I. Introduction: everyone knows what a swatch is for
In Ways of Worldmaking, Nelson Goodman told the story of Mrs Mary Tricias, who chose an upholstery fabric from a sample booklet and ordered enough of it for her overstuffed chair and sofa, insisting that it be exactly like the sample. When the bundle arrived, several hundred small pieces with zigzag edges fluttered to the floor (Goodman 1978, 63–64). The story is funny because everyone knows what a swatch is for. It exemplifies its colour, weave and texture, and it does not exemplify its size, its shape or its pinked edges. Nobody needs to be told this, and nobody could list in advance all the conditions that make it obvious. Goodman's answer to the question of how the relevant properties are selected was that the selection rests on established practice: a sample is a sample of some of its features and not others because of the ways in which samples are habitually used.
Appeals of this kind run through the whole of twentieth-century ordinary language philosophy. Ludwig Wittgenstein grounded meaning in use, in language-games and finally in forms of life. Goodman grounded inductive validity in the entrenchment of predicates through their past projection. J. L. Austin grounded philosophical method in the distinctions that ordinary words had accumulated over generations of use. G. E. Moore grounded certainty in common sense, and Gilbert Ryle exposed category mistakes by appealing to how words are ordinarily deployed, as when a visitor shown the colleges, libraries and laboratories of a university asks where the university is (Ryle 1949). In each case, practice does the decisive work, and in each case the practice in question is treated as good: good enough to settle what words mean, which predictions are valid, which distinctions are worth drawing, and which philosophical puzzles are confusions.
This article asks what exactly that practice is. Its thesis is that the appeal to good practice in ordinary language philosophy draws on two different things without systematically distinguishing them. One is the established practice of symbolic classification, the stabilized ways in which words are applied, projected and agreed upon. The other is the living coordination of beings across five irreducible mediations, multisensorial embodiment, multispecies being-with, multiversal dwelling, multimateriality and multisymbolization, within which differences become consequential at all. The two are continuously entangled and never interchangeable. Because the methods of ordinary language philosophy were designed to investigate the first, they cannot by themselves establish what matters in the second. The argument can be compressed into a single formula: common sense is not the same as common life.
Three further distinctions give the formula analytical precision. A classification has symbolic validity when it works according to established linguistic and conceptual practice. It has mesocosmic adequacy when it helps to disclose the consequential processes of the living coordination it classifies. And it acquires recursive efficacy when it becomes consequential by reorganizing the coordination it describes. These distinctions organize the readings that follow. Wittgenstein is the great investigator of symbolic validity, of what it is to use a word competently within a form of life. Goodman asks how particular symbolic projections become established and how relevant properties are selected. Austin examines the conditions under which symbolic utterances are appropriately performed. Each illuminates a different achievement of symbolic practice, and the question Living Value Theory adds is how those achievements relate to mesocosmic adequacy and recursive efficacy, which none of their methods was built to establish. The point is less that three very different philosophers made the same mistake than that the strengths of their methods share a common boundary.
The argument begins further back than ordinary language philosophy usually does, with what living beings do before there is anything to be true or false, and with the claim that publicly detachable propositional truth is a distinctively human achievement. It then reconstructs the differentiation of three kinds of symbolic fit, distinguishes common sense from common life, and reads Wittgenstein, Goodman and Austin in turn. It closes with the radical reductiveness of symbols, the difference made by what kind of counterpart a classification classifies, and a reformulation of what good practice could mean.
II. Before truth: the octopus does not lie
An octopus settles against a reef and within a fraction of a second its skin takes on the mottled colours and raised textures of the coral around it. A predator swims past without stopping. The Indo-Malayan mimic octopus goes further, flattening itself into the shape of a sole, trailing its arms like the venomous spines of a lionfish, or burying all but two arms in the sand and waving them in alternating bands like a sea snake (Norman, Finn and Tregenza 2001). Primatologists have documented a large repertoire of what they call tactical deception, in which apes and monkeys conceal food, feign indifference, or lead a rival away from something they want (Whiten and Byrne 1988).
Is the octopus lying? The answer is no, and the reason is precise. To lie is to assert a proposition one takes to be false, and the octopus asserts nothing. There is no hidden sentence in its skin, no crypto-proposition reading "I am a coral formation" or "I am a lionfish" that the predator fails to decode. Camouflage and mimicry are forms of mesocosmic coordination: the octopus's body coordinates with light, texture, the visual systems of predators and the configuration of the reef, and the coordination succeeds or fails. When the predator is misled, what occurs is a mesocosmic miscoordination, a successful exploitation of the way another living being discerns its surroundings. Most cephalopods are thought to have only a single visual pigment, and yet they match the colours of their backgrounds with remarkable accuracy (Hanlon 2007). Whatever the octopus is doing, it is not describing anything to itself, let alone to the predator.
