Aristotle definition of poetry, by robert j. yanal
Aristotles Definition of PoetryAuthor(s): Robert J. YanalReviewed work(s):Source: Noûs, Vol. 16, No. 4 (Nov., 1982), pp. 499-525Published by: Wiley-BlackwellStable URL: http://www.jstor.org/stable/2215204 .Accessed: 23/09/2012 13:44Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact firstname.lastname@example.org.. Wiley-Blackwell is collaborating with JSTOR to digitize, preserve and extend access to Noûs.http://www.jstor.org
AristotlesDefinition of Poetry ROBERT J. YANAL WAYNESTATEUNIVERSITY I. THE PROBLEMIn Chapter IX of the Poetics, Aristotle proposes this definition of"poetry": It also follows from what has been said that it is not the poets business to relate actual events, but such things as might or could happen in accordance with probability or necessity. A poet differs from a histo- rian, not because one writes verse and the other prose (the work of Herodotus could be put into verse, but it would still remain a history, whether in verse or prose), but because the historian relates what hap- pened, the poet what might happen. That is why poetry is more akin to philosophy and is a better thing than history; poetry deals with general truths, history with specific events. The latter are, for example, what Alcibiades did or suffered, while general truths are the kind of thing which a certain type of person would probably or inevitably do or say. (1451a36-51b 10)1One natural interpretation of this definition is this. The poet, in orderto make poetry and not something else, must write about possible, notactual, things. These possible things must be of a sort which could becharacterized as types of persons doing things which those types wouldprobably or necessarily do. Non-poetry-history, for example-mustdeal with actual, not merely possible, things. The historian must relatewhat actually happened to particular, actual people. This interpreta-tion requires further analysis, but it will do for the moment, since I nowwant to bring out a feature of the Poetics which casts doubt not only onthis natural interpretation but on the consistency of Aristotles theoryas well. Immediately after writing the above lines, Aristotle admits that the"writers of iambic lampoons" are "concerned with a particular individ-ual" (1451b14), and says this of the tragedians: The tragedians cling to the names of historical persons. The reason is that what is possible is convincing, and we are apt to distrust what has not yet happened as not possible, whereas what has happened is obvi- 499
500 NOUS ously possible, else it could not have happened.... [Someone] is no less a poet if he happens to tell a true story, for nothing prevents some actual events from being probable or possible, and it is this probability or possibility that makes the (tragic) poet. (1451bl5-32)Assuming that the lampooner and the tragedian are poets, the im-mediately preceeding passage suggests that the poet can write poetryabout actual specific persons and events. Yet the essential differencebetween poetry and non-poetry, on the preceeding natural interpreta-tion, seemed to rule this possibility out. The main focus of this essay willbe an attempt to resolve this apparent inconsistency. There are, how-ever, other problems which we must raise in order to get a better focuson the main question. I turn to the first, and easiest, of them. II. LITERARY ARTWORKSWhat is Aristotle defining? The translators answer is "poetry," but thisterm, a translation of poiesis, is misleading. Aristotle takes pains todisabuse his readers of the view that poiesis must be in a poetic rhythmor meter: It is true that people join the word poet to the meter and speak of elegiac poets or epic poets, but they give the same name to poets merely because they use the same meter, and not because of the nature of their imitation. The same name is applied even to a work of medicine or physics if written in verse; yet except for their meter, Homer and Empedocles have noth- ing in common: the first should be called a poet, the second rather a physicist. (1447b 14-20)More importantly, when Aristotle cites some examples of poiesis heincludes certain sorts of texts which are clearly not covered by theEnglish term "poetry": Now the art which imitates by means of words only, whether prose or verse, whether in one meter or a mixture of meters, this art is without a name to this day. We have no common name which we could apply to the mimes of Sophron or Xenarchus and the Socratic dialogues, and also to any imitations that may be written in iambics, elegiacs, or other such meters. (1447a28-47b 14)So, Homer could have written poesis in prose, and Plato wrote pozesisnote the dig at Book X of the Republic-despite the fact that he wrotethe Socratic dialogues in prose. Thus, Aristotle would not have hadmuch trouble accomodating the prose novel and the prose drama,along with poetry in its standard English sense, as polesis; indeed, hemight be said to have anticipated them.
ARISTOTLES DEFINITION OF POETRY 501 By poiesis Aristotle means "the art which imitates by means ofwords only," and his complaint that "this art is without a name to thisday" is remedied in English. The word he wants is "literature" or moreaccurately, "literary artwork," for there are senses of "literature" whichare non-artistic. "The literature on a given topic" is such a non-artisticuse; and there are evaluative senses of "literature," roughly meaning"beautiful writing" or "elegant style," as in "Freuds last books areworks of literature." I take it that Aristotles attempts to prypoiesis loosefrom its connotations of rhythm and meter are intended to pry it loosefrom its connotations of beautiful writing or style as well. Aristotlesadvice on diction (that it "should be clear without being common," etc.(1458a18)) is addressed to both the poet and the non-poet, and one ofhis examples of metaphor is drawn from Empedocles ("old age is thesunset of life" (1457b24)). Even if Empedocles wrote beautifully, he isstill a physicist and not a poet, and a tragedy written in clumsy prose isstill literature and its author a poet. P. F. Strawson once classified Aristotle as a descriptivistmetaphysician. I think he was a descriptivist philosopher of art as well.Tolstoy, the paradigm revisionist philosopher of art, knew his ownconcept of art was not shared by his contemporaries, and Whatis Art? isa diatribe against the current concept. In contrast, Aristotles tone is nothortatory. He expects his contemporaries to share his concept of poiesis,and his definition of this concept was intended to be enlightening notrevolutionary. If it is to be acceptable, the suggestion of "literaryartwork" as the nearest and best English equivalent of poiesis ought topreserve Aristotles descriptivist intention. Subsequent argument willmake this case. III. THE OBJECTS OF LITERARY IMITATIONI shall argue that there is a hierarchy of objects of artistic imitation. Themost general term used by Aristotle to denote such objects is "men inaction," but when Aristotle speaks of tragedies he says that they imitate"plots." I shall argue that the most complex entities, plots, are consti-tuted by men in action (I call these "persons in action"), and thatpersons in action are constituted by simpler entities which I call "artistickinds." Aristotles first statement on the objects of artistic imitation isthis: Since those who make imitations represent men in action, these men must be superior or inferior, either better than those we know in life, or worse, or of the same kind. For character is nearly always derived from these qualities and these only, and all mens characters differ in virtue and vice. (1448a1-5)
502 NOOS In his definition of literature (quoted at the beginning of this essay) Aristotle tells us that literature deals with "the kind of thing which a certain type of person would probably or inevitably do or say," while history talks of specific persons, e.g., "what Alcibiades did or suffered." From these remarks we can initially make these two points about the objects of literary imitation: they are some sort of moral entity, and they are a type of thing (or person). One theory familiar to analytic philosophy is to construe fictionalentities as possible things individuated by the properties ascribed tothem by the author of a work of fiction.2 To set out the bare structure ofsuch a view, we first make the assumption that all the sentences of awork of fiction which describe a fictional entity are reducible to theform "Fa," where "F" denotes some property or relation and "a" is asingular term. It is posited that "a" denotes that possible entity which isindividuated by the set of all properties, F1 ... Fn, such that "Fla" . . . Fna" are directly written or indirectly implied by the author of thework of literature in which the fictional entity, a, appears as a char-acter. This theory has it that the sentence "Fa" may actually occur in thetext, in which case F is directly ascribed to a. Or, F may be somehowimplied by the properties directly ascribed to a, in which case it isindirectly ascribed. Fictional entities are possible individuals which pos-sess all and only the properties directly or indirectly ascribed to them bythe author of the text. A variation of this view holds that fictionalentities are not possible individuals, but rather sets of properties.3 Onthis variation, the fictional character named by "a" is not that possible awhich has all and only F1 . .. Fn;he (it) is F1 . .. Fn itself. This latter sortof fictional entity lays claim to being a type of person. I think a case can be made that Aristotle holds a variation on thepreceeding variation. That is, he holds that the objects of imitations aretypes of persons, but that these types are limited to the moral qualitiesof fictional entities. Let us make a brief excurses into AristotlesNicomachean Ethics before setting out this interpretation in more detail. The virtuous act must have a certain quality, namely the familiar"Golden Mean": So virtue is a purposive disposition,lying in a mean that is relativeto us and determinedby a rationalprinciple,and by that whicha prudent man would use to determine it. (I106b36-1107a2)4The virtuous man must also have certain qualities: But virtuousacts are not done in ajust or temperateway merelybecause theyhave a certainquality,but only if the agent also acts in a certainstate,
