Democracy Beyond a Common Language - First Discussion of a Generative Relational Theory of Multilingual Democracy

ENGLISH

Democracy Beyond a Common Language

First Discussion of a Generative Relational Theory of Multilingual Democracy

Wanhong Huang · huangwanhong@serendip.ngo

---

Abstract

Democratic theory has generally taken a shared public sphere as a condition of legitimacy, and a shared public sphere to require a common language. The European Union removes the appearance of naturalness that the nation-state gave that assumption. The standing responses divide between a lingua franca with compensating redistribution and a pluralism of protected language rights; this paper argues that both rest on a premise they share, that democratic capacity is a property distributed over individuals, and that the deadlock between them follows from the premise rather than from any disagreement about it. It proposes instead that democratic capacity resides in the arrangement between participants, so that commonality of meaning is a product of democratic relation and not its precondition, and translation is a mechanism by which a common meaning is generated rather than a transfer of meaning across a boundary. Neither the presence of power nor the alienation of a claim into terms not its author’s can serve as the criterion of failure, since a standard indicting either would indict every arrangement. What the paper identifies instead is closure: the point at which an arrangement ceases to generate the means of its own revision. A candidate criterion is offered, with alternatives noted and not compared, together with a set of questions the argument opens and does not settle.

Keywords: multilingual democracy; linguistic justice; translation; relational generation; interpretive systems; European Union; deliberation.

A note on the standing of this paper. This is a first discussion, and it is offered as one. The argument is set down at the stage where it has become statable rather than at the stage where it has been secured: its central notions, interpretive system, conversion closure, the generative and assimilative forms of translation, are defined well enough to be argued against and not well enough to be applied by a procedure, and §8 sets out at some length the questions the argument opens and does not answer. Publication at this stage is deliberate, on the view that a criterion stated early and contested is worth more than one refined in private, but the reader should hold the whole to the standard appropriate to a first statement. It is also a work of political theory, not of European policy analysis. The European Union appears throughout as the setting in which an assumption of democratic theory becomes visible and can be tested, not as the object to be assessed; no verdict on the Union’s language regime is intended, and where its practice is described the description is meant to be recognizable to those who study it rather than novel to them. The order of exposition follows the order of the argument: the standing positions are set out first and at their strongest, and the relational account is introduced only as what they leave unaddressed. Should the existing literature on plurilingual competence and democratic translation be shown already to supply the criterion sought here, the later sections lose their ground.

A word on the summaries of others’ work. The positions set out in §2 are compressed, in some cases from books, and compression distorts. Where a position is summarized here it is summarized for the single feature the argument turns on, which is not the feature its author would foreground; readers should take the citations as pointers to what those authors argue and not the summaries as substitutes for it. Misattribution is the error the author would most regret and is the correction most urgently wanted.

Objection, correction, counter-evidence, and extension are all warmly welcome, and are the reason the paper is circulated in this form. Corrections of fact, on the European institutions, on the comparative cases, on the literatures the argument leans upon, are particularly wanted, as are objections that the central distinction does not hold, arguments that one of the open questions of §8 admits a settled answer the author has missed, and extensions of the account to cases it does not consider. Correspondence to huangwanhong@serendip.ngo.

§1 Introduction

Democratic theory has generally held that legitimacy requires a public sphere in which citizens can address one another, and has generally supposed that addressing one another requires a language held in common. The supposition is rarely defended, and the reason it is rarely defended is that it has rarely been visible. For the greater part of the theory’s history the polity in view was the nation-state, and the nation-state had already produced a dominant public language, through compulsory schooling, administrative practice, military service, and the terms on which political membership was extended, before the theory arrived to describe the result. Linguistic commonality thus presented itself to theory as a background condition of political life rather than as one of its products. It is worth being exact about the historical claim, since an overstated version invites a correct objection: no nation-state was ever linguistically homogeneous in fact, and the internal linguistic diversity of France, Italy, and Spain at the moment of their consolidation was considerable. The claim is not that homogeneity obtained but that a dominant public language was institutionally produced and thereafter appeared natural. That appearance is what made the assumption invisible, and the appearance is itself a governance achievement.

The European Union removes the appearance. A polity of twenty-four official languages, several hundred regional and migrant languages, a working practice that leans heavily on English, and an elaborate apparatus of interpretation and translation cannot present linguistic commonality as a background given. It presents it as a problem, and the problem has a recognized political expression: among the complaints assembled under the heading of the democratic deficit, the linguistic complaint recurs, namely that a public sphere conducted in a language most citizens did not grow up in is not their public sphere, and that a public sphere conducted in twenty-four is not one public sphere at all.

The standing responses to this problem divide, and divide sharply. One holds that a common vehicle is what a demos requires, accepts English in that role, and treats the resulting advantage to its native speakers as an injustice calling for redress in the form of transfers and of territorial protection for other languages in their own domains. Another holds that no such vehicle should be imposed, that languages carry rights, and that the just arrangement is one of recognition, accommodation, and institutional multilingualism. A third holds that the problem is insoluble as posed: absent a common public sphere there is no European people, and therefore no European democracy, only democracy in the member states.

