Abstract
Conceptual metaphor theory, as developed by Lakoff and Johnson, established that abstract domains are systematically structured by concrete mappings. The key flaw of their theory is that it overattributes their ground to the stable aspects of embodiment, like bones, hands, or eyes. This article argues that metaphorical sources are multimediated, arising from the mesocosm of embodiment, dwelling, multimateriality, being-with, and multisymbolization. In this article, I distinguish nonrecursive, selfrecursive, and interrecursive counterparts and argue for a strong anchoring asymmetry: the vast majority of metaphors recruit stable, portable regularities such as boundaries, trajectories, patterned transformations, schematized turns and polarities in order to make sense of less stable processes. Most metaphors are sourced from entities that do not respond to being addressed. Reanalysis of emotion, argument-is-war, and time metaphors supports the claim. A provenance-analysis method is proposed to test the hypothesis across corpora, genres, and languages.
I. The problem with a single ground
Cognitive linguistics achieved something rare in the human sciences. It took an intuition that everyone had half-noticed, that we speak of abstract things in concrete terms, and turned it into a research programme with predictive reach. George Lakoff and Mark Johnson showed that when we say an argument has no foundation, that a relationship is going nowhere, that prices are rising, or that someone is feeling down, we are thinking through the figure rather than decorating a literal thought with optional ornament (Lakoff and Johnson 1980). The metaphors are systematic, they cluster into families, and they organise inference. If time is a resource, it can be spent, saved, wasted, borrowed, and run out of. The entailments travel together because the mapping travels together.
Nothing in what follows disputes that finding; the question concerns its ground. Where does the concrete come from, and why does the reservoir from which metaphor draws have the contents it has rather than others? Lakoff and Johnson's answer has hardened over four decades into something close to a slogan: the source is the body. Image schemas such as container, path, balance, force, and verticality are said to arise from the recurring structure of bodily life, from having an inside and an outside, moving through space, standing upright against gravity, pushing and being pushed (Johnson 1987). In later work the claim becomes a philosophical programme in which reason itself is shaped by the flesh (Lakoff and Johnson 1999).
This answer identifies something important. Bodily regularities do underwrite a very large part of the metaphorical record. The argument of this article is that embodiment does not exhaust the provenance of metaphorical anchors, and that what metaphor recruits from embodiment is often a relatively stable relation abstracted from bodily life rather than the phenomenological richness of embodied experience. Lakoff and Johnson's own recurrent phrase, the body in space, already names more than a body. It names a body dwelling among found conditions and made things, which is to say a multimediated situation.
Living Value Theory, within which the article is written, offers a vocabulary for both corrections: the five mediations of mesocosmic coordination, and the distinction among nonrecursive, selfrecursive, and interrecursive counterparts. With that vocabulary a more precise hypothesis can be stated. Conceptual metaphor disproportionately recruits relatively stable, recurrent, and portable features of mesocosmic coordination to articulate domains that are harder to stabilise symbolically, especially selfrecursive and interrecursive processes. This can be called the anchoring asymmetry. It is a hypothesis about a particularly important class of conceptual metaphors rather than a law of all symbolisation, and it is developed here from canonical cases whose systematic testing across corpora remains to be done. The final sections specify that test, including what would count against the hypothesis.
II. Mediations, recursivity, and the anatomy of a metaphor
The mesocosm is the middle-scale world between the microphysical and the cosmological in which living beings coordinate. Living Value Theory describes that coordination through five mediations. Multisensorial embodiment, being-with, and multiversal dwelling are the zoetic mediations; together they constitute a complete living mesocosm, sufficient for any organism to coordinate with its own body, with other living beings, and with the found conditions of its world: ground and sky, gravity and weather, water, light, and the cycles of day and season. Multimateriality and multisymbolization are the poietic mediations. They allow recursive life to deposit prior coordination into relatively durable, nonrecursive ground, material in the case of tools, paths, buildings, and vessels, and symbolic in the case of words, numbers, records, and rules.
