1

Metaphors: What is Said or What is Implicated?

ESTHER ROMERO AND BELÉN SORIA

Abstract: A variety of theorists have recently argued against the explanation of metaphor
as particularized conversational implicature. Although we agree with them in defending
that the result of metaphorical interpretation, the metaphorical meaning, is involved in
what is said, we do not believe that the arguments they have given against the proposal of
metaphor as an implicature were conclusive. Indeed, many theorists still defend a
conception of metaphor as implicature. In this context, the main aim of this paper
appears: to develop a theory on metaphor as implicature that solves the main criticisms to
the identification and interpretation criteria in classical theory of metaphor as implicature
in order to defend not only that this theory on metaphor is compatible with the proposal
that metaphorical meanings form part of what is said but also that viewing metaphorical
meaning as part of a propositional content leads to a view on which it possesses the
properties that are usually attributed to what is said.

Financial support for this research, which has been carried out in the project ‘Phrasal Pragmatics’ (HUM 2006-

08418), has been provided by Spanish Ministry of Science and Education (DGICYT) and European Funds
(FEDER). This paper has benefited from comments in the Lisbon ECAP-V Conference (2005), the 15

th

Annual

Meeting of the ESSP (2007) in Geneva, and the Riga Conference on Metaphor (2007).

Addresses for correspondence: Esther Romero, Departamento de Filosofía I, Facultad de Psicología, Campus
de Cartuja, Granada, 18071, España. Belén Soria, Departamento de Filologías Inglesa y Alemana, Facultad de
Filosofía y Letras, Campus de Cartuja, Granada, 18071, España.
Emails:

[e-mail]

and

[e-mail]

2

1. Introduction: Theories of Metaphorical Meaning

We believe that a theory of metaphor must be able to account for its peculiarities.
Specifically, what we think is distinctive about metaphor is that it involves a specific human
ability: analogical reasoning. Regardless of whether metaphorical utterances are sentential or
non-sentential, referential or non-referential, negative or affirmative, etc.: in all cases,
metaphor involves a type of analogy. Analogical reasoning, in metaphor, involves using
knowledge about something that is not connected to the subject under scrutiny to give us
information about the matter we are really interested in. A theory of the interpretive process
for metaphorical utterances must specify how this analogy, the very essence of metaphor,
takes form. Furthermore, it must specify how, in the metaphorical interpretation,
metaphorical meanings appear.

Having said this, here we will not go into details of the process used to interpret

metaphor, rather we will provide arguments to show that the result of metaphorical
interpretation, the metaphorical meaning, is more naturally located in what is said than in
what is implicated. Our interest in defending this proposal stems from the fact that, if we take
into account the current theories of metaphorical meaning, we do not find agreement about
what the status of metaphorical meaning is, even if an identical metaphorical process were
defended.

Generally, it has been held that metaphorical content depends on the emergence of

metaphorical meanings, meanings that make this content itself special. Nevertheless, there are
philosophers that, following Davidson’s (1978) proposal, maintain that there are no
metaphorical meanings, no matter what characteristics might be attributed to them.

Proposals that argue for the existence of metaphorical meanings maintain that there is

some process of metaphorical interpretation that changes the meaning of some part of the
metaphorical bearer,

1

resulting in a metaphorical meaning different from the literal meaning

of that part. These proposals can be classified in terms of the role that is played by this
meaning in the propositional content that is communicated; metaphorical meaning can be
considered either as an ingredient of what is implicated or of what is said. According to the
first position, the speaker makes as if to say one thing in order to mean another. According to
the second, the speaker means what she metaphorically says.

When metaphorical meaning is understood as part of the content that is implicated by

the metaphor, the pragmatic process of interpretation is always considered to be an inferential
one. In the metaphorical utterance, words do not change their meanings to take on different
meanings that contribute to what is said. Words do not obtain a semantic value different from
the one they usually have; rather, what is meant has a content distinct from the proposition
literally expressed by the utterance, a content that results from pragmatic inferences which
depend on a pragmatic principle. The principle is normally taken to be either some version of
the Cooperative Principle or the Relevance Principle. When some version of the Cooperative
Principle is involved, different authors resort to different mechanisms to explain how the
inferential process reaches metaphorical implicature or how metaphorical implicature is
calculated: resemblance for Grice (1975), some heuristic principle for Searle (1979), or a
process of interaction for Kittay (1987). In general, we could say that the authors that have

1

A metaphorical bearer is understood as an utterance (or an expression in some theories) that is identified as

metaphorical in natural language and that conveys a metaphorical content. Nowadays it is usual to distinguish
between metaphorical bearer and metaphorical vehicle and the latter is reserved for the word or words used
metaphorically in a metaphorical bearer.

3

defended literalism in what is said have also argued for metaphor as implicature.

2

The

outstanding exception is Stern (2006) since he calls himself ‘literalist’ but considers that
metaphorical meanings belong to what is said. When Relevance Principle is involved, as
illustrated by Sperber and Wilson (1986/95) in Early Relevance Theory (ERT), the inferential
process is guided by this principle and results in the loosening of a concept where one or
more features of this concept are dropped in the process of arriving at the intended
interpretation.

Opposed to the view of metaphorical meaning as part of an implicated content, it has

been maintained that metaphorical meaning is part of what is said. On this view, with a
metaphor the speaker means what she metaphorically says. Thus what is said is not always
what is literally said. Supposedly, to argue that what is said is not always what is literally said
is automatically to become a contextualist, at least in the following sense:

According to contextualism, the contrast between what the speaker means and what
she literally says is illusory, and the notion of ‘what a sentence says’ incoherent. What
is said (the truth-conditional content of the utterance) is nothing but an aspect of
speaker’s meaning. That is not to deny that there is a legitimate contrast to be drawn
between what the speaker says and what he or she merely implies. (Recanati, 2004, p.
4).

Admittedly, for the contextualist what is said by the speaker is not always literally said. The
most obvious position in favour of what is non-literally said is the defence of the legitimacy
of what is metaphorically said, but this is not sufficient to be a contextualist. Indeed, it does
not seem plausible for contextualism to defend, as classical rhetoricians did, that some words
have, in addition to their literal meaning, a metaphorical linguistic meaning which can
coincide with the literal meaning of some other word, where this latter meaning is involved in
what is metaphorically said; on this view, the interpretation of a metaphor merely requires a
process of disambiguation between these two fixed meanings. Instead, the contextualist
maintains that what is said is not always literally said because contextual information intrudes
what is said. Because disambiguation is a process where the context only serves to choose
among options that are already established, the proposal of classical rhetoric would not be
acceptable to contextualists. In addition, the classical rhetoric proposal, which is a type of
substitution view, was rejected long ago because, as Black (1954) showed, it reduces the
metaphorical contribution to an ornamental value or a case of catachresis, thereby denying the
cognitive value of metaphor.

