ISSN 1393-614X Minerva - An Internet Journal of Philosophy Vol. 5 2001.


Ross Macleay 


This paper considers the charge that fictions and metaphors are non-normal, non-serious, non-primary uses of communication. This charge was made when Plato undertook the first critique of narrative art; and, ever since Aristotle’s Poetics, what there is that might be called a philosophy of narrative art has always seemed duty-bound to worry about it. This apologetic predicament has limited the development of a philosophy of fiction.

The complementary charge—that literal, non-fictional uses have normal, serious and primary status in human communication, and that fictions and metaphors are parasitic uses—is widely held; and, as expressed by John Austin in his theory of speech acts, it drew a response from Jacques Derrida, setting off a curious little debate between two philosophical traditions that have normally been as oil and water to one another.

This paper investigates the relation between literal and non-fictional uses on the one hand, and metaphorical and fictional uses on the other. In the matter of determining the status of priority, it looks to the history of selection—both natural and social selection—of communicative traits and functions. Though it deals mainly with language, it argues that the questions should be generalised to cover human symbolic communication generally, if only because human symbolic communication has not evolved in isolation from mimetic acts and linguistic acts are themselves so often mimetic. And it argues that, because of the nature of conditional inferences involved in symbolic communication, metaphorical and fictive forms are inseparable from normal, serious, primary use.

This ‘defence of poesy’ as contemporary with and inseparable from reason is offered not simply as a contribution to the philosophy of language, but primarily as a contribution to a philosophy of fiction.

The Apologetic Predicament

It is a strange thing to tell stories about non-existent people doing things that never happened. Alone and writing before the abyss of such non-existence, few novelists would have entirely avoided the occasional sense of despair. In fact, making such despair unfounded is just one challenge for the art of fiction: fictions that don’t impress us with their necessity don’t even manage to entertain us. Workers in film, theatre and television at least have the consolation of working in company or what Helen Garner (1996, 119) calls "the slightly crazed pleasure of collaboration"; meanwhile for audiences, fiction is usually relegated to the less important, recreational department of life. Even for an age when fiction can claim to be one of the great spiritual projects, and when the fiction business is big business, the common sense persists that fiction is neither a normal, essential, primary nor serious kind of narrative endeavour, but just a bit of pretending. And something like this has persisted at least since Plato saw how fascinating all this idle, pretentious mimesis could be, and duly expressed his reservations about it. Narrative art continues to worry the workaday conscience.

Certain theories of fiction are symptoms of this worry. Take John Austin’s view of fiction, and of other literary uses of language, in his influential lectures on the theory of speech acts, How To Do Things With Words: "These are aetiolations, parasitic uses, etc., various ‘non-serious’ and not full normal uses (104)." Replicated through the tradition of speech act theory, as well as through popular common sense, this assumption eventually prompted Jacques Derrida (1972, 307-330) to respond across the gulf that separated English language analytical philosophy from European critical philosophy. Austin held that a promise spoken by an actor on stage was not a serious or binding promise, but that it was a derivative or parasitic use of language. In Austin’s terms, a fictional or stage promise suspends the binding illocutionary force of a normal promise—the illocutionary force being the binding, interpersonal warranty that a speaker, in saying something, gives to a listener about how something is to be taken and how the truth of what is said can be validated or made good if need be. If Austin’s assumption was one of those periodic attacks on art, in the Platonic tradition, Derrida’s response belonged to the tradition of apologetics. Indeed together, both sides make up a kind of apologetic syndrome—a kind of self-perpetuating conflict system. Derrida went on the attack—with a kind of exaggerated turning of the tables. He claimed language was essentially fictive and that so called serious linguistic behaviour or normal usage was itself a kind of role-playing.

Actually there are two levels of apologetics. Firstly there is the apology for poetry or fiction, inaugurated by Aristotle’s Poetics; but then there is criticism of narrative art, and the discipline of criticism is a kind of second order parasite. It is parasitic on fiction, which is in turn parasitic on serious, literal communication. So criticism is especially ashamed of its parasitism and especially given to an apologetic syndrome of its own. Its symptoms are various, ranging from journalistic triteness (deliberately light, supposedly unpretentious, bantery reviews, especially of film and television) to would-be-artistic bombast (pseudo-profound, frisson-seeking stylistic displays, especially in matters of literary taste) to busy, academic desperation (brave, grave, attempts at critical theory, popularly ridiculed for their postmodern jargon). All this is another story, except to say that most of what we might call the theory of fiction has been shaped by this doubly apologetic predicament.

Ever since Aristotle’s appeals to ‘catharsis’ and ‘universal truths’, the apology for fiction has typically resorted to claiming some kind of useful (often moral) function for it. This applies even to Derrida’s claim that so-called serious communication is actually parasitic on fictional role-playing. Critics and theorists of fiction have liked to equip themselves with some of the prestigious props of their canonised art, flashing a bit of showy literary style, or the odd cadence or metaphor to signal the ineffable profundity of literary taste or experience, a bit of fictive licence to free theoretical speculation from the mean spirit of literalness or empiricism. For academic literary critics especially—still perhaps worried by their unconscionably frivolous object, not to mention attacks on university funding or the sensed demise of literary narrative—Derrida’s view was probably comforting. Their job, rather than being primarily scientific or even scholarly could be, or had to be literary, and therefore blessed with a certain aesthetic-cum-philosophical dignity (and terminology) of its own.