The same holds at L1 and L2 throughout the living world. At L1 coordination proceeds without explicit attention; at L2 something is felt as misaligned before it can be named. Neither level contains sentences waiting to be spoken. Treating such coordination as tacitly propositional projects the structure of human symbolic articulation into processes that operate without it.
Animals certainly respond to miscoordination, and some respond to unreliable signals in ways that look like sanction. Client reef fish chase away or abandon cleaner wrasse that bite off mucus instead of removing parasites, and the cleaners respond by cooperating more (Bshary and Grutter 2005). Paper wasps whose facial markings were experimentally altered to signal a higher status than their behaviour supported received more aggression from nestmates (Tibbetts and Dale 2004), and some primate studies have been read in similar terms, though their interpretation is contested. These findings matter, and they show that responding to dishonest signalling is not a human monopoly. What they do not show is anyone being held to account for a proposition. The clients respond to the bite and the wasps to a mismatch between marking and conduct within the encounter itself. The distinctive human achievement lies elsewhere: in holding an individual accountable for an explicitly articulated proposition whose correctness can be examined independently of the interaction in which it was made, by people who were not there, at a later time, against evidence that the speaker never saw. The prey animal that escapes an ambush does not reproach the octopus for having lied about being coral. Reproach for lying presupposes that something was said, and that it can be held against a world that was otherwise.
Humans, by contrast, have built some of their most elaborate institutions around the answerability of what is said. Children are taught early that lying is wrong, and are punished for it in ways that are distinct from punishment for hitting or taking. One of the Ten Commandments forbids bearing false witness against one's neighbour, which is to say, testifying falsely about someone in a dispute. Courts administer oaths and punish perjury, scientific communities retract papers and end careers over fabricated data, and reputations rise and fall on whether a person's word can be relied upon. None of these institutions has any counterpart in the punishment of cleaner wrasse, because what they sanction is a symbolic act measured against a world, and the cleaner never performs one.
The claim of this article is therefore a theoretical one about a particular kind of truth: publicly detachable, retrospectively assessable propositional truth enters the world with humans, through symbolically articulated propositions at L3 to L5 that can be detached from the immediate pressures of the coordination in which they arise. A proposition can be uttered, recorded, repeated in another place, held against later evidence, and found wanting. Only then does the question whether it is true become askable in this sense. The claim does not depend on proving the absence of every sophisticated symbolic capacity throughout the animal kingdom, and it would be strengthened, not refuted, by evidence of rudimentary forms elsewhere. Wittgenstein noticed the edge of the matter when he asked whether an animal can hope. "One can imagine an animal angry, frightened, unhappy, happy, startled. But hopeful? And why not?" His answer was that only those can hope who have mastered the use of a language (Wittgenstein 1953, II.i). Hope requires a future that can be articulated, and truth in this sense requires a world about which claims can be made and later checked. Both belong to beings who live partly in symbols. Far from diminishing the coordination of other animals, which is extraordinarily subtle, this locates propositional truth as one achievement of one mediation, which is the first thing a philosophy of ordinary language should have to explain and the last thing it usually considers.
III. Three kinds of fit and their differentiation
Once symbolic articulation is in play, its claims can be answerable to different things. Living Value Theory distinguishes three relations of symbolic fit. Mesocosmic fit concerns how adequately a symbolic articulation discloses and helps to coordinate living coordination across the five mediations. Intersymbolic fit concerns relations among symbolic articulations themselves, the coherence of a proof, a legal code or a scientific theory. Cosmic fit concerns the relation between symbolic articulation and realities that do not depend on how any mesocosm organizes them, the length of a sofa as much as the orbit of a planet.
The three kinds of fit are best understood as increasingly differentiated symbolic capacities, and what follows is a reconstruction of that differentiation, not an established empirical chronology. Its starting point is a particular kind of detachment. Vervet monkeys give acoustically distinct alarm calls for leopards, eagles and snakes, and listeners respond appropriately, running into trees, looking up or scanning the ground (Seyfarth, Cheney and Marler 1980). Later research has shown that such calls are more context-dependent than a simple one-call-one-referent reading suggests, and that listeners integrate them with much else they know about callers and circumstances (Seyfarth and Cheney 2003). The calls remain absorbed in the coordination of the moment: a call coordinates escape now, and nothing suggests that vervets discuss yesterday's leopard, compare it with last year's, or argue about whether the caller was right. Human language differs less in specificity than in what can be done with an utterance once the pressure of the moment has passed. It can be repeated to someone who was not there, retold the next day, doubted, corrected and remembered as wrong. Propositional truth became possible when symbolic articulation could be used without immediate mesocosmic coordination pressure, when people could talk about absent things at leisure and hold what was said against what turned out to be the case.
