This paper considers two post-Gricean attempts to provide an explanatory account of verbal irony. The first treats irony as an echoic use of language in which the speaker tacitly dissociates herself from an attributed utterance or thought. The second treats irony as a type of pretence in which the speaker “makes as if” to perform a certain speech act, expecting her audience to see through the pretence and recognise the mocking or critical attitude behind it. The two approaches have sometimes been seen as empirically or theoretically indistinguishable, and several hybrid accounts incorporating elements of both have been proposed. I will argue that the echoic and pretence accounts are distinguishable on both theoretical and empirical grounds, and that while echoic use is essential to standard cases of verbal irony, pretence is not. However, the term irony has been applied to a very wide range of phenomena, not all of which can be explained in the same way, and I will end by briefly mentioning some less central cases where varieties of pretence or simulation do indeed achieve ironical effects.
1. Introduction Here are some typical examples of verbal irony:
Mary(after a difficult meeting): That went well.
As I reached the bank at closing time, the bank clerk helpfully shut the door in my face.
Tim Henman is not the most charismatic tennis player in the world.
The point of these utterances is not to claim what they would be taken to claim if uttered literally (that the meeting went well, the bank clerk behaved helpfully, and there are more charismatic tennis players than Tim Henman), but to draw attention to some discrepancy between a description of the world that the speaker is apparently putting forward and the way (she wants to suggest) things actually were. A hearer who does not recognise this will have misunderstood, and a speaker who doubts the hearer’s ability to recognise it on the basis of background knowledge alone may provide additional clues (for instance, an ironical tone of voice, a wry facial expression, an incongruity or exaggeration, as in (2), or a superlative, as in (3)).1 The ability to understand simple forms of irony is normally present from around the age of 6, and is known to be impaired in autism and certain forms of right hemisphere damage.2 The goal of pragmatics is to describe this ability and thus explain how irony is understood.
According to classical rhetoric, verbal irony is a trope, and tropes are utterances with figurative meanings which relate to their literal meanings in one of several standard ways. In metaphor, the figurative meaning is a simile or comparison based on the literal meaning; in irony proper, as in (1) and (2), it is the opposite of the literal meaning; and in ironical understatement, as in (3), it is a strengthening of the literal meaning. These definitions are part of Western folk linguistics and can be found in any dictionary. To turn them into an explanatory theory, we would need, first, a definition of figurative meaning, second, a method of deriving figurative meanings from their literal counterparts, and third, some rationale for the practice of substituting a figurative for a literal meaning. If figurative meanings are assigned by the grammar, we need an explicit mechanism for deriving them; if they are pragmatically inferred, we need an account of how the inference is triggered, what form it takes, and what types of outputs it yields.
In a few cases, what starts out as a creative use of irony may become fully lexicalised or grammaticalised.3 However, the interpretation of tropes in general is so highly context-dependent that it is most unlikely to be dealt with entirely in the grammar. Grice’s brief discussion of tropes (Grice, 1967/1989: 34) was the first serious attempt to analyse them using pragmatic machinery independently needed for the analysis of ordinary literal utterances. As is well known, he treats irony, metaphor, hyperbole and meiosis as blatant violations of the first maxim of Quality (“Do not say what you believe to be false”), designed to trigger a related true implicature: in the case of metaphor, this would be a simile or comparison based on the literal meaning, in the case of irony it would be the contradictory or contrary of the literal meaning, and in the case of understatement it would be something stronger than the literal meaning. On this approach, the implicatures of (1)-(3) above would include (4a)-(4c):
(4) a. That meeting didn’t go well.
b. As I reached the bank at closing time, the bank clerk unhelpfully shut the door in my face.
c. Tim Henman is far from being the most charismatic tennis player in the world.
The proposal to replace encoded figurative meanings by pragmatically derived implicatures is a step in the direction of a genuinely explanatory account of tropes. It is only a first step, though: in other respects, Grice’s account of tropes is simply a modern-dress variant of the classical account, and shares many of the same weaknesses. In particular, it does not explain why a rational speaker should decide to utter a blatant falsehood in order to convey a related true implicature which could just as well have been literally expressed. In later work, Grice acknowledges that his account of irony is insufficiently explanatory (although he does not seem to have had similar worries about his parallel accounts of other tropes), and mentions some additional features of irony which may be seen as intended to supplement his account or point in the direction of an alternative account; I will touch on these briefly in discussing Grice’s approach to irony in section 2.
However, my main concern in this paper is with two post-Gricean attempts to provide a rationale for irony in which the blatant violation of a pragmatic maxim or principle of literal truthfulness play no explanatory role (although, as noted above, the fact that an utterance would be blatantly false or inappropriate if literally understood may be a useful clue to the presence of irony). One approach, first proposed by Sperber and Wilson (1981), treats verbal irony as a type of echoic allusion to an attributed utterance or thought. On this approach, the speaker of (1) is not herself asserting that the meeting went well, but expressing her own reaction to a thought or utterance with a similar content which she tacitly attributes to someone else (or to herself at another time), and which she wants to suggest is ludicrously false, inadequate or inappropriate. Thus, Mary might use (1) to communicate that it was ridiculous of her to think that the meeting would go well, stupid of her friends to assure her that it would go well, naïve of her to believe their assurances, and so on. Mary echoes a thought or utterance with a similar content to the one expressed in her utterance, in order to express a critical or mocking attitude to it. More generally, the main point in typical cases of verbal irony such as (1)-(3) is to express the speaker’s dissociative attitude to a tacitly attributed utterance or thought (or, more generally, a representation with a conceptual content, for instance a moral or cultural norm), based on some perceived discrepancy between the way it represents the world and the way things actually are (Sperber and Wilson, 1981, 1986, 1990, 1998; Wilson and Sperber, 1992).
The second approach, which is suggested by the etymology of the word irony and has a much longer history, treats verbal irony as a type of pretence. On this approach, the speaker of (1) is not asserting but merely pretending to assert that the meeting went well, while expecting her audience to see through the pretence and recognise the critical or mocking attitude behind it (see, for instance, Clark and Gerrig, 1984; Currie, in press; Recanati, 2004; Walton, 1990). Similarly, the speaker of (2) is merely pretending to have found the bank clerk’s behaviour helpful, and the speaker of (3) is merely pretending to give serious thought to the possibility that Tim Henman might not be the most charismatic tennis player in the world.
