Introduction

One of the most controversial areas in the analysis of meaning is the issue of metaphor. While the existence of metaphor clearly reveals the creative aspect of human language, linguists and philosophers of language disagree over the extent to which it can be handled within a purely semantic framework. According to Grice metaphor manisfests itself in the violation of the linguistic rules, and, therefore, is a feature of pragmatics rather than semantics. He incorporates it into his notion of conversational implicature because, as he claims, it represents a ‘categorial falsity’ and, therefore, should be studied exclusively within the domain of implied meaning. To begin with, it may be useful to examine Grice’s notion of meaning, which he introduces by first drawing a distinction between the ‘natural sense or senses of an expression’ (meaning n) and its non-natural sense or senses’ (meaning nn) (Grice 1971: 53) respectively in the following examples :

1. Those spots mean measles

2. Those three rings on the bell (of the bus) mean that the bus is full

By performing the action of ringing three times the bell of the bus, the conductor non-naturally means that the bus is full. As communication is rule-governed, the conductor of the bus does not need to produce the linguistic utterance “The bus is full; don’t get on it”, to get the message across. The utterance would non-naturally mean something similar. Grice formulates an analysis of meaning nn along the following lines:

… for x to have meant nn anything not merely must it have been uttered with the intention of inducing a certain belief but also the utterer must have intended an audience to recognize the intention behind the utterance. (ibid. 1971 :53)

In other words, meaning nn relies upon the utterer’s meaning in that the ’M intended effect” is that the hearer should believe something or that the hearer should do something. Instances of meaning nn as presented by Grice (1975) consist of such utterances as “a police officer waving to a car driver to cause him to stop”, “a conductor ringing the bell three times to stop passengers getting on the bus”, or “an individual showing a painting depicting Mrs. X showing familiarity to Mr. Y”.

Grice attempts to capture the notion of meaning nn in terms of utterer’s meaning, which relies upon the utterer’s intention to do something as well as the audience recognition of that intention. This amounts to saying that what an utterance means is definable in terms of what speakers mean on occasions. Another instance of meaning nn takes the form of conversational implicature.

1. Conversational Implicature 

An account of conversational implicature (henceforth CI) will be presented in this section, followed by examples as well as conditions and characteristics of CI.

1.1 What is conversational implicature?

Prior to elaborating on the notion of conversational implicature (henceforth CI), it is necessary to briefly mention Grice’s rules of conversation. He argues that talk-exchanges have a communicative purpose; we use words and sentences to contribute to a given conversation, and to provide required information. Of course, there are cases where this is not altogether true: people sometimes talk with no particular intention of communicating something. More than that, they may not even consider the question of providing information. We shall ignore these possibilities and proceed to the description of typically communicative talk-exchanges. Grice’s major contribution takes the form of a “Cooperative Principle”. In order to be cooperative, therefore, you should

… make your converastional contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk-exchange in which you are engaged. (Grice 1975:  45)

Grice notes that a number of maxims follow from the Cooperative Principle (CP):

1. Quantity: Make your contribution as informative as is required.

2. Quality

• Do not say what you believe to be false.

• Do not say that for which you lack adequate evidence.

1. Relation and Manner: Be brief, orderly, and perspicuous.

1.2 When is a conversational implicature born?

The deliberate violation of one or more maxims results in a conversational implicature, which rests upon the distinction between what is said and what is meant (or, more precisely, conversationally implied). Let us consider Grice’s own example.

A tutor, who is requested to give a reference for a former student applying for a lecturing post in philosophy, writes:
“Dear Sir, Smith’s handwriting is beautiful, and his attendance at tutorial has been regular. Yours faithfully.”

The recipient of the letter above is unlikely to offer the post to Smith because he understands that the tutor implies that Smith is hopeless at philosophy and, therefore, should not be offered the post. To the question of how the recipient of the letter of recommendation infers such information, the following steps are suggested:

1. 1. Given that the tutor is asked to inform someone about Smith’s competency in philosophy, and not his handwriting, for instance;

   2. The tutor is able to provide this information;

   3. Given also that the tutor is being cooperative (otherwise, why should s/he bother to write the letter at all?);

   4. The tutor has blatantly flouted a maxim: that of quantity; s/ he did not say enough. If s/he makes no mention of Smith’s philosophical abilities, s/he probably wants the recipient of the letter to infer that Smith is no good at philosophy.