The zoetic mediations precede the poietic ones derivationally, but the relation is recursive rather than linear. Once multimateriality and multisymbolization emerge, they transform embodiment, being-with, and dwelling in turn. Language in particular becomes L1 ground: coordination that proceeds without becoming separately salient, as a native speaker's language functions without being noticed. There is therefore no one-directional pipeline from nonsymbolic mediations to ordinary language to conceptual thought. The mediations are also always co-present. No lived situation occurs in one mediation at a time, and the right analytical question about any metaphor is which mediation or feature its source foregrounds within a multimediated situation, never which single mediation it comes from.
The three types of recursive counterparts can be defined once. A nonrecursive counterpart does not recursively modify its mediation in response to its own or another being's recursive mediation: a stone, a river, a storm, a written sign. A selfrecursive counterpart is one in which the consequences of its own prior mediation alter its subsequent mediation, as in a living body regulating itself, or a skill refined through practice. Interrecursive counterparts respond to one another's responses, so that the course of their coordination is jointly generated, as in conversation, courtship, negotiation, or combat.
Nonrecursivity must be kept distinct from stability. A storm is nonrecursive and wildly unstable; a river is nonrecursive and constantly changing; an interrecursive relationship can remain remarkably stable for decades. What metaphor appears to require is extractable regularity: a recurring spatial relation, causal sequence, topology, polarity, trajectory, or material affordance that can be lifted out of a situation and carried elsewhere. Stability here is always relative to a timescale. A road is stable relative to a conversation, a mountain relative to a lifetime, a bodily orientation relative to an emotional episode. Something can serve as an anchor because it changes slowly, or in a patterned way, relative to the process it helps to articulate.
A metaphor also involves three things that are easily run together. The source domain is the field in which an expression does its literal work, and it is always multimediated; it may well include recursive processes. The selected feature is what the mapping actually projects from that field onto its target. The symbolic deposit is the metaphor once articulated, a nonrecursive symbolic form that persists in a language and is reactivated in each new use. A source domain can be interrecursive while the metaphor selectively recruits a relatively stable feature from it, and the resulting deposit has a stability of its own that neither the source nor the target possesses.
Several further premises follow from Living Value Theory's broader account of symbolisation. Recursive processes exceed any symbolic deposit made of them: a recursive process can deposit a distinction into a stable symbolic form, but the deposit does not exhaust the living process. Recursive life therefore places especially severe demands on symbolisation, though it holds no monopoly on the need for it; humans name mountains, spoons, stars, and chemical elements constantly. Nor are metaphorical targets always recursive. The foot of a mountain, the mouth of a river, the eye of a storm, and the leg of a table map between nonrecursive or mixed domains. And multisymbolization can itself serve as a source domain for further symbolisation: we read the room, turn over a new chapter, rewrite our lives, follow a script, frame an issue, label a person, and decode a signal.
Two further points bear on how metaphor works. First, metaphor often articulates an L2 configuration, a felt disturbance or pull that has become salient without yet having an adequate literal L3 formulation. Someone who says they feel boxed in is giving portable form to a configuration they could not otherwise state, and the source need not capture the living process; it supplies an articulation of some salient relation within it. Second, metaphors operate in all three symbolic modes. In coordination mode a metaphor participates in the situation it names: telling an opponent that they are moving the goalposts changes the argument. In reference-to-absence mode it makes an absent situation matter to present coordination, and in re-presentation mode it makes present again what is imagined or remembered. A metaphor is therefore something more than a stable description of an unstable event. Finally, language is a radically sparse symbolic trace within a much richer multimediated situation, and this helps explain why metaphors work so economically. A metaphor need not encode its source completely, because speaker and listener already bring bodily, material, social, and dwelling knowledge to it.
III. The body in space
Johnson's canonical inventory of image schemas includes container, path, source-path-goal, link, force, blockage, balance, support, contact, surface, centre and periphery, verticality, and cycle. Two observations change how this inventory should be read.