The rejection of this substitution view gives rise to proposals intended to account for

the production of metaphorical meaning in a way that does justice to metaphor’s cognitive
value. On such a view the content that the metaphorical vehicle contributes to the proposition
metaphorically expressed differs from any of its linguistic meanings and from any of the
literal contents that the expression might fix. The metaphorical meaning does not coincide
with the conventional meaning of some other word either. Contextual factors will always be
present in the elaboration of metaphorical meaning and the contextual intrusion is determined

2

In literalism the sentence linguistic meaning, understood as a compositional meaning that results from the

combination of the conventional meaning of sentence terms, is closely related to what is said by an utterance of
the sentence. Their differences depend on the contextual information that is involved in the latter, taking into
account that, in general, pragmatic contributions demanded by the linguistic meaning itself are kept to a
minimum. So, the pragmatic contribution needed to interpret metaphor intervenes in an implicature and not in
what is said. Authors such as Stanley (2005) argue for this proposal.

4

by the mechanism responsible for its production. However, within this approach we can
distinguish, not only several different explanations of the mechanism responsible for the
production of metaphorical meaning, but also several different views about how the
interpretive process is triggered: either as something indispensable for the expression of
propositional content or as something dispensable. This distinction is relevant because only
when the metaphorical process is considered dispensable (optional) can the proposal be
classified as contextualist.

Contrary to what one might think, it is not easy to draw the distinction between

mandatory and optional demands for contextual information. Indeed, there are at least two
distinct senses of ‘mandatory’. On the one hand, from a linguistic point of view, a process or
its result is mandatory (mandatory

L

) when it is, as Recanati (2004, p. 98) says, ‘required in

virtue of a linguistic convention governing the use of a particular construction (or class of
constructions).’ On the other hand, from a truth-conditional point of view, an interpretation
process or its result is mandatory (mandatory

T

) when it is necessary for a propositional

content to be generated in the interpretation of an utterance (Recanati, 2004, p. 62).
Sometimes, a process is mandatory in both senses, as when pronouns are involved.
Sometimes, the interpretation process is mandatory

L

but not mandatory

T

, as in the process

required for recovering the conventional implicature of ‘but’.

3

In other occasions, the process

is mandatory

T

but not mandatory

L

, as in the process required to interpret incomplete definite

descriptions (Romero and Soria, forthcoming). To avoid the presupposition failure associated
with the incomplete definite description and thus a failure to express a proposition (Glanzberg,
2005), a mandatory

T

pragmatic process is triggered. Finally, there are processes that are not

mandatory in any of these senses, as in the cases of free enrichment of the circumstances of
evaluation (Recanati, 2004, pp. 115-30).

The process of metaphorical interpretation cannot always be mandatory

L

since there is

no need to consider lexical items in a particular construction as requiring a metaphorical
meaning, as it happens in a metaphorical utterance of ‘My cat is on the mat’ to refer to an
infant.

4

But, if the process of metaphorical interpretation is required in order to obtain a

propositional content, then the metaphorical process is mandatory

T

. This position is

compatible with a literalist account of what is said and could be attributed to Stern (2006).
For him, the metaphorical interpretation depends on a mandatory

T

process which recovers a

deictic operator that triggers a mandatory

L

interpretive process which is itself context-

dependant.

The operator semantically demands contextual information. Stern’s literalist

position posits a richer underlying linguistic representation whose meaning will determine the
truth-conditions of each metaphorical utterance in context. In Romero and Soria (2007), we
also argue for a mandatory

T

process, although we claim that it results in a transfer which is

achieved by a mapping and not by a deictic operator. Our differences with Stern lie in our
views about the particular metaphorical mechanism rather than in the mandatory

T

character

of this mechanism.

5

If Stern and we are right, then the minimal proposition expressed by a

metaphorical utterance must be non-literal. This means abandoning the assumption that the
input to what is implicated is always what is literally said. Nevertheless, it does not entail

3

The meaning of ‘but’ sets up a slot to be contextually filled and when it is saturated (mandatory

L

), it

determines non-truth-conditional aspects of the utterance meaning (non-mandatory

T

).

4

For a metaphorical utterance of this sentence, see example (3) below. The process of metaphorical

interpretation would be mandatory

L

for a metaphorical utterance of ‘The sky is crying’.

5

If, for us, what makes a process obligatory or optional depends on the way in which it is triggered, then the

same process, against what Recanati (2004) maintains, can be obligatory or optional. The obligatory or optional
character of a pragmatic process is extrinsic to the process itself.

5

embracing contextualism. Contextualists about metaphor hold that neither the process nor its
result are mandatory in any sense. The process either results in an optional

LT

ad hoc concept,

as Wilson and Carston (2006) in Current Relevance Theory (CRT) would argue, or it is itself
an optional

LT

process of modulation, in the case of Recanati. A summary of our classification

can be seen in Figure 1.

Figure 1.Theories of metaphorical meaning

No metaphorical meanings: Davidson

Metaphorical

Meaning

What is
implicated

Cooperative
Principle

Grice: resemblance
Searle: heuristic principles
Kittay: interaction

Relevance Principle: Sperber and Wilson (in ERT)

What is

said

= a metaphorical linguistic meaning (Classical rhetoric)

≠ any of the
linguistic
meanings

Mandatory

T

-an operator that triggers a mandatory

L

process (saturation): Stern
-transfer: Romero and Soria

Optional

LT

- an ad hoc concept: Wilson and Carston
(in CRT)
- modulation: Recanati

This brief presentation of the accounts on metaphor that represent the different

proposals that may appear in a classification of theories of metaphorical meaning, one
articulated on the basis of the central discussion of whether metaphorical meaning is involved
in what is implicated or in what is said by an utterance, permits us to show that both positions
are currently defended.

Still, although the theory of metaphor as implicature has been attacked from its very

beginning, the theorists of metaphor as implicature do not abandon their proposal because
they understand that these attacks can be resisted.

6

As we have seen, many reputed authors

nowadays still consider metaphor as implicature while a growing group of equally reputed
ones argue for metaphor as part of what is said. We argue for the latter position but we take
the other proposal seriously and find that, for the sake of the argument, we need to explain
what theory can best account for metaphor as implicature. To do this, we will see first, in the
next section, the Gricean proposal of metaphor as implicature and will enumerate the
problems that have been attributed to it. Then, we will see, in section 3, how the theorist of
implicature might escape from these criticisms. This leads to a new explanation on metaphor
within the theory of implicature. Nevertheless, this will not be the end of the story because
the measures that implicature theorists have to adopt to solve the problems typically
attributed to it lead to a proposal that is compatible also with the one in favour of what is
metaphorically said. Thus we will move, in section 4, to the arguments which allow choosing

6

Camp (2006) has claimed that the arguments that have been recently elaborated against the explanation of

metaphor as particularized conversational implicature are also not conclusive. She considers four criteria for
distinguishing what is said from what is merely meant that, according to contextualists, support classifying
metaphor in what is said and argues that when rightly understood, these criteria do not support this claim. This
has left a space, according to her, for defending the conception of metaphor as implicature. Nevertheless, we
argue in this paper that this space is not justified. When we get an explanation on metaphor within the theory of
implicature that solves the problems traditionally attributed to metaphor as implicature, we obtain a proposal
that is also compatible with a theory of what is metaphorically said. As metaphorical contents described in this
way have the characteristic features of what is said, they are more naturally located in what is metaphorically
said.