Conversational implicature stands for any implied meaning which the conventional, literal usage of words and sentences does not carry at face value. One prominent characterisation of CIs can be summarised as “what is meant is not what is said”. Their existence, however, assumes the speaker’s observance of the CP, which, although permitting the speakers to violate one or more maxims, still urges them to be cooperative in a given talk-exchange. Cases of deliberately misleading the hearer, despite their being examples of the breach of maxims, fall outside the range of conversational implicatures because of their unobservance of the CP.

1.3 Examples of conversational implicature : Irony and Metaphor

According to Grice, irony and metaphor represent a case of conversational implicature: they are examples of the breach of the maxim of quality because, in issuing a metaphor or an ironical statement, the speaker has made an untrue statement.

1.3.1 Irony

Suppose an individual A is talking about a person B who has recently betrayed him. In the presence of an audience who knows that, A may say “B is a good friend”. The audience immeditely sees what A implies: in using his ironical utterance, A actually implies that B is not a good friend at all. This implicature is calculated from the blatant violation of the maxim of quality, in that A says what he believes to be false. Given sufficient knowledge of the circumstances – sufficient knowledge of the context and participants – the implicature is successfully worked out. Likewise, an ironical statement is, by definition a false statement, as in John is brilliant when, in fact, John is not brilliant at all.

1.3.2 Metaphor

Saying “John is a pig” is making a false statement because John is literally not a pig.
In the metaphorical statement “You are the cream in my coffee”, the speaker implies that the addressee is his/her “pride and joy” (Grice 1975 : 53).

Grice classifies metaphor and irony as cases triggering a conversational implicature because they violate the maxim of quality in that their production involves a categorial falsity’(ibid). Assuming that the speaker is still being cooperative but is blatantly violating one of the maxims, the hearer must work out that the speaker implies something different from what s/he says.

1.4 Characteristics of conversational implicature

Conversational implicature stands for any implied meaning which the conventional, literal usage of words and sentences does not actually bear. Grice, who considers irony and metaphor as indisputable examples of conversational implicatures, suggests the following criteria for testing for conversational implicature.

1. Calculability: The generation of a CI assumes the observance of the CP and its maxims. In other words, it is only on the assumption that the CP is being respected that the flouting of one or more maxims is calculated as an attempt by the speaker to imply a given meaning (or a particular covert message).

2. Context: CI is context-sensitive: an utterance will give rise to a variety of interpretations, depending on a number of parameters such as context and shared beliefs and values between conversationalists.

3. Indeterminacy: The exact implicature resulting from an utterance is not always straighforward (i.e., it is indeterminate) as it depends on context and situation, namely.

4. Cancellability: Unlike conventional meaning, a CI may be cancelled without leading to a contradiction. For example, a speaker, without sounding odd, may deny a CI that may have arisen from his utterance, as in the case of someone saying “John is impossible”, in which case the hearer may infer that the speaker dislikes John. Yet the speaker is still in a position to deny this implicature by adding “but I like him”.

5. Non-conventionality: Since conversational implicatures do not correspond to the direct meaning of language, they are unlikely to be conventionally carried out. This means that they do not depend on sentence meaning.

There are, on the other hand, other types of implicatures, overtly handled by logical connectors such as “and”, “but”, which Grice labels conventional implicatures. For example, in issuing the statement “He is British, so he is generous”, one conventionally implies that British people are, in general, generous.

2. Challenging Views on Grice’s Implied Meaning

Grice’s account of meaning has been criticized by many philosophers of language, namely Searle (1969) and Wright (1978), Sadock (1979), Ortony (1979), Rumelhart (1979), and Levin (1971).

2.1 Searle’s view on Grice’s implied meaning

Searle observes that implied meaning, as suggested by Grice, fails to account for the idea that meaning is governed by rules and conventions. He notes that

… this account of meaning does not show the connection between one’s meaning by what that one says actually means in the language. (Searle 1969 : 43)

This remark is derivable from Searle’s original distinction between sentence meaning and utterance meaning. To substantiate his claim, Searle cites the example of the W.W. II American soldier uttering “Kennst du das Land wo die Zitronen blûhen”, in order to cause his audience (Italian troops who did not have much knowledge of German themselves) to think that he meant “Ich bin ein deutscher Soldat” (I am a German officer). What the sentence actually means is “Knowest thou the land where the lemon trees bloom?”(Searle 1969 :4). If, according to Grice (1975), what a sentence means is derivable from utterer’s meaning (the intended effect the utterer wants to achieve), then “Kennst du das Land wo die Zitronen blûhen” would be said to mean “I am a German officer” because that is what the speaker intends his audience to believe (i.e., that he is a German officer). Grice’s notion of meaning is defective in various other respects, as discussed at great length by Searle (1971). One of Searle’s basic criticisms may be briefly summarized as follows : first, sentence meaning cannot be defined purely in terms of what speakers mean on occasions; secondly, the intended effect of meaning something is derivable on the basis of the propositional content of a given sentence, and given Grice’s original account (1975), there are no obvious ways of doing so. It is crucial to Searle’s theory of meaning that sentence meaning (propositional content) be distinguished from utterance meaning. The success of an intended utterance depends on at least two factors: conventions and context. Whereas sentence meaning can be handled in purely semantic terms, utterance meaning is subject to an analysis that goes beyond semantics (or the conventionally conveyed), and this is known as pragmatics, an analysis that involves speakers, hearers, context, and so on.