Both echoic and pretence accounts reject the basic claim of the classical and standard Gricean accounts, that the hallmark of irony is to communicate the opposite of the literal meaning. Both offer a rationale for irony, and both treat ironical utterances such as (1)-(3) as intended to draw attention to some discrepancy between a description of the world that the speaker is apparently putting forward and the way things actually are. These similarities have provoked conflicting reactions. On the one hand, the two approaches are sometimes seen as empirically or theoretically indistinguishable; several hybrid versions incorporating elements of both echoic and pretence accounts have been produced, and the boundaries between them have become increasingly blurred. On the other hand, some defenders of both echoic and pretence accounts see their own approach as proving the key to irony and the other approach as offering at best an incidental sidelight.4 I want to consider whether this is a largely terminological debate of interest mainly to sociologists of academic life, or whether there is some genuine substance behind it.
In rhetorical and literary studies over the years, the term irony has been applied to a wide variety of loosely related phenomena ranging from Socratic irony, situational irony, dramatic irony, Romantic irony, cosmic irony and irony of fate to verbal irony and various forms of parody, wit and humour.5 Not all of these phenomena fall squarely within the domain of pragmatics, defined as a theory of overt communication and comprehension. Some are clearly forms of echoic allusion, others are more closely related to pretence; some involve both echoing and pretence, while others have no more in common with (1)-(3) than the evocation of a similar attitude or the presence of some perceived discrepancy between representation and reality. There is no reason to assume that all these phenomena work in the same way, or that we should be trying to develop a single general theory of irony tout court, based on either the pretence or the echoic account: in other words, irony is not a natural kind. What I do want to argue is that the echoic account of irony is both theoretically and empirically distinguishable from most versions of the pretence account, and that typical cases of verbal irony such as (1)-(3) are best analysed as cases of echoic allusion and not of pretence.
2. Grice on verbal irony
For Grice, the interpretation of tropes depends on the hearer’s ability to recognise that the speaker has overtly violated the first maxim of Quality (“Do not say what you believe to be false”) in order to convey a related true implicature, which in the case of irony is the contradictory of the proposition literally expressed (Grice, 1967/1989: 34, 120). In the last twenty-five years, this approach to tropes in general, and to irony in particular, has been questioned on both descriptive and theoretical grounds.6
One problem is that in order to reanalyse figurative meanings as implicatures, Grice had to extend both his notion of implicature and his account of how implicatures are derived. A speaker’s meaning typically consists of what is said, together with any implicatures. Regular implicatures are added to what was said, and their recovery either restores the assumption that the speaker has obeyed the Co-operative Principle and maxims in saying what she said (in those particular terms), or explains why a maxim has been violated (as in the case of a clash). In Grice’s account of tropes, however, nothing is said. The speaker’s meaning consists only of an implicature, and the recovery of this implicature neither restores the assumption that the Co-operative Principle and maxims have been obeyed (if the speaker has said something she believes to be false, the situation cannot be remedied by the recovery of an implicature) nor explains why a maxim has been violated. In order to accommodate tropes, Grice thus had to abandon the basic idea that an implicature is an elaboration of the speaker’s meaning required to bring the overall interpretation of the utterance as close as possible to satisfying the Co-operative Principle and maxims.
There are more specific problems with the analysis of tropes as overt violations of the first maxim of Quality. One has to do with how the maxim itself should be understood. Does saying something amount simply to expressing a proposition, or does it amount to asserting a proposition, with a commitment to its truth? This makes a difference in the case of tropes. If saying something is simply expressing a proposition, then the first maxim of Quality is certainly violated in Grice’s own ironical examples (5a) and (6a), which he treats as implicating (5b) and (6b) (Grice, 1967/1989: 34, 120):
(5) a. He is a fine friend.
b. He is not a fine friend.
(6) a. Palmer gave Nicklaus quite a beating.
b. Nicklaus vanquished Palmer with some ease.
However, if saying something is asserting a proposition, with a commitment to its truth, then the first maxim of Quality is not violated in (5a) and (6a), since the speaker is patently not committing herself to the truth of the propositions literally expressed. Elsewhere in his framework, Grice treats sayingas not merely expressing a proposition but asserting it. He regularly describes the speaker in tropes not as saying something, but merely as “making as if to say” something, or as “purport[ing] to be putting forward” a proposition (Grice, 1967/89: 34). But if nothing is said, then the first maxim of Quality is not violated, and Grice’s account of tropes does not go through.7
The analysis of tropes as overt violations of the first Quality maxim is inadequate for other reasons. For instance, ironical understatements such as (3) above (Tim Henman is not the most charismatic tennis player in the world) are not blatantly false, but merely blatantly uninformative, or under-informative. The same point applies to negative metaphors (e.g. The agenda for the meeting is not written in stone), and to many cases of verbal irony proper. Suppose Bill is a neurotically cautious driver who keeps his petrol tank full, never fails to indicate when turning and repeatedly scans the horizon for possible dangers. Then his companion’s utterance of the imperative in (7a), the interrogative in (7b) or the declarative in (7c) could all be ironically intended and understood, although none of them is blatantly false:
(7) a. Don’t forget to use your indicator.
b. Do you think we should stop for petrol?
c. I really appreciate cautious drivers.
Notice, too, that (7a)-(7c) cannot be analysed as implicating the opposite of what they say. While the implicatures of (5a) and (6a) above might well include (5b) and (6b), no corresponding implicatures are conveyed by (7a)-(7c). More generally, the definition of irony as the trope in which the speaker communicates the opposite of the literal meaning does not do justice to the very rich and varied effects of irony. The standard Gricean approach to irony thus fails to explain not only what triggers the pragmatic inference process, but what its output is.
Some of these problems could be avoided while preserving the spirit of Grice’s account by claiming that what is overtly violated in tropes is not the first maxim of Quality but the first maxim of Quantity (“Make your contribution as informative as is required”) or the maxim of Relation (“Be relevant”). After all, if nothing is said, then the speaker’s contribution is neither informative nor relevant, and the maxims of Quantity and Relation are certainly violated. Moreover, these maxims (unlike the Quality maxims) apply not merely to what is said but to the speaker’s whole contribution (what is said, plus what is implicated), so that the recovery of an appropriate implicature could restore the assumption that these maxims have been obeyed. The general pattern for the interpretation of tropes would then be “The proposition literally expressed is blatantly irrelevant; by deriving an appropriate implicature, I can preserve the assumption that the maxims of Quantity and Relation have been obeyed”. However, this move would create a range of further problems that are much harder to solve. For one thing, whatever implicature is derived, the resulting interpretation would irrevocably violate the Manner supermaxim (“Be perspicuous”), since the most straightforward way of conveying this implicated information would have been to express it directly. For another, the proposed pattern of derivation is so widely applicable that it would vastly over-generate, predicting potential uses of irony that would never in fact occur. Moreover, since it applies equally well to the interpretation of metaphor and hyperbole (which are also treated in Grice’s framework as blatant maxim violations designed to convey a related implicature), it would give no insight into the intuitive differences between irony and other tropes. (For further discussion of this point, see Wilson and Sperber, 2002.)