6

one of these positions. We will raise the question of what features metaphorical propositional
content has, no matter whether it is explained from one or the other standpoint. By comparing
the characteristic features of implicated propositional content and the features of
metaphorical propositional content, we will reach the conclusion that metaphorical
propositional contents possess the properties that are usually attributed to what is non-
conventionally said. So the result of metaphorical interpretation is more naturally located in
what is said than in what is implicated. Metaphors should be conceived as cases in which
what is said is metaphorically said. Finally, in section 5, we will highlight that the
subpropositional process required to reach what is metaphorically said is truth-conditionally
mandatory and so a contextualist approach on metaphor should not be defended.

2. Grice’s Theory of Metaphor as Particularized Conversational Implicature and its
Problems

If we go along with the Gricean approach (Grice, 1975, 1978), we should explain the
behaviour of the metaphor (1),

(1) [In wondering whether to take an umbrella or not, A asked B what the weather is

like today and B utters:] The sky is crying.

as a case of particularized conversational implicature. Resorting to the Gricean distinction
between what the speaker literally says and what the speaker implicates, it could be said that
with (1) the speaker literally says that the sky is crying, a proposition that involves a
categorial falsity, something she believes to be false. Thus, the speaker is flouting the first
maxim of quality of the Cooperative Principle, ‘Do not say what you believe to be false’. The
speaker cannot plausibly mean that the sky is crying, a categorial falsity. So, with (1) the
speaker has just made as if to say literally that the sky is crying.

7

Consequently, what the

speaker means is only what the speaker implicates, and what she implicates depends on the
attribution to the sky of some features in respect of which the sky resembles more or less an
object that can literally cry. Furthermore, what the speaker implicates metaphorically with (1)
is that it is raining, thus the situation is re-established and the speaker’s behaviour becomes
cooperative. In this way, the interpretation of metaphorical utterances always proceeds in two
propositional stages: first, the speaker makes as if to say literally something that flouts the
first conversational maxim of quality, which leads, second, to the search for an implicated
content that will re-establish the cooperative situation.

Grice said that conversational implicatures, whether particularized or generalized,

have several characteristic features. If metaphorical propositions are implicatures, then these
propositions should share them. In particular, they should be cancellable and nondetachable,
and their truth conditions should be independent of the truth of the utterance.

7

According to Grice, what is said is part of speaker’s meaning and it is always intended by the speaker. Indeed,

when a proposition is expressed by an utterance and is not intended by the speaker, it is a case of making as if to
say. ‘One may distinguish, within the total signification, between what is said (in a favored sense) and what is
implicated; and second, one may distinguish between what is part of the conventional force (or meaning) of the
utterance and what is not. This yields three possible elements―what is said, what is conventionally implicated,
and what nonconventionally implicated―though in a given case one or more of these elements may be lacking.
For example, nothing may be said, though there is something which a speaker makes as if to say.’ (Grice,
1978/1989, p. 41). Metaphor is one of those cases in which something is made as if to say and nothing is said
and, thus, the total signification of a metaphorical utterance lacks one of its components: what is said.

7

A conversational implicature is always cancellable either explicitly or contextually. It

is explicitly cancellable if it is admissible to add but not p (or I do not mean to imply that p)
to the form of words the utterance of which putatively implicates that p (Grice, 1978, p. 44).
It is contextually cancellable if one can find situations in which the utterance of the form of
words would simply not carry the implicature (Grice, 1978, p. 44).

8

The metaphorical

implicature of (1), that it is raining, is cancellable if (2)

(2) [In wondering whether to take an umbrella or not, A asked B what the weather is

like today and B utters:] The sky is crying but it is not raining.

is inteligible. This property is a necessary condition for all conversational implicatures; but it
is not sufficient. There is meaning that is cancellable and does not form a part of an
implicature, as in the possibility of using a word or form of words in a loose or relaxed way.

9

Strictly speaking, the cancelled linguistic meaning does not form a part of speaker’s meaning.
Indeed, what is said is not cancellable. If we try to cancel explicitly what is said, then we
make the utterance unintelligible.

10

Another characteristic feature of conversational implicatures is nondetachability and it

is conceived as an identity criterion for what is said. Two utterances in the same context do
not coincide in what is said if they launch different implicatures. It is a necessary condition,
except when the implicature depends on the exploitation of a maxim of manner or when there
is no alternative way of saying what is said. Nondetachability does not apply to what is said.
So, if the metaphorical content is an implicature that depends on the explotation of the first
maxim of quality, it must be nondetachable, that is, there is no way to make as if to say, in the
same context, that the sky is crying without suggesting that it is raining.

The last feature of conversational implicature we want to highlight is the

independence of the truth conditions of the implicature from the truth conditions of the
utterance. An implicature (conventional or conversational) does not fix the truth conditions of
the utterance of a sentence because implicatures, as Grice taught us, are not asserted but
merely suggested. What makes an explicit proposition different from an implicature is that
the former has been asserted and so our utterance cannot be true if the explicit proposition is
not. The truth conditions of an utterance are fixed by what is said. A proposition fixes the

8

To understand these two ways of cancellation, consider the following examples. If in a normal context

somebody utters ‘Mary is in the kitchen or in the bathroom’, the speaker implicates that he does not know where
Mary exactly is. This content is explicitly cancellable because this speaker can add to his utterance ‘but I do not
mean to imply that I do not know where she is’ without rendering the utterance unintelligible. If the first
sentence is uttered in a context in which the participants of the communicative act are playing hide and seek, the
speaker contextually cancels the implicature that the speaker is ignorant because he is just giving a clue.

9

In the utterance of the sentence ‘Macbeth saw Banquo’ when Macbeth hallucinated, the speaker cancels the

requirement of the existence of Banquo that is required by part of the conventional meaning of ‘saw’ and,
nevertheless, this cancellation does not show that this content must be considered as an implicature (Grice,
1978, p. 44).

10

There are different reasons to think that Grice understood that cancellation must produce an intelligible

utterance. One of them is provided in Grice (1975, p. 39) where he argued that cancellation depends on the
possibility of opting out of the observation of the Cooperative Principle when it affects the production of the
implicature, which is compatible with keeping cooperation in part by means of what is said. Another can be
found in Grice (1978, p. 44) where he affirmed that an implicature is cancellable if it is admissible to add ‘but
not p’ to the utterance that implicates that p. Finally, it is even clearer when he explains why presuppositions are
not cancellable. In Grice (1961, p. 128), he maintains that presuppositions are not cancellable because if we
cancelled them we would run the risk of unintelligibility. The result of cancellation must be admissible and
intelligible.

