited implicature" for illocutionary standardization).
Performatives are distinctive because they involve the use of the very
verb that names the type of act being performed, e.g., 'promise' to
make a promise. This distinctive feature might suggest that
performativity requires a special explanation, and some have appealed
to illocutionary conventions or even to special performative meanings.
Harnish and I argued against such approaches (Bach and Harnish 1979,
pp. 173-192), but a new appeal to conventionalization has been made in
a recent article by John Searle (1989), who claims that performatives
are a kind of "declaration." He is led to this view on account of
certain alleged difficulties with our standardization thesis. We have
since replied by showing that these difficulties are illusory and that
standardization provides the more economical explanation of
performativity (Bach and Harnish
In our view, theorists have been misled into thinking that
performativity requires a theoretically special explanation by the fact
that in making an explicit performative utterance the speaker is saying
what he doing (after all, one can promise, for example, without doing
so explicitly, without using the performative verb). Some suppose that
performativity is a matter of linguistic meaning, that there is a
special semantic property of performativity, so that it is part of the
meaning of words like 'promise,' 'apologize,' and 'request' that one
can perform an act of the very sort named by the verb by uttering a
performative sentence containing that verb. An obvious problem with
this view is that it implausibly entails that such verbs are
systematically ambiguous. For a performative sentence can be used
literally but nonperformatively, e.g., to report some habitual act (one
might say 'I apologize ...' to describe typical situations in which one
apologizes). As for the view that a special sort of convention explains
why uses of the performative form count as promises, apologies,
requests, etc., not only is it gratuitous but it misses the fact that
performativity is but a special case of a more general phenomenon.
There are all sorts of other forms of words which are standardly used
to perform speech acts of types not predictable from their semantic
content, e.g. 'It would be nice if you ...' to request, 'Why don't you
...?' to advise, 'Do you know ...?' to ask for information, 'I'm sorry'
to apologize, and 'I wouldn't do that' to warn. In particular, there
are hedged and embedded performatives, such as 'I can assure you ...,'
'I must inform you ...,' 'I would like to invite you ...,' and 'I am
pleased to be able to offer you ...,' utterances to which the alleged
conventions for simple performative forms could not apply. Could such
conventions be suitably generalized? The variety of linguistic forms
standardly used for the indirect performance of such speech acts seems
too open-ended to be explained by a convention, which would need to
specify just those linguistic forms whose utterance counts as the
performance of an act of the relevant sort. Consider, for example, the
difference in potential indirect force between 'Where do you think
you're going?' (or 'Who do you think you are?' or 'What do you think
you're doing?') with 'When do you think you're going?' Standardization
avoids all the above difficulties. Rather than attributing special
meanings to performative verbs or supposing that performativity depends
on special conventions, the standardization thesis says simply that the
speaker's performative intention is inferable in Gricean fashion but
that precedent for the performative use streamlines or shortcircuits
the inference required on the part
of the audience.
The only cases in which performativity does involve convention
are those associated with specific institutional situations in which a
specific form of words is designated, and often required, for the
performance of an act of a certain sort. Examples include adjourning a
meeting, sentencing a convicted criminal, or christening a ship. But
ordinary performative utterances are not bound to particular
institutional contexts. Like most speech acts, they are acts of
communication. As such they succeed not by conformity to convention but
by recognition of intention; they are performed with an intention "whose
fulfillment consists in its recognition"
(Bach and Harnish 1979, p. 15).
2. Standardized Nonliterality
There are many sentences whose standard uses are not strictly
determined by their meanings but are not implicatures or figurative
uses either. For example, if one's spouse says 'I will be home later'.
she is likely to mean that she will be home later that night, not
merely some time in the future. Using the word 'tonight' would have
made what she meant fully explicit. An important analogous type of case
involves quantified noun phrases, which are often used as if their
domain is more restricted than their component words suggest, as in
(1) Everyone must attend class.
(2) Most people will vote Republican.
Something like this occurs with perfect tenses, like the present perfect in
(3) I have been in a meeting.
In these cases, what is meant is an "expansion" of what is said, in
that the speaker would have had to use additional words to make what he
meant fully explicit (Bach 1994a, 1994b). In my view it is gratuitous
to suppose that there are hidden domain markers that restrict the
"universe of discourse." Rather, what is meant is something the speaker
could reasonably expect his audience to figure out, in Gricean fashion.
In (1), for example, obviously the speaker could have inserted 'in this
class' after 'everyone,' and in (3) obviously the speaker could have
inserted 'just now' after 'meeting.' Here are some other examples, with
the implicit material in brackets:
(4) Jack and Jill are married [to each other].