In later work, Grice acknowledges that his original account of irony is descriptively inadequate. He considers an utterance which satisfies his proposed conditions on irony but would not normally be intended or understood as ironical:
A and B are walking down the street, and they both see a car with a shattered window. B says, Look, that car has all its windows intact. A is baffled. B says, You didn’t catch on; I was in an ironical way drawing your attention to the broken window. (Grice 1967/89: 53)
Here, B’s utterance is blatantly false and is allegedly intended to implicate the opposite of the proposition B has expressed. However, as Grice points out, even if uttered in an ironical tone of voice in a culture where irony is familiar and frequently used, this exchange would be absurd. He suggests that what was missing from his original account is the idea that irony involves the expression of a certain sort of critical judgment or attitude:
The absurdity of this exchange is I think to be explained by the fact that irony is intimately connected with the expression of a feeling, attitude, or evaluation. I cannot say something ironically unless what I say is intended to reflect a hostile or derogatory judgment or a feeling such as indignation or contempt. (ibid: 54)
What makes it hard to interpret the utterance Look, that car has all its windows intact as ironical is that it is hard to see what could have justified this critical judgment or attitude in the circumstances described.
The idea that irony is linked to the expression of a certain type of derogatory, hostile or contemptuous attitude raises several further questions. What is the object of this attitude, and what is the connection between communicating such an attitude and expressing a proposition that is patently false, under-informative, or irrelevant? These questions are not trivial. If the object of the attitude is a person or a piece of behaviour, why is it possible to convey it by saying something blatantly false, under-informative or irrelevant? When some hooligan breaks my car window, I may well feel critical of him (or his behaviour). However, in normal circumstances, I could not rationally attempt to convey this feeling by saying, in an ironical tone of voice, Look, my car has all itswindows intact. But if the object of the critical attitude or judgment is not (or not primarily) a person or a piece of behaviour, what is it? And why can it be conveyed by producing an utterance that would be pragmatically inappropriate if literally understood?
One possible connection between the presence of a critical, derogatory or mocking attitude and the expression of a false, under-informative or irrelevant proposition was suggested by Sperber and Wilson (1981). They argued that the speaker in irony does not use the proposition expressed by her utterance in order to represent a thought of her own which she wants the hearer to accept as true, but mentions it in order to represent a thought or utterance she tacitly attributes to someone else, and which she wants to suggest is ludicrously false, under-informative or irrelevant. On this account, the interpretation of irony does not depend on the ability to recognise that the speaker means the opposite of what she has said; it depends on the ability to recognise that she is mentioning or echoing a thought she attributes to someone else (or to herself in the past) in order to express a certain type of dissociative attitude to it. Early experiments by Jorgensen, Miller and Sperber (1984) confirmed that irony is easier to understand when there is an explicit prior utterance that the speaker can be taken to echo and reject.8 And indeed, all that is needed to rescue Grice’s example Look, that car has all its windows intact is an appropriate prior utterance of this type. Suppose that before A and B set out on their walk, B has complained that her street has become a dumping ground for broken-down cars, and A has reassured her that he sees no evidence for this. In these circumstances, when they come upon a car with a broken window, B’s remark Look, that car has all its windows intact would be easily understood as an ironical echo of this prior utterance, intended to express a mocking or critical attitude to it. Thus, the interpretation of irony is facilitated by the presence of an obvious source for the echoic utterance – a result predicted by the echoic account of irony, but not by the classical or Gricean accounts.
Another way of connecting the presence of a mocking or critical attitude to the expression of a blatantly false proposition involves the idea that irony is a type of pretence. Grice (1967/1989: 54) suggests that this might explain why a metaphorical utterance can be prefaced by the phrase To speak metaphorically, but an ironical utterance cannot be prefaced by the phrase To speak ironically:
To be ironical is, among other things, to pretend (as the etymology suggests), and while one wants the pretence to be recognised as such, to announce it as a pretence would spoil the effect. (ibid: 54)
Replying to Jorgensen, Miller and Sperber (1984), Clark and Gerrig (1984) develop a pretence account of irony as an alternative to the echoic account. Consider (8) below:
(8) Trust the Weather Bureau! See what lovely weather it is: rain, rain, rain
Jorgensen, Miller and Sperber (1984: 114) treat this as an echoic allusion to a forecast from the Weather Bureau that the speaker wants to reject as ludicrously false. Clark and Gerrig treat it as a type of pretence:
With See what lovely weather it is, the speaker is pretending to be an unseeing person, perhaps a weather forecaster, exclaiming to an unknowing audience how beautiful the weather is. She intends the addressee to see through the pretense – in such rain she obviously could not be making the exclamation on her own behalf – and to see that she is thereby ridiculing the sort of person who would make such an exclamation (e.g. the weather forecaster), the sort of person who would accept it, and the exclamation itself. (Clark and Gerrig, 1984: 122)
On this approach, the interpretation of irony depends on the hearer’s ability to recognise that the speaker is pretending to be a certain sort of person seriously producing an utterance, and simultaneously expressing her own attitude to it and to the sort of person who would produce or believe it.
Replying in turn to Clark and Gerrig, Sperber (1984) defended the echoic account against their criticisms and went on to raise some objections to Clark and Gerrig’s version of the pretence account. Other versions of the pretence account have been developed in the recent philosophical and psychological literature (see, for instance, Currie, in press; Colston and Gibbs, in press; Kreuz and Glucksberg, 1989; Kumon-Nakamura, Glucksberg and Brown, 1995). Several of these respond to Sperber’s objections by combining elements of the pretence and echoic accounts, and I will look at them more closely in section 4. My main claim will be that unless the notion of pretence is stretched incredibly thin, pretence accounts of irony, with or without an additional echoic element, are both descriptively and theoretically inadequate. While echoing and pretence can combine to produce occasional ironical effects, echoic use is essential to typical cases of verbal irony such as (1)-(8), and pretence is not.
3. Irony and echoic use Echoic use is a technical term in relevance theory. Echoic use is, in the first place, an interpretive rather than a descriptive use of language.9 An utterance is descriptively used when it is used to represent a possible or actual state of affairs; it is interpretively used when it is used to represent another representation (for instance, a possible or actual utterance or thought) that it resembles in content.10 Interpretive uses of language require a higher order of metarepresentational ability than descriptive uses. In order to understand an interpretively-used utterance, the hearer must recognise that the speaker is thinking not directly about a state of affairs, but about another utterance or thought.11 This may be explicitly communicated by use of parenthetical comments such as I think, he claims; where no overt linguistic indication is given, it must be inferred.
Some interpretive uses of language are based on rather abstract properties of the metarepresented utterance or thought. In (9b) and (10b), for instance, the speaker is tacitly metarepresenting a purely logical or conceptual content rather than a datable thought or utterance that she wants to be understood as attributing to someone:
(9) a. Some propositions are tautologies. b. For instance, a tall man is a man.