Mesocosmic fit is the first answerability of such articulation. The earliest human symbolization, whatever form it took, served coordination within the mesocosm: where the water is, who is coming, what the weather is doing, whom to trust. Its success was measured by whether it helped living beings coordinate. Here a distinction is needed. A claim that there is water over the hill was already answerable to cosmic reality, since the water was either there or not, whatever anyone thought. The possibility of cosmic answerability therefore accompanies even elementary symbolic articulation. What came much later was the development of specialized means of establishing cosmic fit independently of immediate mesocosmic experience, for things no one could walk over the hill to check.
Intersymbolic fit requires more than one symbolic system, or at least symbols that can be held against other symbols. Here the comparative evidence is suggestive. Chimpanzees have learned to label sets of objects with Arabic numerals (Matsuzawa 1985), and bonobos have learned to communicate through keyboards of lexigrams (Savage-Rumbaugh and Lewin 1994). These are remarkable achievements, and they remain achievements within a single system mapped onto perceptible situations. There is no evidence known to this author of an ape squaring one symbolic system with another, checking a count of numerals against a sequence of lexigrams, or noticing that two symbolic records of the same event disagree. Intersymbolic fit, the possibility that a symbol is right or wrong because of its relation to other symbols, appears to be a human achievement.
Those specialized means depend on the other two kinds of fit. Consider Pluto. Nobody has ever seen Pluto in the way one sees a tree. Clyde Tombaugh found it in 1930 by comparing two photographic plates taken six nights apart and noticing a point that had moved. Its orbit was computed from positions recorded over years. Its mass could only be estimated reliably after the discovery of its moon Charon in 1978, from the moon's orbit, again on photographic plates. The images of its icy plains that circulated in 2015 were radio signals from the New Horizons spacecraft, transmitted across billions of kilometres over more than a year and assembled into pictures by software. Pluto as an object of knowledge is a composite of traces, images, measurements and calculations, every one of them mediated through symbols and squared with one another through intersymbolic coherence. Establishing cosmic fit beyond the reach of the mesocosm became possible only when multisymbolization grew massive enough to support instruments, writing, mathematics and institutions that allow symbols to be checked against other symbols across generations. In this sense the systematic pursuit of cosmic fit is the latest of the three developments, and the most demanding, because it rests on the largest architecture of deposits.
This reconstruction matters for what follows. Ordinary language philosophy studied the medium in which all three kinds of fit are negotiated, and its methods, which attend to how that medium works in ordinary use, are best placed to illuminate symbolic validity and least placed to settle the other kinds of answerability.
IV. Common sense is not the same as common life
In "A Defence of Common Sense", G. E. Moore listed propositions he claimed to know with certainty, beginning with the truism that "there exists at present a living human body, which is my body", and continuing through the facts that it was born, has been in contact with the earth's surface, and has had experiences of many kinds (Moore 1925). Later he held up his hands and offered as proof of an external world the claim that here is one hand, and here is another (Moore 1939). Moore's common sense consists of propositions. That is its strength and its limitation. It is a stock of symbolic stabilizations so secure that doubting them seems absurd.
Wittgenstein's last notes, published as On Certainty, took Moore's propositions as their starting point and moved beyond them. Some propositions, he suggested, function like hinges on which our questions and doubts turn, exempt from doubt because doubt depends on them (Wittgenstein 1969, §341). And at the bottom of the language-game, he wrote, lie neither self-evident propositions nor a kind of seeing: "it is our acting, which lies at the bottom of the language-game" (§204). This is the closest that ordinary language philosophy came to Living Value Theory. Wittgenstein saw that beneath common sense lies something that is no proposition at all. He did not ask what that acting consists of.
Living Value Theory distinguishes the two explicitly. Common sense is the stock of familiar symbolic stabilizations through which activities and things are recognized, named and spoken about: that this is a hand, that football is a game, that buying bread is an exchange. Common life is the ongoing living coordination across five mediations within which hands grip and tire, matches are played in rain and wind, and households keep themselves going. Common sense is part of common life, since multisymbolization is one of its mediations, and it is a small and highly selective part. The two are continuously entangled. They are not interchangeable.
The three distinctions introduced at the outset can now be stated more fully. A classification has symbolic validity when it works according to established linguistic and conceptual practice: it is correctly applied, generally accepted, and causes no confusion. It has mesocosmic adequacy when it helps to disclose the consequential processes of the living coordination it classifies. And it acquires recursive efficacy when it becomes consequential by reorganizing the coordination it describes. The three are distinct and closely related. A classification can have complete symbolic validity and little mesocosmic adequacy. "Students are consumers" is perfectly intelligible English, and it captures very little of what happens when someone learns to think. It can also acquire recursive efficacy, as when universities reorganize teaching, assessment and funding around satisfaction scores, so that the classification becomes increasingly consequential without ever having been shown to capture what is valuable about education.