This paper argues that the first two positions, for all that separates them, share a premise, and that the third draws its force from the same premise. The premise is that democratic capacity is a property distributed over individuals, a language one possesses, a right one holds, a competence one has acquired, so that the political question is how such properties should be allocated. Grant the premise and the disagreement follows: one may allocate a single language efficiently and compensate the losers, or allocate rights over many languages and protect the holders, and there is no third option, because on the premise there is nothing else to allocate. If this is right, the difficulty is not one that further argument between the positions would resolve, since it follows from something they agree on rather than from what they dispute.

Two bodies of existing work have already begun to refuse it, and the paper’s debt to them should be stated at the outset rather than discovered later. The Council of Europe’s framework documents describe plurilingual competence in terms of what a speaker can do across a repertoire that is partial, uneven, and receptive as well as productive, rather than in terms of parallel mastery of several languages. Work on democratic translation locates democratic capacity in the deliberative arrangement itself, in the presence of interpretation and in what interpretation does to the distribution of speaking. Neither move is claimed here as new. What is suggested, with the qualification that both literatures are large and the author will not have read all of either, is that neither has yet said what makes such an arrangement democratic, that is, what distinguishes an arrangement in which meaning is generated across difference from one in which it is imposed across difference, both being arrangements and both involving translation. That criterion is what the present account undertakes to supply.

The account may be stated as a single claim.

Central claim. Democratic capacity resides in the arrangement between participants rather than in an endowment they severally bring to it. A polity is democratic in the measure that its arrangements leave interpretations open to revision across linguistic difference, and fails in the measure that such revision is closed off. Commonality of meaning is accordingly not the precondition of democratic relation but among its products: a shared public language, where one exists, is what sustained democratic relation may generate, and not what it presupposes. Translation, on this account, is not the transfer of a meaning across a boundary but one mechanism by which a common meaning is generated, and its democratic significance lies in what it does to the arrangement rather than in the fidelity it achieves.

The claim unfolds in four steps. The first sets out the standing debate at its strongest and locates the premise the positions share. The second gives the notion of an interpretive system the argument requires, and does so restrictively, since the notion is the one on which an overextended version of the argument would founder. The third treats translation as a mechanism of generation rather than of transfer, distinguishes its generative from its assimilative form, and asks who holds the authority to certify a conversion. The fourth, the section on which the account stands or falls, separates the permanent conditions of relation from the point at which an arrangement closes around them, since a theory that can describe only success is not a theory of anything, and a theory that finds failure everywhere is not one either.

The remainder proceeds as follows. §2 sets out the no-demos thesis, the lingua franca solution, and the pluralist solution, and argues that the division among them is a consequence of a premise they share. §3 defines an interpretive system and states the asymmetry that default-language selection installs. §4 treats translation as generation. §5 distinguishes the ineliminable conditions of relation from the point of closure, and offers a candidate criterion. §6 tests the distinction against four settings outside Europe. §7 draws the institutional consequences, reads existing practice against them, and states the cost. §8 sets out the questions the argument opens and does not settle. A closing section marks the limits of the account.

§2 The Standing Debate and the Premise It Shares

Three positions occupy the field. They are set out here at their strongest, since an account that enters by weakening its opponents has not earned its entry.

2.1 The No-Demos Thesis

The first position holds that the preconditions of democratic legitimacy are not presently in place at the European level. Dieter Grimm’s is a careful statement of it (Grimm 1995). Arguing against constitutionalization of the Union at the time he wrote, Grimm held that a democratic polity requires a public sphere in which opinion is formed and mediated to political institutions, that such a sphere depends on media, parties, and associations operating across the polity, and that these were not then present at European level; among the obstacles he identified was the absence of a shared communicative infrastructure, language being one component of it. Two things should be kept in view, since the position is often reported in a stronger form than its author gave it. Grimm’s argument was directed at a particular proposal rather than at the possibility of European democracy as such, and he allowed that the conditions he found absent might develop over time. Language is one element in his account and not the whole of it; the summary here isolates the linguistic element because that is what the present paper is about, and the isolation is the paper’s doing rather than his. The thesis has been contested at length (Habermas 1995), and it is worth registering that it has been worked around more often than refuted: several of the proposals that answer it, a European public sphere emerging through networked national ones, or a demos constituted politically rather than culturally, accept a good deal of Grimm’s description of what was absent and argue that something else may serve in its place.

2.2 The Lingua Franca Solution

The second position accepts that a common vehicle is required and asks what justice demands once one is adopted. Philippe Van Parijs’s is the developed statement (Van Parijs 2011). On his account English is spreading as a European lingua franca through a dynamic no one designed, and its spread is on balance to be welcomed, in part because a shared language gives the worse off a means of coordinating with one another. He identifies two forms of unfairness that accompany it, a free ride enjoyed by those whose first language others must pay to learn, and pressure on the domains other languages occupy, and proposes to address them directly rather than by resisting the spread: a linguistic tax addressed to the first, and a territoriality principle, under which each language is protected within its own region, addressed to the second. The summary is compressed and the argument is developed at length in the book.