(10) a. Most lexical concepts are atomic. b. telephone, electron, cabbage.
In other cases, the metarepresented thought or utterance is chosen not purely for its logical properties, but for the fact that it has been, or might be, produced or entertained by a particular person or type of person (or by people in general), and a hearer who fails to recognise this will have misunderstood. Free indirect speech and thought, as in (11b) and (12b), are obvious illustrations of this tacitly attributive use of language:
(11) a. The Dean spoke up. b. The university was in crisis.
(12) a. The students were thoughtful. b. If they didn’t act now, it might be too late.
A plausible interpretation of (11)12 is that the claim that the university was in crisis (or some claim similar enough in content for (11b) to be regarded as an appropriate paraphrase or summary) is being tacitly attributed to the Dean. Similarly, a plausible interpretation of (12) is that the thought that if the students didn’t act straight away it might be too late (or some thought similar enough in content for (12b) to be regarded as an appropriate paraphrase or summary) is being tacitly attributed to the students. Here, (11b) and (12b) are not descriptively used: the speaker is not asserting them, and does not take responsibility for their truth, but is metarepresenting a thought or utterance with a similar content that she attributes to some identifiable person or group of people. According to the echoic account, verbal irony is a tacitly attributive use of language.
Echoic use is, in the second place, a particular sub-type of attributive use. The main point of an echoic use of language is not simply to report the content of the attributed thought or utterance, but to show that the speaker is thinking about it and wants to inform the hearer of her own reaction to it (Sperber and Wilson, 1986, chapter 4, section 9). Consider Jack’s utterance in (13) and the possible echoic responses in (14a-c):
(13) Jack: I had dinner with Chomsky last night.
(14) a. Sue: You had dinner with Chomsky! What did he say?
b. Sue: You had dinner with Chomsky? Is he in England?
c. Sue: You had dinner with Chomsky. Don’t make me laugh.
In each case, the point of Sue’s response is not to remind Jack of what he has only just said, but to show that she is thinking about it, and to convey her attitude to it: surprise and pleasure in (14a), puzzlement perhaps tinged with scepticism in (14b), and outright mockery and disbelief in (14c), where Sue echoes Jack’s claim in such a way as to indicate that she does not believe it, and finds it absurd. The range of attitudes that a speaker can express to an echoed thought or utterance range from acceptance or endorsement of its descriptive content, as in (14a), through questioning, puzzlement or desire for confirmation, as in (14b), to various shades of scepticism, mockery or rejection, as in (14c). Just as attributions may be more or less explicit, so in echoic use the speaker may give an overt linguistic indication of her attitude, or leave the hearer to infer it from paralinguistic or contextual clues. The main claim of the echoic account is that verbal irony is a sub-type of echoic use in which the speaker (generally tacitly) expresses one of a range of dissociative attitudes (scepticism, mockery, rejection, etc.) to a (generally tacitly) attributed utterance or thought. The main point of irony is to dissociate the speaker from an attributed thought or utterance which she wants to suggest is more or less obviously false, irrelevant or under-informative.
To illustrate, let’s consider how examples (1)-(7) might be analysed on this approach. Mary’s utterance in (1) (That went well, said after a difficult meeting) might be understood as echoing Mary’s earlier hopes or expectations, or the reassurances of her friends, that the meeting would go well, in order to show that she now finds them ridiculously over-optimistic or ill-founded. The utterance as a whole is echoically used: Mary is not asserting that the meeting went well, but, on the contrary, tacitly dissociating herself from a thought or utterance with a similar content (e.g. a hope, desire, expectation or reassurance that the meeting would go well). To claim that irony is echoic is to claim that it is closely related to other tacitly attributive uses of language such as free indirect speech or thought. If verbal irony is a case of echoic use, the speaker of (1) above should no more be automatically committed to the claim that the meeting went well than would be the speaker in cases of free indirect speech or thought such as (11b) or (12b) above (The university was in crisis; If they didn’t act now, it might be too late). Attributive uses of language in general are constrained by considerations of faithfulness rather than truthfulness. When a whole utterance such as (1) above is interpretively used, the question of whether the speaker has obeyed a maxim, norm or convention of literal truthfulness should not arise.
By contrast, (2) (As I approached the bank at closing time, the bank clerk helpfully shut the door in my face) is a regular (if hyperbolic) assertion in which only the word helpfully is echoically (and dissociatively) used. The speaker commits herself to the claim that the bank clerk closed the door before she got there, but not to the claim that this behaviour was helpful. Here, the word helpfully might be seen as ironically echoing the expectations we have when choosing a bank, the claims to helpfulness and consideration that banks often make in their adverts, or the more general norm that people should behave helpfully to each other, which is widely shared and has clearly been violated in this particular instance. As noted above, ironical understatements such as (3) (Tim Henman is not the most charismatic tennis player in the world) could well be literally true, and this presents problems for the standard Gricean account. However, (3) could be regarded as informative or relevant only by the most devotedly blinkered fan or publicist of the gifted but very English Henman. When ironically used, it can be seen as echoing and dissociating the speaker from just such blinkered claims as ludicrously irrelevant or under-informative, and making fun of the people who would make or accept them.
Grice’s (5a) (He’s a fine friend) is a classic case of verbal irony. Although not traditionally analysed as involving the tacit expression of a dissociative attitude to a tacitly attributed thought, it is intuitively quite closely related to (1) (That went well), which is plausibly seen as echoing a more or less specific hope, desire or expectation that Mary might have had before going into the meeting. When we make a friend, we have certain hopes, desires or expectations about how things will go: we may think we have made a fine friend who will treat us well, imagine people congratulating us later on having made such a fine friend, and so on. As a classic case of verbal irony, (5a) echoes hopes or expectations of friendship which are very widely shared, and requires no particular scene-setting to be understood as ironical. By contrast, Grice’s (6a) (Palmer gave Nicklaus quite a beating) does not echo universal human hopes or aspirations, and to be appropriate as a case of verbal irony, it must be understood as echoing something more specific: for example, the predictions of sports commentators, the boasts of Palmer’s supporters, and so on. According to the echoic account, there is no principled distinction between the examples in (1), (5a) and (6a), but merely a difference in the extent to which the hopes or expectations being echoed are universally shared.