ARISTOTLES DEFINITION OF POETRY 503 viz, (1) if he knowswhathe is doing, (2) if he chooses it, and chooses it for its own sake,and (3) if he does it from a fixed and permanentdisposition. (1 105a26-33)Let us say that moral character in the narrow sense (I will shortlyintroduce a broad sense) is constituted by virtues and vices and isdescribable by the virtue and vice terms, such as "courage" and "cow-ardice." Someone is virtuous if he acts from the appropriate moral state(one possessing the three qualities Aristotle mentions), and if his actionfalls on a mean between two extremes; otherwise he is vicious. (Onecomplication I shall mention is that someone can fail to be virtuouswithout being vicious. A child or amoral adult may be incapable ofchoosing from a fixed and permanent disposition, but it might bebetter to withhold the appelation "vicious" from such a person. I shallleave this as a caveat for the discussion which follows.) Aristotles remarks on tragedy force us to broaden our conceptionof moral character: ... [Elor tragedy is an imitation, not of men but of action and life, of happinessand misfortune.These are to be found in action, and the goal of life is a certain kind of activity,not a quality. Men are what they are because of their characters,but it is in action that they find happinessor the reverse.The purpose of actionon the stageis not to imitatecharacter, but characteris a by-productof the action. (1450a15-21) Happiness is the virtuous achievement of what we call a life-plan. A life-plan is a set of typically human ends ordered and weighted with respect to one another. Among the ends humans typically try to attainare health, wealth, power, friendship, family life, benevolence, plea-sure, and contemplation. Person A, for example, might prefer wealthand power to friendship and family life, while person B might preferfriendship, family life, and benevolence to pleasure, wealth, andpower. To be fully happy, Aristotle thinks that one must not only fulfillones life-plan, one must be virtuous in the pursuit of it. But we mustnote that describing someones virtues and vices does not therebydescribe his life-plan, and describing someones life-plan need notdescribe his moral character in the narrow sense. Thus, we cannot limitmoral character, as it relates to the objects of artistic imitation, to virtueand vice. We have to include a persons life-plan in the description ofhis moral character. I shall call this the broadsense of moral character-"broad" because describing the ends a person seeks is not as closely tiedto our concept of the moral as his virtues and vices. Moral qualities in the narrow sense are describable by such virtueand vice terms as being courageous,being angry in the right way and at therighttime,beingangry in the wrongwayand at thewrongtime,beinggenerous,
504 NOOSetc. Moral qualities in the broad sense are describable by such life-planterms as maintaininga stablefamilylife, pursuing wealthto theexclusionof allelse, mixing workand pleasurein equalamounts,etc. As an initial attempt toformulate what the objects of literary imitation are, let us delimit theset, M, which is the set of moral qualities in both the narrow and broadsenses it is logically possible for persons to have. I shall introduce thenomenclature "artistic kind" to capture Aristotles claim that artworksare about these entities: X is a artistic kind of person - df* X is a consistent sub-set of M-properties such that for every M property P, either P or not-P is a member of X. Aristotles insight, as I see it, was to restrict the kinds of propertieswhich constitute an artistic kind to those properties which constitutemoral character. Homer, let us say, writes: "Achilles is fleet of foot."Now, beingfleet offoot may well be considered part of the fictional entitydenoted by "Achilles," at least on the familiar view from analyticalphilosophy adumbrated above. If Homer had written, "Achilles raninto battle faster than anyone else," we might say that he indirectlyimplies that Achilles was fleet of foot, and so beingfleet offoot is part ofthe fictional entity, Achilles. But beingfleet of foot is not primafacie amoral quality, in either the broad or narrow sense, and if the objects ofliterary imitation are partly constituted by artistic kinds, beingfleetolfootis not part of the object of imitation of Homers epic. This result might strike one as counter-intuitive. After all, if Sher-lock Holmes is anything, he is a detective or contains the property beinga detective.Yet being a detectiveis not primafacie a moral quality, and socannot count towards constituting the artistic kind named by "SherlockHolmes." Intuitions are only as good as the insights they offer, and maybe remedied by a striking theory. Aristotles thrust is to direct thereaders attention away from such contigent properties of fictionalentities as the color of their hair, their occupations, and their place ofresidence, towards the more essential properties of their moral char-acter. It is not Holmes profession that should interest us as much as histenacity and his indifference to the feelings of others. Persons areessentially moral agents, and Aristotle thinks that the business of art isto represent what is essential to persons. How does one determine which properties make up an artistickind of person? One engages in a process we can call moral evaluationof character, which I suggest is similar to moral reasoning in real life. Inthe Ethics Aristotle tells us: But this[hittingthe mean]is presumablydifficult,especiallyin particular cases;becauseit is not easy to determinewhatis the right wayto be angry,
ARISTOTLES DEFINITION OF POETRY 505 and withwhom, and on whatgrounds, and for how long ... [I]t is not easy to define by rule for how long and how much a man maygo wrongbefore he incurs blame ... (1109b14-21) The difficulty of determining whether Achilles is virtuously angry or viciously angry is similar to determining whether a real person is virtuously or viciously angry. Similarly, determining the end(s) which a literary character seeks will pose many of the same difficulties as determining the ends real life people seek. There will, of course, be differences. Authors of literature usually provide access into both the exterior and interior lives of their characters, access which we do not, except under extraordinary or first-person circumstances, have in real life. But this is an epistemic difference, not a difference in the sort of reasoning involved. Moral evaluation of character is a process which treats the work of literature as so much information about a (fictional) persons doings in order that moral reasoning may be applied. Evalua- tion of character in literature is moral reasoning, not towards an action as in real life, but rather towards the construction of an artistic kind. The question in reading literature is not what shall I, the reader, do about this (since the situation will usually be either past or fictitious), but rather, what shall I, the reader, say about the moral character of this fictional entity. The results of such inquiries will be moral characteris- tics which will be eventually built into an artistic kind. We might disagree about the constitution of the kind, Archilles. Some might say he was virtuously angry to leave Agamemnon; some might say he was angry in the wrong way or for the wrong reasons.Such disputes about artistic kinds are resolved, or not, in the way real-life moral disagreements are resolved, or not. The importantpoint is that the properties constituting artistic kinds be moral qualities,that they be the results of moral evaluation. Calling Othello stupid orLear senile counts as moral evaluation if and only if one is prepared tosay of "stupidity" or "senility" that it is a virtue or vice term (andperhaps have a moral theory to back this up). We have seen that Aristotle has stressed that tragedy is an imitationof an "action," and that his first term for the objects of artistic imitationis "men in action." Now, artistic kinds are not the most appropriate sortof entity to count as actions. An action, even in the somewhat meta-phorical sense Aristotle is using it in the Poetics, must be at least somesort of change. That is, it must have parts which are simultaneouslyincompatible with one another. The tragic hero, for example, must behappy and then miserable. Artistic kinds seem more like a static photo-graph of someones moral qualities at a given moment than a motionpicture of their action. There is another element of men in action, as the objects of artisticimitation, which the concept of artistic kinds does not capture, and this