A response to Derrida came from John Searle, one of Austin’s most thoughtful heirs in the theory of speech acts. Searle insisted again that fictive uses—simulations, quotations, irony—were parasitic, because, as Habermas (1985, 195) repeated it in his own support of Searle’s contention, "logically they presupposed the possibility of serious, literal, binding use" of language. Meanwhile, Derrida insisted that calling something parasitic was really just a kind of pejorative, ethical or political claim dressed up as a theoretical one.

There is no doubt that art worries all sorts of consciences in the context of apologetics, but for someone supposed to be advocating a literary, rhetorical reading of texts, Derrida was strangely unappreciative of the ironic deprecation in the term parasitic. I suspect that Austin, for instance, actually liked his example of parasitism—John Donne’s song about catching falling stars and getting mandrake roots with child—and that he originally used the word parasitic both provocatively and fondly. Unfortunately, the divide between analytical and critical-theoretical philosophy created a context for ignoring irony and, for the sake of arguing against the other, compelled misunderstanding the other. Austin’s use of the term parasitic wasn’t just a bit of British phlegm. It was parasitic too. It involved the kind of irony Nietzsche complained of in Socrates, and promptly employed himself. It was the kind of irony employed by wisdom when it plays the raven, inspired by the intoxicating whiff of carrion coming from its dearest fellow creatures. And as for How To Do Things With Words, even its own didactic function was parasitic on the kind of measured pace of novelistic disclosure that we find in fictional entertainment forms like the detective novel.

Perhaps the disapproval, expressed by Habermas (1985), of the violation of the divide between the literal and the literary genres is justified: perhaps Austin’s own parasitism on fiction damaged his argument, for in the inbred world of language philosophy, the kind of essayistic freshness of insight that Austin seemed to have, may have been hybrid vigour, replete with the instabilities of fiction. Ironically it was a case of Derrida the literalist for rhetoric against Austin the rhetorician for the literal. By the time Habermas had prolonged the same critique of fiction and non literal communication well into the 1980s—especially in the context of what he saw as Derrida’s dissolving the distinction between fact and fiction—the vexed relation of fictive and literal communication had been embedded in so much self-perpetuating debate that the reference of the debate—the actual relation—could scarcely be traced back beyond all the recursions of discursive toing and froing.

Perhaps the debate limped on for so long because the two sides were arguing about different things, or else neither side was clear what the argument was about. Arguments can be like this, each side deluding itself about the other by self-interestedly and self-deceptively mistaking the other’s intentions—an easy task when irony is being employed and meaning’s functions are changeable, and a task that exemplifies the ironic (?) Derridean thesis that every reading is a misreading. These contending theories of fiction are best seen as symptoms of the limitations of the apologetic syndrome. The contender’s theoretical position has usually already been taken, either by unconscious psyche, or in unreflected processes of affective life—like taste or conscience— while arguments scrape together theses to provide a retrospective validation.

History of Communicative Functions

There is a selection pressure for the apologetic syndrome: bewitchment by ontology. Science has emerged from this, and so, long ago, did fiction—it’s just that this enlightenment has long been ignored or trivialised by the naive ontological sensibility for the concrete. In turn, the apologetic syndrome is a kind of cultural selection pressure for the persistence of certain arguments in the theory of fiction and critical theory. Rooted in the task of justification, they confuse norms and facts, hardly pausing to think about the historical interactions between the two. Both sides of the apologetic syndrome located the problem extratemporally in the context of questions about what communication essentially is, and what it ought to be, all too readily assuming that what it is is something essential and obvious to intuitive introspection, and that what it ought to be is to be observed by honouring what it already authentically is. Instead, as is usual in questions of historical function and significance, the problem lies in the fact that functions change, and in how and why they do. Most significantly, biological functions like communication (i.e. functions for an organism), and social functions like symbolic communicability (i.e. functions for a symbol or meme) belong to the great class of historical phenomena (e.g., originality, memorability, promises, the artistic canon, friendship, beginnings, origin, etc.) whose actual nature is only confirmable after the fact, and whose inherence through time is always subject to future changes and disconfirmation.

In the context of the two thousand year old tradition of apologetics, the difficult questions about the functions of communication—its means and its ends—and the historical developments and transformations of those functions, both phylogenetic and cultural, remain. These questions of the historical function and meaning of communication are all too commensurate with the human capacity for wonder, but they might seem to be beyond the historically supplied means of empirical research. This hot combination of fascination and aporia was well known to scepticism, which was duly suspicious of any speculation on the origins of language or of fiction. But there is no escaping these questions. The difficulty of addressing them, and even of properly asking them, is not the kind of predicament that can be turned into an injunction not to ask them. This though was the luxury the philosophy of language long thought it enjoyed—and the predicament it could scarcely avoid.

Thus when Time came on stage in descriptions of human communication, it was scarcely recognised. Usually it entered, as it did in traditional metaphysics, dressed as Logic. The term priority is used as a relation in both time and logic, that is, in both narrative and deductive arguments. Speech act theorists like Searle and Austin used the concept of parasitism to direct philosophical priorities according to seemingly logical priorities in communicative function. As Habermas said, serious, literal usage had to be a logical presupposition of the parasitic uses. But what is parasitism if not a functional relation in which the literal and binding is seen as the means to fictive, non-binding ends? Austin’s, Searle’s and Habermas’s assumption was a procedural move really, with little empirical support, based on an unexamined, metaphysical intuition, and motivated by the need to direct "initial spadework" at the supposed foundation of language. This foundation was imagined as the essence of language, language in its authentic form, and the ground of any other—and therefore non-essential—use of language. Such a move was evidence of the residue of first philosophy in the philosophy of language and the theory of speech acts: demonstrating its genesis in the supersession of origin myths, metaphysics was long grounded on questions of priority detemporalised, asked and answered with steadfast avoidance of historical reflection. If first things can be worked out first, then everything else should logically follow! And the foundation of language, the "normal," "binding", "primary" "literal" use was obvious, wasn’t it?