The utterances in (7a) (Don’t forget to use your indicator) and (7b) (Do you think we should stop for petrol?), like Grice’s example Look, that car has all its windows intact, do not echo widely shared hopes or expectations, and require a certain amount of scene-setting in order to be understood as echoic or ironical. If we imagine them addressed to Bill, the neurotically cautious driver described in section 2 above, it is easy to see them as ironically echoing thoughts which the speaker attributes to him (“It always helps to indicate”, “It’s never too early to stop for petrol”), and which she wants to suggest he is taking much too seriously. (7c) (I really appreciate cautious drivers) can be interpreted along similar lines, as a mocking echo of a specific thought of Bill’s (“People will really appreciate me for taking such care”). In different circumstances, however (for instance, when used to comment on a particularly reckless piece of driving), it would be understood as an ironical echo of a more general hope or expectation that other drivers will not take risks on the roads.
This account is based on a clear-cut theoretical distinction – reflected in truth or satisfaction conditions – between descriptive and interpretive use. Within the range of interpretively-used utterances, however, the borderline between attributive and echoic use, between ironical and non-ironical attitudes, and between tacit and overt attributions and expressions of attitude are much less clear-cut. In the first place, an utterance which is primarily intended as a report of speech or thought may be incidentally used to convey some information about the speaker’s attitude: the borderline between reporting and echoing is a gradual one. In the second place, the prototypical ironical attitudes shade off into other types of dissociative or sceptical attitude, and a single utterance can convey a quite complex mixture of attitudes: the borderline between irony and other types of echoic use is a gradual one. In the third place, the gap between fully explicit conceptual encodings of attribution and attitude and purely tacit attributions and expressions is filled by a wide variety of paralinguistic and peripheral linguistic forms (intonation, facial expressions, gestures, interjections, discourse particles, quotation marks, parentheticals, etc.): the borderline between overt and tacit attributions and expressions of attitude is a gradual one. All this suggests that irony is not a natural kind, and belongs together with other forms of echoic, attributive and interpretive use, which must all be treated in the same way. The implication for pretence accounts of irony is that either all these forms must be analysable as cases of pretence or none are.
To illustrate the gradualness of the borderline between reporting and echoing, and between the various types of attitude expressed in echoic use, consider the exchange in (15), where Bumpers, the narrator of a novel by Peter de Vries, is defending his PhD on Causes of Divorce in Southeastern Rural Iowa against the criticisms of the chief examiner, Timken:
(15) Bumpers: What I’m trying to say, gentlemen, is that divorce is as complicated as marriage, and that is a relationship inconceivably intricate.
Timken: Which a bachelor like myself can be only hopelessly unequipped to understand.
(Peter De Vries: 8)
To understand Timken’s response as ironical, we would have, in the first place, to decide that he is not making an assertion in his own right but interpreting what he takes to be an implication of Bumpers’ remark.13 We would have, in the second place, to decide that his utterance is intended to achieve most of its relevance by expressing Timken’s attitude to this attributed thought, rather than merely reporting its content. We would have, in the third place, to decide that he is dissociating himself from this attributed thought rather than endorsing it or remaining neutral about it (and if the ironical attitudes are only a subset of the dissociative ones, we would have to decide that he is expressing one of this subset of attitudes).14 None of this information is linguistically encoded. From the pragmatic point of view, this introduces a massive element of underdeterminacy into the utterance. From the theoretical point of view, it makes it inadvisable to treat irony as belonging in a separate category from other types of echoic, attributive or interpretive use.
To illustrate the gradualness of the borderline between overt and tacit attribution and expression of attitudes (and also the gradualness of the distinction between reporting and echoing), consider the following scenario. Mary and a friend have been watching Peter lose very badly at tennis. At the end of the match, Peter comes up to them and says (seriously) I almost won. Mary turns to her friend and says, wryly, one of the following:
(16) a. He says he almost won.
b. He almost won, he thinks.
c. Poor fool. He thinks he almost won.
d. He almost won. Allegedly.
e. He almost won. Not.
f. He almost won. Huh!
g. He almostwon.
In each of these utterances, Mary can be understood as expressing a dissociative attitude to an utterance or thought that she attributes to Peter. The main differences between them are in how explicitly the attitude is expressed and the attribution made. In (16a), (16b) and (16d), the attribution is linguistically indicated (by use of the words he says, hethinks and allegedly) and the attitude tacitly conveyed. In (16c), (16e) and (16f), by contrast, the attitude is linguistically indicated (by use of the expressions poor fool, huh! and not) and the attribution is tacitly conveyed. In (16c), both attitude and attribution are linguistically indicated, and in (16g) both attitude and attribution are tacitly conveyed. Only (16g) is a typical case of verbal irony: this is the only example involving the tacit expression of a dissociative attitude to a tacitly attributed utterance or thought. However, as illustrated in (16a), (16b) and (16d), more explicit forms of reported speech and thought may also tacitly convey a dissociative attitude, and thus achieve ironical effects; moreover, as illustrated by (16c)-(16f), fully conceptual forms of encoding shade off into interjections (which themselves shade off into various paralinguistic and non-linguistic cues): the borderline between overt and tacit attributions and expressions of attitude, and hence between typical and less typical cases of irony, is a gradual one.
I have treated (16g) as typical case of verbal irony. However, this is true only if it is uttered with the flat, low-key intonation generally known as the “ironical tone of voice”.15 Mary might have expressed her sceptical reaction to Peter’s remark (and thus achieved ironical effects) by uttering (16g) in at least two other ways. In one, she would parody or imitate Peter, using a tone of voice and manner of articulation similar to his, perhaps combined with a mocking or contemptuous facial expression. In the other, she would adopt an exaggeratedly bright, convinced tone of voice and the manner of articulation that someone would have if genuinely convinced by what Peter said. Both can legitimately be seen as cases of pretence: in the first, Mary is pretending to be Peter (or to speak in the way Peter does), and expecting her audience to see through the pretence; in the second, Mary is pretending to believe Peter, and expecting the audience to see through the pretence. This raises the question of whether the utterance of (16g) in an ironical tone of voice – and more generally, typical cases of verbal irony such as (1)-(8) – might not also be legitimately seen as cases of pretence.
4. Irony and pretence
The central idea behind pretence accounts of verbal irony is that the speaker is not herself performing a speech act such as making an assertion or asking a question, but pretending to perform it (or, in more elaborate versions, pretending to be a certain type of person performing it). This idea has been fleshed out in various ways, often within broader theories of mimesis, simulation or pretence (Clark and Gerrig, 1990; Currie, 2002, 2004; Recanati, 2000, 2004; Walton, 1990; see also Nichols and Stich, 2000). My concern here is not with these broader theories, which provide valuable insights into the ways in which the perception of resemblances may be exploited in communication and expression, but about the much more limited issue of whether irony is best analysed as a type of simulation or pretence. I will argue that it is not.