506 NODSis that "men in action" must not only signify a change, it must signify achange which happens "in accordance with probability or necessity." Suppose a fictional character-Lord Jim, say-goes from cowar-dice to courageousness in the course of a novel. This action cannot berepresented as an artistic kind because, although courageousness andcowardice count as moral qualities, both cannot be included in a consist-ent set of such properties. Also, a mere concatenation of artistickinds-Lord Jim as cowardly, Lord Jim as courageous-will not pos-sess the sort of inevitability which Aristotle demands for men in action. I therefore suggest the following definition for the objects Aristo-tle calls "men in action," here called "persons in action," again letting Mproperties be the moral characteristics, in both the broad and narrowsenses, which it is logically possible for persons to have: X is a person in action = df* X is a set of properties, A, such that (1) A consists of subsets of M properties, M1 ... Mn; (2) each subset, Mi, of A is a consistent set of M properties such that for every M property, P, either P or not-P is a member of Mi; and (3) A is ordered by the rules of probability or inevitability.Action in literature is more akin to the pastactions of persons than it is tothe actings of persons. Action in literature does not take place over time:it exists, all at once, in metaphysical space. Action in literature is, atleast, change in the moral qualities of the fictional character, but achange that, so to speak, has already happened. Whatever their tempo-ral settings, stories have the flavor of a world that always was. Although Aristotle sounds as if he is aiming at one sort of object forall art and in particular all literature to imitate, this, I think, is mislead-ing and for two reasons. First, in Chapter VIII Aristotle says that a storyis an imitation of "plot," and the question is whether plot is to beidentified with persons in action. Second, intuitions about art-that is,intuitions cast within the Aristotelean framework of the Poetics-suggest that not all art, not even all literature, is an imitation of personsin action; and that a special sense should be assigned to plot to distin-guish it from persons in action. Let me take up the second point. Some kinds of art-painting andsculpture, for example-seem to be best construed as imitations ofartistic kinds rather than of persons in action. After the quote that"those who make imitations" represent men "either better than thosewe know in life, or worse, or of the same kind," Aristotle says, "We cansee this in painting." This quote spawned the definition of "artistickinds," and it seems that the most appropriate Aristotelian sorts of
ARISTOTLES DEFINITION OF POETRY 507 entities for painting to imitate are artistic kinds. Rembrandts Lucretiais more a static picture of moral qualities than it is of change. There are dramatic paintings, Davids Death of Socratesfor example, which mightbe said to imitate a person in action. But even here I would argue that such works imitate an artistic kind-Socrates nobel acceptance ofdeath-rather than a person in action. The same holds true for non-narrative poetry, the lyric for exam-ple. Shakespeares twenty-ninth sonnet ("When in disgrace with For-tune and mens eyes . . . ") is a picture of the acceptance of thecomforting power of love. It is not about any change which has to beorganized according to the rules of probability or inevitability. Insaying what such poems are about we tend to use nominals-the ac-ceptance of love, the acknowledgement of Gods grace, and the like-appropriate names for the sort of entity specified by artistic kinds. It ishard to pin Aristotle down on this, for he skips meagerly through hisdiscussion of the non-narrative, non-literary artforms. His main inter-est is in the tragedy (and, but only to draw contrasts, the epic). Thus, heanticipates himself, and in leading us to narrative or dramaticliterautre, where persons in action may appropriately serve as theobjects of imitation, he fails to give thorough consideration to whetherall art imiates persons in action. We must advance to a discussion of "plot" for this entity is made tobear the burden of tragic imitation-and, by extension, of other narra-tive and dramatic literary imitation as well. Aristotles most sustainedremarks on plot take up Chapter VIII of the Poetics, and I shall quotethem at length: A story does not achieve unity ... merely by being about one person. Many things, indeed an infinite number of things, happen to the same individual, some of which have no unity at all. In the same way one individual performs many actions which do not combine into one action. ... Now Homer ... seems to have perceived this clearly, whether as conscious artist or by instinct. He did not include in the Odysseyall that happened to Odysseus-for example, his being wounded on Parnassus or his feigning madness when the troops were being levied-because no thread of probability or necessity linked these events. He built his plot around the one action which we call the Odyssey;and the same is true of the Iliad. As in other kinds of imitative art each imitation must have one object, so with the plot: since it is the imitation of an action, this must be one action and the whole of it; the various incidents must be constructed that, if any part is displaced or deleted, the whole plot is disturbed and dislocated. For if any part can be inserted or omitted without manifest alteration, it is no true part of the whole. (1451a17-34) It is here suggested that plot is the object of imitation for narrativeand dramatic literature, though this is not completely clear. It would bemost natural to say that works of literature have plots, and it is by virtueof their having plots that they imitate their objects. Aristotle adopts this
508 NOOS natural mode when he says that "it [plot] is the imitation of an action," as if to say that the Odysseyhas a plot and it is by virtue of its plot that it imitates an action. Yet at other points Aristotle comes close to identify- ing plot with the object of imitation. Homer "built his plot around the one action which we call the Odyssey," and this "one action" is under the same constraints which plot is under: "this must be one action ... ; the various incidents must be so constructed that, if any part is displaced or deleted, the whole plot is disturbed and dislocated." "Plot" seems best reserved as a name for the unified sort of action which Aristotle insists that works of literature imitate. What is wrong with "person in action" to signify this sort of object of imitation? Above, Aristotle claims that being about one person is notsufficient for having a unified plot. Presumably it is not necessaryeither, for a story can be about more than one person and still have therequisite unity. However, Aristotle does not discuss such cases. Hisexamples are drawn from works of literature-a Heracleid, a Theseid,the Odyssey-which, as a matter of fact, are about one person. Yet howare we to account for the unity of Anna Karenina, with its cast ofcharacters? Anna, Vronsky, Karenin, Levin, Kitty, and the rest seemindependent persons in action, though their actions coincide in impor-tant ways. Just as LordJimshould not be represented as a concatenationof artistic kinds (Lord Jim as cowardly, as courageous), so too, AnnaKarenina should not be represented as a concatenation of persons inaction, each with its own set of moral qualities changing according tothe constraints of probability or inevitability. Such a description isappropriate for a book of sketches-Lillian Hellmans Pentimento, forexample-but not for a unified novel. So, we ought to allow for a still more complex object of imitationfor the developed narrative or dramatic literary artwork. I shall call thisentity "plot" and shall define it as follows: X is a plot =df. X is a set of properties, P, such that (1) P consists of sub-sets of properties, P1 ... Pn; (2) each sub-set, Pi, of P is a person in action; (3) P is ordered by the rules of probability or inevitability; and (4) P is "whole and complete" (1450b24) or has the requi- site "unity".We have preserved Aristotles insight that the objects of imitation are tobe restricted to the moral qualities of fictional characters, and haveallowed room for his insistence on the unity of changing moral qual-ities.