It should be recorded that this position is the one the present paper has most reason to take seriously, and not a foil. Van Parijs does not treat generalized plurilingualism as the remedy, and his reasons are not ones an appeal to the value of diversity meets: on his account, expecting everyone to acquire several languages is costly, and the cost of language learning does not fall evenly, so that a policy of general plurilingualism can bear hardest on those with least. Any argument that democratic capacity is better located in plurilingual repertoires must answer this, and answering it by asserting the intrinsic worth of linguistic diversity does not meet the objection, which was not about worth.

2.3 The Pluralist Solution

The third position declines the common vehicle and elaborates what justice owes to languages and their speakers. It is a family of views rather than one, and its members differ. Alan Patten has argued for a principle of equal recognition, on which the state extends comparable support to the languages of its citizens, and has set out the difficulties of the alternatives, including linguistic neutrality on the model of religious disestablishment (Patten 2003, 2014); Denise Réaume has argued that a language may have value to its speakers that a purely instrumental account misses, and that official-language rights answer to that value (Réaume 2000). They do not agree on the grounds, and the summary here is concerned only with what the family shares: that the arrangements justice recommends are those of institutional multilingualism, in the form of official status, provision in several languages, and protection for regional and minority languages. A reader wanting the disagreements among these authors will not find them here.

The characteristic difficulty of this position is not injustice but parallelism. Recognition and provision may be extended fully to every group and leave the groups beside one another rather than in relation, coexisting under a common institutional roof, each conducting its own public life, with no process in which a common meaning is generated between them. It is not obvious how such an outcome would be named as a failure in terms of recognition and provision alone, since on those terms nothing has failed: every right is honored and every language provided for. This is a remark about a criterion and not about what any of these authors believes; nothing in the works cited commits them to indifference toward the outcome. That it nonetheless looks like a failure of democracy rather than of justice is a datum the position, on the reading offered here, has no obvious place for; a defender of it may well reply that democracy was never what the theory undertook to supply.

2.4 The Hidden Premise

The two constructive positions oppose one another on what should be allocated. They agree on the prior question of what kind of thing is at stake. In both, democratic capacity is a property distributed over individuals: for the lingua franca solution, a competence in the shared vehicle, unequally costly to acquire and hence calling for compensation; for the pluralist solution, a right held in respect of one’s language, unequally honored and hence calling for recognition. Both are distributive theories over individually held goods, and their disagreement is a disagreement about the distribuendum.

This may be why the exchange has been durable. Grant that democratic capacity is individually held and the options are exhausted by the two: hold one language in common and redress the cost, or hold many and protect each. There is no third arrangement to argue for, because the premise has already fixed what an arrangement can be, namely a distribution. The no-demos thesis draws its force from the same source: it observes that the individually held goods do not add up to a public sphere, which is correct, and infers that no public sphere is available, which follows only if a public sphere must be constituted out of individually held goods.

Claim 1. The division between the lingua franca and pluralist solutions is a consequence of a premise the two share, that democratic capacity is a property distributed over individuals, rather than a disagreement about that premise. So long as the premise holds, the positions exhaust the field, and the no-demos thesis follows from observing that neither distribution constitutes a public sphere. If that is right, the way out is not a better distribution but a different location for democratic capacity, in the arrangement between participants rather than in what they severally hold. Whether the relocation succeeds is what the rest of the paper is for, and the diagnosis offered here may be wrong about the premise without either position being wrong about anything else.

2.5 The Existing Displacements of the Premise

Two existing bodies of work have made part of this move, and the account that follows is a continuation of them rather than a departure. The Council of Europe’s framework documents describe plurilingual competence in terms of a single interrelated repertoire on which a speaker draws, in which competences may be partial and uneven and in which receptive skill in a language one does not speak counts as competence (Council of Europe 2001, 2020). That description locates competence in what a speaker can do across a repertoire rather than in a stock of separately mastered languages. The reading offered here, that this is formally a relational rather than a bearer-shaped notion, is the present author’s and is not a claim made in those documents, which are instruments for language education and were not written to settle questions in political theory. Peter Kraus has developed the political implications of European plurilingualism at greater length (Kraus 2008). Work on multilingual deliberation locates democratic quality in the arrangement itself: Nicole Doerr’s studies of transnational social forums examine what happens to deliberation when interpretation is present (Doerr 2012, 2018). A wider literature on linguistic justice, surveyed by Alcalde (Alcalde 2018), takes up the institutional questions from several disciplinary directions.

Neither, however, states what makes such an arrangement democratic. A plurilingual repertoire may be put to work under domination as readily as under democracy; the colonial subject fluent in the metropolitan language has a repertoire and operates across it. Translation is present in imperial administration as much as in deliberation. That an arrangement is multilingual, and that its participants operate across repertoires, does not by itself distinguish it from an arrangement in which meaning is imposed across difference. Supplying that distinction is the work of §5, and the two sections before it prepare the terms in which it can be stated.

§3 Language as an Interpretive System

3.1 Against the Instrumental View

The instrumental view treats a language as a channel: a thought is formed, encoded, transmitted, and decoded, and the choice among channels is a choice among efficiencies. On this view the selection of a public language is a coordination problem with a distributive residue, which is exactly how the lingua franca solution treats it, and the treatment is coherent.