The first concerns multimediation. The schemas are presented as structures of bodily experience, yet their examples are bodies in rooms, bodies on paths, bodies crossing boundaries, bodies under gravity. The container schema is learned from mouths and bellies, but also from cups, pockets, burrows, houses, and bags, which is to say from embodiment together with multimateriality. The path schema is learned from walking, but walking always proceeds across terrain given by dwelling and very often along tracks and roads made by others. Verticality is anatomical uprightness held against gravity, a found condition of terrestrial dwelling. When Lakoff and Johnson describe the ground of the schemas as the body in space, they have found multimediation and named it embodiment. This correction does not depend on any further thesis. What conceptual metaphor theory calls embodied grounding repeatedly turns out, on inspection, to involve embodiment-plus-dwelling, embodiment-plus-multimateriality, or a wider multimediated coordination.
The metaphor that more is up shows the pattern in miniature. Prices rise, numbers climb, output falls, and a fortune dwindles to a low. On one influential account this belongs among the primary metaphors that arise from recurrent correlations in early experience, in this case the correlation between quantity and height (Grady 1997). The correlation is real, but consider where it is encountered. A child sees the level rise as liquid is poured into a cup and watches a pile grow as things are stacked on it. The correlation requires vessels and heaps, which are made or gathered things; it requires gravity, which keeps liquid at the bottom and makes piles rise upward rather than spread in all directions; and it requires a perceiving body that attends to both. Take away the vessel and the gravity, and the embodied experience of quantity would not correlate with height at all. The primary scene is multimediated from the start, and the schema that emerges from it is a relation abstracted from a coordination among body, found conditions, and made things.
The second observation concerns abstraction. The body is selfrecursive living embodiment. Its skin is continuously renewed, its posture continuously adjusted, its orientation continuously maintained by recursive biological processes. Rather than recruiting anatomy as opposed to living, as though anatomy were a nonrecursive thing set apart from metabolism, metaphor abstracts relatively stable spatial and topological regularities from recursively living embodiment. When happiness is up and sadness down, or a thought is inside me, the mapping draws on relations that remain stable at the timescale of the episode being articulated: the upright axis, the boundary of the skin. The metaphors are notably insensitive to the phenomenological variation of the experiences involved, to how confinement or height actually feels from moment to moment and person to person. What they carry is the relational pattern that the varying experience presupposes.
The inventory also contains schemas that are plainly dynamic. Force, blockage, balance, link, cycle, and motion are relations of change rather than structures that hold still, and forcing them into a thesis about stability would misdescribe them. The more productive question is what makes them metaphorically portable. Several candidate properties suggest themselves. They are repeatable, in that the same kind of push or obstruction recurs across countless situations. They are recognisable at a glance. They have directional structure, from a source toward a goal or against a resistance. They are causally legible: a push produces movement, a blockage stops it, a counterweight restores equilibrium. And they are spatially simple, involving few elements in clear relation. What this requires is patterned change rather than the absence of change: change organised in a way that can be grasped quickly and transferred without loss of its organising form.
This suggests that the operative contrast may lie between open-ended recursive responsiveness and comparatively tractable patterned transformation, rather than between the recursive and the nonrecursive as such. A force meeting a blockage is a transformation whose course is legible from its conditions. A conversation is a transformation whose course is jointly generated by counterparts who respond to one another's responses. Both change; they differ in whether their unfolding can be anticipated from a small number of recognisable elements. The anchoring asymmetry, stated at this level, predicts that the first kind of process will be recruited as a source far more readily than the second, even where the second is more frequently and intensely experienced.
A related observation deserves more weight than it has received. Recurrence in experience cannot by itself explain the inventory. People experience conversations, negotiations, embarrassments, reconciliations, and teaching daily, and these recurring experiences appear far less often among canonical source domains than containers and paths. This is a claim about canonical inventories that corpus work would need to confirm, but if it holds, frequency of experience would not be the operative variable, and tractability of structure might be.
IV. Three tests
The anchoring asymmetry can be tested first against conceptual metaphor theory's own showcase cases. Three are especially instructive: emotion, argument, and time.