Well, no! This intuition involves the kind of native idealisation made by intentional idioms. It feels right, but it is not the kind of judgement that science and philosophy can be happy with. At the very least, science and philosophy are not just a matter of the self-observations of psyche, but of the social observation of psyche’s self-observation. Common sense intuition about the normal or basic function of language may come from the horse’s mouth, but what comes from the horse’s mouth is no royal road to science. Opinions and norms about the function of language are the object of philosophy, not philosophy itself as a kind of natural birthright. Despite Peirce’s (131) call for a "barbarous" terminology to distinguish philosophy from the "natural language" that is its object—an object that makes itself mysterious to itself by what Wittgenstein (#4.002) called the unconscious "silent adjustments" of the common idioms used to refer to intentional experience—philosophy has been and remains a discourse in and about intentional idioms. This has always been a particular problem for the analytical philosophy of language, which, for all its achievements, has always been imperilled and inspired by a deceptive mix of half understood intuitions and a myopic bemusement at the banal minutiae of intentional idioms. This is why philosophy is a self-critique of "natural language" and its familiar intuitions—the difficult predicament of the reflexive science par excellence.


Of course, explicit propositional assertion— the literal—is of the utmost importance in linguistic communication. Its meanings are rapidly processed in the search for relevance, because the information is available in an inferentially accessible form. But speech act theory was always rightly concerned with language use as a way of acting in speaking—of promising, declaring, announcing, demanding and so on, as well as asserting. It thus helped direct philosophical attention away from the narrow concentration on the purely denotative, assertoric use of language, and emphasised the fact that there is a whole range of symbolic, communicative actions and that these are governed pragmatically by the intersubjective norms and social institutions in which they are embedded. It should also be emphasised that philosophical attention needs to be directed to all those symbolic, communicative actions that are not spoken but sung or written or screened or staged or whatever.

None of Austin, Searle, Derrida or Habermas was especially concerned with reflecting on the history of communicative functions. Their concern was with the functional state of human language, as if what language is now for intentional consciousness is language in its essential form once and for all. At the very least this ignored the way that language, and symbolic communication generally, change their functional workings in different or new media. To determine relations of functional priority without reflecting on the history of functions was a way of thinking about an historical phenomenon that remained reassuringly and misleadingly ahistorical: "logically", the so called parasitic uses presupposed the possibility of normal, serious, literal, binding use. However just why promising, declaring, announcing, and asserting were normal, serious, binding and primary uses, while quoting, copying, being ironic or metaphorical or fictive were non binding, parasitic, johnny-come-lately uses, is unclear. Unless there just is a natural and original norm of binding, literalness.

Take first the question of binding and non-binding uses. The illocutionary force that binds the parties to a speech act so that they know how to take what is said (or how to take it as relevant) seems to be just as important and fundamental a function in fiction as in other communicative actions. Mary Louise Pratt (1977) applied the theory of speech acts to literary fiction just to argue this point. Habermas thought that fiction suspended or bracketed the illocutionary binding force. But in doing so it sets up its own binding force; it doesn’t say all semantic, intersubjective bets are off. A promise made on stage in a play may be null and void as far as the actors are concerned when they go home after the performance. But the fiction played on stage, in which the promise is embedded, is certainly not null and void as a fiction in the world of the actors and audience; meanwhile, in the world of the fiction, the promise is a promise to be made or broken as promises are. Someone who takes fiction otherwise—as a fact, say—will have wrongly ignored its binding force as fiction.

All this of course can be framed in another already anticipated way, without relying on the speech act theory concepts of illocutionary binding force, and normal, literal use. In Sperber and Wilson’s (1986, 122) terms, communication, whether fact or fiction, normal or parasitic—and, I might add, regardless of medium—involves inferences from a principle of relevance: the audience interprets the text on the basis that it is relevant in the context, and therefore has some psychologically significant effect in that context. This relevance is determined by the degree of contextual effect relative to the degree of information processing effort (that is, to the degree of inferential accessibility) involved in inferring the contextual effect. This characterisation of degrees of effort is determined by shared human psychology (and so may depend on fast, content specific processing as well as on general deductive processing), and the inferences are undertaken using premises available as a shared culture of mutually manifest propositions. According to this characterisation, relevance may be assessed intuitively by effect, and may be represented in comparative judgements. The principle of relevance is thus a way of understanding illocutionary binding force in terms of the amount and benefit of information processing involved. Parties to dialogue assume relevance, and then, based on that constant, infer the variables such as what kind of communicative act is being enacted. That is, the audience infers the precise intersubjective force of the communicative action—whether it is literal or ironic or metaphorical or fictive or whatever—and what manifest, cognitive context it implies. Take a case of metaphor—William Grey’s (2000) example ‘Richard is a gorilla’. If it is manifest that Richard is a man and we were to interpret this sentence on the assumption that it is a statement about primate taxonomy, this sentence would rapidly be judged as false. Instead though, we process it for its relevance (and truth) by attempting to find what the context and the case would be if it were true. What Grey calls the ‘semantic depth’ of a metaphor is a metaphor for the inferential accessibility of its meaning.