What the instrumental view cannot register is that the selection determines which experiences are readily expressible at low cost, which concepts are already available for public use, and which participants bear the burden of formulating what they mean in terms not their own. These are not effects on the transmission of a message. They are conditions on what can be said at all in a given public setting, and they are unequally distributed by the selection itself rather than by anything the participants do afterward.

3.2 The Definition of an Interpretive System

The argument requires a notion more precise than “language,” and the notion is stated restrictively here because the overextended version is where an argument of this kind fails.

Definition. An interpretive system is a practice that imposes conditions on what counts as a valid continuation, such that a competent participant can be wrong within it, and wrong by the practice’s own standards rather than by an observer’s.

Two features of the definition do the work. The first is normativity: an interpretive system is not merely a set of habits or a style but a practice with standards, so that a move within it may be corrected, and corrected by other participants who need not agree about anything else. The second is that the standards are internal: what makes a continuation valid is fixed by the practice and not by its consequences elsewhere.

Natural languages satisfy this straightforwardly. Whether other symbolic practices satisfy it, disciplinary registers, legal argument, mathematical notation, professional idioms, the codified registers of bodily and affective expression, is a real question and is left open here. Some plainly do; others may satisfy it only under specific conditions; and the difference between mutual unintelligibility among natural languages and the difficulty of moving between a legal and a lay register is a difference of kind that a hasty generalization would efface. The definition is offered so that the argument’s later claims are checkable, and so that any extension beyond natural language must be earned case by case. Nothing in what follows depends on the extension.

3.3 Asymmetry and Its Absorption

The selection of a default language installs a standing asymmetry: some participants formulate at low cost in a system that is theirs, others at high cost in a system that is not. This asymmetry is not eliminable by good intentions and is not eliminable by the removal of any particular language from the default position, since some default will obtain, if only by attrition.

The framework’s position on asymmetry is that asymmetry always generates power, and that the question is never whether power is present but whether the arrangement continues to absorb it through generation or lets it consolidate into a fixed grammar. Applied here: the presence of a dominant working language is not by itself the failure. The failure is the arrangement in which the dominance ceases to be revisable, in which the question of who bears the burden of formulation can no longer itself be raised in the public sphere whose language is at issue. What distinguishes the two cases is the subject of §5.

§4 Translation as Generation

4.1 Against the Transfer Model

The transfer model treats translation as the conveyance of a meaning from one language to another, with fidelity as its standard and loss as its characteristic failure. As an account of what a translator is professionally obliged to attempt, this is unobjectionable. As an account of what happens to political concepts in a multilingual polity, it misdescribes.

A concept entering another interpretive system is subject to that system’s conditions on valid continuation. It acquires the inferences that system licenses and loses those it does not; it acquires neighbors, connotations, and a history it did not have. The result is not the same meaning in a new dress, nor is it simply a degraded copy of the original. It is a new item in a new system, standing in some relation to the first, and the relation is one the participants must work out rather than read off.

4.2 The Union’s Own Evidence

The European vocabulary supplies instances, and they are documentary rather than illustrative. Solidarity, solidarité, and Solidarität carry different institutional and doctrinal freight; subsidiarity has a Catholic social-teaching lineage in its German and Italian reception that its English use does not carry; sovereignty is not the same item in a French constitutional register as in a British parliamentary one. These terms now have European senses. Those senses were not found by better translation, and they are not the intersection of the national senses. They were produced through institutional practice, litigation, negotiation, drafting, the accumulated interpretation of texts that exist equally in all versions, over decades, and they are still moving.

The legal case is the sharpest. Union law knows no original: every language version is equally authentic, and where the versions diverge the Court interprets by reference to the purpose and general scheme of the instrument. There is, formally, no meaning behind the versions of which they are renderings. The meaning is what the interpretive practice across the versions produces. Whatever one thinks of the doctrine, it is an institution built on the denial of the transfer model.

4.3 The Empirical Result

One empirical finding bears directly on the claim, and it should be given with its limits. Doerr’s studies of multilingual deliberation in transnational social forums argue, on the basis of extended observation of those settings, that deliberation conducted with volunteer interpretation could in certain respects be more inclusive than deliberation conducted in English alone: interpretation slows the exchange, interrupts the advantage of the fluent and the confident, requires that a contribution be restated, and creates a position, the interpreter, from which a contribution may be clarified before it is answered (Doerr 2012, 2018).

The limits are substantial and should not be minimized. The setting is a social-movement forum with participants selected by their commitment to the process, not a legislature or an electorate; the interpreters were politically engaged volunteers, not neutral professionals, and that is part of the mechanism rather than a flaw in it; the finding has not been reproduced at the scale of formal institutions, where the incentives differ. What her work supports is narrower than the paper’s thesis and is nonetheless the right shape: that the presence of translation in an arrangement can change the distribution of effective participation, and can change it in the direction opposite to the one the efficiency argument predicts. The efficiency argument holds that translation is friction and that friction is cost. Here the friction was doing democratic work.

