The fourth exchange takes place within a tool that both sides must rely on: language. The constructor needs shared words in order to coordinate; the opposition points out that every act of naming loses experience. We cannot hand complete experience over to another person just as it is. Once a moment of seeing, a stretch of pain, or a hesitation enters language, it must choose its object, its time, its causality, and its emphasis. This choosing causes loss, yet it is also what allows experience to leave the one who had it and meet those who were not present. The public value of language and its capacity to conceal therefore appear at the same time. Imagine a fictional small town repairing an old bridge. During construction the bridge deck keeps one narrower passage open, and a notice reads "pedestrians may pass normally." The maintenance crew uses this sentence to indicate that the bridge is not fully closed; a resident pushing a stroller takes a detour and concludes that she cannot pass normally at all; the meeting minutes later record the dispute as "individual residents have differing perceptions regarding the convenience of the construction."
All three sentences use everyday words, and all three can find some factual support. The bridge is indeed not fully closed, the passage does exist, and the resident did express inconvenience. The difficulty does not lie in identifying one pure lie, but in the fact that each act of retelling changes the object of the question: "a passage exists" replaces "who can use it," "cannot pass" may omit the particular equipment and time window, and "differing perceptions" folds observable width and conditions of action back into subjective experience.
This chapter traces how the same dispute passes through on-site expression, working language, and public record. The bridge, the people, and the retellings are all a thought experiment and prove nothing about any real institution having the same problem. What we want to discuss is this: how shared words make collaboration possible, and how they leave experience that has not yet entered a classification standing outside; and what kind of power stands in need of critique once language connects with the right to record, the right to name, and the qualification to respond.
Shared Words Let Dispersed Experience Reach One Another
If each person could keep only his own sensations, the maintenance crew could not know which section of the passage needs inspection, and residents could not compare changes across different times. To say "the east entrance," "eight in the morning," or "a stroller cannot turn" is already to break continuous experience into parts that can be located again. Language does not copy the scene, yet it creates points of entry for shared verification. Shared words also lower the cost of collaboration. When the construction team calls a set of steps a "temporary passage arrangement," it need not restate the entire background each time; when a resident says "detour," others can already expect to walk a longer way. Concepts compress many events into one addressable node, allowing memory, division of labor, and commitment to persist.
Compression is not a secondary defect but the condition of symbolic work. A word that contained every detail of every event would be difficult to use in a new situation. RC describes language as the underlying framework of the deterministic set and of cognition, while acknowledging that abstraction causes loss of experience. This is a theoretical positing together with its internal implications, not an empirical law about all linguistic activity.
Abstraction Is Not a Betrayal of Concrete Experience
The maintenance crew must schedule work for multiple entrances, multiple time windows, and multiple kinds of users, and cannot rely on any single person's complete narrative. Summarizing the situation as "passage capacity restricted" can support scheduling and resource judgments. If we refuse all generalization, the people who are able to keep attending meetings and repeating details are precisely the ones more likely to occupy public attention.
Whether an abstraction is defensible depends on which question it serves. "Is the bridge fully closed?" and "Does the current passage allow common mobility aids to get through?" require different classifications. The first question can be answered with open or closed; the second requires dimensions, turning positions, and patterns of actual use. When a coarser classification is used to answer a finer question, the loss is no longer merely unavoidable; it becomes an overreach of inference.
The resident's account is itself an abstraction. When she says "I simply cannot get through," she may mean that with her current stroller she cannot complete the crossing safely from the east entrance, not that no one can cross at any time. If the emphasis of colloquial speech is interpreted as a universal fact, and her statement is then refuted by finding one pedestrian who got through, the object of verification has been quietly enlarged. Effective dialogue does not require every sentence to meet report standards from the outset. Everyday expression can first indicate a problem and then, through follow-up questioning, form statements that can be compared. The aim of follow-up questions should be to learn what happened, not to prove that the reporter was insufficiently precise the first time and therefore forfeits the standing to say more.