Calling certain uses of language normal may have been a way of deciding the procedural priorities of philosophical inquiry, but it pre-emptively "solved" certain practical problems for the inferential processing of communication: it hypostatised so called literal, normal, serious and binding uses of language, and relegated the processing of other functions to a matter of inferring departures from the norm (Sperber and Wilson, 230). Positing a formal solution to a problem may be convenient, but whether humans actually communicate this way is another matter. This formal solution has little empirical support, and little to do with the historical relation and development of different communicative functions, literal, fictive or whatever.

Two Useful Metaphors

Derrida, Grey (2000) and many others have described the literal by using metaphor—that the literal is frozen or dead metaphor. This itself is an ailing metaphor, one that, as Grey would say, is "on the way to expiring". I would prefer to emphasize metaphor’s historical, memetic career and say that the literal is a kind of persistently replicated and selected metaphor. As such, the function for which it has been socially selected has become rapid generation and interpretation in the majority minds. Significantly, the term literal is itself just such a term. Literally, it meant according to the letter. It has been used so often to refer to a certain kind of rapidly interpreted sense of a text that this has become its most inferentially accessible semantic function. In the process of the selection of psychically and socially adapted self-descriptions of human communication, this function has been referred to by emphasising its difference from other semantic functions that have been designated metaphorical, fictional, and non-normal. This socially selected semantic function has been acknowledged after the fact by giving it the title of the literal (and no longer the metaphorical) meaning of literal. Even so, the most commonly offered definition of the new literal meaning of literal is the older, metaphorical one. Other much-replicated, clichéd metaphors are often socially selected for what Roman Jakobson (71) called the phatic function of keeping lines of communication open, or for padding out one’s speech to keep and extend one’s turn in dialogue, or to hold the floor as the socially acknowledged communicator, or to impress people with one’s mastery of eloquent repetitions. Social and psychic functions that were paramount at the initial coinage of a subsequently much-replicated metaphor, including wit, entertainment, amusement, relevance and truth for a particular audience, may no longer be important. The literal has no mortgage on truth. As the slightest reflection on literal lies reveals, truth has much to do with intersubjective intent, as falsehood has with misapprehension. Meanwhile, dead metaphors have a life of their own. At best, for us, they have a literal truth function or maybe just a degenerate phatic function. For themselves though, ease of generation and interpretation is an adaptation to human psyche and society that may evolve to be quite contrary to our interests, particularly to truth. Such an adaptation may lead to their epidemic explosion or persistence. As Daniel Dennett (1991) has said, it’s a case of "how words do things with us."

Many narratives have similar self-serving adaptations. Good fiction is like good metaphor. It uses narrative for our interests—like truth—rescuing it from the self-serving functions of received, clichéd narrative. Significantly, like the term literal, the term fiction has a selection history characterized by metaphorical use. Etymologically, its primary English reference, like that of poetry, may have been to a fashioned or fabricated narrative composition, and its references to things feigned, imaginary, fanciful, misleading, deceitful, dissimulating and so on, may be johnny-come-latelies. Even allowing appeals to the Latin etymon fingere (to fashion), just what sense is primary and what is metaphorical is a question about the contingencies of the selection history of a term bandied about in the psychic and social environment of the self-descriptions of human communication. Like questions of origin in the context of any selection history, it is not going to discover any authentic, original meaning, nor the truth about the phenomenon of fiction. Any truth about fiction is going to have to be about fiction right up to now, when the term has various literal meanings ranging from fabricated untruth to a work or kind of modern (usually literary, cinematic or televisual) narrative art. For a modern artistic project designed to do such things (among many others) as test the adequacy of factual reference, any theory of fiction is likely to be sorely tested by time, if only because fictions themselves will continue to test the adequacy of theories of fiction. No wonder then that theories of fiction resort to fiction in order to be true. We might be tempted say that one of the most frequently voiced theories of fiction is a lie: namely, that fiction is a lie. However, it is a fiction; and read as such, quite true.

The Selection of Functions

Questions about functional relations in communication really need to be considered in the historical context of the coevolution of the brain and symbolic culture. In this context, functional analysis must be directed at the result of what are two interanimating selection processes: genetic and memetic. Bearing in mind that the actual terms literal and fictional are both etymologically specific to written culture and somewhat anachronistic (and metaphorical) when extended beyond that context, let’s assume that they may be extended, metaphorically and as they are, to non literary linguistic acts and then make the following assumption: the claim that fictional language is parasitic on what is deemed to be the normal, binding, literal use of language implies that there is an evolved neurophysiological and social symbolic use for language in its literal form and this in turn provides the functional means for communicating in fictional forms; further, it implies that we could have literal usage without fictional usage but not vice versa. It might mean something else depending on whether we take parasitism to be a logical, ethical or functional relation, or one of various other relations of historical priority such as evolutionary or teleological. In this assumption I have taken it to be the functional relation of a functional means to an end, a relation which, from the micro-perspective of the functional process, is historical, causal and asymmetrical, even if, from a greater time scale, functional relations might be effectively detemporalised and treated as the functional states of systems. None of Austin, Searle, Derrida or Habermas reflects on this all too handy ambiguity that infects the concept of functional relation. Consequently the ambiguity infects claims and counterclaims about priority and parasitism.

It might also have been better to have considered a functional relation between literal and fictive uses of human symbolic communication generally. However this would have been a less intuitively clear assumption. Even so, when dealing with the restricted claims about language, it is important to reflect on the general case; and it should be noted now that, despite the assumed authenticity and historical precedence of spoken language, the ancient multimedia nature of human symbolic communication is something that, right from the outset, would seem to speak against the assumption of any special priority or normality of the literal.