The emotion metaphors are the received theory's clearest case of embodiment, because emotion is bodily if anything is. He is about to burst. She is letting off steam. He is boiling with rage. She was frozen with fear. He exploded. She simmered for days. Their anger evaporated. Lakoff and Kövecses reconstructed from such expressions a cultural model of anger as heated fluid in a container (Lakoff and Kövecses 1987). Every one of these sources is a change: boiling, bursting, exploding, freezing, and evaporating are transformations rather than states. So the emotion metaphors refute any version of the thesis that equates anchoring with stasis. What they recruit is the recognisable causal organisation of the process: heat raises pressure, pressure strains containment, containment fails or is relieved through a vent. Boiling itself varies with pressure, composition, container, and heat, and what matters is that people know a sufficiently dependable mesocosmic pattern linking heat, agitation, pressure, and change of phase. The target, by contrast, is selfrecursive: anger rehearsed becomes anger enlarged, and dread compounds itself. The source is a patterned transformation; the target is a process that feeds on its own course. Other emotions show the same structure with different sources. Sadness is heavy: a heavy heart, weighed down by grief, a burden lifted. Here the source is a load borne against gravity, which combines a found condition of dwelling with the carried things of multimateriality and the strain of an embodied posture. Fear freezes, chills, and paralyses, recruiting the transformation of matter under cold and the arrest of movement. In each case the source changes, and in each case the change has a legible course: a weight presses down and can be set down, cold stiffens and can thaw. The emotions themselves have no such reliable course, and that difference may be exactly why the sources are useful.
The emotion case also displays multimediation. The felt heat and agitation of anger, an embodied correlation, plausibly explain why heat is available as a mapping at all. But the elaboration draws on vessels, lids, valves, pressure, and boiling points, a domestic and industrial technology of containment that belongs to multimateriality. The anger metaphor is embodiment-plus-multimateriality, and much of its conceptual work happens in the material part. It also operates in coordination mode: calm down, don't blow up is an intervention in an unfolding episode, and it may alter that episode by offering the angry person a portable picture of where it is heading.
Argument is war is the first extended example in Metaphors We Live By and the most famous conceptual metaphor in the literature. He shot down my argument. Your claims are indefensible. I attacked every weak point. We defended our position. He came under fire. We had to retreat. He ceded ground. The target is interrecursive in nearly its purest ordinary form: each move in an argument changes the space of possible next moves for everyone involved. The source is no less so. War is itself interrecursive, and it is also massively multimediated. It involves bodies that are wounded and exhausted, terrain and weather, weapons, fortifications and supply, commands, laws of war, and institutions. It would be evasive to claim that the metaphor recruits only war's material and spatial furniture. Many of its most productive terms are intrinsically relational and often interrecursive: opponent, attack, counterattack, defend, strategy, outflank, retreat, surrender, win, lose.
The case is more interesting treated as evidence of selective abstraction within a multimediated source. The question becomes which features of war travel into argument, and which do not. What travels includes the spatial model of positions and ground, the material model of weapons and targets, and a particular form of interrecursive structure: move and countermove, attack and defence, advance and retreat, organised into legible turns with a clear polarity of winning and losing. What largely stays behind includes death, mutilation, grief, occupation, the suffering of civilians, famine, supply chains, and the long aftermath of a war whose outcome is unsettled. The hypothesis this suggests is refined rather than protected. Interrecursive structure can be recruited when it has been schematised into turn-taking, polarity, and position. What appears less readily recruited is interrecursivity in its open-ended form, the jointly generated uncertainty that no schema anticipates. Whether this holds is an empirical question, and war is precisely the kind of rich source against which it can be checked.
Lakoff and Johnson themselves offered a thought experiment that bears on this. They asked the reader to imagine a culture in which argument was viewed as a dance, the participants seen as performers aiming at a balanced and aesthetically pleasing performance, and suggested that we would hardly recognise what such people did as arguing (Lakoff and Johnson 1980). The imagined alternative is also a source drawn from being-with, and it too would recruit a schematised form of interrecursivity: steps, partners, leading and following. Neither war nor dance would enter the mapping as the open-ended mutual responsiveness that actual fighting or actual dancing involves.