8

truth conditions of an utterance of a sentence if it is absolutely impossible for the utterance to
be true without the proposition being true. By contrast, the implicature may be false while the
utterance may be true. The truth or falsity of the implicated proposition that it is raining does
not affect the evaluation of (1). If metaphorical propositions are implicatures, then
metaphorical utterances such as (1) do not express propositions on whose basis their truth is
evaluated, since they are merely cases of making as if to say.

This Gricean theory of metaphor has been attacked from its very beginning. Among

the attacks, we will focus on problems concerning the specific Gricean proposals relating to
the identification and the production of implicature, and on the incompatibility of certain
empirical results with the explanation of metaphor as involving two propositional stages.
Although, as we will see in the next section, the first two problems require some changes in
the conception of metaphor as implicature, the other is merely apparent.

With respect to metaphorical identification, the implicature theorists maintain that the

triggering condition for metaphor is a categorial falsity, which is a special type of flouting of
the first maxim of quality of the Cooperative Principle, ‘Do not say what you believe to be
false’. From this position the identification of metaphorical utterances cannot be explained.
As Recanati (1987) argued, this type of flouting a maxim entails that the speaker has just
made as if to say the literal propositional content and thus the speaker has not said anything
that really violates the Cooperative Principle. In addition, not all metaphors involve a
categorial falsity as the metaphorical utterance (3)

(3) [A is at home. Her only daughter, who is a two-year-old girl, is playing with a

woollen ball on the mat. B, a good friend of A, enters the room, asks A where her
daughter is, and, A answers:] My cat is on the mat.

Besides, there are non-metaphorical utterances that involve a categorial falsity as the
metonymic utterance (4)

(4) [In a restaurant, looking at the customer of the ham sandwich, a waitress tells

another:] The ham sandwich is waiting for his check.

Furthermore, it is argued that not all metaphors can fix a literal propositional content. The
speaker of (1) does not make as if to say any proposition literally because (1) cannot be
interpreted literally as far as our linguistic competence is concerned.

With respect to implicature production, the problem we are considering is posed by

Recanati (1987, p. 228). He claims that it is not possible to determine what the speaker
implicates from (1) as there is no proposition that can reconcile the utterance with the
apparently flouted conversational maxim.

In addition, the proposal that the interpretation of metaphorical utterances proceeds in

two propositional stages has been criticized by cognitive metaphor theorists since the late
seventies.

11

According to cognitive metaphor theorists (Gibbs, 1983; Keysar and Glucksberg,

11

Theorists of cognitive metaphor such as Ortony et al. (1978), Clark (1979), Gibbs (1984, 1986, 1992), Gildea

and Glucksberg (1983), Inhoff et al. (1984), and Keysar and Glucksberg (1992) have performed some
experiments in which the results, as they interpret them, are incompatible with the explanation of metaphor as
implicature. The idea is that metaphorical utterances should be included in ‘what is said’ because its
comprehension is direct, in the sense of coming first in the order of interpretation. This proposal has also been
defended by contextualists, such as Recanati (1995), Bezuidenhout (2001), etc. although they argue for it by
invoking sub-personal processing.

9

1992), several psychological experiments have demonstrated that the interpretation of literal
and metaphorical utterances exhibit equivalent processing times. On the assumption that
processing effort is directly proportional to processing time, and that two processes should
take more effort than one, they thus reject implicature theories of metaphor, which appeal to
an additional process, as inconsistent with the empirical data. Nevertheless, as we have just
said, this problem is merely apparent.

The implicature theorists have three lines of defence for this criticism. They can, as

suggested by Recanati (1995, p. 208), defend their position by saying that the results
involving equivalent processing times depend, in most of the cases, on choosing examples of
conventional metaphors where processing times might be expected to be equivalent, but that
nothing is proved about novel metaphors, which would require processing the literal
proposition first, as the Gricean account suggests. Second, in Romero and Soria (2003, pp.
176-82), we have argued that the assumption that the processing effort is always directly
proportional to the processing times is erroneous. Third, implicature theorists no longer need
this strategy because, according to some recent empirical evidence (Onishi and Murphy,
1993; Giora, 1999), the processing times for the interpretation of literal and metaphorical
utterances of the same sentence are not equal.

These strategies, however, do not provide defence for the implicature theory in

particular but for the more general thesis that there is an asymmetric dependence of
metaphorical meaning on literal meaning: the thesis that the meaning conveyed by an
expression is ‘metaphorical’ only if it is derived from some literal meaning which must be
processed in order for the metaphorical meaning to be accessed.

This thesis can be articulated in two distinct ways (Recanati, 1995, p. 230): one results

in the conception of metaphor as implicature, while the other requires a conceptual
mechanism whose results intervene in what is said. Thus, the problem posed by theorists of
cognitive metaphor against the implicature theory of metaphor does not seem to touch this
conception, although its defence is also compatible with a conception of metaphor as what is
said.

3. Grice’s Proposal Revisited

We now turn to consideration of the problems related to the identification criterion for
metaphor and the production of metaphorical implicature. This set of problems requires some
changes in the Gricean explanation of metaphor as implicature. In order to construct a
defensible account of metaphor as implicature, it is necessary to determine the criteria for
identifying metaphors; we argue that to identify metaphor we need to recognise (albeit
generally not consciously) both a contextual abnormality, and a source domain and a target
domain (Romero and Soria, 1997/98, pp. 377-80). We will also defend the thesis that the
interpretation process is subpropositional. But these proposals, we will argue, not only permit
us to have a theory of metaphor as implicature but a theory of metaphor in which
metaphorical meaning forms a part of what is said. The reason why these two possibilities are
open is that metaphorical identification and interpretation do not depend on processing the
entire literal proposition, they depend on features of subpropositional components instead.
When the conception of metaphor as implicature changes in order to avoid the problems
related to metaphorical identification and interpretation, the notion becomes compatible with
the type of explanation that includes metaphorical meaning in what is said.

10

3.1 Identification Criteria
When metaphor is considered an implicature, the triggering condition for metaphor is a
categorial falsity, a special type of flouting of the first maxim of quality of the Cooperative
Principle, ‘Do not say what you believe to be false’.

The first problem with this triggering condition is, as we have said before, shown by

Recanati (1987, p. 230). If with (1) the speaker has just made as if to say that the sky is
crying, the speaker has not said anything; thus the maxims are not really violated and it is not
necessary to suppose that the speaker has implicated anything in order to maintain the
Cooperative Principle. This criticism, however, is not conclusive. The implicature theorists
might argue that when a categorial falsity is present, the speaker merely makes as if to say
that p but does not actually say anything, and thus, although she does not flout the first
maxim of quality, she does flout an alleged maxim of quantity related to the amount of
information: ‘Make your contribution informative’.

12

This maxim violation is then repaired

by the generation of the implicature. This position would entail a complex identification
criterion, as both a categorial falsity and a flouting of the alleged maxim of quantity are
required.