One way of reconciling Grice’s original account of irony with his later remark that irony is a type of pretence (see section 2 above, and Grice, 1967/1989: 34, 53-54, 120) is to assume that he saw “making as if to say” as a type of pretence. Then “making as if to say” that Paul is a fine friend would amount to pretending to say that he is a fine friend, and Grice’s account of irony would have been a pretence account all along. Recanati (2004: 71) interprets Grice along these lines, and appears to endorse a similar version of the pretence account:
Suppose the speaker says Paul really is a fine friend in a situation in which just the opposite is known to be the case. The speaker does not really say, or at least she does not assert, what she “makes as if to say” (Grice’s phrase). Something is lacking here, namely the force of a serious assertion. … What the speaker does in the ironical case is merely to pretend to assert the content of her utterance. …By pretending to say of Paul that he is a fine friend in a situation in which just the opposite is obviously true, the speaker manages to communicate that Paul is everything but a fine friend. She shows, by her utterance, how inappropriate it would be to ascribe to Paul the property of being a fine friend.
Clark and Gerrig (1984) also trace their version of the pretence account to Grice’s original remark that irony is a case of “making as if to say” something.
As noted in section 2 (and as several of these authors recognise), non-echoic versions of the pretence account do not explain why a speaker cannot produce any blatantly false or inappropriate utterance and expect it to be understood as ironical. One can pretend to be anyone at all, asserting or believing anything at all. So why can’t the speaker in Grice’s example Look, that car has all its windows intact be understood as pretending to be the sort of person who would assert or believe (in the face of clear counter-evidence) that the car has all its windows intact? For Grice, the solution to this problem was connected with the hostile or derogatory attitude that the speaker is taken to express. But the most plausible way of linking the expression of a hostile or derogatory attitude with the production of a manifestly false, under-informative or irrelevant utterance is to assume that the speaker is expressing this attitude primarily to a thought or utterance with a similar content to the one she has expressed, and only secondarily to a person. Moreover, the expression of a hostile, derogatory, or more generally dissociative attitude to a possible thought or utterance must have a point. As Sperber (1984: 131) puts it,
Absurdity of propositions per se is irrelevant. The absurdity, or even the mere inappropriateness, of human thoughts, on the other hand, is often worth remarking on, making fun of, being ironic about. In other words, in order to be successfully ironic, the meaning mentioned must recognisably echo a thought that has been, is being, or might be entertained or expressed by someone.
Thus, what is missing from non-echoic versions of the pretence account is precisely what is emphasised by the echoic account: that the attitude expressed in irony is primarily to a thought or utterance that the speaker attributes to some identifiable person or type of person, or to people in general.
Adding an echoic element to the pretence account helps to explain why Grice’s example Look, that car has all its windows intact requires a certain amount of scene setting in order to be understood as ironical, while other ironical utterances can be uttered “out of the blue”, in any discourse context. According to the echoic account, an ironical utterance must be recognisable as echoing a thought or utterance (or more generally a representation with a conceptual content) attributable to some identifiable person or group of people, or to people in general. Cultural norms are widely represented in human minds, and are always available for ironical echoing. This is what happens in (2) (As I reached the bank at closing time, the bank clerk helpfully shut the door in my face), where the bank clerk’s behaviour (which clearly violates a cultural norm) is ironically described as helpful. By the same token, it might also be relevant, on seeing a car with a broken window, which has patently not been well looked after, to say ironically, The cars are well looked after around here. Both these utterances are ironical allusions to cultural norms, and can be used without any scene setting, to echo a widely shared norm that has been broken in a particular instance. By contrast, the fact that a car has a broken window doesn’t violate any cultural norm, and for the utterance Look, that car has all its windows intact to be echoic, some more specifically identifiable source for the metarepresented thought or utterance is required.
Kumon-Nakamura, Glucksberg and Brown (1995) accommodate these points by proposing an “allusional pretence” theory in which an ironical utterance must not only be “pragmatically insincere” (that is, a case not of saying but of “making as if to say”) but also allude to a “failed expectation or norm” (1995: 19). A fuller and richer account on similar lines is proposed by Currie (in press: 116), who argues that in irony, “one pretends to be doing something which one is not doing: speaking seriously and assertively, seriously asking a question, seriously expressing distaste”, in order to target “a restrictive or otherwise defective view of the world”:
… what matters is that the ironist’s utterance be an indication that he or she is pretending to have a limited or otherwise defective perspective, point of view or stance, F, and in doing so puts us in mind of some perspective, point of view or stance, G (which may be identical to F or merely resemble it) which is the target of the ironic comment. (Currie in press: 118)
Currie’s account addresses many of the objections made by Sperber (1984) to earlier versions of the pretence account. As he notes, it has much in common with the echoic account (as well as several differences of substance or detail which I will have to leave to another time).16 Both recognise that irony involves the attribution of a thought (or perspective, or point of view) to a specific person or type of person, or to people in general, and the expression of a dissociative attitude to the attributed thought. Both note that the thought that is the object of the ironical attitude need not be identical to the proposition expressed by the ironical utterance but may merely resemble it in content. Both recognise that a genuine speech act may contain a single constituent which is ironically used, as in (2) (As I reached the bank at closing time, the bank clerk helpfully shut the door in my face), a genuine assertion in which only the word helpfully is ironic. However, although allusional pretence accounts deal well with many of the objections to earlier pretence accounts, I want to suggest that they still encounter a significant problem: unless the notion of pretence is stretched incredibly thin, the standard forms of verbal irony illustrated in (1)-(8) above are not cases of pretence.
Both echoic and pretence accounts are agreed that the speaker of an ironical utterance does not perform the speech act she would standardly be taken to perform if her utterance were literally understood. But it does not follow from this alone that they are cases of pretence. Consider the metaphor in (17a), the hyperbole in (17b) and the approximation in (17c):
(17) a. That office is a viper’s nest.
b. The article contained millions of typos.
c. The chairs formed a circle.
None of these asserts the proposition literally expressed: the speaker of (17a) is not genuinely asserting that the office is a viper’s nest, and so on for the other examples. Grice did indeed analyse metaphor and hyperbole, like irony, as cases of “making as if to say” (i.e. as pretending, in a very general sense): on a Gricean account of (17a) and (17b), no speech act of assertion is performed and the speaker’s meaning consists solely of implicatures. However, he seems to have drawn the line at approximations, describing cases similar to (17c) as making genuine assertions in which a word is used “loosely, in a relaxed way, which the nature of the context of utterance makes permissible” (Grice, 1967/1989: 45). Recent accounts of metaphor and hyperbole treat them as forming a continuum with loose use and rough approximation, and hence as making genuine (though not strictly literal) assertions (Carston, 2002; Recanati, 2004; Sperber and Wilson, in press; Wilson and Sperber, 2002). If these accounts are on the right lines, there is no valid argument from the premise “This utterance does not have the force of a serious, literal speech act” to the conclusion “This utterance is a case of pretence”.