Gilbert Ryle came close to the distinction from another direction. His contrast between knowing how and knowing that showed that intelligent practice cannot be reduced to the possession of propositions: someone who can ride a bicycle, tell a joke or play chess well knows something that no list of facts captures (Ryle 1949). Knowing how is the nearest ordinary language philosophy came to naming coordination at L1. Yet Ryle analysed knowing how in terms of dispositions describable in ordinary language, and his category itself remained undifferentiated. Riding a bicycle is a coordination of balance, gravity, momentum and a machine; telling a joke is an interrecursive coordination with listeners whose laughter or silence alters the telling as it proceeds; playing chess well depends on a memory of symbolic patterns. All three are knowing how, and the operations through which each is achieved have little in common. The category that rescued practice from propositions grouped very different coordinations under one heading.
The danger in appeals to ordinary practice is that symbolic validity is the only one of the three that ordinary use can directly demonstrate. A word that is used fluently, understood immediately and never questioned has passed the test of ordinary language. Whether it discloses what matters in the coordination it names is a separate question, and the fluency of its use gives no answer to it.
V. Wittgenstein: forms of life and the undifferentiated practice
Wittgenstein's later philosophy is the most sophisticated attempt to ground language in life, and the most searching investigation of symbolic validity, of what it is to use a word competently. "The speaking of language is part of an activity, or of a form of life," he wrote (1953, §23). Language-games are embedded in human activities, and the activities are themselves part of nature: "Commanding, questioning, recounting, chatting, are as much a part of our natural history as walking, eating, drinking, playing" (§25). When asked whether human agreement decides what is true and false, he answered that "it is what human beings say that is true and false; and they agree in the language they use. That is not agreement in opinions but in form of life" (§241). And when justification runs out, "I have reached bedrock, and my spade is turned. Then I am inclined to say: 'This is simply what I do'" (§217). What has to be accepted, the given, is, as he put it, forms of life (1953, II.xi).
Several of these remarks anticipate the argument of this article. Section 241 locates truth precisely where Living Value Theory locates it: in what human beings say. Section 25 places speaking in natural history alongside walking and eating, which is a decisive break with philosophies that treat language as a disembodied system of signs. And the bedrock of section 217 is something done, not something stated.
Far from claiming that commanding, walking and eating are constituted in the same way, the list in section 25 locates language within the diversity of human activity, and Wittgenstein would have had no difficulty acknowledging that walking and commanding are very different things. The question this article puts to him concerns what his method can establish. Walking is coordinated through embodiment and dwelling, through gravity, terrain and the body's balance, in ways that no rule of a language-game specifies. Eating is a metabolic coordination with materials and other living beings. Commanding is an interrecursive coordination in which one person's words are taken up by another who may obey, refuse or misunderstand. Chatting may do almost nothing but sustain being-with. Wittgenstein's method investigates the role of words within such activities. It does not, and was not designed to, differentiate the mediational operations through which the activities themselves are accomplished, and the concept of a form of life leaves those differences unthematized.
The games passage shows why this matters. Wittgenstein asks what is common to the activities we call games and finds family resemblances where an essence was expected (§§66–67). His famous instruction was "don't think, but look!" (§66). The instruction is excellent, and it invites the question of what to look at. Wittgenstein looked at the uses of the word, at what we call a game and how we explain the word to someone who does not know it, and he could reasonably reply that he never intended the classification to provide a theory of the activities grouped under it. The practice of calling football and chess games is a practice of multisymbolization. The practices of playing football and chess are coordinations of bodies, opponents, pitches, boards, clocks, weather and anticipation, and they differ profoundly. The practice of classifying two activities together is no evidence that the activities themselves are organized through the same mediational practices. The question is therefore whether a method that establishes how a word is competently used can also establish which differences are consequential within the coordinations the word classifies. Living Value Theory gives grounds for answering no. A shared word can be applied with complete success to coordinations whose operations have little in common, and the success of its application settles nothing about which of their similarities matter.
The qualification that preserves the ontology of Living Value Theory is essential here. The five mediations are not five independent realms, each containing self-sufficient practices of its own. Every living coordination is continuously co-mediated: a football match is embodied, shared with others, played on a pitch in weather, conducted with a ball and boots, and organized by rules and words, all at once. But the operations of the mediations remain irreducible to one another. Every practice is co-mediated, but not every mediational operation performed within a practice is equivalent. Wittgenstein's forms of life connect meaning to how people live. Without a differentiation of how people live, the concept cannot by itself tell the success of a word apart from the organization of the life it names.
Wittgenstein's remark that "if a lion could talk, we could not understand him" (1953, II.xi) is a case in point, though its interpretation remains disputed, and recent readings caution against treating it simply as a claim that animal lives are too different from ours to be understood. Living Value Theory offers an alternative analysis rather than a refutation. Much of the lion's life can be understood by a human without any talk at all, through shared embodiment, shared dwelling and the shared coordinations of hunger, fatigue and threat. What would separate a talking lion from us would lie in the symbolic mediation through which its talk would be organized, while the zoetic life the two share would remain a ground of understanding that no difference in language removes.