Translation Brings Different Tasks into Contact and Shifts Responsibility
For on-site experience to enter the repair schedule, one translation is needed: from "I cannot turn around" to "an obstacle exists for a certain kind of mobility at a certain entrance." For the work plan to enter a public notice, another translation is needed: from a set of construction restrictions to a statement residents can act on. Translation is not the substitution of synonyms but a choice, between two tasks, of what to preserve. The person responsible for translation may act without ill will and still shape the public shape of reality. He decides whether "cannot turn" belongs to safety, to convenience, to an individual need, or to a construction exception; different categories lead to different places of handling. If a classification offers no path back to the scene, the translator has distributed the fate of the problem without ever claiming the right to decide.
Translation can also correct partial perspectives. A resident may not know that another entrance is available, and the maintenance crew can supply that; the crew may not know the specific limits a stroller faces on the ramp, and the user can supply that. The information held by the two sides is neither automatically equal by identity nor automatically complete on one side. Good translation preserves the source, so that new material can indicate which step needs rewriting.
Language thereby establishes a limited publicness: it does not require all subjects to share the same experience, only that they be able to compare statements around some object, supply missing conditions, and state what remains unknown. Chapter 4 argued that shared values need not expand into a complete view of life; in the same way, a common language need not expand into a single way of experiencing.
One Dispute Through Three Retellings
The first retelling comes from the scene. The resident says: "This morning I pushed the stroller to the east entrance, the front wheel jammed at the turn, and I could only back out." This contains time, position, action, and result, and it also preserves the fact that the speaker made no claim about the maintenance side's intentions. It remains incomplete, for example saying nothing about other entrances or on-site signage, but others know what they can check. The second retelling comes from the work group. The liaison writes: "A resident reports that east-side passage is inconvenient; construction team please take note." This sentence is convenient for assignment, yet it demotes "could not complete" to "inconvenient" and omits the stroller and the time window. If the construction team judges only by this, it may understand the problem as an acceptable extra nuisance rather than as the current arrangement failing for a certain way of passing.
The third retelling enters the meeting minutes: "During construction, passage is normal overall; individual residents raised convenience concerns, which have been forwarded to the relevant personnel." The minutes accurately record that concerns were raised, and they may accurately describe that most people can get through; but they do not state whether the concerns were verified, who was to respond, or whether conditions in reality changed. "Forwarded" becomes the full stop that closes the process.
Each retelling can be defended. The on-site narrative is too long for a task list; the task list is too specific for an overview; the minutes must let absent people grasp the overall situation quickly. The problem lies in whether, once a later text acquires higher formality, the earlier text is still allowed to change it. If the public record can cite only its own summaries, the loss is amplified by authority.
How Subject, Verb, and Scale Move
"A stroller cannot turn at the entrance" places the obstacle within a relation of action; "residents find passage inconvenient" places it within a person's experience; "convenience concerns exist" turns the actor into a kind of opinion. As the subject becomes progressively more abstract, the object of responsibility fades. Not one sentence says outright that the resident was wrong, yet the final text no longer needs to answer whether the passage is usable. Verbs also distribute states of completion. "Reported" says only that expression occurred; "forwarded" says only that information moved; "handled" may include everything from reading a message to changing the facility. When an organization uses "handled" to cover these differences, language lends the appearance of completion to a reality that is not yet complete.
Changes of scale determine whether the experience of a minority remains relevant. That most people can pass in general does not contradict the claim that some common way of using the bridge is blocked. If the question at hand is whether the bridge remains basically open, the overall scale is meaningful; if the question is whether the repair arrangement excludes a portion of users, small numbers cannot automatically cancel the object.
"Individual" is sometimes a necessary piece of quantity information and sometimes a hint that the problem may be ignored. Telling the two apart requires knowing how the number of people and the severity of consequences enter the decision. One person suffering a slight delay and one person losing the only path she can reach are the same in number but different in public significance. If language offers only frequency, the intensity of consequences hides behind statistical modifiers.
Omissions Accumulate Rather Than Averaging Away
One summary omits the stroller, the next omits the entrance, and the one after that omits the status of what has not been verified. Each step loses only a little, yet the relation between the final record and the scene changes in kind. The loss has a path: later texts usually take the previous summary as their source rather than revisiting the original material. This does not prove that more retelling necessarily means less truth. A new reteller may check the scene, add the experience of other users, and correct the initial misunderstanding. The decisive difference is whether the chain of retelling merely copies conclusions or can make contact with the object again. The number of sources does not equal independent observation, as earlier chapters have already shown.