The canonical metaphor that love is a journey offers a useful contrast. Look how far we've come. We're at a crossroads. This relationship is a dead-end street. We're spinning our wheels. Here an interrecursive target, a relationship, is articulated through a source that is mostly not interrecursive at all: a vehicle travelling along a road across terrain, which is to say multimateriality and dwelling organised by an embodied trajectory. What is projected is shared direction, progress, obstacle, and decision point. Even the crossroads, where the metaphor comes closest to the jointly generated character of a relationship, reduces that character to a choice among given routes. Argument is war and love is a journey therefore arrive by different paths at a similar result: in both, what gets carried into the interrecursive target is a legible structure of positions, directions, and turns.
Time is the domain through which Lakoff and Johnson argue most explicitly that some concepts can only be grasped metaphorically. Christmas is coming. We are approaching the deadline. The weeks flew by. Time is a moving object, or the observer moves through time, and either way the ground is motion along a path. This is well attested, and it fits the anchoring asymmetry: lived temporality is recursive, the present continuously reshaped by memory and anticipation, and it is articulated through the relatively simple trajectory of motion. Cross-linguistic work shows that the spatial construal can vary strikingly; in Aymara, for instance, the known past lies ahead and the unknown future behind (Núñez and Sweetser 2006).
The body's motion is only one of time's sources. The dawn of a new era, the twilight years, the spring of life, winter is coming, and the year is turning all recruit the cycles of dwelling: the transit of the sun, the round of the seasons. The recruited feature is a recurrent sequence, relatively stable at the scale of a human life though each cycle is itself change. Time also recruits multimateriality and multisymbolization: we carve out time, spend it, save it, and run out of it, drawing on stocks and on money, the latter a symbolic deposit. A distinction is needed here. Expressions such as three moons ago, before the harvest, and after the thaw are probably literal systems of temporal reckoning rather than conceptual metaphors. They are excellent evidence that human temporal symbolisation draws on dwelling as well as embodiment, but they should not be counted as metaphors unless that status can be established. Taken together, the time case shows temporal articulation drawing on several mediations, with the recurring feature being a trajectory or sequence that is tractable at the relevant timescale.
One further family of time metaphors points ahead to the next section. Time heals, time will tell, time is a thief, and time waits for no one all personify time as an agent. Personification is one of the largest metaphor families there is, and it draws its source from the domain of acting beings, which is to say from being-with. It therefore needs to be taken seriously as a possible counterweight to the anchoring asymmetry.
V. Being-with as a source: an open question
If the anchoring asymmetry holds, open-ended interrecursive processes should appear as metaphorical sources less often than their experiential prominence would lead one to expect. Being-with is the mediation in which such processes are most concentrated: trust and betrayal, friendship and rivalry, teaching and following, negotiation and reconciliation are the domains human beings attend to most constantly. It is tempting to conclude from canonical inventories that being-with supplies almost no source domains. That conclusion is premature.
There are many obvious candidates. Life is a game. We dance around an issue, follow someone's lead, and court disaster. Nations court one another, sever ties, and divorce. Firms enter marriages of convenience. Disciplines are said to be in dialogue. The market punishes a company. Kinship is an especially rich source: mother tongue, mother country, sister ships, daughter cells, parent companies, families of languages. States and colonies have long been described through relations of parent and child. Any serious test must begin from such cases rather than from a list prepared to exclude them.
Examining them suggests a pattern that is worth testing, without deciding its outcome in advance. Kinship metaphors recruit the relatively stable topology of descent and derivation, which is why they serve so well to name relations of origin among languages, cells, and firms; they draw far less on the open-ended interrecursive life of families. Games recruit rules, turns, winners, and losers, much as argument recruits war. Dances recruit steps and the relation of leading and following. Ritual and institution introduce relatively stable forms into social life, and those forms travel: an argument can be a trial, a courtship a dance. On the other hand, some candidates seem to project mutual responsiveness itself. Disciplines in dialogue, nations courting one another, and flirting with an idea draw on tentative, reciprocal approach and adjustment. These may resolve partly into more schematic features, or they may be real instances of open-ended interrecursivity serving as a source. Their status depends on what is actually projected and on how frequent such mappings prove to be.