VI. Goodman: entrenchment, the swatch and the question of which practice
Goodman's appeal to practice took a different form, and it addresses a different achievement: how particular symbolic projections become established. In Fact, Fiction, and Forecast he posed the new riddle of induction. Define "grue" as applying to all things examined before a certain time just in case they are green, and to other things just in case they are blue. Nothing has to change colour for the definition to work. Every emerald examined so far is green, and every one of them is therefore also grue, so the same observations support two incompatible predictions about emeralds first examined after that time: that they will be green, and that they will be blue. Why do we project "green" and not "grue"? Goodman's answer was entrenchment. "Green" has a long history of actual projection in our inductive practice, and "grue" has none. Predicates earn their projectibility through the frequency with which they have in fact been projected, so that, in Goodman's phrase, the roots of inductive validity are to be found in our use of language (Goodman 1955). His later critique of similarity extended the point. Since any two things share indefinitely many properties, similarity cannot itself justify a classification, and which similarities count depends on established practices of selection (Goodman 1972).
The question to put to Goodman is which practice he means. Entrenchment is a history of symbolic projection, of predicates used and reused in predictions. It belongs to multisymbolization, and its record is an intersymbolic deposit. Goodman's entrenchment explains which predicates have acquired authority through prior symbolic projection. Living Value Theory asks a different and logically prior question: what living coordinations made certain distinctions consequential enough to be repeatedly articulated and projected in the first place? Living beings who could see colour coordinated with ripening fruit, changing leaves and the difference between healthy and sick plants long before there were predicates to project, and the predicate "green" stabilized distinctions that had already proved consequential in mesocosmic coordination. The history of symbolic entrenchment, far from self-explanatory, is a selective deposit of earlier mediational histories.
This does not amount to a solution of the riddle. Goodman constructed grue so that the same observations would support both predictions, and no history of mesocosmic relevance turns that logical situation into a deductive warrant for preferring green. What Living Value Theory adds is an account of what entrenchment records without naming. The predicates that became entrenched are those that tracked differences on which living coordination depended, and the riddle's power lies partly in showing that symbolic practice alone, considered as a history of projections, cannot say why those differences and not others came to be projected at all.
The swatch makes the same point more concretely. Goodman located the selection of exemplified properties in established practice. But the practice that makes the swatch intelligible involves far more than the symbolic convention that samples exemplify colour and texture. The customer and the upholsterer share embodied familiarity with fabrics, expectations about how sofas are sat on and how cloth wears, material knowledge of looms and dyes, the dwelling in which the sofa will stand, and an anticipated future object in a particular room. The symbolic convention rests on this distributed mesocosmic coordination, and the coordination does most of the work. The woman in Goodman's story did not misunderstand a symbolic rule so much as fail to share, or pretend not to share, the coordination in which the rule makes sense. That is why the story is a joke and not a puzzle.
The ambiguity runs through Goodman's later work. In Ways of Worldmaking he replaced truth with a broader notion of rightness, and treated rightness largely as a matter of fit with other versions and with established practices of categorization (Goodman 1978). Living Value Theory can accept much of this as an account of intersymbolic fit and of the role of practice in classification. What it cannot accept is the assimilation of all fit to fit with practice, because that leaves no room for cosmic fit, the answerability of some symbolic claims to realities that no practice makes, and no account of mesocosmic adequacy, the answerability of classifications to the living coordinations they select from. When the practice to which rightness answers is left undifferentiated, symbolic practice can silently stand in for both.
VII. Austin: the first word and the distinctions worth drawing
Austin offered the most explicit defence of ordinary language as a philosophical resource, and his work illuminates a third achievement of symbolic practice: the conditions under which utterances are appropriately performed. "Our common stock of words embodies all the distinctions men have found worth drawing, and the connexions they have found worth marking, in the lifetimes of many generations," he wrote in "A Plea for Excuses". These distinctions are likely to be "more numerous, more sound, since they have stood up to the long test of the survival of the fittest, and more subtle" than any that a philosopher could devise in an armchair. Austin was careful to add that "ordinary language is not the last word: in principle it can everywhere be supplemented and improved upon and superseded. Only remember, it is the first word" (Austin 1956). He called the resulting method linguistic phenomenology: by attending to what we should say when, we sharpen our perception of the phenomena themselves.
In attending to words, Austin insisted, "we are using a sharpened awareness of words to sharpen our perception of, though not as the final arbiter of, the phenomena" (Austin 1956). The qualification is admirable, and it raises the question of which phenomena words can sharpen our perception of. Words give access to what has been articulated at L3 and stabilized at L4. They give much less access to coordination at L1, which proceeds without articulation, and to the felt misalignments of L2 that never found a word. A linguistic phenomenology can be extraordinarily illuminating about the distinctions a language has drawn, and it is structurally limited in what it can reveal about the coordinations from which those distinctions were drawn.