If the liaison attaches the original message to the task, the construction team inspects the entrance, and the minutes record "the east side is unusable for a certain kind of stroller; the west side is usable but requires a longer walk," public understanding may become more complete than the first sentence. Linguistic loss is not unrepairable; repair comes from the renewed meeting of sources, objects, and different observations.
Conversely, to demand that every summary contain all the original words would make public records unusable and might also disclose information unnecessarily. Completeness is not maximal word count; it means that material relevant to the current judgment does not systematically disappear because the speaker's position is weak, and that what was omitted can be retrieved when the question changes.
How Naming Changes the Public Position of a Problem
When the liaison names the event a "passage obstruction," it may enter the handling process for safety and accessibility; named a "convenience concern," it may enter general suggestions; named an "emotional complaint," the first response may be communication and appeasement. The name does not create the event by itself, yet it changes who takes over, what evidence is required, and when the matter counts as closed. Naming also produces similarity. If multiple different reports are all filed as "construction complaints," the organization can count them, yet it may fail to see that among them are different objects: mislabeled signage, scheduling conflicts, entrance obstructions, and noise. Categories help materials meet one another, and they can also let events that should not be identified overwrite one another.
Technical vocabulary has a real function. "Accessibility" connects design, review, and responsibility in a way that "is it easy to walk" cannot; but those who use the technical term do not thereby acquire ownership of all experience. A resident who does not know the terminology may still accurately describe the failure of an action. Terms should help her statement reach the place of handling, not serve as a prior qualification for the problem to exist.
Concept Ownership Forms a Hidden Threshold
Suppose the maintenance side requires that every report be classified as "passage interruption," "capacity reduction," or "subjective inconvenience," and that the system automatically returns any report whose category the resident has chosen wrongly. Classification can improve the efficiency of handling, but if ordinary people cannot know the boundaries, the point of entry hands all the interpretive labor to those with the least information. Those able to name are more easily able to make their experience into official reality. RC's account of language can be pushed from here toward a dark use: since experience that can be stably said more easily enters common cognition, those who command the vocabulary and the classifications may be considered more qualified to decide what exists. A description of representational capacity in the theory is expanded into a normative judgment about the standing of subjects; this step is not automatically guaranteed by the original proposition, yet it can borrow from that proposition an appearance of consistency.
Stronger control need not forbid speech. It can keep updating the vocabulary, so that critics are always speaking in "outdated," "unprofessional," or "wrong-level" terms; it can also acknowledge experience while admitting it only into categories that carry no power of decision. The channels of expression remain, while the bridge from language to action is cut. Whether such mechanisms occur in reality cannot be established by theoretical possibility. One must examine the classification rules, the records of returned reports, the responses given to different phrasings, and who can modify the lexicon. To interpret every term as control would equally make professional communication impossible and turn critique into a closed narrative that accepts no counterexample.
Concept ownership also runs in the opposite direction. Critics may declare that only those who have lived through something may use a certain word, and thereby refuse external verification; or they may use an ever-expanding critical vocabulary to explain every response in advance. Protecting the source of experience does not mean letting the source monopolize conclusions about cause and responsibility. Whoever is affected possesses irreplaceable material, but that material still needs to be compared with other conditions.
Ambiguity Preserves Room but Also Shifts Responsibility
A notice reading "passage may be adjusted when necessary" can cope with the unknowns of construction. If all changes cannot be listed in advance, keeping some elasticity is reasonable. Ambiguity is not intrinsically deceptive; it can also express honestly what is not yet settled. The question is who enjoys the elasticity of interpretation. A resident plans her trip around "may pass normally," and when blocked is told that "normal" never promised every mode of passing; the manager extends restrictions on the strength of "may be adjusted," and when held to account says the text fixed no meaning. The same ambiguity offers options to whoever holds more power and transfers predictive risk to whoever depends on the text.
A darker inference would say that since language necessarily contains an interpretive gap, no promise can ever truly be broken, and every dispute is only a difference of projections. This inference converts an epistemic limitation into an exemption from responsibility. That expression cannot exhaust reality does not mean that any interpretation equally fits the text at the time, the shared background, and the foreseeable consequences.