Searle observes that both sentence meaning and utterance meaning are closely related in cases of “indirect speech acts”. For example, in “Can you pass the salt?” the sentence meaning is a question about the addressee’s ability to pass the salt. In terms of utterance meaning, it represents an indirect request addressed to someone to pass the salt. Clearly, both sentence meaning and utterance meaning are related by means of a conventional tie: the propositional content (i.e., “pass the salt”). Indirect speech acts are characterized as such because, while sentence meaning is still acceptable (i.e., it represents a yes/no question about someone’s ability to do something) but is irrelevant in that particular context, the hearer has to look for a possible conveyed meaning (such as that related to requesting him/her to do something). This means that, in indirect speech acts, the sentence produced has meaning but is insufficient in providing the intended effect the speaker wants to achieve. As a result, the hearer has to look for the intention of the speaker beyond the meaning of the sentence used, and this may be achieved with the aid of the context in which the utterance occurs. In other words, for Searle, an indirect speech act is what the sentence says, and more.

2.2 Wright on the role of intentions in Grice’s implied meaning

In a similar vein, Wright observes:

… the speaker cannot simply intend; there must be a common convention between speaker and his audience, and there must also be appropriate circumstances. (Wright 1978 : 372)

For blurring the distinction between conventions and intentions results in a “Humpty Dumpty meaning of meaning”(ibid. 372) where anyone with a given intention can utter anything, regardless of the conventional meaning of words and sentences.

3. Conversational Implicature and Metaphor: Semantics or Pragmatics?

Grice’s statement that metaphor is a case of conversational implicature has met several reactions on the part of theorists and philosophers of language. Proponents of metaphor as a feature of utterance meaning – pragmatics – suggest that it is an elliptical simile while those maintaining a semantic reading of metaphor state that, given its conventionality, it belongs to sentence meaning.

3.1 Sadock on Grice’s non-conventionality of conversational implicatures

Sadock argues that the criteria proposed by Grice, and which are expected to throw light on the distinction between the conventional and the non-conventional are circular. Furthermore, non-conventionality is a feature of conversational implicature “by definition”, and were this phenomenon classified better, “there would not be any need for other criteria” (Sadock 1979: 285). Non-conventionality cannot be a criterion used for testing for conversational implicature; instead, it represents what a conversational implicational should be. For conversational implicature is, by definition, non-conventional. Sadock also argues that cancellability turns out to be an indisputable feature of ambiguity, too, with this difference that while conversational implicatures are socially based, ambiguous sentences emanate from sentences themselves and, as a result, belong to the study of language per se.

3.2 Sadock: conventionalisation of metaphors

Sadock’s analysis is consistent with Searle’s claim that metaphor has to be studied within the theory of language use. He remarks that figuration, like any other non-literalness, has social and psychological correlates, and therefore, cannot be studied in purely linguistic terms. Despite his characterisation of figurative speech, including metaphors, as not being linguistic but human in nature, Sadock insists that, “locative deictics”, for example, have undergone a “freezing” (conventionalisation) process in practically all languages, starting from English to Eskimo. Such metaphors are so deeply institutionalized that their interpretation as metaphors is beyond our reach. He does not believe it is possible to draw a clear line between convention and figuration. Rather, they represent a continuum.

3.3 Ortony: Metaphor as an elliptical simile

The statement that metaphor is an elliptical simile has raised a number of issues. If this claim is true, then the principal feature of metaphor as a semantic falsehood which makes it a case of CI, is open to doubt. The reason for this is that similes, or statements of similarity, could be true, and are, therefore, literally interpreted. Ortony (1979) remarks that “the reduction of metaphor to similes will contribute nothing to their solution (p. 189). He goes on to argue that, in any case, similes are literally false, considering, for example, the following (p. 191) :

1. Encyclopedias are like dictionaries.

2. Encyclopedias are like goldmines.

He labels (1) as ‘literal comparison’ and (2) as non-literal comparison” because, whereas the former is literally true, the latter is literally false.