Austin's claim is at bottom a claim about relevance. Words embody the distinctions people have found worth drawing, and finding something worth drawing is a matter of what has proved consequential in living. Living Value Theory can agree with much of this. Ordinary words are L4 deposits, stabilizations of distinctions that were once felt as snags at L2 and articulated at L3, and they are excellent evidence that a distinction mattered somewhere, to someone, at some time. Austin's own subject, excuses, is an ideal case, since the vocabulary of doing things by accident, by mistake, inadvertently or unintentionally grew up around the occasions on which people had to answer for what they had done, which are precisely the occasions on which coordination had gone wrong.
Four qualifications follow, and together they show why the first word carries less authority than its survival suggests. First, deposits are selective. Words record the distinctions that became articulable, and much of what matters in living coordination never does. Second, the archive is asymmetrical. Language records misalignment far better than alignment, because snags demand words and smooth coordination does not. There are many words for what goes wrong in walking, stumbling, tripping, limping, slipping, and very few for the ordinary success of keeping one's balance on uneven ground. A method that reads phenomena through words will systematically overrepresent breakdown. Third, deposits outlast the relevance that produced them. A distinction worth drawing in one mesocosm can persist in the language long after the coordination it served has changed, and its survival proves only that it has not been displaced. Fourth, some deposits become essentially uncontested classifications, so familiar that their selectivity has become invisible. Austin's own example of excuses shows the point. The ordinary vocabulary distinguishes acting by mistake from acting by accident, and Austin's illustration, of shooting a neighbour's donkey when one meant to shoot one's own, makes the distinction vivid. The distinction matters greatly for blame, and it was stabilized in exactly those coordinations in which people had to answer to one another for harm. It says almost nothing about the embodied coordination of aiming, the tremor of a hand, the light in the field or the shape of two donkeys at a distance, which is where the harm actually came from. The test of survival that Austin praised is a test of symbolic validity. It confirms that a distinction has been usable. It does not confirm that the distinction discloses what matters in every coordination to which the word is now applied.
Austin's later account of statements deserves a generous reading, because it comes close to the argument of this article. Asked whether "France is hexagonal" is true, he answered that it is a rough description, good enough for a general and not for a geographer, and concluded that true and false name a general dimension of being a right or proper thing to say in given circumstances, to a given audience, for given purposes (Austin 1962). Austin was not claiming that truth is simply whatever works. He was showing that descriptions can be rough, selective, exaggerated or apt for some purposes and not others, and that a bare opposition of true and false cannot capture these differences. That is close to the Living Value Theory account of the selective adequacy of symbols. What Austin did not develop is a systematic distinction among the things to which a description can be answerable. The shape of France's borders is a matter of cosmic fit; the adequacy of "hexagonal" to the purposes of the general or the geographer is a matter of mesocosmic adequacy and relevance; the coherence of the description with an atlas is intersymbolic; and the appropriateness of saying it on a given occasion is a matter of felicity. Austin identified several of these dimensions along which an assertion can be evaluated. By gathering them under a single dimension of rightness, he left them unsorted, and the world the description speaks about risks receding behind the practice of describing it.
VIII. The radical reductiveness of symbols
Behind all three philosophers lies a feature of symbols that ordinary language philosophy, for all its attention to words, did not fully confront. Symbols are radically reductive. A person standing in a cold kitchen at night may simultaneously register the chill on her forearms, a stiffness in her lower back, a faint hunger, the hum of the refrigerator, the weight of the day, the anticipation of sleep and a thread of worry about something said earlier. Try to describe even two or three of these simultaneous differentiations with precision, together and in their relation to one another, and the difficulty becomes obvious. Language is sequential and categorical; embodied coordination is simultaneous and continuous. A description must choose an order, a grain and a vocabulary, and each choice discards most of what it describes.
This is no defect of language. Symbolic stabilization is valuable precisely because it is selective. A word that tried to carry everything would carry nothing, and the capacity to compress a dense coordination into "I'm tired" is what makes it possible to tell someone else, to be helped, and to remember tomorrow how today went. The point concerns what the success of such a word shows. If symbolic stabilizations are necessarily selective, their successful use cannot establish that they adequately represent the living processes they classify. "I'm tired" can be said truly, understood perfectly and acted on appropriately while leaving almost everything about the state it names unarticulated.
Wittgenstein described the origin of one such compression with great precision. A child hurts himself and cries, and adults teach him exclamations and later sentences; the words are connected with the natural expressions of the sensation and used in their place, so that the verbal expression of pain replaces crying instead of describing it (Wittgenstein 1953, §244). The clinic adds a further layer. Asked to rate her pain from one to ten, a patient compresses a coordination of throbbing, stiffness, fear, fatigue and the memory of last night into a single number, which then travels through records and decisions about medication. Each layer is useful, and each is a further selection. The number is answerable to the scale, the sentence to the language, and the cry to the injury, and only the cry was part of the coordination itself.