(5) John broke his leg [unintentionally].
(6) Willard is not [what I would describe as]a weatherman but a meteorologist.
In each case the nonliterality is standardized, because each is an
instance of a general pattern of nonliteral use. I term this "sentence-
nonliterality," as opposed to constituent nonliterality, because the
nonliterality is not attributable to particular words or phrases in the
sentence (Bach 1987, p.71). For example, compare (3)
(3) I have been in a meeting. [just now]
with the relevantly similar (7),
(7) I have been in the Army.
which is likely to be used literally, i.e., without any implicit time
restriction. Clearly the difference between them is not attributable to
anything in (3) being used nonliterally.
3. Semantic Incompleteness and
There is another kind of case in which what a speaker means is not made
fully explicit by what he says. The inexplicitness is due to the fact
that sentence being used is "semantically incomplete"--it does not
express a complete proposition. That can be because there is a missing
argument, as in (8),
(8) I'm not ready.
which cannot be used to mean that one is not ready simpliciter. One
might mean more than that, e.g., that one is not ready to leave one's
house. And if one's spouse yells back,
(9) We'll be late.
she probably means that they will be late for a certain event. In
neither case is there anything in the sentence that corresponds to the
implicit argument. Although the speaker means something definite, the
sentences themselves lack determinate truth conditions, even after the
references of the indexicals 'I' and 'we' are fixed. Because they are
semantically incomplete, what the speaker means must be more than what
the sentence means; since the sentence does not express a complete
proposition, understanding its utterance requires a process of
"completion" (Bach 1994a,
In some cases, the linguistic meaning does not fix the scope
relation required for interpreting the utterance. This occurs with
mixed quantifiers, as in
(10) Five boys washed ten cars.
which admits of different interpretive possibilities, and in
(11) I love you too.
which has four possibilities that I can think of (Bach 1982). Some
cases of semantic incompleteness are of special philosophical interest,
including counterfactual conditionals, contrastive explanations, and
reports of propositional attitude.
Utterances that require either completion or expansion (section
2) involve what I call "conversational impliciture" (Bach 1994a),
because part of what is meant is communicated not explicitly but
implicitly. They are not cases of figurative speech or indirection, as
in Gricean implic-a-ture. Sperber and Wilson coined the word
"explicature" for this in-between category (1986, p. 182), since part
of what is meant explicates what is said, but I prefer "impliciture,"
since part of what is meant is communicated not explicitly but
implicitly, by way of expansion or
4. Singular Reference and Literal
Standardization plays an important role in my account of singular
reference. Consider, for example, the distinction between referential
and attributive uses of definite descriptions. If it were of semantic
significance, then sentences of the form 'the F is G' would be
systematically ambiguous--they could express either general
propositions (of the sort given by Russell's theory of descriptions),
to the effect that there is exactly one F and it is G, or singular
propositions concerning the particular individual that the speaker has
in mind. It is now widely accepted that the referential/attributive
distinction is not semantically significant, owing largely to Saul
Kripke's (1977) argument that the difference in use could arise even in
a language that is stipulated to be Russellian (I used a similar
argument in Bach 1975 against the ambiguity of performative sentences).
In my view (Bach 1987, chs. 5-6), the referential use is analogous to
standardized indirection. Also, uses of incomplete definite
descriptions, like 'the woman' and 'the table,' which are obviously not
satisfied uniquely, are cases of standardized nonliterality. A standard
use of a sentence like 'The table is dusty' concerns a particular
table, but this fact can be explained without positing a semantic slot
for a restriction on the universe of discourse or for a completion of
the description. Strictly speaking, the utterance of such a sentence is
not literal--what the speaker means
is an expansion of what is said.
As for proper names, I suggest that the intuitions supporting the
thesis that they are rigid designators and are directly referential can
best be explained in terms of certain pragmatic facts (Bach 1987, chs.
7-8). In particular, although in my view a proper name literally
applies only to an individual that bears the name (I claim that a name
'N' is semantically equivalent to the metalinguistic description 'the
bearer of "N"'), thinking of the property of bearing the name is rarely
the means by which a hearer is intended to think of the referent;
generally, a proper name is used merely to indicate what one is
speaking about. Accordingly, most uses of proper names are like
referential uses of definite descriptions. That is, the speaker is
expressing a singular proposition even though that proposition is not,
strictly speaking, determined by the linguistic meaning of the sentence
containing the name.