Claim 2. Translation in a multilingual polity is not the conveyance of a meaning across a boundary but one mechanism by which a common meaning is generated, since a concept entering another interpretive system is re-specified by that system’s conditions on valid continuation. Its democratic significance lies accordingly in what it does to the arrangement, to the distribution of the burden of formulation and to the occasions on which a contribution may be restated and revised, rather than in the fidelity it achieves.

Two things this claim does not license. It does not dissolve untranslatability: that some items resist re-specification, and that the resistance is itself informative, is compatible with the account and possibly necessary to it, and no position is taken here on the philosophical literature that has made untranslatability its subject (Cassin 2014). And it does not make translation sufficient. Translation is present in arrangements no one would call democratic, and the imperial administrations were, among other things, formidable translating institutions. What distinguishes the cases is the subject of the next two subsections.

4.4 Generative and Assimilative Translation

Two operations answer to the word translation and stand in different relations to the criterion of this paper.

In the first, both systems move. A concept entering another interpretive system is re-specified by it, and the receiving system is in turn altered by what it has received: a category is stretched, a distinction is introduced, a term acquires an application it did not have. The European legal vocabulary of §4.2 exhibits this, unevenly: subsidiarity entering English legal usage changed the receiving usage as much as the English usage changed the term. Call this generative translation: the relation is symmetrical in the specific sense that both systems are liable to revision by the encounter.

In the second, one system moves and the other does not. A claim formulated in one interpretive system is rendered into the categories of another, becomes politically legible in the process, and the receiving categories are unaffected. The claim enters the public sphere; the terms on which it entered do not enter with it. Call this assimilative translation. Its mark is not inaccuracy, an assimilative rendering may be scrupulous by every professional standard, but the direction of the liability to revision, which runs one way.

The distinction is not a distinction between good and bad translation, and it is not decided by the translator’s care. It is a property of the arrangement in which the translating occurs, and specifically of whether the receiving system stands to be altered by what it receives. This is what allows the obvious objection to the whole account, that the Union already has translation, and at great expense, to be answered rather than deflected. Translation infrastructure is necessary and does not settle the question. What settles it is whether the infrastructure permits peripheral systems to alter the central one, or only permits them to enter it.

4.5 Interpretive Authority

Behind the distinction stands a question the transfer model does not raise: who performs the conversion, and under whose conditions of validity.

In a regime of equal authenticity no version is the original, and formally no one owns the meaning. In practice the rendering is certified: by a translation service working to established equivalences, by a drafting committee operating in a working language, by a court construing divergent versions, by a bilingual stratum whose competence in both systems makes it the passage between them. Certification is not a clerical function. Whoever certifies determines which formulation counts as what was said, and therefore which of the possible renderings becomes the claim that the institution answers.

This has a consequence for the claimant that should be stated plainly, since it is the mechanism by which conversion becomes a wrong rather than a cost. The claim returns to its author in a form authored elsewhere, and he is required to accept it as his own in order to be answered at all. His standing as the person who said it has been separated from his authority over what was said. Where the arrangement admits that separation, where he may say that is not what I meant and be heard as making a claim rather than as obstructing a procedure, the conversion remains an operation he is party to. Where the arrangement does not admit it, he has been made the author of a statement he did not make. The framework’s general account of subject production applies here without adjustment: the arrangement has produced the participant it required, and the question is whether what it produced can still revise it.

§5 The Conditions of Relation and the Point of Closure

An account of democratic arrangement that describes only the successful case describes nothing, since the successful case is recognized in advance. This section identifies what goes wrong. It begins, however, with what does not: a great deal of what a critical account would be tempted to indict is an ordinary condition of relation, and a criterion that indicted it would sort nothing.

The section accordingly runs from the conditions to the closure. The first subsection establishes that neither the presence of power nor the presence of alienation can serve as the criterion. The next two describe arrangements that exhibit both and remain open. The fourth identifies the point at which the arrangement closes, which is the failure the account names. Two further forms are then distinguished, and a candidate criterion is offered.

5.1 The Ineliminability of Asymmetry

Every relation is asymmetric in some respect, and asymmetry generates power: this is the framework’s standing position and it is not a concession made reluctantly here. Two parties do not stand in relation without one of them being, in some dimension, better placed: more fluent, better resourced, more practiced, or closer to the institution that will decide. The dimensions are many and no party is advantaged in all of them, but the sum is never zero, and an arrangement that eliminated it would be an arrangement in which nothing was being done.

Alienation is likewise a consequence of the symbolic as such and not a defect of particular arrangements. To state a claim is to commit it to a system of signs that is not the claimant’s private possession, whose terms carry histories he did not author and inferences he did not intend, and which will be understood by others under conditions he does not control. This is the condition of saying anything at all to anyone. It obtains in the monolingual case as fully as in the multilingual one; what the multilingual case does is make it visible, by giving the moment of commitment a name, translation, and a location.

The consequence for the criterion is decisive. If the presence of power made an arrangement unjust, every arrangement would be unjust; if the presence of alienation made it so, every act of speech would be. A criterion that condemns everything discriminates among nothing, and in practice licenses whatever the person applying it was already disposed to condemn. The account here therefore does not ask whether power is present in a conversion, nor whether something of the claimant was lost in it. Both are always true.