It is my contention that the assumption of the functional priority of the literal is wrong for the following reason: The communicative means supplied by physiological and social evolution are capable of both literal and fictive communicative uses, and both these usages depend on each other by virtue of the peculiar nature of human symbolic communication. The interdependency of the literal and the fictive—and of other uses of language whether deemed to be normal or non normal—arises because the kind of conditional inferences that underlie the processing of symbolic reference and narrative argument demand counterfactual as well as factual assumptions. For a symbolic, social and intentional animal, the world is not only a totality of facts; it is a totality of facts and counterfacts. It is a totality of manifest meanings not of things, a totality of possible representations of events and states.

In the context of the coevolution of the brain and symbolic communication, a particular neurophysiological design or capability has been selected because it was the means to useful symbolic representational and inferential processing and communicative transmission. This new symbolic ability was the functional, neurophysiological means upon which a number of psychically and socially distinguishable communicative uses were parasitic. Most important amongst these uses in terms of phylogenetic selection advantage are things like the communicative representation of goal-oriented processes in a contingent environment, in particular social goals in a social environment, and especially communicative goals themselves. It is precisely such representations, in the face of time and others, that construct the world as a totality of contingent meanings. As I argue below, in order to infer what others mean from what they say we have to entertain a number of counterfactual conditions.

The kinds of communications produced by such a neurophysiology have their own social evolution. And, indeed, present neurophysiology has probably evolved in a selection environment where symbolic communications had already become a major phylogenetic selection pressure. Deacon (345) has suggested that the emergence of symbols as the principle source of phylogenetic selection is "the origin of humanness", "the diagnostic trait of Homo symbolicus." Social symbolic evolution (or physiological evolution for that matter) is complicated enough, but this coevolutionary process would seem to be especially tangled. At the very least we need to distinguish functional designs of organisms for genetic selection from functional designs of communications for cultural selection. While communication has its phylogenetically selected "biological" functions, other "social" functions such as communicability are culturally selected and supervene on the former. The former are physiological adaptations that ultimately favour the survival and reproduction of the organism’s genes. The latter are social adaptations that ultimately favour the survival and communicative reproduction of social entities and include things like memorability, learnability, ease of generation and interpretation, replicability, communicative relevance (e.g. truth, beauty, entertainability) and textual persistence for human psyche. Still other individual "psychic" functions are those personalised communicative functions that a subject uses or takes as her or his own meanings. The scare quotes are to indicate that when we are talking about humans, the biological, the social and the psychic subsume one another: society is biological; and psyche is both social and the environment of society (see Luhmann 1984). Especially in the context of social history (but I suspect also to some extent in the context of phylogeny in a cultural environment) functional designs are being adapted and transformed willy-nilly to new social, psychic and organismic purposes. In the social sphere, they may well have been leap-frogging one another throughout their evolution, just as, now, the literal, fictive, ironic and metaphoric functions of communication may all use and leap frog one another in a single sentence or narrative. In such a context, which function comes first hardly seems to matter compared to which comes last. Besides, whatever came first was itself parasitic on something else anyway. Parasitism, like mediation, is universal.

Naive Categories in the Self-description of Linguistic Communication

The functional differentiation of literal, ironic, metaphorical and other uses of symbolic communication is a culturally evolved categorisation resulting from social selection processes. As a range of categories that divide up linguistic functions in particular at around the micro discursive level of the proposition rather than the argument (which includes dialogue), they are not mutually exclusive nor sharply divided categories, nor does any one category exhaust another. We might say, though, that they parasitise one another. They are somewhat arbitrarily chosen, according to such localised social selection processes as have applied when naïve psychic and social self-descriptions have replicated throughout various historically specific contexts. Except for important differences in communicative technologies, the selection pressures for the categorisation are largely inherent in linguistically constructed psyche and society, so they are largely pancultural, and to that extent the evolution of the categorical structure has been convergent in the various human societies. This is most evident in the case of society insofar as it is born in a pancultural medium like speech, although once other media take up an important place in communicative transmission, new selection pressures begin to operate with consequent changes in the precise structure of the functional differentiation. Traces of the cultural/technological differences in the categorical structure lie in features like the etymology of the term literal itself, which is literally applicable only to written society.

Insofar as these categories are dependent on the history of communicative media and the history of scientific concepts, they are culturally determined and historically specific. They have different institutional forms for modernity in comparison to their ancestral forms in oral societies without written science and history. While the historical importance of speech and speech acts is undoubted, new media transform earlier functions. They selectively parasitise the functions of earlier media. Even if there were normal and parasitic uses of language in speech at a particular stage in its history, the norms are not the same in written language, and they are different again in, say, cinematic communication. What would be the normal, literal, binding uses of film? Does normal, non-parasitic, filmic action lie in documentary or fiction? And what are the different literalities or factualities involved in, say, the direct telecast of court proceedings, a staged re-enactment of the proceedings, a quoted transcript of the proceedings, a silent movie of the proceedings, an edited replay of the proceedings, a summary, diegetic report of the proceedings telecast from the courthouse steps and a feature film of the trial?