The reductiveness extends across all five mediations. Being-with is compressed when a couple who have spent an evening in a complex silence say that everything is fine. Dwelling is compressed when a farmer's discernment of a change in the air, the light and the behaviour of animals becomes the forecast "rain later". Multimateriality is compressed when a carpenter's feel for how a board will split becomes the word "grain". Even multisymbolization is compressed when a whole tradition of argument is summarized as "Wittgenstein's view". In each case the word works. In each case its working leaves the coordination it names largely unarticulated.
The danger is greatest when a symbolic stabilization becomes essentially uncontested. It can then appear to give us the activity itself, when it is in fact a highly selective classification of that activity. This is also why the qualification about symbols must be kept in view. Language is neither unreal nor superimposed on a reality outside it. Multisymbolization is one of the five mediations through which life is coordinated, and its stabilizations are as real as a footpath worn into a field. The point is that its operations are irreducible to the other four, and that its deposits cannot exhaust what happens within them. A philosophy of practice that grounds meaning in use must therefore distinguish the use of words from the uses of bodies, places, things and relationships that the words select from.
IX. What kind of counterpart is classified?
The argument so far suggests a further distinction that clarifies where ordinary language philosophy is most and least reliable. Classifications differ according to the kind of recursive counterpart they classify. Some classify nonrecursive counterparts, beings and things that do not respond to how they are classified: rocks, planets, the length of a sofa. Others classify living coordinations among selfrecursive and interrecursive participants who respond to their own states and to one another, and who can respond to the classifications themselves.
When the counterpart is nonrecursive, the classification can eventually be held against cosmic fit. Whether Pluto clears its orbital neighbourhood is settled by measurements that do not care how astronomers feel about the result, and a classification of Pluto that contradicted them would be corrected however entrenched it had become. Symbolic validity and cosmic fit can come apart, and when they do, there is a court of appeal beyond the practice.
When the counterpart is living coordination, the situation is different, though not because such classifications are insulated from evidence. Living processes have many properties that can be investigated independently: whether a patient's temperature is raised, whether someone attended a lesson, whether money changed hands, whether a particular conversation took place. A classification of friendship, teaching or exchange is answerable to such observations, to participants' testimony and to the consequences of particular arrangements, and false descriptions can be corrected by them. The deeper difficulty is that no fixed, nonrecursive criterion can exhaustively determine what matters within an ongoing recursive coordination. Whether a relationship is best understood as an exchange, whether listening to music is usefully called consumption, or whether a friend is a friend, are questions that evidence can inform and no measurement can close. The coordination is itself recursively open, so that its own organization keeps generating differences that no fixed classification can specify, and its participants can take up the classification and change in response to it. Under these conditions symbolic validity can pass for adequacy more easily, because nothing in the fluency of a word reveals what it leaves out, and its recursive efficacy, the way it reorganizes the coordination it names, can then seem to confirm it.
Ian Hacking came closest to this distinction in his analysis of what he called the looping effects of human kinds. Classifications of people, such as diagnoses, can change the people classified, who then change in ways that require the classification to be revised (Hacking 1995). Hacking later contrasted such interactive kinds with indifferent kinds, which are unaffected by being classified, such as quarks or plutonium (Hacking 1999). Living Value Theory agrees with the distinction and develops it more generally. What matters is the type of recursive counterpart being classified, and the difference concerns more than the classified being's awareness of the classification. The forest is the clearest case. A forest does not recognize the word forest, and nothing it does responds to being so called. It nevertheless remains an ongoing organization of living processes, trees, fungi, insects, birds, soil and water continuously responding to one another and to their conditions, whose development no fixed classification can exhaustively establish. Recursive openness, and not awareness of being classified, is what limits the reach of classification.
This helps to explain why ordinary language philosophy has been so persuasive about some matters and less reliable about others. When it dissolved puzzles about knowledge of material objects, it was working near classifications of nonrecursive counterparts, where ordinary practice is continually corrected by cosmic fit. When its methods were carried into social theory, they encountered classifications of living coordination, where evidence constrains description without closing the question of relevance, and where the success of saying can more easily be mistaken for understanding. The same analysis explains why exchange, transaction and consumption, words with impeccable symbolic validity, have been so difficult to dislodge from the study of human life. The sequence can be stated briefly. Living beings coordinate across five mediations. Selected aspects of that coordination are stabilized symbolically. The stabilizations become so familiar that their selectivity is invisible. Phenomena classified together come to seem obviously comparable. The classification is then treated as identifying what matters. And once institutions organize activity around it, its growing consequences seem to confirm its authority. Not every ordinary classification follows this course, and none is compelled to. The point is to see where the course begins.
X. Conclusion: what good practice could mean
Ordinary language philosophy was right that meaning is not a private mental item, that words get their sense from use, that use is embedded in life, and that philosophers who ignore how words are ordinarily used will invent problems that do not exist. Its appeal to practice was a decisive advance over theories that treated language as a self-contained system of signs. Wittgenstein showed what competent use involves, Goodman showed how projections become established, and Austin showed what makes utterances felicitous. Their methods share a boundary: they illuminate the achievements of symbolic practice, and they do not by themselves differentiate the living coordination within which those achievements occur. Moore's common sense meets the same boundary from the other side, since it consists of propositions.