3.4 Rumelhart : “two ends of a scale”

Between Searle’s and Cohen’s two opposite poles, there exists an intermediary position which is represented by Sadock (1978) and Rumelhart (1979) when the latter states that “literal and figurative usage are at two ends of a scale” (Rumelhart 1979: 78). Rumelhart (1979) is aware of the difficulty involved in the attempt to distinguish literal from figurative usage. His claim is based on a language-acquistion process which breaks down the distinction between convention and figuration when he observes that, “[f]or the child, the production of literal and non-literal speech may involve exactly the same process” (p. 78). He, therefore, does not comply with the Gricean pragmatic analysis of metaphor.

3.5 Searle

Searle argues that metaphor is a feature of utterance meaning, and, therefore, should belong to the realm of pragmatics.

3.5.1 Metaphor as a feature of utterance meaning

In his paper “Metaphor” (1978), Searle advocates the characterization of metaphor as a feature of utterance meaning, and one of the main questions raised is concerned with the processes whereby a metaphorical interpretation is made. Searle draws an explicit distinction between sentence meaning – semantics – and utterance meaning – pragmatics. In his view, metaphor is best understood within the realm of pragmatics. Sentence meaning is linked with the notion of the direct locution, which is the conventional, literal meaning. Utterance meaning, or the indirect illlocution, carries not only the literal meaning but also any further implications of it. In order to understand the indirect illocution of an utterance, the hearer must take into account contextual parameters. He goes on to observe that, if we can define the indirect speech act by stating that it is the literal meaning and something more, the situation for metaphors is quite difficult to pin down. Therefore, one of the most important characteristics that Searle ascribes to metaphor is its “semantic nonsense” (Searle 1978: 114): a metaphor is detectable by its obvious falsehood. For example, Smith is a pig, is literally false because Smith is not a porcine mammal’, but it may be true if taken in a metaphorical sense. Searle observes that knowing how metaphor works involves prior knowledge of the distinction between sentence meaning and utterance meaning.

Literal meaning corresponds to sentence meaning, which, in its turn, is analyzed in terms of truth-conditions. Stating, for example, that “X is on P” would be an assertion whose meaning – truth – depends on the actual context in which it occurs; it is, more specifically, related to given factual background information. A metaphor, in Searle’s view, manifests itself in exactly the opposite direction: it does not correspond to a specific background information, but it is, instead, attributed the quality of semantic nonsence. A sentence such as “Smith is a pig’is semantically deviant because it is false – it fails to refer in the actual world (i.e., Smith is a porcine mammal’ is false). By this, however, the speaker means that ‘Smith has unpleasant habits’ and his/her use of ‘Smith is a pig’ is only metaphorical. This amounts to saying that, if sentence meaning does not apply, then a metaphorical reading is required. Searle defines the meaning of indirect speech acts as what the speaker says and more. The basic difference between indirect speech acts and conversational implicatures is the absence in CIs of the propositional link between sentence meaning (what the speaker says) and utterance meaning (what the speaker means).

3.5.2 Searle’s strategy for understanding a metaphor

There are, according to Searle, three stages which the hearer must undergo to understand a given metaphor. Detecting a metaphor represents the first stage, which may be achieved in the following manner: ‘Where the utterance is defective taken literally, look for an utterance meaning that differs from sentence meaning’ (Searle 1978: 114). In interpreting ‘P is R’, for instance, the hearer knows that, taken literally, the sentence is false, and should, therefore, be taken metaphorically. Secondly, once a metaphorical reading is required, the hearer must find out some strategy of discovering the similarities between ‘P’ and ‘R’. Thirdly, as there may be various similarities between ‘P’and “R”, the hearer must make use of some principles wherein he could restrict the values of R that are identical in P.

3.6 Morgan on Searle’s Metaphor as a Semantic Deviance

Morgan (1979) argues that the “semantic deviance” advocated by Searle as a feature for detecting metaphors is inadequate for the interpretation of a considerable fraction of metaphors. For example, the sentence “The king Tut exhibit is a pharaoh’s burial treasure” may be interpreted metaphorically despite its being semantically (i.e., literally) functional. He further states that the strategy used for detecting metaphors – i.e., by way of “semantic falsehood” – does not apply to examples of the type “The rock is becoming brittle with age” (Morgan (1979): 137) since the same sentence may be interpreted either literally – in the case of a geological expedition – or metaphorically – in the context of a student referring to some old professor. Now it seems that context plays a crucial role in detecting some metaphors: had the sentence been uttered during a geology expedition, there would have been no need for a metaphorical interpretation. If, on the other hand, the same sentence had been produced out of its “normal” context, it would have represented a semantic falsehood, and would have, therefore, required a metaphorical reading. As a result, semantic falsehood and context are useful criteria for detecting metaphors.