As opposed to such categories as literal, ironic, and metaphoric, the categories of fiction on the one hand and history, theory, myth, and poetry on the other (i.e. non fiction) are more strictly applied to the macro level of argument structure, rather than to that of the proposition. Though metaphorical and ironic use of language may be thought of as somewhat fictive, the term fiction is not usually applied to use at the propositional level. Fictions are narratives, and narratives are arguments. A proposition is only fictional insofar as it is, as Peirce (154) suggested, a kind of rudimentary argument. This functional categorization at the level of a work’s argument structure is also a socially (i.e. memetically) selected one. Though it might seem to correspond to the divide between the literal and the other uses at the micro level of propositional communication, this is not a strict correspondence. Rather, during the course of the social selection of the categorisation, the division at the micro level of propositional use has served as a ready analogy for categorisation at the macro level of argument. But this analogy is a bit of a Procrustean bed. These oh so familiar terms demand reflection on their own memetic constitution. As categorisations of the functional differentiation of the use of communicative symbols, whether at the propositional or argument level, they get replicated and selected because, to the intuitions of socially (particularly linguistically) formed psyche, they are easily learnt and communicated; and they are adequate for most everyday (if not for scientific) purposes. In particular, of course, they are perfectly adequate, and therefore all too seductively replicable, for self-descriptions of their own culturally selected, normatively construed ontology of linguistic and symbolic use.


Non-fiction—factual, referential discourse—though it makes an overall validity claim for its representation of actuality, does this by employing propositions and sub-arguments that do not necessarily make such claims. Factual and literal discourse can and must use counterfactual propositions, consider possible worlds and display assumptions without asserting them, if only to represent the fact that people think or might think about these things. Something as important to factual discourse or normal, literal discourse as the interrogative, may involve the display of an assumption, a likeness of which, if true, would be regarded by the questioner as relevant (Sperber & Wilson, 25). Similarly, the imperative—including Donne’s "Go and catch a falling star"—does not represent an actuality. Irony, metaphor, examples, jokes, allegory and so on are all non literal elements of declarative, factual discourse, and all function as premises in the inferential processes of factual discourse. Metaphor, in particular, though the epitome of what is not literal, is, even in its classically lyric context, strongly referential. As Richard Wilbur asked, "Why is a thing most itself when likened?" Meanwhile in fiction, factual and literal premises are employed as part of the fiction’s argument, while fictional works, as social facts, are to be read as significant historical documents.

All these actions that transform literal and factual communication, and that fiction shamelessly exploits, involve the kind of leaping in function or meaning upon which a selection process might eventually confer its imprimatur. In poetic metaphor and in the art of fiction we rehearse the transformation of literal and factual forms insofar as they are self-replicating social norms or symbolic forms that may be antagonistic to human interest—especially the human interest in truth. At the propositional level, ironic and metaphorical uses may involve the attenuation of reference: the inference of the reference is not immediately accessible. However of all the categories, fiction is the one that persists in maintaining the attenuation of reference throughout its entire argument. Whereas irony eventually refers mimetically to something—whether to an earlier speaker of a text or to an earlier expression of a text or to a socially replicated attitude to a text or ambiguously to all three of these things— a fiction’s attenuation of reference achieves a kind of self-reference to its own would-be narrative autonomy. A work of fiction is a kind of monad, but one in which all the windows are left open.

Given the prevalence of imitation in primate nature, it is unlikely that the evolution of communication would ever have been free from mimetic, proto-fictive communicative actions. Such imitative actions may well have copied the empirical form of a communicative action, but in the service of a new function or a new meaning—perhaps a learning function, or a socially binding function like that of ritual, or (as in irony) reference to the first meaning or to its meaner. Making copies and likenesses is an action with the potential for extraordinary utility: the terms replication, imitation and mimesis each suggest slightly different functional values within this field. Language is not alone among symbolic systems in remaining deeply imitative or iconic: the relation between each of the empirical event, the perception of it, the concept, the utterance and the report of the utterance, The cat sat on the mat, is one of likeness. As Peirce (158) and Wittgenstein (#4.01 - 4.022) suggested, the sense, including the literal meaning, of the utterance involves or refers to a likeness of its schematic propositional form—namely, to the perception or concept of the empirical event. That is, as likenesses, they share logical implications. And this means that we still have to process the proposition in the context of some assumptions about a possible world even in order to reveal its explicit, literal meaning or what Sperber and Wilson call its explicature. The literal usage demands that we entertain assumptions about which cat and which mat, and when, and what kind of sitting.

This is simply in accordance with Frege’s (1892) insight into the truth conditions of a sentence: to understand a sentence is to know what would be the case if it were true (Wittgenstein #4.024). Such a counterfactual way of putting it immediately suggests that even a literal meaning depends on assumptions about a possible world or context. For even if the sentence is about an ostensively indicated empirical phenomenon—even if a cat has just sat on a mat before our very eyes—we could still take the utterance to be referring to another cat or mat, and it would still be doing this literally, because literal meaning is not restricted to immediate, ostensive reference. If someone said to us "The cat sat on the mat" when this was empirically obvious and redundant information, our search for the relevance of the utterance might, depending on context, lead us to assume that the speaker is talking—and still literally—about another cat or mat. The point might be that the speaker is trying to tell us that this is not the mat we had assumed it was, but one that we particularly didn’t want to get cat fur on. Stress on the verb might be used to indicate such a sense. If the speaker uttered the sentence without such intonation, as if it were simply an announcement of a particularly mundane event that was manifest to both of us, then most likely, and less literally, the relevant sense might lie in the speaker’s noting that a stock phrase in English pronunciation had just been enacted before us, or the speaker might even be giving us a lesson in English pronunciation, or being sarcastic about stock phrases used in pronunciation lessons. If I were to write, "The dog lay on the log," you would be right to assume that I was referring to the sentence "The cat sat on the mat."