None of this requires abandoning ordinary language philosophy's best insight. Use remains the right place to begin an inquiry into meaning, ordinary distinctions remain valuable evidence of what has mattered, and common sense remains a resource that philosophers ignore at their peril. What changes is the status of these resources. They become evidence to be weighed alongside other evidence about living coordination, and they lose the role of a final court. An appeal to what we say has to be followed by an inquiry into what we do, in all five mediations, and into how the saying selects from the doing.
Living Value Theory proposes a different answer to the question in this article's title. A practice of symbolization is good when it is answerable to what it articulates. Its symbolic validity shows that it can be used. Its mesocosmic adequacy, investigated across embodiment, being-with, dwelling, multimateriality and multisymbolization, shows whether it discloses what matters in the coordination it classifies. Its recursive efficacy shows what it does to that coordination once it is taken up. And where it makes claims about realities beyond the mesocosm, its cosmic fit shows whether those claims are true. None of these can be read off the fluency of ordinary use.
Truth itself, on this account, is one of the late achievements of a species that learned to articulate propositions detachable from the coordinations in which they arose, to square symbols with other symbols, and finally to reach through instruments and calculations toward a cosmos that owes nothing to anyone's practice. The octopus on the reef does not lie, because it says nothing. Humans can lie, and can tell the truth, because they say things that can be held against a world. What ordinary language philosophy called good practice was the practice of saying. Ordinary language may be the first word, as Austin said, and living coordination was going on long before anyone spoke it.
References
Austin, J. L. 1956. "A Plea for Excuses." Proceedings of the Aristotelian Society 57: 1–30.
Austin, J. L. 1962. How to Do Things with Words. Edited by J. O. Urmson. Oxford: Clarendon Press.
Bshary, Redouan, and Alexandra S. Grutter. 2005. "Punishment and Partner Switching Cause Cooperative Behaviour in a Cleaning Mutualism." Biology Letters 1 (4): 396–399.
Goodman, Nelson. 1955. Fact, Fiction, and Forecast. Cambridge, MA: Harvard University Press.
Goodman, Nelson. 1972. "Seven Strictures on Similarity." In Problems and Projects, 437–446. Indianapolis: Bobbs-Merrill.
Goodman, Nelson. 1978. Ways of Worldmaking. Indianapolis: Hackett.
Hacking, Ian. 1995. "The Looping Effects of Human Kinds." In Causal Cognition: A Multidisciplinary Debate, edited by Dan Sperber, David Premack and Ann James Premack, 351–383. Oxford: Clarendon Press.
Hacking, Ian. 1999. The Social Construction of What? Cambridge, MA: Harvard University Press.
Hanlon, Roger. 2007. "Cephalopod Dynamic Camouflage." Current Biology 17 (11): R400–R404.
Matsuzawa, Tetsuro. 1985. "Use of Numbers by a Chimpanzee." Nature 315: 57–59.
Moore, G. E. 1925. "A Defence of Common Sense." In Contemporary British Philosophy, second series, edited by J. H. Muirhead, 193–223. London: George Allen and Unwin.
Moore, G. E. 1939. "Proof of an External World." Proceedings of the British Academy 25: 273–300.
Norman, Mark D., Julian Finn, and Tom Tregenza. 2001. "Dynamic Mimicry in an Indo-Malayan Octopus." Proceedings of the Royal Society of London B 268: 1755–1758.
Ryle, Gilbert. 1949. The Concept of Mind. London: Hutchinson.
Savage-Rumbaugh, Sue, and Roger Lewin. 1994. Kanzi: The Ape at the Brink of the Human Mind. New York: John Wiley.
Seyfarth, Robert M., Dorothy L. Cheney, and Peter Marler. 1980. "Monkey Responses to Three Different Alarm Calls: Evidence of Predator Classification and Semantic Communication." Science 210 (4471): 801–803.
Seyfarth, Robert M., and Dorothy L. Cheney. 2003. "Signalers and Receivers in Animal Communication." Annual Review of Psychology 54: 145–173.
Tibbetts, Elizabeth A., and James Dale. 2004. "A Socially Enforced Signal of Quality in a Paper Wasp." Nature 432: 218–222.
Whiten, Andrew, and Richard W. Byrne. 1988. "Tactical Deception in Primates." Behavioral and Brain Sciences 11 (2): 233–273.
Wittgenstein, Ludwig. 1953. Philosophical Investigations. Translated by G. E. M. Anscombe. Oxford: Blackwell.
Wittgenstein, Ludwig. 1969. On Certainty. Edited by G. E. M. Anscombe and G. H. von Wright, translated by Denis Paul and G. E. M. Anscombe. Oxford: Blackwell.