The echoic account does not treat irony as forming a continuum with loose use, metaphor and hyperbole; however, as argued in section 3, it does treat ironical utterances as forming a natural class with other types of interpretive, attributive or echoic use. If this account is on the right lines, then either all interpretively used utterances must be treated as cases of pretence or none can. I want to argue that none are appropriately analysed as cases of pretence.
The pretence account is particularly inappropriate for interpretive uses in which the speaker tacitly or overtly metarepresents an abstract logical or conceptual content rather than an attributed utterance or thought. Consider the tacitly metarepresentational (9) and (10) (repeated here for convenience) and the more explicit versions in (18) and (19):
(9) a. Some propositions are tautologies. b. For instance, a tall man is a man.
(10) a. Most lexical concepts are atomic. b. telephone, electron, cabbage.
(18) The following proposition is tautological: a tall man is a man.
(19) Lexical concepts, including telephone, electron and cabbage, are atomic.
In (9b) and (18), the speaker is not pretending to make an assertion, or imitating some other person or type of person. Her behaviour, tone of voice, manner of articulation, facial expression, etc. are not intended to resemble those of any other person or type of person: she is speaking in her own voice, and using language purely to pick out a proposition that she wants to bring to her hearer’s mind. The same point applies to (10b) and (19), in which the concepts telephone, electron and cabbage are mentioned rather than used. My claim is that when the main point of an interpretively used utterance is to pick out a content or meaning – whether a purely abstract meaning, as in these examples, or the meaning of an attributed thought or utterance – this is not appropriately described as a case of mimicry, simulation or pretence. One cannot mimic or simulate a content, a meaning or a thought. The pretence account of irony works only for cases where an element of mimicry or simulation of behaviour is involved.
To illustrate this point, consider the tacitly attributive utterances in (11b) and (12b) (repeated here for convenience) and their more explicit counterparts in (20) and (21):
(11) a. The Dean spoke up. b. The university was in crisis.
(12) a. The students were thoughtful. b. If they didn’t act now, it might be too late.
(20) According to the Dean, the university was in crisis.
(21) The students were thinking that if they didn’t act now, it would be too late.
The speaker of these utterances chooses an indirect rather than a direct form of quotation, which gives her audience some idea of the content of the thoughts or utterances she is metarepresenting, but not necessarily of their form. She is not mimicking the Dean or the students, or pretending to make an assertion, but simply drawing the audience’s attention to a meaning or content that she wants to attribute to the students or the Dean. Some pretence theorists (for instance, Barker, 2004; Walton, 1990) seem tempted to treat such utterances as cases of pretence. According to Walton (1990: 224),
In quoting a person indirectly (“He said that...”) one does not use the very words he did. But it may be fictional that one endorses a certain thought,17 thereby indicating that the quoted person, using his own words, endorsed it. Such participation may occupy less than whole sentences, even a single word or phrase. ... The scare quotes, or an obviously exaggerated, sarcastic tone of voice serve both to make it clear that the speaker is engaging in this pretense and to betray it, to indicate that the speaker is only pretending.”
However, indirect reports of speech and thought such as (11)-(12) or (20)-(21) need involve no mimicry or “pragmatic insincerity”: the speaker is genuinely reporting a content or meaning rather than pretending to do something else.
In fact, not even all echoic utterances are plausibly analysed as cases of pretence. The most obvious problems for the pretence account are raised by echoic questions such as (22b):
(22) a. Peter: I’ll be arriving there around six.
b. Mary: You’ll get here at sixish?
Here, Mary echoes Peter’s assertion to show that she is thinking about it, and perhaps to ask him to confirm that she has heard and/or understood it correctly (Noh, 1998, 2001; Wilson, 2000). She is not pretending to make an assertion or to ask a question (if anything, she is asking a genuine question), and in the interpretation I am considering here, she is not mimicking Peter or imitating his utterance, but merely interpreting it and asking him to confirm her interpretation. Her utterance does not repeat a single word that he used, but paraphrases the content of his utterance: the intended resemblances between her utterance and his are entirely in content and not in form. Similar points apply to many echoic questions that closely resemble a preceding utterance in form as well as content. Consider Sue’s echoic response to Jack in (14b) above (repeated here for convenience):
(13) Jack: I had dinner with Chomsky last night.
(14) b. Sue: You had dinner with Chomsky? Is he in England?
It is hard to see this as a case of “pragmatic insincerity” or “making as if to say”. Sue is echoing Jack’s immediately preceding utterance in order to show that she is thinking about its content, and to express her reaction to it. In the interpretation I am considering here, Sue does not intend to be understood as imitating Jack or simulating his behaviour: the only relevant resemblances between her utterances and his are in content rather than form.
Now consider echoic responses such as (14a) (repeated below):
(13) Jack: I had dinner with Chomsky last night.
(14) a. Sue: You had dinner with Chomsky! What did he say?
In (14a), Sue echoes Jack’s preceding remark in such a way as to indicate that she accepts it and wants to express her surprise and pleasure at the fact that it is true. It seems entirely inappropriate to treat this as a case of “pragmatic insincerity” or “making as if to say”. Sue is not pretending to assert anything: she is accepting Jack’s assertion and expressing her reaction to it. Or consider cases of delayed acceptance or endorsement such as (23b) (from Sperber and Wilson 1986: 239):
(23) a. Peter: It’s a lovely day for a picnic.
[They go for a picnic and the sun shines.]
b. Mary: It’s a lovely day for a picnic, indeed.
Here Mary echoes Peter’s utterance in such a way as to make it clear that she is endorsing it and complimenting him on his suggestion. She need not be imitating Peter or mimicking his behaviour: what matters (in the interpretation I am considering here) is the resemblance in content between her utterance and his.
Echoic utterances can convey a wide variety of attitudes. For instance the echoic questions in (14b) (You had dinner with Chomsky?) or (22b) (You’ll get here at sixish?) may convey a touch of scepticism, and thus a touch of irony, while the main point of ironical questions such as (7b) (Do you think we should stop for petrol?) is to dissociate the speaker entirely from the attributed thought. As noted above in section 3, the borderline between genuine speech acts with a tinge of irony and utterances whose main point is to dissociate the speaker from the attributed thought is a gradual one. It would be hard to argue that ironical questions such as (7b) are cases of pretence while ordinary echoic questions such as (14b) and (22b) are not.
According to the echoic account, ironical utterances such as (24b) are interpreted on exactly the same pattern as echoic endorsements such as (23b) (It’s a lovely day for a picnic):
(24) a. Peter: It’s a lovely day for a picnic.
[They go for a picnic and it rains.]
b. Mary: It’s a lovely day for a picnic, indeed.