Personification is the most important test here. Lakoff and Johnson noted how readily agency is ascribed to abstractions: inflation attacks the foundations of the economy, life has cheated someone, cancer finally caught up with him (Lakoff and Johnson 1980). Such expressions project an agent onto a target, and agency belongs first to living beings. Yet the agents projected are usually of a particular kind. They act on us, in one direction and with a recognisable intent: the thief takes, the attacker strikes, the healer mends, the judge decides. Much rarer, at least in canonical collections, are personifications that project a responsive counterpart, one whose next move depends on ours and whose response we must read as it unfolds. People seldom speak of negotiating with time, of time misunderstanding them and then trying again, or of reaching a gradual accommodation with inflation. If this contrast holds up across corpora, personification would confirm the hypothesis in an unexpected way: even when being-with supplies the source, what travels is agency schematised as a directed role rather than interrecursivity in its open-ended form. If it does not hold up, personification would be the place where the hypothesis most clearly fails.
The hypothesis, then, is that being-with supplies source structures chiefly where its relations have been stabilised into roles, rules, genealogies, or choreographies, and less often where open-ended mutual responsiveness is the projected feature. If that is so, it would be a striking finding, because it would show that the mediation richest in recursive coordination contributes to metaphor mainly through what social life has itself deposited in relatively stable form. If corpus work instead shows that mappings projecting mutual responsiveness are common once literal frequency is controlled for, the hypothesis should be revised or abandoned.
VI. Historical drift, inheritance, and new deposits
An account of metaphorical sources should explain the history of the metaphorical record as well as its synchronic state. An embodiment account implies that the deep sources of metaphor are broadly constant, since human bodies change little over historical time. A multimediational account implies that the inventory will drift with the mesocosm, because what is recursively relevant to coordination changes as material and symbolic worlds change.
The relevant concept is recursive relevance. At any moment some features of the mesocosm are load-bearing for coordination: life is scheduled by them, work organised around them, safety dependent on them. These saturate attention and are disproportionately available for recruitment. Mesocosms organised around craft and cultivation deposited forging alliances, weaving narratives, sowing doubt, and shouldering burdens. The industrial mesocosm filled the languages of feeling and knowledge with pressure, output, flows, running on steam, and cogs in the machine. The digital mesocosm is now supplying bandwidth, storage, platforms, networks, processing, and rebooting as sources through which minds, relationships, and institutions are articulated. These are best understood as new poietic deposits becoming available as source domains, and they give the account a clear diachronic prediction: newly consequential material and symbolic technologies will enter the metaphorical record as sources with a lag that corpus history can measure.
Recruitment is only half the story, because symbols sediment. Repertoires have inertia, and dead technologies leave living metaphors behind. People still dial numbers, hang up phones, copy colleagues on emails under a heading inherited from carbon paper, and press a save icon shaped like a floppy disk that many users have never seen. The historical claim should therefore be one of biased recruitment plus symbolic inheritance rather than direct tracking. New sources are recruited disproportionately from what has become recursively relevant, while earlier deposits persist, sometimes as live metaphors and sometimes as expressions whose source has been forgotten. The result is a stratified lexicon with live metaphors above and fossils below.
Multisymbolic sources show the same stratification with particular clarity, and they confirm that symbolisation regularly takes other symbolic practices as its source. The deposits of literacy are everywhere: we read a situation, turn the page, speak of a closed book, try to get everyone on the same page, and begin a new chapter. Print and bureaucracy added the footnote, the fine print, and the rubber stamp. Cinema and recording added fast-forwarding, rewinding, zooming in, the flashback, and the freeze frame, and several of these have already outlived the technologies that gave them their literal sense. Dated attestations in historical dictionaries and corpora make it possible to measure how long each deposit took to become a source and how long it has survived as one, which turns the account of drift into something that can be checked rather than merely illustrated.
The same reasoning accounts for much cross-cultural variation. Different mesocosms make different features recursively relevant, and their metaphorical repertoires differ accordingly, alongside the substantial commonalities that shared bodily and dwelling conditions produce (Kövecses 2005). A theory that locates the source of metaphor in a universal body can accommodate the commonalities. A multimediational account can accommodate both.