Nevertheless, metaphors cannot be identified with this complex identification

criterion because it is not free of the other problems related to the simpler criterion, the
criterion of the categorial falsity, which is included in the former. As we have said, not all
metaphors involve a categorial falsity. In example (3),

(3) [A is at home. Her only daughter, who is a two-year-old girl, is playing with a

woollen ball on the mat. B, a good friend of A, enters the room, asks A where her
daughter is, and, A answers:] My cat is on the mat.

there is, at best, a simple false proposition, then there is no special type of flouting the first
maxim of quality. If we widened the criterion to any type of falsity and not just of a
categorial type in order to be able to identify (3) as a metaphor, then all implicatures that
depend on flouting the first quality maxim would have to be classified in the same way, even
non-metaphorical cases such as (5)

(5) [Kent tells his son who is crying because of a minor cut:] You are not going to die.

In addition, there are non-metaphorical utterances that present both a categorical falsity and a
flouting of the alleged maxim of quantity. In example (4),

(4) [In a restaurant, looking at the customer of the ham sandwich, a waitress tells

another:] The ham sandwich is waiting for his check.

the speaker says that the ham sandwich is waiting for his check, which is a categorial falsity;
thus she merely makes as if to say this and flouts the alleged maxim of quantity.
Nevertheless, (4) is considered a case of metonymy and not of metaphor.

12

We say ‘alleged’ because, strictly speaking, Grice does not defend this maxim. Grice’s maxims of quantity

make reference to the conversational contribution being as informative as is required for the purposes of the
exchange and not more informative than is required. Still, in these two maxims of quantity, it is always
supposed that there is some informative contribution. This is just what does not happen when the proposition is
false. As Grice (1987, p. 371) says: ‘False information is not an inferior kind of information; it just is not
information.’ Thus the more general assumption that there is some informative contribution is violated.

11

More radically, not all metaphors can fix a literal content. When a categorial falsity is

involved, the speaker cannot make as if to say something literally. The speaker of (1) does
not make as if to say any literal proposition at all because (1) cannot be interpreted literally as
far as our linguistic competence is concerned. ‘To cry’ is the type of action that requires an
animate subject with eyes, and the sky does not fulfil this requirement. (1) cannot fix a literal
proposition or some truth conditions determined merely by the linguistic meaning of the
sentence. Since (1) cannot fix a literal content because of the categorial falsity, it cannot be a
literally false utterance. A categorial falsity stands opposed to well-formedness, not to true
propositions.

To overcome the last problem, the identification criterion should not depend on

processing a literal proposition if (1) is to be counted as a metaphor. In addition, to avoid the
problem that not all metaphors present a categorial falsity, we must admit that there are
pragmatic categorial falsities, which are involved in (3) but not in (5). (3) does not express a
simple false proposition rather it presents a categorial falsity as a result of using ‘My cat’ in a
context in which the speaker does not refer to a cat, this type of categorial falsity is not
involved in (5). Kittay’s theory of metaphor as implicature with its Incongruity Principle
(1987) satisfies these two demands, but her principle needs something else to exclude (4), a
metonymy, as a case of metaphor. This is achieved by means of our conceptual contrast
(Romero and Soria, 1997/98, 2006). For us, hearers are able to identify metaphors because
they involve both a contextual abnormality (our technical notion for categorial falsity,
pragmatic or not) and a conceptual contrast.

The notion of contextual abnormality must be understood as the use of an expression

in an unusual linguistic or extra-linguistic context; it differs from other notions of anomaly
(Loewenberg, 1975; Kittay, 1987). There are two ways in which abnormality can be
manifested:

(a) An oddity in the relation between the terms uttered, and
(b) An oddity in the relation between the occurrence of an expression in the actual

unusual context and the implicit context associated with a normal use of this
expression.

Mode (a) can be illustrated by examples (1) and (4). In (1) the normal interpretation of

‘is crying’ as the predicate of ‘The sky’ is not allowed. In (4) the normal interpretation of the
predicate, ‘is waiting for his check’, is incompatible with the normal interpretation of the
noun phrase ‘the ham sandwich’.

Mode (b) can be exemplified by (3) as infants are not the kind of thing that we call

‘cats’ according to our conceptual system. The contextual abnormality is produced by the
confrontation between the actual and unusual context and a possible normal context of the
expression. It is not normal to use the expression ‘my cat’ for the speaker to refer to an infant.
This is an unusual use of the expression which entails a contextual abnormality, a pragmatic
categorial falsity.

Contextual abnormality is a necessary but not a sufficient condition for metaphor.

Some additional identification criterion is needed: a conceptual contrast. The conceptual
contrast is the recognition that the speaker is talking about a topic (target domain) using terms
which normally describe another (source domain). In (3), we identify that the speaker is
talking about an infant (

INFANT

is the target domain) using a term that normally describes a

feline (

CAT

is the source domain). Similarly, in (1) we detect a conceptual contrast, now

between the target domain,

WEATHER CONDITIONS

, and the source domain,

EMOTIONAL

12

STATES

.

Novel metonymies, such as (4), are very different: there is no source domain and no

conceptual contrast. Both the customer and the ham sandwich belong to the target domain,
we refer both to the customer and to the ham sandwich although the only explicit expression
is ‘the ham sandwich’, which is used to trigger the recovery of the veiled restricted nominal
element ‘customer’.

13

So, the implicature theorist might solve the problems of the complex identification

criterion not only by accepting the requirement of a conceptual contrast, in order to avoid
non-metaphorical utterances such as metonymies, but also by replacing the requirement of
categorial falsity with that of contextual abnormality that gives form to the requisite that there
are also pragmatic categorial falsities. When there is a contextual abnormality and a
conceptual contrast, the literal interpretation is blocked because either it is not possible

14

or it

would be of no use.

15

In these cases, the implicature theorist might argue, the speaker flouts

the alleged maxim of quantity because there is nothing less informative than having no literal
proposition at all.

The details of these proposals make it manifest that it is possible to have a theory of

metaphor as implicature that includes accurate identification criteria. Nevertheless, they also
make it manifest that once the criteria for metaphorical identification do not depend on
processing the entire literal proposition and depend on features of subpropositional
components instead, it is possible to argue that what is said may be metaphorically said. This
is relevant in the main discussion of our paper because, given the accurate identification
criteria, both the proposal that metaphorical meaning is involved in an implicature and the
proposal that it forms a part of what is said are viable.

3.2 Implicature Production
The proposals on metaphorical implicature production is also attacked. On the Gricean
approach, metaphorical implicature must be calculated from what the speaker has just made
as if to say. What the speaker implicates depends on resemblance between two topics. Thus,
once (1) is identified as metaphor, its interpretation is made, according to this proposal,
calculating the implicature, that it is raining, from what the speaker has made as if to say,
that the sky is crying. The implicature depends on the attribution to the sky of some features
in respect of which the sky resembles more or less the object that can literally cry.