ARISTOTLES DEFINITION OF POETRY 509 I shall make some brief remarks on the "rules of probability orinevitability" which order persons in action and plots. For all the weightthis concept has to bear in the Poetics, Aristotle is vague as to exactlywhat it is a concept of. However, we can distinguish at least two aspectsof probability in literature. Persons in action and plots must be realisticand plausible. The rules of realism are equivalent to the rules whichgovern the behavior of real-life human beings. Aristotle says that"characters must be appropriate or true to type" (I454a22-23), and that"when the poet is imitating men who are given to anger, indolence, andother faults of character, he should represent them as they are . . .(1454b1 1-13). The rules of plausibility are those which lead the reader to acceptor believe-in whatever appropriate sense there is to believing a storyone knows to be fiction-the events in the narrative. "What is impossi-ble [presumably, according to the rules which govern the behavior ofhuman beings] but can be believed should be preferred to what ispossible but unconvincing" (146Oa26-27). Aristotles view, then, seemsto be that the poet must follow the rules which govern the behavior ofreal-life human beings, except when deviation from these rules isrequired to produce conviction. Genius, we might say, without knowl-edge of the laws governing the behavior of real-life humans is blind;mere psychology without genius is empty. On "wholeness and completeness" or "unity" Aristotle is alsovague. Homer is praised for leaving out "all that happened to Odys-seus," for example Odysseuss being wounded on Parnassus, and thisbecause "no thread of probability or necessity linked those events." Incontrast, the historian "has to expound not one action but one period oftime and all that happened within this period to one or more persons,however tenuous the connection between one event and the others"(1459a21-24). Apparently, then, a plot may follow the rules of realismand of plausibility-Odysseus being wounded could be convincinglyand realistically described-but still fail to have the requisite unity.Aristotle gives two "tests" for this unity: if the insertion or omission ofan incident causes "manifest alteration" in the plot; and if events followone another with a "common end in view" (1459a24). Now, any addi-tion or omission will cause, trivially, an alteration in the plot, so Aristo-tle must have in mind some special sort of alteration, but he never tellsus which sort. It is possible for Aristotle to specify a "common end" fortragedy-catharsis-but it is unclear what would count as the commonend of, say, the novel. A full account of unity remains one of the glaringomissions of thePoetics. It should, however, be noted that unity is not anessential aspect of plot, but only of superior plots. Epic poetry, inAristotles estimation, is like history in that it has episodes with nocommon purpose (see 145lb33-37).
510 NOuS IV. LITERARY IMITATIONHow does literature come to "imitate" persons in action and plots? Allart-Aristotle mentions epic poetry, comedy, dithyrambic poetry (i.e.,choral odes), music on the flute and lyre, dance, and painting-is saidto be "imitative." At the beginning of the Poetics Aristotle embarks on aproject of distinguishing artforms according to their means, manner,and object of imitation. Aristotle never defines what "means" and"manner" are. He gives examples. Manner seems peculiar to literature,for one manner of imitation is first-person narration, another is dra-matic impersonation and a third is a combination of the first two. Thesemanners seem to have no sensible application to music or painting.What could it mean to paint or play the lyre, sometimes in the first-person, sometimes in the third? In Chapter I we are told that there are two general means ofartistic imitation: "color and shape" and the human voice. The latterhas three species: melody, rhythm, and speech. We can assemble thefollowing categorizations from Aristotles remarks: (1) Painting imitates (artistic kinds) by means of color and shape. (2) Music imitates (artistic kinds? plots?) by means of mel- ody and rhythm. (3) Dance imitates (artistic kinds? plots?) by means of rhythm alone. (4) Theatre (i.e., either a staged play, or what we would call a dramatic recitation or reading) imitates (plots) by means of melody, rhythm, and speech (in the manner of dramatic impersonation). But Aristotle has forgotten that there is at least one other generalmeans of artistic imitation, and the problem to confront us here is howto analyze "imitation" in: (5) Literature imitates (plots) by means of "words alone" (either in the manner of first-person narrative, or dramatic impersonation, or a combination of both).Aristotle never provides an explicit analysis of what "imitates" means,nor does he say whether the means of imitation modify the sense of"imitates" or whether this relation retains a univocal sense in all of (1)through (5), regardless of what means are employed to do the imitat-ing. Commentators on the Poetics make Aristotle out to hold an iconictheory of imitation. S. H. Butcher interprets Aristotle to mean, "A work
ARISTOTLES DEFINITION OF POETRY 511of art is a likeness or reproduction of an original, not a symbolicrepresentation of it" (: 124), D. W. Lucas concurs: "Mimesismeans tomake or do something which has as a resemblance to something else"([ 13]: 53), Grube comments that sense of "imitation" as "resemblance"is used by Aristotle "not as a view that has to be argued, but as one thatcan be taken for granted" (: xix). Now, there is a certain plausibility that (1) means: (1) Painting resembles (artistic kinds) by means of color and shape.And the same plausibility carries over to (2) through (4). After all, colorand shape, and rhythm, melody, and diction are plastic or auditorymaterial objects, and so have some chance of resembling their objects ofimitation. There are problems with this, of course. Nelson Goodman([ 10], chs. 1 and 2) has launched some powerful objections against theidea of a natural resemblance between art and its objects; and Aristotlewould have to face some particularly severe problems of how particu-lar, material objects could resemble artistic kinds, persons in action, orplots, which are, among other things, abstract objects. However, what-ever plausibility there may be with analyzing (1) through (4) along thelines of (1) vanishes when it comes to (5). In what sense could Emma-aseries of words and sentences-be said to resemble person in action, theEmma? Butcher notes this difficulty: Poetry unlike the other arts produces its effects (except such as depend on metre) through symbols alone. It cannot directly present form and colour to the eye; it can only employ wordsto call up imagesof the objectsto be represented;nor need these wordsbe audible;they may be merely written symbols. (: 137)Here Butcher seems to turn the iconic analysis of imitation into anexpression or causal analysis. This, however, seems to be a theoryentirely of Butchers own invention, since nothing in the Poetics sug-gests it and Aristotles remarks on catharsiscontraindicate it. The spec-tator of a (good) tragedy is supposed to be filled with pity and fear andto undergo a catharsis of those emotions. That is, a (good) tragedycauses the spectator to have feelings of pity and fear. Yet tragedy is animitation of a certain plot, not of pity and fear. Moreover, Butcherpresents an implausible phenomenology of the act of reading andunderstanding a literary text. Here is the opening of Jane AustensEmma: Emma Woodhouse, handsome, clever and rich, with a comfortable house and happy disposition seemed to unite some of the best blessings
512 NOOS of existence and had lived twenty-oneyearsin the worldwithverylittleto distress or vex her. I do not form "images of the objects to be represented"-mental pictures of a handsome, clever, rich, undistressed, twenty-one year old woman-when I read this passage. I may imagine the state of affairs described above to be true, but this is a different thing from imaging that state of affairs. For these reasons Butchers expression- interpretation of literary imitation in the Poetics is to be rejected. "Words alone" can be said to resemble aspects of their objects of imitation. Gibbons sentences get longer and longer as he describes provinces further and further away from Rome. Joyces prose in the "Nighttown" episode of Ulyssesbecomes disjoint and scattered, just as the incidents he is creating become disjoint and scattered. Joan Did- ions prose style is as "cold" as the characters she describes (A Book ofCommonPrayer). As a complete theory of literary imitation, resemblance can only goso far. Gibbon is describing more than just the distance of the farprovinces, Joyce more than disjointness, Didion more than coldness.Even if we take the bare bones of the structure of tragedy as put forthby Aristotle, we can see failures in accounting for this structure bymeans of resemblance between the words alone and the tragedy. A(good) tragedy is to take a noble hero from happiness to misery via areversal and discovery. Can the words or sentences be said to be happyor miserable? Can they undergo a reversal, a discovery? In their de-scription of the hero, need the words be noble? The fact is that literaryrefer to or denote their objects of imitation. Exactly how they do this willbe the problem for the next two sections of this essay; and this, indeed,is the crux of the problem of interpreting (away) the inconsistencyadumbrated at the beginning of the essay. Aristotle does not seem to have bothered much with this problemin the Poetics, perhaps because literary artworks were presented asstaged-by means of "melody, diction, and rhythm"-and the post-Gutenburg fashion of taking a text itself as the artwork had not beenfully anticipated by Aristotle. The modern philosopher has to seewhether this lacuna in the Poetics can be successfully and consistentlyfilled in. V. THE PROBLEM RESTATEDWe are in a better position to return to the chapter IX definition andthe apparent inconsistency adumbrated at the beginning of this essay.Before setting out this definition and making the inconsistency morethan merely apparent, we should make sure Aristotle intends the
ARISTOTLES DEFINITION OF POETRY 513chapter IX definition to be the canonical definition of literature, forthere seem to be competing definitions in the Poetics and, as we shallsee, the chapter IX definition is not a definition but a statement of anecessary condition for a text to be a literary artwork. To facilitate discussion, I shall say that a person-nameis a singularterm such that, were it to denote an existent particular, it would denotea person. Person-names are to be distinguished from place-names,animal-names, etc. Animal allegories may appear to contain animal-names, but I would urge that the characters in such stories are personswith non-human shapes. In chapter VIII, Aristotle strongly implies that the definition ofliterature is this: (A) For an text, T, T is a literary artwork if and only if T denotes (or is about) a plot.To be sure, the highest literary artform for Aristotle, the tragedy,imitates a plot. Yet he admits that some epics lack one of the ingredientshe refers to as "unity", and from an inspection of the various genres ofwhat we would acknowledge to be literature, it seems that (A) is toostrong. Some epics will lack a plot, and so will the lyric poem. Moreover, a text will denote a plot only if all of its person-namesdenote persons in action; and a person-name will denote a person inaction only if it denotes an artistic kind. The chapter IX "definition"contrasts the literary artist who writes about a person in action ("thekind of thing which a certain type of person would probably or inevit-ably do or say") with the historian who writes about particular persons("what Alcibiades did or suffered"). Here, Aristotle seems to focus onhaving person-names denote persons in action as a necessary conditionfor literature; that is: (B) For an text, T, T is a literary artwork only if T contains person-names and all of the person-names in T denote persons in action.(B) seems to be the best interpretation of what Aristotle says in chapterIX, but not the best interpretation of the literary artwork in general. (B)leaves out the lyric poem whose person-names-here I count "thedramatic speaker" as a person-name-will denote artistic kinds. Abetter statement would be: (B) For any text, T, T is a literary artwork only if T con- tains person-names and all of the person-names in T denote either persons in action or artistic kinds.