Demonstrative reference clearly involves expressing singular
propositions. The question arises whether accompanying demonstrations
are semantically significant, a view which David Kaplan (1989) formerly
held but later renounced. Marga Reimer (1991a, 1992b) has argued that
Kaplan was right the first time, but in reply I have claimed (Bach
1992a, 1992b) that he was right to change his mind. I argue that the
intention is decisive and that the demonstration plays merely the
pragmatic role of facilitating communication. Reimer's arguments that
the demonstration is decisive misconstrue the nature of the intention
involved in demonstrative reference. It involves more than having
something in mind and intending to refer to it; as part of a speaker's
communicative intention, a distinctively referential intention is
specifically the intention that one's audience think of a certain item
as that which one is pointing at and thereby intending to be talking
about. The procedure for identifying the referent is a matter not of
the semantics of demonstratives but of the standardization involved in
their use. For the reference to be successful, the intended individual
must be uniquely salient in such a way that the hearer can reasonably
suppose the speaker to have intended him to single it out as the
I have defended a similar point regarding anaphoric pronominal
reference (this point does not apply to lexical anaphors, such as
'himself' and 'each other,' whose referential properties are a matter
of semantics). In a sentence like (12),
(12) Jerry thinks that he is clever.
the pronoun 'he' may or may not be used to refer to Jerry. That is
because being previously mentioned is but one way of being salient. In
my view (Bach 1987, ch. 11), using a pronoun to refer to someone
previously mentioned is no different in kind from using it to refer to
someone otherwise salient, and there is no need to mark the difference
by means, e.g., of identical or distinct indices, as in Binding Theory.
The main objection to this assimilationist approach is that there are
cases in which the anaphoric option is blocked, as in
(13) He thinks that Jerry is clever.
Indeed, there seems to be a grammatical reason for this, namely that
the pronoun c-commands the noun. I tentatively defend the
assimilationist approach on the grounds that what is ruled out is not
coreference but intended coreference and that this is not a properly
grammatical notion. The phenomenon in question is pragmatic, I suggest,
despite its sensitivity to a syntactic
5. Delimiting What Is Said
The conception of standardization illustrated by the various examples
given above requires a strict notion of what is said, whereby what is
said is limited to what is closely correlated with the linguistic
material in the utterance. Although this limitation seems reasonable,
it has the consequence that the phenomenon of sentence-nonliterality,
as characterized above (section 2), does not pass the intuitive test of
nonliterality proposed by François Recanati (1989). However, his
objection to my notion of sentence-nonliterality (1989, pp. 313-315)
overlooks the way in which it differs from figurative speech. With
sentence-nonliterality the linguistic content of the utterance does not
make fully explicit what the speaker means, and yet nothing in the
sentence is being used nonliterally. It is simply a matter of intended
additional conceptual material to be read into one's audience, a
process that generally occurs so routinely as not to be noticed. That
is why it does not pass Recanati's intuitive test. Intuitively, people
would not classify as nonliteral typical utterances of sentences like
'Let's go to Chez Panisse' [together] and 'I haven't taken a shower'
[today]. In my view, however, even if we do not intuitively regard such
utterances as nonliteral when we reflect on them metalinguistically, in
practice we take them nonliterally when we hear them. They are not
literal even though they may seem to be because, as in the above
examples, the conceptual material that we unreflectively insert into
the utterance does not correspond to anything in the sentence being
This strict notion of what is said is the notion with which Grice
contrasted what is implicated. For Grice what is said is "closely
related to the conventional meaning of the ... sentence ... uttered"
(1989, p. 25), but it is not identical to conventional meaning because
there can be ambiguity or context-dependent reference--usually only one
conventional (linguistic) meaning is operative in a given utterance,
and indexical reference is context-sensitive. So not just linguistic
knowledge but (salient) contextual information can play a role in
determining what is said. Grice gives the impression that the
distinction between what is said and what is implicated is exhaustive,
making it seem that if something being conveyed is not implicated, it
must be part of what is said. He overlooked the intermediate case of
impliciture, in which the identification of what is conveyed requires
expansion or completion of what is said. Grice should have distinguished
not only the implied from the explicit
but the implicit from the implied.
6. Conventional Implicature?
Grice may have disallowed inexplicit saying, but he did recognize a
category of explicit nonsaying. For in his view there can be elements
in what is meant that correspond directly to elements in the sentence
uttered but do not enter into what is said. Because of this
correspondence, they result in "conventional" instead of conversational
implicatures, propositions which are merely "indicated" (1989, p. 25).
In my view, the category of conventional implicature can be explained
away (Bach 1994a, pp. 144-149). Grice's examples of "problematic
elements" are connectives, such as 'therefore' and 'but', which make a
certain contribution to what the speaker means by indicating a certain
relation between the two items they connect, e.g. that one is a
consequence of the other or that there is a contrast between the two.