What it asks is what becomes of the power once generated. An asymmetry may be absorbed, taken up by an arrangement that continues to generate, and continues thereby to produce occasions on which the asymmetry can be pressed against, or it may consolidate, settling into a fixed grammar that the arrangement no longer revises. The difference is not a difference in the quantity of power. It is a difference in whether generation continues.

One further point belongs here, since it removes a temptation the rest of the section would otherwise invite. The norms under which a conversion is assessed, what counts as an adequate rendering, which categories are available, what may be pressed and by whom, are not imported from outside the relational system they govern. They are generated within it, by the same process whose products they then assess. There is accordingly no neutral standpoint from which a conversion could be certified correct, and the search for one produces the two fantasies this paper has been resisting: a faithful translation that answers to a meaning behind all versions, and a language-neutral public sphere in which no one bears the burden of formulation. The norms are products. The question is only whether they remain answerable to what they govern, or become premises that what they govern can no longer reach.

5.2 Exclusion

In the first mode a participant cannot enter at all: the public language is one he does not have, no interpretation is provided, and the arrangement offers him no position from which to speak. This is the case the existing literature handles well, and both standing positions of §2 address it, the one by universalizing competence in the shared vehicle, the other by extending official status and provision. It is entered here for completeness and not argued, since an account whose contribution lay in noticing exclusion would have no contribution.

5.3 Conversion With an Admitted Remainder

In the second mode entry is available and requires conversion. A claim formulated in one interpretive system must be rendered into the categories of the system in which decisions are made, and the rendering does not carry everything: some inference the original licensed is unavailable in the receiving categories, some distinction it turned on is one those categories do not draw. Call what fails to carry the remainder.

The burden here is real and it is asymmetric. Whoever must convert bears a cost that whoever formulates natively does not: the labor of finding a rendering, the delay it introduces, the risk that the rendering will be answered rather than the claim. The distribution of that burden is a matter of justice and it is the subject of a substantial literature.

It is not yet, however, an injustice of the kind this paper is concerned with, and saying so is necessary if the concept is to do any work. A conversion whose remainder is admitted, where the claimant retains the standing to say that is not what I said, and to have the saying of it treated as a claim about the matter rather than as an objection to the procedure, is an operation he remains party to. It is costly, it is unequal, and it leaves the arrangement open. Conversion as such is not the wrong. If it were, every act of political translation would be one, and the concept would sort nothing.

5.4 Conversion Closure

In the third mode the conversion is treated as complete. The remainder is not admitted as a remainder: what did not carry is not held to exist, the rendering counts without residue as what was said, and the converted claim cannot revise the categories that received it. The claimant may repeat himself, and the repetition is converted again by the same operation. The arrangement has closed around the conversion.

This is the closure, and it should be described accurately: not a strong party wronging a weak one, but an arrangement that received something and generated nothing from having received it. Domination may be present and often is; it is not what the criterion identifies, since it is present in the open cases too. The instance that displays it most cleanly is the encounter between an indigenous claim and a settler legal system. A relation to land involving ancestry, obligation, and collective standing is admitted to the proceeding only after being asked which of the available categories it is: property, in which case it must have an owner and a boundary; environmental interest, in which case it must be balanced against development; or religious freedom, in which case it is a belief held by individuals and protected as such. Each rendering makes the claim legible. Each also decides in advance the question the claim was raising, since what was at issue was precisely whether the relation is the kind of thing that has owners, or is a belief. The claim is heard, converted, and answered, and the categories that answered it are not altered by having done so. Nor does the proceeding generate any occasion on which the party who brought the claim could make them alter. Nothing was excluded. The proceeding was scrupulous. And the interpretive system in which the decision was made emerged from the encounter exactly as it entered.

The distinction from the second mode is the presence or absence of that alteration, and the remedies differ accordingly. Conversion with an admitted remainder is remediable within the arrangement: the burden can be redistributed, the sequencing changed, the drafting stage brought into the multilingual regime, the remainder given a procedural place. Conversion closure is not remediable by those means, because the means by which one would argue for the remedy have been made subject to the same conversion.

The relation to Fricker’s hermeneutical injustice should be stated rather than left for a reader to notice (Fricker 2007). Fricker locates a wrong in a gap in the collective interpretive resources, so that a person’s experience is unintelligible because the concept that would render it does not exist, and the wrong may obtain with no one at fault. The wrong described here is not a gap. The receiving categories are fully formed and perfectly adequate to their own purposes; the claim is rendered into them without difficulty. What is closed is the arrangement’s liability to revision by what it received, and closure of that kind is maintained by someone, in the ordinary sense that a court’s construction, a drafting convention, or a standing equivalence is maintained. The two accounts are neighbors and are not the same, and the difference is that one names an absence and the other names a closure.

5.5 Symbolic Closure

A related form deserves naming because it operates within a single system as well as between two. A term may cease to be revisable and begin to function as a test of belonging, no longer a concept whose application is arguable but a marker whose acceptance signals membership and whose questioning signals exteriority. Ever closer union, on some usages; European values, on some usages; and, in the other direction, sovereignty as it functions in certain national registers. The mark of closure is that the term’s application can no longer be argued and its use is instead a signal of position, so that a participant who questions it is heard as declaring an allegiance rather than as making a claim.