514 NOOSHowever, nothing is lost if, in the following discussion, we confineourselves to the simpler (B). The same problems arise with (B) as theywould with (B). There is another competitor for "the" definition of literature inthe Poetics. At the very end of chapter IX, Aristotle says that "it is thisprobability or possibility that makes the (tragic) poet." This jettisonsartistic kinds from literature, for it says that the poet can write about-use person-names which denote-actual, particular persons as long asthe poet orders their doings according to the rules of probability andinevitability. What Aristotle seems to be advocating there is: (C) For any text, T, T is a literary artwork if and only if T contains person-names, and whatever sorts of entities that are denoted by these person-names are ordered according to the rules of probability or inevitability.I think Aristotle has made a slip of the pen. (C) would make artistickinds accidental features of some literary artworks, yet the Poetics ispermeated with talk of how the artist imitates moral types. Moreover,Aristotle admits that "nothing prevents some actual events from beingprobable or possible." Accepting (C) would obliterate his distinctionbetween literature and history. Suppose a poet happened to choose areal-life incident which, as a matter of fact, conformed to the rules ofrealism, plausibility, and unity. How would he be distinguished fromthe historian who happened to write about the same incident? For thesereasons, (C) is to be rejected as Aristotles canonical defintion of litera-ture. The inconsistency can now be made more apparent. (B) impliesthat OedipusRex is a literary artwork only if "Oedipus" denotes a personin action. Intuitively, OedipusRex is a literary artwork, and so "Oedipus"should denote a person in action. The lampooners, Aristotle tells us,are "concerned with a particular individual," and the tragedians "clingto the names of historical persons." This is an admission that a text cancontain person-names which denote actual, particular persons and stillbe a work of literature. We have a choice of contradictions; eitherOedipus Rex is and is not a literary artwork (intuitively, it is, but"Oedipus" is alleged to denote a real person); or "Oedipus" bothdenotes and does not denote a person in action (it should denote aperson in action because OedipusRex is a literary artwork, but the fact isalleged to be that "Oedipus" denotes the historical Oedipus). Weshould not, of course, abandon the intuition that OedipusRexis a literaryartwork. We will, in due time, abandon (B), but first it behooves us tosee whether we can resolve the inconsistency that "Oedipus" does anddoes not denote a person in action, while holding onto (B).
ARISTOTLES DEFINITION OF POETRY 515 Some commentators have pointed out that Aristotle uncritically swallows the doubtful view that the great tragedies are about real persons. This may be so, but it doesnt obviate the possibilitythat the literary artist may write about real persons. For one thing, Aristotles remark on the lampooners is correct. The lampoon is necessarily about someone real. For another, his reason for such artistic practices in tragedy sounds good: "we are apt to distrust what has not yet happened as not possible, whereas what has happened is obviously possible, else it could not have happened." Writing about real persons and their actual doings is a good rule of plausibility. Renaissance historical drama was defended by one of its contemporaries "with the argument that a plot about a supposedly great king of whom the audience had never heard would be laughed out of the theatre" (: 159). Wouldnt Napoleons retreat in War and Peace from Moscow into the Russian winter and defeat strike one as implausible or unrealistic-such a blunder from a military genius-if one didnt know that the actual, historical Napoleon really did it? Some commentators on Aristotle think that the particular is some- how changed into the universal by art. Butcher writes, "Greek tragedies, though founded on fact . . . transmute that fact into im- aginative truth" (: 170); and Gerald Else says, "The poet . . . is required to make something for himself, namely that structure of events in which universals may come to expression. . . " (: 320). It is hard to see how facts can become "imaginative truth," or how univer- sals can "come to expression" via the particular. The metaphysical wallseparating the two is impenetrable. Perhaps what can happen is thatthe names in literature shift reference,from denoting some actual par-ticular to denoting a person in action. There is a twentieth-century tradition which holds that the refer-ence of names in literature is different from the reference of thosesame names outside literature. Some philosophers have countenancedthat names in literature are used in ways that are radically differentfrom their use outside literature. Ogden and Richards (), forexample, have claimed that names and terms are used in literature toevoke emotional attitudes. The anti-intentionalists (see Wimsatt andBeardsley  and Macdonald ) hold that it is an error to go"outside" the work, and that, accordingly, all names in literature mustbe taken to be purely fictional. What such views must hold is that there is some property whichliterature has and which non-literature lacks, a property which some-how prevents an author from making reference to real persons in awork of art. Call this property Q. If a text has Q. then try as he might, anauthor cannot succeed in making reference to any real person on anyoccasion of the use of any person-name in that text. Such a view is
516 NOUS unsupportable. Is there the slightest reason to believe that the refer- ence of "Napoleon" in the sentence "Napoleon marched from Auster- litz to Moscow" must change from the historic Napoleon to a person in action, depending on whether that sentence occurs in A History of theModern Worldor in War and Peace? Authors have often used the oppor- tunity to make real reference to real persons in works of literature,whether to edify, establish plausibility, or just to get even, and there isevery reason to believe that they have succeeded in making such realreference, which gives us every reason to believe that there is no suchproperty Q. As a final stab at trying to hold interpretation (B) while dissolving the inconsistency, we might hold that person-names in literature are sometimes systematically ambiguous. While it is usually held to be the essence of proper names that they denote one and only one thing (if they denote anything at all), it might be the case that, at least for art, some person-names are bi-denotative, denoting botha particular and a person in action. I see nothing incoherent about bi-denotative names. If I invite Simone de Beauvoir and Simone Signoret to dinner and want to summon them both to table, I suppose I can use the name "Simone"on this occasion to call both. That is, I intend "Simone" in the sentence,"Simone, a table," to refer to both women. The Simones may becomeconfused, but this may be because they never thought that a namecould be bi-denotative, or that this use of "Simone" is bi-denotative.Similarly, I suppose that it is possible that Sophocles used "Oedipus" torefer both to the historic Oedipus (assuming there was such a person)and to a certain person in action. The problem here is not with any inherent incoherency in thenotion of a bi-denotative person-name, but with using this as a criterionfor distinguishing literature from history. Are tragedians known forusing names in a bi-denotive fashion, just as I might be known forreferring to both Simones on occasion as "Simone"? This is unlikely.We would have to assume that Sophocles held Aristotles theory (B) andused "Oedipus" to refer both to the historical Oedipus and to a certainperson in action. But Sophocles need not have held Aristotles views-probably no one did before Aristotle-and he still could write a tragedywhich Aristotle ought to be able to account for without second-guessingthe dramatists intentions. Suppose we have the text of King Lear in front of us, and we dontknow whether it is literature or not. Suppose we also know that Shake-speare used "Lear" to refer to that historic king which Holinshed used"Lear" to refer to in his Chronicles. Is there any way to tell whether"Lear" functions in a bi-denotative way in Shakespeares play, denotingboth the historic Lear and a person in action? Suppose a letter ofShakespeares were produced in which Shakespeare claimed to use