Grice denies that this linguistically specified relation enters into
what is said, evidently to allow for an element of literal content that
is not truth-conditional. He denies that an utterance of (14), for
(14) He is an Englishman; he is, therefore, brave.
"would be, strictly speaking, false should the consequence in question
fail to hold." However, the speaker does seem to be saying that the
second claim is a consequence of the first--he is just not using a
separate clause to say it. Just because connectives like 'therefore'
and 'but' are not truth-functional does not mean that they do not enter
into truth conditions.
Grice also suggests that conventional implicature is involved in
the performance of what he calls "noncentral speech acts" (1989, p.
122), such as qualifying, contrasting, or concluding, performed using
phrases like 'loosely speaking,' 'in contrast,' and 'all in all.' Such
phrases (other examples include 'frankly,' 'to digress,' and 'if I may
say so') are used to comment on the very utterance in which they occur,
as in the following examples.
(15) In contrast, George would never do a thing like that.
(16) All in all, inflation is the biggest economic threat.
However, it seems to me (Bach 1994a, pp. 144-149) inaccurate to call
these second-order speech acts (conventional) implicatures. In uttering
(15) or (16), one is not implying that he making a contrast or drawing
a conclusion. In using a locution like 'in contrast' or 'all in all',
one is saying something about (providing a gloss or commentary on)
one's utterance or its conversational role. There is a straightforward
explanation why these locutions do not fit comfortably into
specifications of what is said: they are in construction syntactically
but not semantically with the clauses they introduce. Syntactically
they are sentence adverbials but they function as "illocutionary
adverbials" (Bach and Harnish 1979, pp. 219-228), interpreted as
modifying not the main clause but its utterance. The result is as it
were a split-level utterance. The abundance of illocutionary adverbials
suggests that this is a case of standardization.
In conclusion, it appears that standardization is a widespread
phenomenon. It is present in a number of philosophically or
linguistically interesting cases. Characterizing the phenomenon
requires recognizing a strict notion of what is said, on which what is
said must be correlated with features of the uttered sentence.
Understanding the phenomenon requires resisting wholesale appeals to
special meanings or conventions. Such appeals misconstrue the role of
precedent. Where there is standardization rather than convention,
precedent serves not to enable a form of words to have a standard that
yet not strictly literal use but rather to facilitate the audience's
understanding of that use by way of identifying the speaker's
Bach, Kent (1975), "Performatives
are statements too," Philosophical Studies 28: 229-36.
_____ (1982), "Semantic nonspecificity
and mixed quantifiers," Linguistics and Philosophy 4: 593-605.
_____ (1987), Thought and Reference,
Oxford: Oxford University Press.
_____ (1992a), "Paving the road
to reference," Philosophical Studies 67: 295-300.
_____ (1992b), "Intentions and
demonstrations," Analysis 52: 140-46.
_____ (1994a), "Conversational
impliciture," Mind & Language 9: 124-62.
_____ (1994b), "Semantic slack,"
in S. Tsohatzidis (ed.), Foundations of Speech Act Theory, London: Routledge.
_____ (1995), "Standardization
vs. Conventionalization," Linguistics and Philosophy 18: 677-686.
Bach, Kent and Robert M. Harnish
(1979), Linguistic Communication and Speech Acts, Cambridge, Mass.: MIT
_____ (1992), "How performatives
really work: a reply to Searle," Linguistics and Philosophy 15: 93-110.
Grice, Paul (1989), Studies in the
Way of Words, Cambridge, Mass.: Harvard University Press.
Kaplan, David (1989), 'Demonstratives'
and 'Afterthoughts,' in J. Almog et al (eds.), Themes from Kaplan, Oxford:
Oxford University Press.
Kripke, Saul (1977), "Speaker's
reference and semantic reference," Midwest Studies in Philosophy 2:
Morgan, Jerry (1978), "Two types
of convention in speech acts," in P. Cole (ed.), Syntax and Semantics
9: Pragmatics, New York: Academic Press.
Recanati, Francois (1989), "The
pragmatics of what is said," Mind & Language 4: 295-329.
Reimer, Marga (1991a), "Demonstratives,
demonstrations, and demonstrata," Philosophical Studies 63: 187-202.
_____ (1991b), "Do demonstrations
have semantic significance?", Analysis 51: 177-83.
Searle, John (1989), "How performatives
work," Linguistics and Philosophy 12: 535-58.
Sperber, Dan and Deirdre Wilson (1986), Relevance, Cambridge, Mass.: Harvard University Press.