VII. Provenance analysis and the test of the hypothesis
The anchoring asymmetry needs a method that allows it to lose as well as win. The core of that method is provenance analysis, and its questions should be neutral with respect to the expected result. For any metaphorical expression, the analyst asks where the source term is used literally; which mediations are involved in that literal use; which counterpart types and which recursive relations occur there; which aspects of that domain are projected into the target; and which aspects are left behind. The answers are recorded as found, including cases in which the projected feature is itself open-ended and interrecursive.
In practice each identified mapping can be coded along a small number of dimensions. The first is the mediations foregrounded in the literal use of the source. The second is the counterpart type that predominates there: nonrecursive, selfrecursive, or interrecursive. The third, and most important, is the type of feature projected, which can be sorted provisionally into static relations such as containment or verticality, patterned transformations such as boiling or blockage, schematised interrecursive structures such as turns, roles, and polarities, and open-ended interrecursive responsiveness. The fourth is the timescale at which the projected feature is relatively stable compared with the target. Two coders working independently should be able to apply these categories with measurable agreement, and categories that fail that test should be refined before any distributional claim is made. The corpus itself should extend well beyond the canonical examples of cognitive linguistics, to conversation, popular song, political speech, scripture, fiction, psychotherapy transcripts, and the technical vernaculars of the disciplines, because canonical inventories were assembled by analysts looking for other things.
The hypothesis can then be stated as a modest prediction. If the account is right, conceptual metaphor should disproportionately recruit relatively stable and recurrent features of mesocosmic coordination, while highly open-ended interrecursive processes should be less frequent as source structures than their experiential prominence would predict. Experiential prominence can be approximated by the frequency of literal discourse about a domain in the same corpus. Metaphorical expressions can be identified with established procedures such as MIP and MIPVU (Pragglejaz Group 2007; Steen et al. 2010), and corpus methods allow distributions of source domains to be compared across genres, periods, and languages (Deignan 2005).
It matters to state in advance what would count against the hypothesis. It would be weakened if, controlling for literal frequency, mappings that project open-ended mutual responsiveness proved as common as mappings that project relatively stable or patterned features. It would be weakened if features projected from interrecursive sources such as war, games, or courtship were not systematically the more schematised ones, the turns, polarities, and positions, rather than the open-ended ones. And it would be weakened if mappings from recursive sources onto relatively stable targets were as frequent as mappings in the opposite direction among targets of comparable salience. To prevent the reclassification of counterexamples after the fact, the coding rules for counterpart type, projected feature, and relevant timescale must be fixed before the data are examined. The present article develops the hypothesis from canonical cases and specifies this test; it does not claim to have carried it out.
Provenance analysis also has applications beyond testing. Any discourse has a profile: the set of anchors through which it habitually articulates its objects, weighted by use. A discourse can be called captured when that profile has collapsed onto one narrow family of anchors, so that a single region of the mesocosm has become the only structure through which its object can be imagined. The contemporary history of science offers an example. Its central object, knowledge, is routinely produced, circulated, exchanged, transferred, stored, accumulated, and mobilised, and it moves through networks and flows. Almost every anchor comes from one historically specific region of multimateriality, the production, logistics, and warehousing of goods in an industrial economy. The vocabulary illuminates what in knowledge behaves like movable stock and obscures what behaves like trust, skill, or apprenticeship. The remedy is repluralisation rather than substitution: a field able to say without strain both that knowledge circulates and that knowledge is cultivated, embodied, and apprenticed would articulate its object less incompletely. Medicine offers a case where the consequences of capture reach patients directly. Susan Sontag showed how thoroughly illness, and cancer above all, came to be spoken of through war: invasion, defence, the body's arsenal, the fight, the battle lost or won (Sontag 1978). The profile is narrow, and its effects are practical. A target that is selfrecursive, a body living with a disease that changes as it is treated, is articulated through a source that recruits siege and combat, and so a person who dies can be said to have lost a battle, as though insufficient fighting had decided the outcome. Alternative anchors exist, including living with an illness, managing a condition, and learning the shape of a new life, and they make different aspects of the same recursive process available. Profiling of this kind makes such narrowing visible and open to revision. Similar profiles could be computed for economics, psychology, management, and the discourse of artificial intelligence.