ARISTOTLES DEFINITION OF POETRY 517"Lear" to refer to the historic Lear and only to the historic Lear. Wouldthis mean thatKingLear is not literature? Or consider what is most likelythe actual case: Shakespeare got the story for King Lear out ofHolinshed, but had no intentions one way or the other, either to referto that historical Lear of Holinshed or to an Aristotelean person inaction or to both. Must we assume that "Lear" in Shakespeares playdenotes both a person in action and the historical Lear, but that "Lear"in Holinsheds Chroniclesdenotes only the historical Lear? Why? Be-cause Shakespears play is literature and Holinsheds Chroniclesare not?Yet, isnt this precisely what we do not know? This discussion of names traded heavily on a Kripke-like accountof denotation as determined by the intentions of the user of the name.Similar quandries would arise with a Fregean sense-determines-denotation theory. The anti-intentionalists would have to claim that thename "Napoleon" changes its sense from its occurrence in A HistoryoftheModern Worldto its occurrence in Warand Peace. Yet, what reason dowe have to believe this? The bi-denotative name gambit would have tohold that in certain texts the sense of a name is such that it denotes botha person in action and a real, particular person. But, how could we evertell? VI. A SOLUTION Interpretation (B) founders on an ineluctable fact. Some literary artworks contain person-names which denote real, particular personsjust as some works of history do, and no conjuring up of eccentrictheories of meaning or denotation for literature can change this. (B)also must compete with a somewhat less ineluctable fact, and that is thatthe theory of reference of fictional names adumbrated in section IIIhas some plausibility. If Homer writes, "Achilles was fleet of foot," then"Achilles," if it does not denote any historical personage, can be rea-sonably said to denote that possible particular which is fleet of foot andhas all the other properties Homer directly or indirectly assigns toAchilles. (B) takes the literary artworkas a text which makes its appearance assuch. I would like to suggest that the literary artwork is almost as mucha construction by the reader as it is by the artist. Texts are not, so tospeak, ready-made artworks or not. They must be handled with care,nurtured into artworkhood. They must be read in a certain way. I shallsuggest that there is a lacuna rather than an inconsistency betweenAristotles definition of poetry and his claims about real denotation inliterature, and that this lacuna is to be filled in with a prescriptiveaesthetic. An aesthetic, as I shall use the term, is a theory of spectator
518 NOCS response to artworks. We can distinguish between a descriptive and a prescriptive aesthetic. A descriptive aesthetic is an empirical theory of what spectators actually experience when confronted with artworks or with artworks of a certain sort. Santayanas account of aesthetic plea- sure (in [ 17]) purports to be a general theory of spectator reactions to good artworks; another descriptive aesthetic (Gutheil, [ 11]) purports to describe the emotional reactions of the auditors of various pieces of music. A prescriptive aesthetic is a set of directions on what spectators ought to do when confronted with artworks, or with artworks of a certain sort. Bullough, for example, thought that if the aesthetic prop- erties of art are to "emerge" one had to behold the artwork with the correct amount of "psychical distance," which method his essay  describes. A descriptive aesthetic is of merely anthropological interest. The philosophical question is what spectators should do when confrontedwith artworks or with artworks of a certain sort. Should spectators try to unpack the cognitive or informative content of artworks as Goodman holds (, ch. VI)? Or, should spectators luxuriate in the unity, com- plexity, and intensity of the essentially non-cognitive aesthetic experi-ence, as Beardsley (, chs. I and XI) has urged? The answer mustcome from a theory of the nature of art. If one believed that art isessentially a cognitive tool, although different in certain ways fromscience, then one would posit a prescriptive aesthetic which directedthe spectator to take certain stances so as to allow art to perform itscognitive task. If, on the other hand, one believed that art was a vehicleto a certain kind of non-cognitive or affective experience, then onewould develop a prescriptive aesthetic to help spectators have this sortof experience. The nature of literature for Aristotle is to bring into sharp focusthe moral qualities of humans. Any Aristotelean prescriptive aestheticwith respect to literature must be such as to allow specators to see thenature of literature. I shall expound on this shortly, but what I shall callthe Aristotelean prescriptive aesthetic-Aristotelean, not Aristotles,for I introduce it to solve a puzzle in Aristotle, not as exegesis of somepassage or other-has already been set out descriptively in this essay.The Aristotelean prescriptive aesthetic directs the readers of certaintexts to ignore any reference to actual, particular persons which someperson-names might make; to set about constructing the appropriateartistic kinds, and from these, the appropriate persons in action foreach ofthe person-names in those texts; and finally to combine thesepersons in action into a plot. Twentieth-century aesthetic attitude theories are the most fullydeveloped prescriptive aesthetics available. It will be worthwhile toreview their essential structure. The pivotal feature of their
ARISTOTLES DEFINITION OF POETRY 519structure-and the feature which has attracted the most criticism (seeDickie , chs. 4-6)-is their introduction of a certain mental state, theaesthetic attitude, variously described as "distanced" (Bullough ),"disinterested" (Stolnitz ), and "prehending" (Aldrich ). Thereason such mental states are introduced is that the nature of art, whichfor the aesthetic attitude theorists is the possession of aesthetic proper-ties, is not revealed until the aesthetic attitude is trained on the artwork.This leads to the second feature of the structure of aesthetic attitudetheories: that the nature of art, here conceived of as the possession ofaesthetic properties, is emergent. What it means for an aesthetic prop-erty to be emergent varies from theory to theory. Bullough and Stolnitzseem to hold that aesthetic properties go, so to speak, unnoticed untilone assumes the proper amount of distance or becomes sufficientlyand correctly disinterested. Aldrich seems to hold that aesthetic prop-erties are created when the mind, in its attitude of prehending (anelaboration of Wittgensteins "seeing as") perceptually interacts withthe artwork. The third feature of aesthetic attitude theories is virtuallyimplied by the first two, and this is that the prescription that theaesthetic attitude should be directed towards artworks, as the correctmode of appreciating them. Aristotle is not concerned with aesthetic properties, at least asthese have been conceived of by aesthetic attitude theorists. The closesthe gets to a modern notion of aesthetic properties is in his discussion ofdiction and metaphor in chs. XIX - XXII of the Poetics. These must betaken as remarks on some of the accidental, albeit interesting, featuresof literature, for literature may lack the aesthetic qualities of dictionand metaphor and still remain literature. Furthermore, the Aristotelean prescriptive aesthetic I have out-lined does not possess the pivotal feature of aesthetic attitude theories,namely their insistence that there is a unique mental state which isuniquely suitable for permitting the aesthetic properties of art toemerge. The mental state which the Aristotelean prescriptive aestheticasks the reader to bring to the literary artwork is his capacity to engagein moral reasoning and moral evaluation and his knowledge of therules of probability and inevitability. There is nothing peculiarly "aes-thetic" about this capacity; in fact, I have been at some pains to indicatethat it is like real-life moral reasoning. But the Aristotelean prescriptive aesthetic is a distant ancestor oftwentieth-century aesthetic attitude theories, for it shares their insist-ence that the nature of art is emergent contingent upon some activeaudience participation, and that this active participation should beengaged in with respect to works of literature as the correct mode ofappreciation. Artistic kinds, persons in action, and plots seem not to be the actual