When language refers to linguistic acts or texts, the mimetic, not the literal, is the norm. Wallace Stevens was pretty much right when he said, "Life consists of propositions about life." Such a lot of communication is about communication, and has had to be throughout its biological and social evolution. The reflexive and mimetic are at least as ‘normal’ or as ‘prior’ as the literal. Primarily, gossip is about gossip, rumour is about rumour, legend about legend, myth about myth, history about historical documents. Fiction is about narration— among other things. And these ‘other things’ follow and are recursively embedded in a spoken or unspoken, written or unwritten Have you heard that…It is said that…The story goes that…The records show that…. Aristotle said that the art of narrative consists in the mimesis of life, and, he might have added, especially the mimesis of communicative, narrative life.

Symbolic Communication

Whether a sentence is literal or a fact is not the primary thing. Language has to represent possibility as much as it has to, and in order to represent, actuality, whether communicative or otherwise, in all its contingency. The representation of what is not actual is one kind of function of human communication upon which the culturally evolved functions of fiction and literal communication both supervene. In the context of an essay ‘On Freedom’, Leibniz wrote about how he eventually came to understand the importance of recognising the contingency of things. Though his thinking is couched in terms of an ontology of possible and existent things rather than in terms of representations of actuality, it is an incipient recognition of the connection between the practical (we might say biological) exigencies of representing contingency and, eventually, the evolved, cultural practice of fiction.

For if certain possibles never exist, then existing things are not always necessary; otherwise it would be impossible for other things to exist instead of them, and so all things that never exist would be impossible. For it cannot be denied that many stories, especially those which are called ‘romances’, are possible, even if they do not find any place in this series of the universe, which God has chosen—unless someone supposes that in the vast magnitude of space and time there exist the regions of the poets, where you could see wandering through the world King Arthur of Britain, Amadis of Gaul and Dietrich von Bern. (106)

All of this is related to the nature of human symbols, and to how a symbol is not simply an index. 1

The symbol-to-symbol relations of language—the syntagmatic and paradigmatic relations of word to words and phrases to phrases, and the inferential relations between sentences, rather than just the indexical relations between words and things or sentences and events—are the kinds of relations that are primary for symbolic language, even for literal use. Certain kinds of words have been selected to have an especially prestigious relation to what has been selected (to some extent culturally) as the especially ontologically relevant category of things: the singular term can be used as a kind of symbolically devised index of an individual thing, and such words can in turn function (symbolically and in a context) to tag a sentence’s indexical relation to its referent—that is, to an event or state involving the particular thing. Demonstratives are also used to signify this indexical use of symbolic reference. Insofar as language is symbolic, syntactic, conceptual and inferential it is always something other than the simple, indexical relation, in terms of which the literal is naively thought. Even those classic examples of an index—the pointing finger and the arrow—are, in their now inevitably social context, mediated by human symbolic nous, as is the indexical, empirical (and therefore social) relation of an effect to a cause. (Empirical science’s methodological rule about the social observability of empirical observations was made possible by the insight that if particular observations weren’t socially observable then that social observability could be made virtually real by the socially selected conventions of repeatability, or, more accurately, quasi-repeatability. An empirical observation is thus a replica of a type, and as such, a symbolic act; and the empirical, like so many things in the natural as well as the cultural world, depends upon the ruse of a virtual reality.)

For humans, the relation of a singular term to a single thing is never as simple as that of a pure index, for concepts inform such terms and what they refer to. The terms have a conceptual constitution, and concepts belong to a web of interanimating concepts. They are not even self-identical. Furthermore, the utterance of a single term like "Cat!" is immediately interpreted as a one word sentence, the sense and reference of which demands interpretation. It might mean, "There is a cat." Or it might mean, "The word that I have been trying to remember is cat." Since the truth of propositions is not just a matter of adequacy to some event or state but also of adequacy for some teleological (including, typically, communicative) purpose, the conceptual breakdown of a term can and must be variable, as metaphor demonstrates, and as the historical, memetic drift of the meaning of terms also demonstrates. The notion of a normal, literal meaning of a term is clearly problematic. But when we speak of literal use we are usually referring to the use of whole propositions to refer to events or states, and propositions have an even less direct indexical relation to their referents.

Roland Barthes (1973, 9) said that literal "denotation is not the first meaning but pretends to be so; under this illusion, it ultimately is no more than the last of the connotations." He wrote about literal denotation turning back on itself and indicating its own existence. Such reflexivity, it may already have been gathered, is a crucial and remarkably useful functional device of human symbolic communication. Indeed, self-reference is a defining characteristic of symbols; any replication or communicative embodiment of a symbol refers to a rule—as Peirce claimed, the symbol itself is actually the rule—which is the condition of interpretation of the communication. Many imagine the directness of the literal to be guaranteed by the direct, unequivocal character of the mutual apprehension of solid empirical objects, but in the context of the dynamics of symbolic nature and the drift of historical meaning, this directness relies on the symbolic self reference of language to reconstitute, as it were, its social pretension to ‘natural’, direct indexical reference. The literal, supposedly spelt out letter-by-letter and word-by-word, is a very useful fiction in a symbolically mediated conceptual web.

If we understand the symbolic reference of a sentence according to the Fregean principle about knowing what the case would be if it were true, we are not dealing with indexical truth. The notion that we are probably lies behind the common sense assumption about the functional priority of the literal. We cannot confirm our knowledge of what would be the case simply by observation of the empirical context of the utterance, especially when the context is that of a philosophical example. Instead, even in literal use, we are dealing with counterfactuals—with what would be the case if it were true. In order to speak literally and to say what we mean, we still have to mean symbolically.