The first problem is, as we mentioned before, that it is not possible to determine what

the speaker implicates from (1) as there is no proposition that can reconcile the utterance with
the apparently flouted first conversational maxim of quality (Recanati, 1987, p. 228). There is
no proposition that added to the false proposition makes the joint contribution true.

This criticism, however, is not conclusive. As we have seen, what forces the speaker,

in cases of metaphor, to make as if to say that p is that this proposition is false. Thus, the
implicature cannot be calculated as something which together with what is said re-establishes
cooperation because, in metaphor, the speaker merely makes as if to say and does not say
anything; she only conveys an implicature. The implicature theorist can say that the
implicature must not be calculated from p, the literal false proposition, because this could not
be the proposition intended by the speaker, as she believes it to be false. The solution would

13

Novel metonymy is identified when the hearer appreciates a non-generic use of a NP, a contextual abnormally

and a veiled restricted nominal element (Romero and Soria, 2006, forthcoming).

14

This is always the case for metaphors in which the abnormality appears among the terms uttered, mode (a). As

far as our linguistic competence is concerned, this type of metaphor cannot fix a literal propositional content.

15

Only in cases of mode (b) can the hearer get a possible literal interpretation, but if he did, the implicature

would not be worked out from, among other things, that proposition.

13

be to calculate the implicature by resorting to, among other things, the conventional meaning
of the words used, without viewing them as expressing a full proposition.

Its production,

Grice would say (1975, p. 34), depends on the resemblance between what we are actually
speaking about and what we are attributing to it.

But, according to Recanati (1995), this understanding of non-literal meaning as being

asymmetrically dependent on the literal meanings of subpropositional components of an
expression strongly suggests that metaphorical meaning is part of what is said rather than of
what is implicated. This proposal, according to Recanati (1995, p. 228), is backed by
examples in which we must compute the non-literal interpretation in order to compute what is
said. In (3), part of ‘My cat’ is used non-literally. In order to know what object the speaker is
talking about, we must saturate a variable corresponding to the relation between the speaker
and a cat. Only if we recognize that ‘cat’ in this noun phrase is being used metaphorically
will we be able to saturate the variable in this context. In order to know what ‘my cat’ refers
to, we have to construct the metaphorical meaning of ‘cat’, which then makes it possible to
saturate the relation between a metaphorical cat and the speaker. The metaphorical process is
previous to the one of saturation, and the latter is a process that is necessary to obtain what is
said.

This argument, again, would not be conclusive for the implicature theorist. The

implicature theorist could explain example (3) by arguing that what the speaker makes as if to
say is that the only cat of the speaker (whatever the relation between them) is on the only mat
of her house. What causes the requirement of the non-literal interpretation to be prior to
saturation on Recanati’s view is his explanation of the referential use of the definite
description included in (3); but this referential use must itself be understood as a case of
implicature. Recanati’s argument depends on a defence of the referential use of the definite
description in what is said, an argument which theorists of metaphor as implicature do not
have to commit themselves to.

Crucially, the fact that there are subpropositional processes of interpretation that

generate non-conventional meanings is also compatible with their result being a part of an
implicature. The implicature may be calculated from the conventional meaning of the
sentence’s constituents together with contextual information as Grice (1975), Sperber and
Wilson (1986/95) and Kittay (1987) say.

The inferential process required for calculating the implicature is, depending on the

particular proposal chosen, characterized differently. In our opinion, the process required for
interpreting metaphor should explain how the analogical reasoning necessary to determine
what is metaphorically communicated occurs. Kittay uses Black’s (1954) interaction view to
improve the resource to resemblance made by Grice. Sperber and Wilson think that the
Relevance Principle leads to an interpretation where the denotation of the term used
metaphorically is loosened or widened radically since what is talked about falls very far
outside the normal denotation. Still, the encoded concept and the communicated one share
some properties. But loosening

or Kittay’s second order meanings are not sufficient to

explain the analogical reasoning required in metaphorical comprehension and, so, they cannot
explain the new properties that characterize metaphors (Romero and Soria, 2007).

Conceptual mappings for metaphors (Indurkhya, 2002) are, in our opinion, the

mechanisms by means of which we get the best explanation of how the analogy takes form in
the interaction view (Black, 1954). In our view, the mechanism of metaphorical
interpretation, which is triggered by recognition that a metaphor has been identified, is a
coherent mapping of a set of features from a source domain to a target domain in order to
obtain a metaphorically restructured target domain; this is a technical way of spelling out

14

analogical reasoning.

16

With these mappings, some properties of the source domain

(specifically, only those which are relevant for obtaining information about the
characterization of the subject matter) are employed as a source of information to describe the
target, so that new properties in the target domain may be generated. These new properties
are not new properties of the source concept or even new properties of the object under
description; they are new just with respect to our conceptualization of the target object. This
process permits us to explain how metaphors communicate new properties that were not
linguistically encodable in the current public language (for the details of the application of
this mechanism in particular examples, see Romero and Soria, 2005). Although no
implicature theorist has, as far as we know, used this approach, it is possible to appeal to
these conceptual mappings in order to elaborate the metaphorical meanings that would form a
part of the implicature.

In order to maintain the position that metaphor is a form of implicature, then, we

should be ready to accept (i) a new way of violating a maxim of the Cooperative principle,
characterized by not requiring the processing of the literal proposition, and (ii) a new way of
working out the implicature that includes the process that explains the analogical reasoning
and that takes the conventional meanings of the words uttered as its starting point, changing
them. As we have just shown, the notion of implicature can be modified until we get an
explanation of the identification and interpretation of metaphors. Even so, an important
question remains: do metaphorical implicatures have the features characteristically
considered essential to the original notion of implicature?

4. Features of Metaphorical Contents

If metaphorical propositions are indeed implicatures, then these propositions should share the
features of implicated propositions. In particular, they should be cancellable and
nondetachable, and their truth conditions should be independent of the truth of the utterance.

However, metaphorical content is, like what is said, not cancellable. The metaphorical

content of (1) is not cancellable because (2)

(2) [In wondering whether to take an umbrella or not, A asked B what the weather is
like today and B utters:] The sky is crying but it is not raining.

is not an admissible and intelligible utterance. Cancellation depends on the possibility of not
having to follow the Cooperative Principle on the level of what is implicated, which is
compatible with going on cooperating by means of what is said. The problem of cancelling
the metaphorical content of (1) is that we cannot cooperate by means of what is said by (1)
because the metaphorical utterance is a case of merely making as if to say. Metaphorical
contents are not cancellable because to deny what is suggested makes the speaker non-
cooperative at all levels.

17

(2) is unintelligible as a whole because nothing is asserted nor

suggested with the first part of the utterance if the second part is really a cancellation of the

16

In the literature, there are other processes or results of processes conceived to explain metaphor such as

saturation to a deictic operator (Stern, 2006) or a semantic descent (Guttenplan, 2006), but as loosening they are
not sufficient to explain the analogical reasoning required in metaphorical comprehension.

17

The same would happen with the explanation of the utterance (5) from the implicature view. The utterance of

‘You are not going to die but you are going to die from that cut’ is unintelligible and with its first part we cannot
signify anything.