520 NOUS semantical properties of texts. They are not implied by the meanings of the sentences in the text. The descriptions of Achilles running into battle do not imply that he was courageous. That he was courageous is a moral evaluation on the part of the reader. In fact, the real semantical properties of literature may not be any different from the real semanti- cal properties of histories. This, I take it, is the final lesson to be drawn from Aristotles almost causal claim that the tragedians wrote about real persons. Rather, artistic kinds, persons in action, and plots are emergent semantical properties. An emergent semantical property is a property which is not part of the ordinary sense or reference of a piece of discourse-a name, a term, or a sentence-but which can be con- structed in a rule-governed fashion and ascribed to that piece of dis- course. These rules are given by the Aristotelean prescriptive aesthetic. Although we can speak loosely about a person-name as denoting a certain artistic kind or person in action, we would do better (at least when philosophizing) to qualify ourselves and to speak of a person- name as A-denoting (for Aristotle) such and such an artistic kind or person in action, and of the work as A-denoting such and such plot. A person-name, N, A-denotes a artistic kind, K, just in case some reader of the text in which N is embedded correctly follows the Aristotelean prescriptive aesthetic and constructs K by applying his moral reasoning to the fictitious incidents constructed around N. A-denotation for persons in action and plots in similarly defined, with the addition of knowledge about the rules of probability and inevitability and unity. We have only solved half the problem, however, for in proposing the Aristotelean prescriptive aesthetic as a method for ascribing per- sons in action and plots to texts I havejettisoned interpretation (B). Weare left without an interpretation of Aristotles definition, and a serious problem. Not every text can turn out to be a literary artwork, for thework of Herodotus, to use Aristotles illustration, is still history. Thereis nothing in the Aristotelean prescriptive aesthetic itself that rulesagainst its being trained on the work of Herodotus and transformingthat into a work of literature. It has been argued (see, for example,Coleman ) that a text is literature or not depending on how it is read.This is not Aristotles position, for he wants to make some sharpdistinction, or as sharp a distinction as can be made, between texts whichare works of literature and texts which are not. It initially appeared thatthis distinction was to be made by examining the semantical propertiesof texts, but this way is no longer open to us. We want to know what it isabout some texts which makes them particularly susceptible to beingread according to the Aristotelean prescriptive aesthetic. I shall refer tothis susceptibility as the "literary potential" of a text, and shall advancethis final interpretation of Aristotles definition:
ARISTOTLES DEFINITION OF POETRY 521 (D) A text is a literary artwork just in case that text has a relatively high degree of literary potential, and a text is a work of non-literature just in case it has a relatively low degree of literary potential. When does a text have a relatively high literary potential? I shall discuss three marks of high literary potential. The first is obvious. For a text to be read as being about persons in action, it must contain person-names. This mark rules out works of economics, physics, math- ematics, most philosophy, geography, chemistry, and the like from being literature. While this is intuitively correct and in line with what Aristotle wants, we are still faced with the problem of separating Warand Peace from A History of the Modern World, given that each contains the same person-name, "Napoleon." On "characterization" Aristotle writes, "Words and action express character ... if they bringout a moral choice, and the character is good if the choice is right" (1454a 17-19). The point here is that words in literature must bring out a moral choice, and I wish to contrast bringing out with stating. We use samples and examples to bring out; we use assertion to state. I can describe to you what an opera is if you have never seen one. Or, I can take you to a performance of an opera.Samples are physical objects which are made to stand for a class ofthings, and a sample is a sample of those things which it both stands forand shares the relevant properties of. Exarnplesaregiven in language.They are descriptions of possible or actual cases. It is sometimes simplymore expedient to describe a case rather than to produce an actualphysical instance of one. Moral philosophy is rife with examples, fromPlatos story of the ring of Gyges to Judith Thomsons case of thekidnapped man who wakes up connected with the famous violinist ashis life support system. Like physical samples, examples are made tostand for a class of things and to share their relevant properties. Platoscase stands for a class of situations where being immoral apparentlymakes one happy, and Thomsons for a class of situations in which onefinds oneself causally though involuntarily responsible for the life ofanother. I can state moral principles ("Never treat a person as a meremeans"), use these principles in moral arguments (". . . and thereforeno one can force you to oblige the violinist"), or express opinions onmoral matters ("I think you are not obligated to remain connected withthe violinist"). Such pieces of discourse are statings, not examples. Samples and examples are exercises in abstraction. To see thepoint of the sample or example-that is, to see what it is a sample orexample of-one must form a hypothesis about the members of theclass which the sample is supposed to stand for. This hypothesis mustbe tested against the sample or example for fit, and some of the specific
522 NOOS features of the sample or example might be ignored in order to fit one hypothesis, or brought in in order to fit another. A red square might be a sample of red things or of squares or of red squares, depending on what specific features of the sample are ignored or not. Understanding a sample or example is an audience participation activity. The hearer or reader of an assertion must exercise his command of the language and bring to bear his previous beliefs in order to understand anassertion, but the process of construction hypothesis about what sam-ples or examples stand for demands a wider range of audience activity.Seeing the sample or understanding the sentences of an example is buta prelude to the process of bringing out the class of things which thesample or example stands for. Science or philosophy will use both samples and examples, but itwill typically tell us what these stand for. It will do our hypothesisconstructing for us. The physicist does not only do an experiment, heinterprets it, tells us what it shows about a class of things, and whichthings it shows this about. The anthropologist, sociologist, or psychol-ogist gathers data and interprets them. He doesnt just describe thebehavior of this or that group, of this or that kind of man; he tells uswhat this shows about this group or that kind. The poet, on the contrary, does not interpret. He shows us a "kindof man" but what kind or what he shows us is left up to the ingenuity ofthe reader to discover. The poet is like the physicist who performs anexperiment in full view of the class, and then leaves it to his students tosay what the experiment shows about what; or like Kierkegaard whothought philosophy was best done by letting the reader figure out forhimself the point of the parable. Shakespeare does not interpretOthello for us, as Freud would have if Othello had come to his couch.Shakesepare gives a marvously rich example of a kind of man, butleaves it to the reader to say what this kind of man is and what thisexample shows about this kind. From these remarks we can see that the second mark of highliterary potential is a lack of interpretation in a text: if, that is, that textinvolves its characters in a series of events but does not tell us the moralsignificance of those events for that character. A text has high literarypotential only if it makes us look for the class which it is giving anexample of, only if it "brings out" a moral choice. This second mark of literary potential suggests a third. For anexample to be worth interpreting, it must be of sufficiently rich com-plexity. An example, like Platos story of the ring of Gyges, whichvirtually wears its meaning on its sleeve, would seem to have lowerliterary potential than Shakespeares Othello, which is endlessly inter-pretable. If literature is to have an interesting moral function, it mustbe prepared to introduce us to kinds of persons which we have perhaps