Closure of this kind is not a linguistic phenomenon. It occurs within a single language as readily as across several, and noting so is important to the account’s honesty: multilingualism is neither a sufficient cause of the failures described here nor a protection against them.

5.6 Parallelism

The pluralist arrangement’s characteristic outcome, groups fully provided for and standing beside one another, should be entered as a distinct failure and not as a mild success. What is absent is not provision but contact: no process in which an interpretation formed in one group can be answered by another, and hence no process in which a common meaning is generated. Parallelism is in this respect the mirror of conversion closure: there, a claim is converted and the receiving categories are unaltered; here, no conversion is attempted and there is nothing for anyone to alter. On the criterion below, parallelism fails not because anyone is excluded but because the arrangement has no site at which revision could occur.

5.7 Toward a Criterion

What the preceding subsections support is narrower than a criterion and is offered as such. They establish where the failure is not located, not in the presence of power, not in the alienation of a claim into terms not its author’s, and they identify a point at which something does go wrong. What they do not establish is that the following formulation is the right way to state that point, or the only one. It is a candidate, and the alternatives to it are set out after.

Claim 3 (a candidate criterion). Neither the presence of power in an arrangement nor the alienation of a claim into terms not its author’s is the failure; both are conditions of relation, and a criterion indicting them would indict every arrangement. What may be proposed in their place is closure: the point at which an arrangement ceases to generate the means of its own revision, so that the interpretive norms it has produced are no longer answerable to what they govern. An arrangement sustains generation in the measure that the act of interpreting within it itself produces the capacity to press against the power that the interpreting generates, so that what is required is not merely that revision is left formally unforeclosed, but that the occasions, the standing, and the resources for revision are among the arrangement’s own products. On this formulation a democracy fails not where some are disadvantaged in conversion, which is always, but where the arrangement receives a converted claim and generates nothing from having received it. The formulation is advanced for examination and not as a definition.

One qualification is owed before the criterion is compared with its neighbors, since without it the account acquires a commitment it does not want. Sustained generation is not an end that arrangements ought to pursue. Nothing here says that a polity should aim at generativity, that more generation is better than less, or that the generative arrangement is the good one; a criterion of that shape would install a telos and would be answerable for why anyone should want it. The claim is negative and narrower: generation is the condition under which an arrangement’s own interpretive norms remain answerable to what they govern, and where it stops, the norms have become premises rather than products. An arrangement may be criticized on that ground without generativity being named as its purpose.

Several other formulations would fit the material of this section, and the paper has not compared them. One could build the criterion on the revisability of interpretive norms, which is close to what is said above but locates the condition in the norms rather than in the arrangement that produces them. One could build it on the distribution of the interpretive burden, which has the advantage of being measurable and the disadvantage of being a distributive criterion, with the difficulties §2 raised. One could build it on the capacity to reopen a settled rendering, which is narrower and might be operationalizable in a way the present formulation is not. One could build it on the preservation of generativity as such, which is more general and correspondingly harder to apply. These are not equivalent, they would sort some cases differently, and choosing among them is work this paper has not done. What is claimed is that the failure lies at the point of closure and not in the conditions that precede it; where exactly to place the criterion around that point is open.

Terminological note. The word generative is already at work in an established literature. Ron Eglash’s generative justice asks how the value produced by participants in a process is retained by them rather than captured by systems that did not produce it. The question here is whether the interpretive norms of an arrangement remain answerable to what they govern. The criterion floated above is left unnamed to keep the two apart.

The candidate should nonetheless be distinguished from a neighboring criterion with which it will be confused. Deliberative accounts in the Habermasian line ask whether the participants can reach a rationally motivated agreement, and specify conditions, among them inclusion, non-coercion, and the force of the better argument, under which such agreement would be reachable (Habermas 1996). The distinction being drawn here is not that such accounts ignore the problem of rendering a claim into terms a public can receive. Habermas’s later essay on religion in the public sphere treats that rendering as a reciprocal burden borne by religious and secular citizens alike (Habermas 2006); its subject is religious and not linguistic difference, and it is cited here only because the structure of the requirement it describes, that a claim be translated into publicly accessible terms and that the burden of doing so not fall on one side alone, is close to what a multilingual critic would press for. Whether the analogy holds is not argued here. The distinction lies elsewhere. The deliberative criterion asks after agreement within a space of reasons; the criterion here asks whether the space of reasons is itself open to revision by those deliberating within it. An arrangement can satisfy the first and fail the second: agreement may be reachable, unforced, and inclusive, within terms none of the participants can any longer contest. It is the second failure that the multilingual case makes visible, because in a multilingual polity the constitution of the space of reasons is not hidden: it is done in public, by choosing which languages the discussion occurs in.