The phylogenetically evolved ability to communicate symbolically is, at the same time, an ability to process and to learn to process quite complex conditionals. Linguistic symbols themselves are not simply directly coded rules or laws for generating interpretations; in order to learn linguistic symbols, a child has to learn by postponing obvious or primary associations like indices. The symbolic association is not so simple as that of sign and object or even sign and event or state, because learning the relation between symbols and what they mean is conditional upon other, higher order information, or higher order patterns of association—that is, those that need not be obvious in such a small sample of instances of usage as an infant is likely to encounter before learning grammar. The particular symbolic associations of linguistic grammar are those that could, in Deacon’s (137) terms, "colonise ever younger brains", because the higher order conditionals involved in learning them demand the very shifts of level that younger brains are most readily given to making. Deacon claims that the traits or rules of linguistic grammar are adaptations to the peculiar memetic selection pressures that are brought to bear by the inferential processing of developing infants. Meanwhile, a competent speaker must still process the complex conditionals of narrative and other discursive inference, suspending primary interpretations, while seeing the propositions as opening onto many possible strings of various and calculable likelihood, whose eventual outcome may well reconfigure some supposedly obvious initial interpretation.

Such reconfiguration of interpretation is a necessity in human communication, and one of which narrative art makes a virtue. As Aristotle said in the Poetics (1452a), a plot in which the narrated events go against expectation and yet, as it turns out, are consequential, thereby causes poetic wonder. The fact that some genres of fiction make a fetish of this only proves its significance to narrative aesthetics. In addition, the gratifying character of such a fetish is probably a case of human emotional experience providing evidence of the evolutionary importance of just the kind of rapid, processing of conditional inferences that symbolic communication demands. What I call gratification is a content specific emotional affect that, however urgently, only engages limited cognitive and emotional resources. In terms of psychological evolution such gratification is an affective encouragement for and indicator of useful and rapid, information processing—the kind of psychological adaptation that is likely to have had a selection advantage during human evolution. Of course there is a trade between speed and accuracy. Mere gratification is emotionally and therefore cognitively restricted, and such an affect may be deceptive. Like the senses, the emotions can mislead, as Robert Gray says, ‘when we rely upon only one of them.’ Good art engages us more sensually and in doing so testifies to its greater intellectual engagement.

Symbolic language and communication, at both its argument level and at its propositional level, relies on a brain that can deal with complex counterfactual conditionals. Symbolic communication and human counterfactual conditional reasoning are coevolved and parasitic on one another. So the notion that fiction is parasitic on some logically or necessarily presupposed normal, literal use of language must be regarded as a superseded conceptual relic left over from the unreflected, folk psychology origins of the philosophy of language.


1.  Throughout this essay the terms symbol and index are used in Peirce's (156-173) sense.


Austin, John 1962. How To Do Things With Words. Edited by J. O. Urmson & Marina Sbisà. Oxford: Clarendon Press, 1975 (2nd edition)

Barthes, Roland 1973. S/Z. (Trans. Richard Miller [Oxford: Basil Blackwell])

Deacon, Terrence 1997. The Symbolic Species. The Co-Evolution Language and the Human Brain. Harmondsworth, England: Allen Lane, The Penguin Press.

Dennett, Daniel C. 1991. Consciousness Explained. Boston: Little, Brown and Company.

Derrida, Jacques 1972 Marges de la Philosophie (Margins of Philosophy.) (Trans. Alan Bass [Chicago: University of Chicago Press, 1982])

Frege, Gottlob 1892. "On Sense and Reference" in Philosophical Writings. (Trans. and ed. Peter Geach and Max Black [Oxford: Basil Blackwell. 1952]) pp.56-78

Garner, Helen 1996. True Stories: selected non-fiction. Melbourne: Text Publishing.

Grey, William 2000 ‘Metaphor and Meaning’, in Minerva – an Internet Journal of Philosophy, Vol 4, November 2000,

Habermas, Jurgen 1985. Der philosophische Discurs der Moderne: Zwölf Vorlesungen (The Philosophical Discourse of Modernity) (Trans. Frederick Lawrence [Cambridge: Polity Press, 1992])

Jakobson, Roman 1987. Language in Literature. Krystyna Pomorska & Stephen Rudy (eds). Cambridge, Massachusetts: The Belknap Press of Harvard University Press.

Leibniz, Gottfried Wilhelm 1995. Philosophical Writings. (Ed. G. H. R. Parkinson. Trans. Mary Morris & G. H. R. Parkinson) [London: Everyman, J. M. Dent]

Luhmann, Niklas 1984. Soziale Systeme:Grundriss einer allgemein Theorie (Social Systems) (Trans. John Bednarz, Jr., with Dirk Baeker [Stanford, California: Stanford University Press. 1995])

Peirce, Charles Sanders 1931-1958. Collected Papers. (Vol II) Cambridge, Mass.: Harvard University Press.

Pratt, Mary Louise 1977 A Speech Act Theory of Literary Discourse. Bloomington: Indiana University Press.

Sperber, Dan, & Wilson, Deirdre 1986. Relevance: Communication and Cognition. Oxford: Basil Blackwell.

Wittgenstein, Ludwig 1922. Tractatus Logico-Philosophicus. (Trans. C. K. Ogden [London: Routledge & Kegan Paul, First Paperback edition, 1981])

Copyright © 2001 Minerva.
All rights are reserved, but fair and good faith use with full attribution may be made of this work for educational or scholarly purposes.

Ross Macleay is an independant scholar who also works in ecological survey and restoration for Greening Australia in New South Wales.

Mail to: Ross Macleay

Return to Minerva (Volume 5) Main Page                          Go to Top of This Page