Morgan argues that reliance on context is by no means true of all metaphors: some metaphors are understood as such with no reference to context. In interpreting “Sam is a pig”, the hearer does not require any contextual clues, except, perhaps, in the case where Sam is a porcine mammal. What is needed, therefore, is the reference of the predicated noun or proper name.

3.7 Cohen and Levin: on the semantics of metaphor

In his The Semantics of Metaphor, Cohen (1978) presents a challenging view to the pragmatic analysis proposed by Searle. He maintains that metaphor, unlike speech acts, is analysed as a “feature of sentence-readings” (p. 64). He posists that there is no dispute concerning the meaning of “inflamed passion”, “rain of blows”, or “feeble argument” (p. 65) because such expressions, previously introduced as live metaphors, have acquired a certain degree of conventionality and are now incorporated into the semantics of language. Cohen’s view is consistent with Levin’s (1977) concept of “dead metaphors”:

Such expressions, created for the nonce, are deviant at their inception and hence are not covered by the grammar; as they become taken up by more and more speakers, however, they after a time move into a set of well-formed expressions and have to be genrated by the grammar – they become aggramatized, we might say (p.31)

4. Metaphors, Cancellabillity, and Conversational Implicature

Metaphorical statements, by virtue of their conventionality, are usually fairly understood and well interpreted by hearers, and as such, they are not cancellable. In this section, we shall discuss what is cancellable (or not) in metaphors used in what is meant to be conversationally implied.

4.1 The conventionality of metaphor

The claim that a metaphor is dead as soon as it is born means that metaphors have been rendered conventional through extensive use and, therefore, should pose no difficulty in being assigned a meaning. This implicates that the feature of non-conventionality that characterises conversational implicature does not apply to metaphor.

4.2 What exactly is cancellable?

While still assuming that, in issuing a metaphor, the speaker is producing a literally false statement, the hearer is able to calculate the implied meaning behind this metaphorical statement, given proper circumstances and a set of inference principles. The problem arises with Grice’s characterization of metaphors as being cancellable.

Consider the two contexts below:

1. Context 1
   (A) Do you like X?
   (B) He’s a pig.

1. Context 2
   (A) What is X like?
   (B) He’s a pig.

In context 1, B says that X is a pig, probably meaning that X is like a pig in some ways (i.e., X has unpleasant habits, etc.); by this, B implies that s/he does not like X, and it is the implicature that B dislikes X that is cancellable since B may still deny it by adding “ … but I like X”.

In context 2, B says that X is a pig, meaning that X is like a pig in some ways or manners, in which case this is not cancellable. In other words, the utterer of “X is a pig” cannt deny s/he said that X is a pig.

The difference in cancellability between B’s replies in each of the contexts (X is a pig) lies, in fact, in A’s question about X. In context 1, B’s answer involves his sympathy with X while in context 2, B’s answer refers to a description of X, in which case the meaning of the metaphor “X is a pig” is not cancellable, at least on the grounds that metaphors of this sort have been fully internalised and are therefore part of sentence meaning. Even considering that, in context 2, A may wish to know B’s opinion about X -or whether B likes X or not – one cannot state that B’s metaphorical production of “He’s a pig” itself constitutes an implicature. It is, rather, the implicature behind the use of the metaphor that is cancellable. In other words, would have B’s reply been a literal sentence (i.e., X has unpleasant habits) or a metaphorical one (X is a pig), the utterer’s meaning is still the same and is, therefore, not cancellable as such. Because implicatures imply and do not implicate. Metaphorical statements clearly implicate a given meaning and, therefore, they are not cancellable.

Conclusion

Grice’s characterisation of metaphor as a feature of utterer’s meaning has triggered a fierce debate. What can be concluded from the controversial views presented in this paper is that the cancellabilty of metaphor is rendered impossible for at least two reasons. Firstly, if, for example, I say of someone that s/he is a pig (a gorilla), the addressee, even without reference to any particular context, knows that I am speaking metaphorically, and because metaphors have been incorporated into language and internalised in the speakers’ and hearer’s minds for so long that many of them even pass unnoticed as metaphors, as in the case for ‘devour a book’, ‘the mouth of the river’, etc. The second argument that supports the non-cancellability of conversational implicatures is that the meaning of the metaphor, what the speaker says in a sentence (i.e., what the sentence implicates), cannot be cancelled. Instead, it is the implicature resulting from the author of the utterance within a given talk-exchange that is cancellable.