Formal notation poses a related but different question. Formal systems can reduce dependence on inherited ordinary-language metaphors by stipulating the relations among their symbols explicitly, which is one reason the exact sciences have developed them. Notation does not thereby leave multimediation behind, since equations are written, perceived, taught, materially instantiated, and socially verified within institutions. What formal notation increases is the possibility of intersymbolic precision and, in some sciences, of cosmic fit with the aspect of the world it formalises.
Conclusion
Conceptual metaphor theory discovered a real and important regularity: abstract and dynamic domains of life are thought through concrete mappings, and those mappings are systematic and organise inference. This article has not disputed that finding. It has argued that embodiment is too narrow a description of its ground. What conceptual metaphor theory calls embodied grounding repeatedly turns out to involve embodiment together with dwelling, with made things, and with the wider multimediated coordination of the mesocosm. The body in space is already a multimediation.
Within that multimediated field, metaphor appears to have a strong bias. It extracts relatively stable, repeatable, and portable relations, such as boundaries and trajectories, patterned causal transformations, turns and polarities, and genealogies and choreographies, and uses them to articulate processes whose unfolding is harder to stabilise symbolically, above all the selfrecursive processes of feeling and the interrecursive processes of living with others. The bias is compatible with sources that change, as the emotion metaphors show, and with sources that are themselves interrecursive, as argument is war shows, because what matters is the feature selected rather than the ontological type of the whole source. It accommodates multisymbolic sources, historical inheritance, and cultural variation.
The account also helps explain why metaphor works as economically as it does. A metaphor often gives portable form to an L2 configuration, something felt as salient before it can be said, and it does so without having to encode its source in full, because speakers and listeners bring to it everything they already know about vessels, roads, weather, and one another. The relation it carries is small. The world that makes the relation intelligible is supplied by the multimediated situation in which the metaphor is used.
The anchoring asymmetry is a hypothesis, and it can now be tested systematically across languages, historical periods, genres, and source domains with methods that allow it to fail. If it survives, it will have shown something about symbolisation in general: that articulating recursive life depends heavily on relations that hold their form long enough to be carried from one situation to another.
Metaphor does not escape living flux; it makes portions of that flux articulable by recruiting relations stable enough to travel.
References
Deignan, Alice. 2005. Metaphor and Corpus Linguistics. Amsterdam: John Benjamins.
Grady, Joseph E. 1997. "Foundations of Meaning: Primary Metaphors and Primary Scenes." PhD diss., University of California, Berkeley.
Johnson, Mark. 1987. The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason. Chicago: University of Chicago Press.
Kövecses, Zoltán. 2005. Metaphor in Culture: Universality and Variation. Cambridge: Cambridge University Press.
Lakoff, George, and Mark Johnson. 1980. Metaphors We Live By. Chicago: University of Chicago Press.
Lakoff, George, and Mark Johnson. 1999. Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. New York: Basic Books.
Lakoff, George, and Zoltán Kövecses. 1987. "The Cognitive Model of Anger Inherent in American English." In Cultural Models in Language and Thought, edited by Dorothy Holland and Naomi Quinn, 195–221. Cambridge: Cambridge University Press.
Núñez, Rafael E., and Eve Sweetser. 2006. "With the Future Behind Them: Convergent Evidence from Aymara Language and Gesture in the Crosslinguistic Comparison of Spatial Construals of Time." Cognitive Science 30 (3): 401–450.
Pragglejaz Group. 2007. "MIP: A Method for Identifying Metaphorically Used Words in Discourse." Metaphor and Symbol 22 (1): 1–39.
Sontag, Susan. 1978. Illness as Metaphor. New York: Farrar, Straus and Giroux.
Steen, Gerard J., Aletta G. Dorst, J. Berenike Herrmann, Anna A. Kaal, Tina Krennmayr, and Trijntje Pasma. 2010. A Method for Linguistic Metaphor Identification: From MIP to MIPVU. Amsterdam: John Benjamins.