Judgments of responsibility can acknowledge degrees of vagueness. First ask which meanings were reasonably available at the time; then ask what information the speaker had, who relied on the text, and whether the ambiguity was clarified promptly once discovered. We cannot demand absolute precision on the grounds that language is imperfect, but neither can we let that imperfection permanently absorb every counterexample on behalf of the side that keeps benefiting.
Detail Is Not Intrinsically Closer to the Truth
Faced with the losses of summary, we are tempted to propose the opposite principle: keep every detail. Yet detail is also selected. A repair report could list every movement of equipment, every group message, and every minute, and still not say for whom the passage is unusable; the material multiplies while the central relation stays absent. Detail consumes attention. If every comment comes with dozens of pages of records, fewer people can read and compare. Formal total disclosure can shift the cost of verification onto residents and then prove the report reliable by the fact that no one pointed to a specific error. The mechanism of information overload is left to later discussion; this chapter only confirms that quantity cannot replace relevance to the question.
Detail can also intrude on lives unrelated to the judgment. To show that a stroller cannot turn, a resident need not disclose her household composition, her physical condition, or why she was crossing at that hour. Demanding more background can sometimes reveal alternative explanations, and it can also turn the reward for a response about passage into an act of self-disclosure.
Precise Numbers Still Require an Account of Their Object
When the construction team states the passage width as a number, this is more comparable than "wide enough." But the number does not automatically answer for turning, slope, companions, and temporary obstructions. Precision about one variable is not precision about the conditions of action. Conversely, the resident's concrete failure does not prove the number false. Measurement and lived experience can both hold, and together expose a relation the original model omitted. If the two sides only fight over which description is truer, they miss the fact that each answers a different question.
Between "the data say the passage exists" and "I could not get through," we need not elect a single true statement. What is needed are further statements: how the data were obtained, at which link the failure occurred, and what the current public commitment is. Theoretical dimensional reduction requires every description to acknowledge its object, its subject, and its temporal boundary. This requirement also binds the critic's own story.
Numbers can also produce a discipline without commands. An organization rules that only reports with timestamps, location codes, and a designated format will be accepted; those with fewer resources can hardly comply; it then proves the problem rare by the small number of valid reports. The convenience of technical format and the qualification to speak adhere at this point. This is not a reason to abolish structured records. The reasonable practice is for the receiving side to bear the necessary conversion, allowing ordinary expression to enter first and then letting the two parties confirm together the fields that can be processed. The purpose of format is to reduce ambiguity about the object, not to filter for those most fluent in institutional language.
Shorthand Builds Trust but Can Also Form Closed Circles
Long-term collaborators constantly use shorthand. When maintenance workers say "the east-entrance plan," they both already know a set of arrangements; when neighbors say "the old route," no repeated explanation is needed. Shared context saves time and also expresses shared memory. Demanding that everything be defined from zero each time would destroy actual collaboration.
A newcomer, however, may not know what a shorthand contains. If asking is taken as ignorance, in-group language converts historical participation into present qualification. The more familiar members are, the more obvious the meaning seems to them; the threshold thereby becomes invisible to those who possess it.
A closed group may also use gentle labels to cover conflict, calling exclusion "keeping rapport" and non-response "respecting the process." The words need not be devised, yet they can make it hard for members to point out consequences, because opposing a behavior is heard as opposing the relationship itself. On the other side, an outside observer may also misread the shorthand and declare, on the letter of it alone, that the group is concealing something. Judgment requires looking at how members actually use the terms, whether dissent can change a meaning, and whether newcomers can obtain explanations. Shared context is not evidence of guilt; only context that cannot be explained and benefits the inside alone constitutes a reason for further examination.
Keeping the Interpretive Gap Returnable
Language cannot eliminate the interpretive gap, but public institutions can decide how the gap is handled. The record of the bridge can be layered: the overview states the general state of passage, the task sheet states the specific entrances and objects, the raw reports preserve the on-site narratives. Each layer compresses on its own, and each can be traced back along its sources. Layered recording is not the permanent preservation of all material. Personal information irrelevant to public judgment should be restricted, and expired material may be dealt with by rule. The point here is that when a summary is challenged, it cannot be proven by the summary alone; and after a correction, the positions that depended on the old wording must learn of the change.