15

implicature of the first part. There is no situation in which the right interpretation of the
normal utterance (1) can do without its metaphorical content, but then (2) is inadmissible.

The nondetachability feature of implicatures, as we have said, consists in the

following identity criterion for what is said: two utterances in the same context do not
coincide in what is said if they launch different implicatures. If the Gricean approach in its
new light is right, metaphorical contents are identified and elaborated from subpropositional
contents and not from a literal proposition. Thus, they do not depend on what is literally said.
In this sense, nondetachablility, a necessary condition for an implicature that depends on the
exploitation of a maxim of quantity, cannot apply directly to metaphorical propositional
contents.

Nevertheless, this criterion can be understood also as an identity criterion for the

subpropositional contents that may be involved in what is conventionally said

18

and not

merely as an identity criterion for what is said. The extended criterion would be: two
utterances in the same context do not coincide in subproposional contents that may be
involved in what is conventionally said if they result in different propositional contents. This
new sense for nondetachability may apply to both what is implicated and what is said if what
is said is non-conventionally said. What is non-conventionally said will be also
nondetachable from the subpropositional contents that may be involved in what is
conventionally said. For example, in the explanation of the utterances involving some
referential indeterminacy, in which what is said depends on some contextual information, the
content non-conventionally said is nondetachable from the content that may be part of what is
conventionally said. From this explanation, there is no way, in the same context, to utter both
‘The client is waiting for his check’ and ‘The customer is waiting for his check’ without
referring to the same person, an ingredient of what is non-conventionally said. This
subpropositional content is nondetachable from the conventional contribution that ‘the client’
or ‘the customer’ may make to the proposition expressed by the speaker.

19

Since metaphor detection does not require elaborating a literal content from which

metaphorical content is nondetachable, this nondetachability can be understood as a
nondetachability from the conventional part of the utterance that represents the contextual
abnormality and conceptual contrast, the type of nondetachability that not only conversational
implicatures but also the cases of what is non-conventionally said have. If the metaphorical
content is nondetachable there is no way, in the same context, to make the same contextual
abnormality and the same conceptual contrast without communicating the metaphorical
content. Metaphorical contents are nondetachable of the subpropositional contents that may
be involved in what is conventionally said. The nondetachability of metaphorical
propositional content is compatible both with the proposal that this content is an implicature
and with the proposal that it is a case of what is non-conventionally said.

Finally, there is independence from the truth conditions of the utterance. If the truth or

falsity of that it is raining does not affect the evaluation of (1), the utterance has no truth
conditions by means of which it is evaluated. Thus, the truth value of (1) does not seem to be

18

The requirement that subpropositional contents ‘may be involved in what is conventionally said’ avoids the

contribution of subpropositional contents that take part in conventional implicatures, contents that are
detachable from what is said. Thus, ‘but’ and ‘and’ contribute the same content to what is said even if they have
a different linguistic meaning.

19

The implicature theorist on referential uses of definite descriptions may well consider that the same

subpropositional content, the same person, is involved in what is implicated and is nondetachable from the
conventional contribution that may be made by ‘the client’ and ‘the customer’ to the proposition expressed by
the speaker.

16

independent of the truth value of the proposition that the implicature theorist would say it
merely suggests. If it is not raining the utterance (1) is false. With metaphor, these
propositions are asserted and not merely suggested.

Figure 2. Features of propositional content

Conversational

Implicature

Metaphorical

content

What is said

non-conventionally conventionally

Cancellability +

–

–

–

Nondeta
chability
of

What is said

+

not applicable

not applicable

not applicable

Subpropositional
contents

+ + +

not

applicable

Utterance truth-conditions
independence

+ – –

–

As can be seen in Figure 2, squeezing metaphorical content into the category of

implicature would entail abandoning features that are considered essential to the original
notion of implicature. If the metaphorical content is not cancellable and its truth-value is
essential to evaluating the truth of the utterance, then these two features of the metaphorical
propositional content are identical to these features of what is said and different of those of
what is implicated. In addition, the nondetachability of metaphorical content can only be
understood as nondetachability from the subpropositional contents, given that
nondetachability of metaphorical content from what the speaker makes as if to say is not
applicable. In this second sense of nondetachability, the implicated contents have this feature,
but it is applicable neither to metaphorical contents nor to what is non-conventionally or
conventionally said. In the former, nondetachability is applied to all kinds of propositional
contents included in Figure 2 with the only exception of what is conventionally said. It would
be a feature that would not allow discrimination between implicated propositional content
and what is non-conventionally said. Nevertheless, taking into account the fact that
metaphorical content coincides with what is non-conventionally said not only in the two
senses of nondetachability but also with regard to the other two features, it seems best to
consider metaphorical content as cases of what is said, if we want to classify metaphorical
contents with regard to either of the two notions involved in this debate.

Given this explanation on metaphorical content, the result of metaphorical

interpretation is more naturally located in what is said, as it does not meet the defining
features of implicatures. As metaphorical contents are elaborated with a process that acts
upon the meaning of constituents, we argue that a process of interpretation, when operating
on the linguistic meanings of the sentence constituents, characteristically produces contents
included in the speaker’s intended proposition expressed or in what is said. The input of
particularized implicatures is content (either propositional or subpropositional), and not
linguistic meaning. Particularized implicatures are propositions that are not recovered from
linguistic meaning but from content and context. By contrast, the input of what is
(conventionally or non-conventionally) said is, as it happens with metaphor, linguistic
meaning. In metaphorical interpretation we get an intended proposition which includes
subpropositional metaphorical provisional contents, a proposition that fixes the truth
conditions of the metaphorical utterance under the mode of presentation imposed by the
metaphorical provisional meaning of the sentence constituents. In this sense, when we argue
for the notion of what is metaphorically said, we challenge the proposal that what is said is

17

always literally said. But this was already suspected by Grice himself when he (1987, p. 361)
claimed that there are cases of dictiveness without formality. What is non-conventionally
signified can also be said.

20

5. Concluding Remarks on Literalism vs. Contextualism

Once we accept that what is said can be metaphorically said, the question is whether the best
explanation leads to a literalist or a contextualist position on what is metaphorically said. As
we said in the introduction, a contextualist argues that what is said by the speaker is not always
literally said, that what is said (the truth-conditional content of the utterance) is an aspect of
speaker’s meaning, and that what is said by an utterance should be explained considering optional
pragmatic processes, processes that are triggered independently of the syntactic and semantic
features of the uttered sentence. A contextualist on metaphor, thus, argues that metaphors are
interpreted by an optional pragmatic process. A literalist, against what might seem, may accept
the two first claims, what she would never defend is that what is said should be explained
considering optional pragmatic processes. Then, our question now is whether what is
metaphorically said is better explained by means of a mandatory or an optional pragmatic
interpretive process.