Here, Mary echoes Peter’s utterance in such a way as to make it clear that she does not believe it, and perhaps to criticise him for his suggestion. In both cases she speaks in her own voice and expresses her own attitude: the only difference between (23b) and (24b) is in the type of attitude expressed. So if echoic endorsements such as (23b) are not analysed as cases of pretence, neither should ironical dissociations be.
Currie (in press: 125) argues that echoic endorsements such as (23b) and ironical dissociations such as (24b) differ in one important respect which makes it legitimate to analyse ironical dissociations and not echoic endorsements as cases of pretence. Suppose that Peter responds to Mary by saying (25):
(25) Yes, I’m so glad we decided to come.
As a response to the echoic endorsement in (23b), Peter’s utterance would be naturally understood as a genuine, non-echoic assertion. As a response to the echoic endorsement in (24b), however, it would be naturally understood as a continuation of the irony. In Currie’s terms, irony “opens the door to pretending”, and Peter’s response to Mary’s ironical utterance “would naturally be seen as an elaboration of Mary’s pretence” (ibid: 125). He is thus prepared to treat Mary’s utterance in (24b) as a case of pretending in an “active”, “substantial” sense, while arguing that echoic endorsements and other types of free indirect quotation are not cases of pretence (or involve pretending only in a “thin, atrophied” sense (ibid: 126)).
Notice, though, that even explicit reports of speech or thought, which clearly make genuine assertions, can be used with ironical overtones which a hearer may pick up and respond to in kind. Recall the scenario above (section 3) where Peter, after a game of tennis, says, seriously, I almost won, and Mary turns to her friend and says, wryly, one of (16a-d):
(16) a. He says he almost won.
b. He almost won, he thinks.
c. Poor fool. He thinks he almost won.
d. He almost won. Allegedly.
It would be quite legitimate for Mary’s friend, noticing Mary’s tone of voice and dissociative attitude, to respond ironically, in a similar tone of voice, as in (26a)-(26c):
(26) a. Yes, it was a really close thing.
b. Such a shame that he didn’t quite make it.
c. Could you imagine anything closer?
These are genuine cases of irony, as evidenced by the speaker’s ironical tone of voice. Neither speaker is simulating a piece of behaviour: in the interpretations I am considering, the only relevant resemblances are in content, not in form.
5. Concluding remarks: parody, allusion and pretence
In this paper, I have raised two main points which I see as presenting problems for pretence accounts of irony. First, in order to explain central cases of irony such as (1)-(8) above, pretence accounts have to be supplemented with something like the notion of echoic use: they are not alternatives to the echoic account but extensions of it. Second, central cases of irony form a natural class with interpretive, attributive and echoic utterances such as (9)-(16) above, some of which may carry ironical overtones: the borderline between central and peripheral cases of irony is a gradual one. It follows that either all interpretively used utterances should be analysed in terms of pretence or none should. Like most pretence theorists, I see pretence as a type of simulation or mimesis which crucially involves the exploitation of resemblances. Unlike them, I have argued that typical ironical utterances such as (1)-(8) above are not cases of pretence. Irony involves the attribution of a thought, a propositional or conceptual content or a meaning. Such abstract objects cannot be mimicked, simulated or imitated. Simulation involves perceptual similarity or resemblance in form. As I have tried to show in section 4, the only type of resemblance relevant to the interpretation of central cases of irony is in content, not in form. In irony, in fact, the speaker gives up the opportunity for mimicry or simulation in order to express her own attitude, in her own tone of voice.
This is not to say that utterances based on the exploitation of perceptual resemblances cannot be used to achieve ironical effects. As noted above in section 3, a speaker may adopt the tone of voice or manner of articulation of some other person or type of person in order to make fun of them, their way of speaking or the thoughts they have expressed (Sperber and Wilson, 1981; Sperber, 1984; Wilson and Sperber, 1992). Such utterances are indeed simulations, and are often used to witty or ironical effect. Consider (27)-(30):
(27) Punctuality is the thief of time. (Oscar Wilde)
(28) Among the smaller duties of life, I hardly know any more important than that of not praising where praise is not due. (Sydney Smith)
(29) A critic is one who leaves no turn unstoned. (George Bernard Shaw)
Each of these parodies or alludes to another utterance that resembles it in form. In (27), Wilde may be seen as dissociating himself from the saying Procrastination is the thief of time, and in (28), Sydney Smith may be seen as expressing some scepticism about the idea that we should always give praise where praise is due. Even (29), which could be seen simply as wordplay, may be intended to make a more serious point about the contrast between a drama critic’s job and an ordinary job. Although these utterances achieve their effects by exploiting perceptual resemblances, or resemblances in form, they are still not appropriately analysed as cases of pretence. In each case, the speaker could be using them to perform a genuine speech act while simultaneously alluding to another one. Here again, ironical effects are achieved without any element of pretence.
The type of irony that does involve pretence is the one sometimes described in the literature as “impersonation irony” (cf. Simonin, forthcoming), where the speaker (or writer) adopts a persona in order to criticise or make fun of those who speak or think in similar ways. The best-known examples are Swift’s “A Modest Proposal” and Defoe’s “The Shortest Way with Dissenters”, both intended to satirise political views of the time. Since I think Neil Smith would enjoy it, and with many thanks to the pretence theorists whose work I find both enriching and provocative, I will end by quoting the first two paragraphs of Stephen Leacock’s essay Are the Rich Happy?, which is indeed a case of pretence usedto achieve ironical effects:
Let me admit at the outset that I write this essay without adequate material. I have never known, I have never seen, any rich people. Very often I have thought that I had found them. But it turned out that it was not so. They were not rich at all. They were quite poor. They were hard up. They were pushed for money. They didn’t know where to turn for ten thousand dollars.
In all the cases that I have examined this same error has crept in. I had often imagined, from the fact of people keeping fifteen servants, that they were rich. I had imagined that because a woman rode down-town in a limousine to buy a fifty-dollar hat, she must be well-to-do. Not at all. All these people turn out on examination to be not rich. They are cramped. They say it themselves. Pinched, I think, is the word they use. When I see a glittering group of eight people in a stage box at the opera, I know that they are all pinched. The fact that they ride home in a limousine has nothing to do with it.
(Leacock 1917/1981: 110)
Many thanks to Neil Smith for discussions on this and other topics over the years, to Dan Sperber for illuminating conversations on irony, parody and pretence, many of which are reflected in this paper, and to Greg Currie, for inspiration, challenge and allowing me to see an early version of his fascinating paper on pretence accounts of irony. Thanks also to Robyn Carston and Vladimir Zegarac for insightful comments on an earlier version, and to two anonymous referees. This paper is part of the AHRC-funded project ‘A Unified Theory of Lexical Pragmatics’ (AR16356); I am very grateful to Robyn Carston and the other members of the project team.