Responses should aim at the level of the statement. When a resident describes one failure, the receiver can first acknowledge the event and then check whether it recurs and whether other entrances are affected; there is no need to accept at once the global judgment that "the bridge is entirely unusable," but neither should the local fact be rejected because the global judgment overreaches. Only when fact, interpretation, and action are kept distinct will language stop forcing every reporter to carry the entire argument at once.
Letting Those Concerned Examine How Their Experience Is Summarized
The liaison can return the task summary to the resident: "I have recorded that the east entrance this morning did not allow this kind of stroller to turn; have I omitted a key condition?" Such confirmation does not give the resident the final engineering judgment, but it can catch serious rewriting before the information enters the next link.
Confirmation must not become endless approval. In an emergency, action may come before verification, and a simple report need not be signed back and forth. The intensity should be proportionate to the consequences and to irreversibility. The more a summary can affect the allocation of resources, the assignment of responsibility, or the standing of subjects, the more its source needs to know how it is being represented.
The interpreter should also state her own added judgments. If the liaison believes the west entrance offers an acceptable alternative, that should be written as an assessment rather than mixed into the resident's words. The dispute can then target whether the alternative is bearable, without first having to prove who distorted whom. The person concerned may also refuse any summary and accept only complete repetition. Public action must still choose its scale of expression. The legitimate limit is to give reasons, to preserve substantive dissent, and to allow specific omissions to be reopened when their consequences appear, not to promise everyone control over the entire public text.
Correcting Language, and Correcting the Arrangements Language Made
Changing the minutes from "convenience concern" to "the east side is unusable for a certain kind of stroller" is only the first step. If the old classification has already closed the task, left the alternative entrance unmarked, or left the relevant hours unanswered, the correction must return to those arrangements as well. Otherwise language receives moral repair while reality stays where it was. Correction need not declare that everyone before acted in bad faith. The summary may have been produced under time pressure or with insufficient information. Responsibility depends on whether the loss was foreseeable, whether it could be repaired once discovered, and who keeps control of the official formulation. The absence of malice can change the form of blame; it cannot make the consequences for those affected disappear on their own.
Language also cannot guarantee the success of action. A new description may still omit other users, and a temporary arrangement may still be constrained by conditions. Responsible public language does not promise ultimate completeness; it makes it possible for omissions to be said again and to have a chance of reaching a position from which arrangements can be changed. From this the chapter draws an affirmative conclusion: the public value of language lies not in lossless copying of experience but in establishing returnable compression. Words must suffice to connect dispersed action; summaries must preserve their relation to sources; naming must remain correctable by its object and its consequences. The interpretive gap will not disappear, but it need not be permanently occupied by one position.
Returnable also means preserving the differences between versions. If a later text modifies the meaning of "normal passage," it should state what condition has been added, rather than letting the new definition quietly replace the old commitment. Version traces let us distinguish what was reasonably understood at the time, what was learned later, and what is now being decided, and they prevent us from reading yesterday's responsibility backward out of today's word meanings.
The common world of civilization forms precisely within this incomplete exchange. It is neither a final dictionary that everyone must share nor a set of private worlds untranslatable into one another. It requires that we use abstractions and can still point out whom they omit; that we accept mediation and do not let mediation monopolize reality; that we allow provisional formulations to stabilize action while preserving the power of correction to change it.
The Responsibility of Translation Cannot Fall to Speakers Alone
If a resident must first master the languages of engineering, administration, and rights before her experience is accepted, the public institution has transferred its own cost of classification onto the source of the material. The receiver should help convert ordinary narrative into processable objects and return key summaries for confirmation. This responsibility has limits: staff cannot complete the entire argument for each person, and speakers should also state, as far as they can, the event, the time, and the change they seek.
Translation can also occur between groups. A language of justice familiar to one group may point to a different history in another. Judging the other side malicious outright escalates a conflict of word meanings into a conflict over the standing of subjects; pretending that all sayings are the same erases real differences in value. Returnable translation preserves the original words, states the working definitions it adopts, and allows consequences to modify those definitions in return. Whoever holds the final administrative decision bears the higher duty of explanation.
The next chapter returns to the concept "civilization," which is larger than every word in this chapter, to examine which phenomena it has actually helped us connect and which differences it may have compressed into an overly tidy whole. This chapter completes only the premise: every grand concept must answer for its own path of compression.