In our opinion, what is metaphorically said is achieved by a subpropositional

pragmatic process which, in spite of not being always linguistically mandated, is never
optional from the truth conditional point of view (optional

T

). Its mandatory character is due to

the fact that most metaphorical bearers cannot fix a literal proposition. The identification of
metaphor achieved by the two criteria discussed in 3.1 blocks literal interpretation in order to
avoid a route that leads to no propositional content (or at best to an irrelevant propositional
content) and triggers the mechanism of metaphorical interpretation. Then approaches that
argue for the mandatory

T

character, such as Stern’s and ours, must be right in this respect.

No

proposition expressed by a metaphorical utterance can ever be literal. With all due respect to
Davidson, our position is that metaphors mean what the words, in their most metaphorical
interpretation, mean.

21

Esther Romero

Department of Philosophy I

University of Granada

Belén Soria

Department of English and German Philologies

University of Granada

Number of words: 9064

20

The cases of dictiveness without formality are cases such as the sentence ‘He is an evangelist’ uttered in a

context in which he is not an evangelist. The speaker meant what in fact his words said; but some of his words
would be dictive and their content would be nonformal (nonconventional) since no part of their conventional
meaning is involved in dictiveness. With this utterance, the speaker says that he is a sanctimonious,
hypocritical, racist, reactionary, money-grubber and the conventional meaning of ‘evangelist’ is not involved in
what is said.

21

The part of the famous passage from Davidson alluded to here is: ‘… metaphors mean what the words, in

their most literal interpretation, mean ….’ (1978, p. 31).

18

References
Black, M. 1954: Metaphor. Proceedings of the Aristotelian Society, 55, 273-94.
Bezuidenhout, A. 2001: Metaphor and what is said: A defense of direct expression view of

metaphor. Midwest Studies in Philosophy, 25, 156-86.

Camp, E. 2006: Contextualism, metaphor, and what is said. Mind and Language, 21(3), 280-

309.

Clark, H. 1979: Responding to indirect speech acts, Cognitive Psychology, 11, 430-77.
Davidson, D. 1978: What metaphors mean. Critical Inquiry, 5, 31-47.
Gibbs, R. 1983: Do people always process the literal meanings of indirect requests? Journal

of Experimental Psychology: Learning, Memory, and Cognition, 9, 524-33.

Gibbs, R. 1984: Literal meaning and psychological theory. Cognitive Science, 8, 275-304.
Gibbs, R. 1986: Skating in thin ice: Literal meaning and understanding idioms in

conversation. Discourse Processes, 9, 17-30.

Gibbs, R.W. 1992: When is metaphor? The idea of understanding in theories of metaphor.

Poetics Today, 13(4), 574-606.

Gildea, P. and Gluksberg, S. 1983: On understanding metaphor: The role of context. Journal

of Verbal Learning and Verbal Behavior, 22, 577-90.

Giora, R. 1999: On the priority of salient meanings: Studies of literal and figurative language.

Journal of Pragmatics, 31, 919-29.

Glanzberg, M. 2005: Presuppositions, truth values, and expressing propositions. In G. Preyer and

G. Peters (eds), Contextualism in Philosophy: Knowledge, Meaning, and Truth.
Oxford : Oxford University Press.

Grice, H.P. 1961: The causal theory of perception. Proceedings of the Aristotelian Society,

35, 121-52.

Grice, H.P. 1975: Logic and conversation. In P. Grice (ed.) 1989, Studies in the Way of

Words. Cambridge: Harvard University Press.

Grice, H.P. 1978: Further notes on logic and conversation. In P. Grice (ed.) 1989, Studies in

the Way of Words. Cambridge: Harvard University Press.

Grice, H.P. 1987: Retrospective epilogue. In P. Grice (ed.) 1989, Studies in the Way of

Words. Cambridge: Harvard University Press.

Guttenplan, S. 2006: The transparency of metaphor. Mind and Language, 21.3, 333-59.
Indurkhya, B. 1992: Metaphor and Cognition: An Interactionist Approach. Dordrecht:

Kluwer Academic Publishers.

Inhoff, A., Lima, S. and Carroll, P. 1984: Contextual effects on metaphor comprehension in

reading. Memory and Cognition, 12, 558-67.

Keysar, B. and Glucksberg, S. 1992: Metaphor and communication. Poetics Today, 13(4),

633-58.

Kittay, E. F. 1987: Metaphor. Oxford: Clarendon Press.
Loewenberg, I. 1975: Identifying metaphors. Foundations of Language, 12: 315-38.
Onishi, K. and Murphy, G. 1993: Metaphoric reference: When metaphors are not understood

as easily as literal expressions. Memory and Cognition, 21, 763-72.

Ortony, A., Schallert, D., Reynolds, R. and Antos, S. 1978: Interpreting metaphors and

idioms: Some effects of context on comprehension. Journal of Verbal Learning and
Verbal Behavior, 17, 465-77.

Recanati, F. 1987: Meaning and Force: The Pragmatics of Performative Utterances.

Cambridge: Cambridge University Press.

Recanati, F. 1995: The alleged priority of literal interpretation. Cognitive Science, 19, 207-32.
Recanati, F. 2004: Literal Meaning. Cambridge: Cambridge University Press.

19

Romero, E. and Soria, B. 1997/8: Stylistic analysis and novel metaphor. Pragmalingüística,

5-6, 373-89.

Romero, E. and Soria, B. 2003: Comunicación y metáfora. Análisis Filosófico, 23.2, 167-92.
Romero, E. and Soria, B. 2005: Metaphoric concepts and language. In J.J. Acero and P. Leonardi

(eds), Facets of Concepts. Padova: Il Poligrafo.

Romero, E. and Soria, B. 2006: Novel metonymy and novel metaphor as primary pragmatic

processes. In Where Grammar Meets Discourse. Functional and Cognitive
Perspectivas. Servicio de publicaciones de la Universidad de Córdoba.

Romero, E. and Soria, B. 2007: A view of novel metaphor in the light of Recanati’s

proposals.

In M.J. Frápolli (ed.), Saying, Meaning and Refering. Essays on François

Recanati’s Philosophy of Language. Basingstone: Palgrave.

Romero, E. and Soria, B. forthcoming: On phrasal pragmatics and what is descriptively referred

to. Croatian Journal of Philosophy.

Searle, J. 1979: Metaphor. In J. Searle (ed.) Expression and Meaning. Studies in the Theory

of Speech Acts. Cambridge: Cambridge Universtiy Press.

Sperber, D. and Wilson, D. 1986/1995: Relevance. Communication and Cognition, Oxford:

Blackwell.

Stanley, J. 2005: Semantics in context. In G. Preyer and G. Peters (eds), Contextualism in

Philosophy: Knowledge, Meaning, and Truth. Oxford : Oxford University Press.

Wilson, D. and Carston, R. 2006: Metaphor, relevance and the ‘emergent property’ issue.

Mind and Language, 21.3, 243-79.

Stern, J. 2006: Metaphor, literal and literalism. Mind and Language, 21.3, 243-79.