ARISTOTLES DEFINITION OF POETRY 523some real-life acquaintance with but which we have not yet consideredtheir moral nature clearly. Examples give us access to situations whichwe would not ordinarily come across, if only because a literary examplegives us so much material that would be inaccessible to us in ourordinary day-to-day dealings with persons. Molly Blooms final sol-iloquoy from Joyces Ulysseslets us hear what a woman of a certain sortthinks, feels, and believes about love and sex and men, thoughts whichare ordinarily (and in the novel as well) denied even to husbands. Aristotle suggests that literature has a pedagogic function in hisremarks on the origins of poetry in Ch. IV: ... [TIhe sight ofcertainthings gives us pain, but we enjoy looking at the most exact images of them, whether the forms of animals which we greatly despise or of corpses. The reason is that learning things is most enjoyable,not only for philosophersbut for others equally,though they have but little experience of it. Hence they enjoy the sight of images because they learn as they look; they reason what each image is, that there, for example, is that man whom we know. If a man does not know the original, the imitationas such gives him no pleasure; his pleasure is then derived from its worksmanship,its color, or some similiarreason. (1448b 10_19) There seems to be two distinct explanations of why we take pleasure in imitations of ugly things. The first is that "we reason what each image is"and then find delight in learning something about an actual particular instance of that type. The second is in delight from its"workmanship"-what we might today refer to as its purely aestheticproperties. Someone who took delight in workmanship would not haveread the text according to the Aristotelean prescriptive aesthetic, andto that extent would not take delight in the text as a work of literature.The relevant literary delight is from "learning as one looks," and thisdelight gets explained by deeming it to be a species of the delight whichone takes in learning anything. What does one learn, and how does one learn it? Aristotle, I think,conflates two sorts of answers to the first part of the question by givingone sort of answer to the last. What one can learn, to use the terminol-ogy developed in this essay, is either that such and such person in actionis possible (e.g., that such heroism as Odysseuss or such intimate self-acceptance as Molly Blooms could be); or that such and such person inaction has an actual instance: "that man whom we know." By "theoriginal" Aristotle might have meant an actual instance of a person inaction, and he might be suggesting that one learns that such and such aperson in action is possible by learning that it has an actual instance-that this actual persons heroism is like Odysseus or that actual personsself-acceptance is like Molly Blooms. This may be a happy conflation, for it suggests that literature is not
524 NOCS the investigation of extreme possibilities of human behavior, but ratherthat what we learn from reading literature in the correct Aristoteleanway is something about real-life poeple. "If a man does not know theoriginal," the imitation as such gives only pleasure "derived from itsworkmanship," not the pleasure which comes from a learning experi-ence of the possibilities of human moral character, possibilities whichare not remote but which are instanced in "that man whom we know." Here Aristotle clearly parts company with Plato, who thought that artwould offer too many of the wrong moral possibilities and possiblysway citizens of the Republic away from the straight and narrow.Aristotle is more the empirical observer of behavior, more willing toseparate the learning experience associated with the moral evaluationof character from the imperative to act on that moral evaluation. The learning potential of a text ought to be a mark of its literarypotential, and so I shall say that the third mark of literary potential iswhether a text offers a person in action or plot of sufficiently richcomplexity so that a person who trained the Aristotelean prescriptiveaesthetic on it would be rewarded by learning something about humanmoral character which he did not know before. I shall say that the three marks of literary potential are individually necessary for a text to have a relatively high degree of literary potential. I do not think them jointly sufficient, for they do not make reference to the sort of unity that Aristotle requires of literature, and on that issue Ishall leave us as much in the dark as Aristotle himself does. The third mark of literary potential may be thought to be toostrong, or even un-Aristotelean, since there is a strong descriptivisttendency in Aristotle to make the classification of texts conform to ourordinary ideas of what a work of literary art is. It might be thought,therefore, that even if the ordinary best-seller does not teach anythingnew about the possibilities for moral behavior-pick any Mickey Spil-lane or Jacqueline Susann at random-still, we should countenance itas literary art, although of a rather impoverished sort. While I agreethat Aristotle is relying strongly on our intuitions about what shouldand should not count as literature, I would urge that our ordinarynotion of the literary artwork contains an evaluative element, and thatthe learning experience which the third mark describes is part of ourordinary concept of literature. As a compromise, I suggest that thethree marks are individually necessary for a relatively high degree ofliterary potential, but that only the first two are individually necessaryfor some degree of literary potential. This completes our investigation into Aristotles definition of po-etry. The goal of this essay was to set out an interpretation of thechapter IX definition which avoided the apparent contradiction ad-duced at the beginning. I feel I have produced an interpretation which
ARISTOTLES DEFINITION OF POETRY 525avoids this contradiction and has the minimal virtue of being notinconsistent with what Aristotle elsewhere says in the Poetics. Whether ithas the higher virtue of eking out what Aristotle "truly meant" no one, Ithink, can say. Whether it has the highest virtue of a theory, truth in itsown right, the correct definition of the literary artwork, I leave as anexercise to the reader.5 REFERENCES  Virgil Aldrich, Philosophyof Art (Englewood Cliffs, NJ: Prentice-Hall, 1963).  Monroe C. Beardsley, Aesthetics:Problemsin the Philosophyof Criticism(New York: Harcourt, Brace & World, Inc., 1958).  Edward Bullough, "Psychical Distance as a Factor in Art and an Aesthetic Princi- ple," BritishJournal of Psychology, 5(1912): 87-98.  S. H. Butcher, AristotlesTheoryof Poetry and Fine Art (New York: Dover Publica- tions, 1951).  Francis X. J. Coleman, "A Few Observations on Fictional Discourse," Language and Aesthetics:Contributionsto the Philosophyof Art, ed. Benjamin R. Tilghman (Lawr- ence, Kansas: The University Press of Kansas, 1973).  George Dickie, Art and the Aesthetic(Ithaca, NY: Cornell University Press, 1974).  Gerald F. Else,AristotlesPoetics: The Argument(Cambridge, MA: Harvard Univer- sity Press, 1957).  Leon Golden and 0. B. Harrison, Jr., AristotlesPoetics (Edgewood Cliffs, NJ: Prentice-Hall, Inc., 1968).  Nelson Goodman, Language of Art (New York: Bobbs-Merrill Co., 1968). G. M. A. Grube,Aristotle on Poetryand Style (New York: Bobbs-Merrill Co., 1968). Emil A. Gutheil, Music and Your Emotions: A Practical Guide to Music Selections Associatedwith DesiredEmotional Responses(London: Liveright, 1952). David Lewis, "Truth in Fiction,"AmericanPhilosophicalQuarterly, 15(1978): 37-46. D. W. Lucas, Aristotle:Poetics (Oxford: Clarendon Press, 1968). Margaret Macdonald, "The Language of Fiction," Proceedings of the Aristotelean Society, Supplementary Volume XXVII (1954): 165-84. C. K. Ogden and I. A. Richards, The Meaning of Meaning 8th ed. (New York: Routledge & Kegan Paul, 1946). Alvin Plantinga, "The Free Will Defense," Philosophyin America, ed. Max Black (Ithaca, NY: Cornell University Press, 1965). George Santayana, The Sense of Beauty (New York: Charles Scribners Sons, 1936). Jerome Stolnitz, Aesthetics and the Philosophyof Art Criticism(New York: Houghton Mifflin Co., 1960). J. A. K. Thomson, The Ethics of Aristotle: The Nicomachean Ethics (New York: Penguin Books, 1953). William K. Wimsatt,Jr. and Monroe Beardsley, "The Intentional Fallacy,"Philoso- phyLooksat theArts, ed. Joseph Margolis (New York: Charles Scribners Sons, 1962). NOTES 1All translations of the Poetics are by G. M. A. Grube [ 10]. Reference for exact quotesis provided only when the quote is mentioned for the first time. Reference to thoughts orparaphrases of Aristotle is generally given by citing the chapter in Roman numerals. 2Such a view has been held by David Lewis [ 12]. 3This variation is given by Alvin Plantinga [ 16], who puts it to a use other than theanalysis of fictional entities. My formulation of the definitions of "artistic kind," "personin action," and "plot" owes much to Plantingas article. 4All translations of the NicomacheanEthics are by J. A. K. Thomson [ 19]. 5I wish to thank the editor and referee of NOUS for taking what I consider extraor-dinary pains to improve earlier versions of this essay.