§6 The Mechanism Outside Europe

The account has so far rested on two settings: the European institutions, and a single worked case of conversion closure drawn from settler-colonial adjudication. That is a narrow base for a claim about arrangements in general, and it invites a specific objection, that closure of the kind described in §5.4 is an artifact of settler legal systems, which have a distinctive interest in receiving indigenous claims without being altered by them, rather than a general feature of multilingual arrangements. This section takes up four further settings to see whether the distinction between an open and a closed conversion does any work in them. The treatments are brief and are written by someone who is not a specialist in any of these cases. They are offered to test whether a distinction discriminates, and not as characterizations of the polities concerned; where they conflict with the accounts of those who study these cases, the latter should be preferred.

6.1 Colonial Language Administration

Colonial administration is the clean case of closure, and it is worth stating why, since the reason is not the obvious one. Colonial regimes were not short of translation. They were formidable translating institutions: they commissioned grammars and dictionaries, trained interpreters, produced legal codes in the vernacular, and staffed courts in which local claims were heard at length. The failure was not that claims could not enter. It was that the categories into which they entered were not liable to alteration by what they received. A land relation was received as tenure, a customary obligation as contract or as custom in the technical sense the code assigned it, a religious authority as an office. Each rendering was administratively serviceable and each settled in advance the question the claim was raising. The interpretive norms governing the renderings were not generated within the relation between the parties; they arrived with one of them.

This is the case in which the distinction of §5 is least contestable, and for that reason it is the case that proves least. It shows that the mechanism is not confined to settler adjudication. It does not show that the mechanism discriminates among arrangements that are not obviously unjust, which is what a criterion has to do to be worth having.

6.2 Post-Soviet Language Legislation

The successor states to the Soviet Union present the harder case, and the account should be tested against it rather than around it. Several of them adopted language laws that established the titular language as the language of public administration, education, and public life, in circumstances where Russian had occupied that position and where large Russophone populations remained. These arrangements have been examined at length in the literature on minority protection, and the summary here is not offered as a contribution to it.

Two descriptions of these arrangements are available and both have something to them. On the first, the legislation responds to a prior settlement that the polity had no part in establishing: a language displaced from public life is restored to it, and on this description the earlier arrangement is the one whose norms were not generated within the polity they governed. On the second, the legislation places speakers who now form a minority in the position of converting their claims into a language and a set of categories they do not control, and on this description the relevant question is what those speakers face under the present arrangement rather than what obtained under the earlier one.

The account here does not settle between these, and it is important to say why rather than to appear to have overlooked the question. The candidate criterion of §5.7 asks whether an arrangement generates the means of its own revision. It does not ask, and cannot ask, whether a closure imposed today is warranted by a closure imposed yesterday, because that is a question about the justification of a settlement rather than about whether the settlement remains open. A reader who wants an answer to the first question will not find one here, and the honest report is that this is a limitation of the criterion and not a feature of it. What the criterion can say about these cases is narrow: an arrangement in which the terms of the language settlement can be raised, contested, and revised by those living under it stands differently from one in which they cannot, whatever the history that produced the settlement, which is a narrower observation than either of the two descriptions above and does not choose between them.

6.3 Quebec and Catalonia

Two cases in which a minority language within a larger state has been made the language of public life in its own territory. Both are frequently cited by the pluralist position as demonstrations that a territorial settlement can protect a language without excluding anyone, and both have been contested within the territory by those the settlement places at a disadvantage.

The interest of these cases for the present account lies in a feature they share and that the criterion picks out. In each, the language settlement is itself a standing item of political contention: it is legislated, litigated, amended, and campaigned on, and the arrangements have been substantially revised more than once within living memory. Whatever else is true of them, the terms are not settled beyond the reach of those living under them. That is what the criterion asks after, and it is worth noticing that the criterion is accordingly less troubled by these arrangements than a distributive account might be, since the burden of formulation in both cases falls unequally and by design. No verdict on either settlement is intended, and the observation is about what the criterion attends to rather than about whether the arrangements are justified. Whether this is a strength of the criterion or an indication that it is insufficiently demanding is a question the author does not regard as settled.

6.4 India

India is the case that most resists the paper’s framing and is included for that reason. The constitutional settlement recognizes a large number of languages, the three-language formula distributes the burden of formulation across schooling in a way no European arrangement approaches, and the arrangement has been revised repeatedly and under considerable pressure, including through the anti-Hindi agitations that reshaped it. By the criterion of §5.7 the arrangement looks unusually open.

It looks less so on other measures. The position English occupies in higher administration, higher education, and the higher judiciary has proved durable, and the relation between the recognized languages and those outside the recognized set has been discussed in terms that resemble what §5.4 described. Both observations can be true, and their coexistence is the useful lesson: an arrangement can be substantially open at one level and closed at another, and a criterion applied to a polity as a whole will return an answer that averages over conditions that ought to be assessed separately. This is a defect in how the criterion has been stated here and not an observation about the polity, which the paper is in no position to assess; it is recorded in §8 accordingly.

6.5 What the Cases Show

Three things, and the third is the least comfortable. The mechanism is not confined to settler adjudication: colonial administration exhibits it clearly, and did so while translating extensively, which is the point on which the account most needs an example. The distinction between an open and a closed settlement does discriminate among cases that are not obviously unjust, which the Quebec and Catalan cases show, and does so in a direction that a distributive account would not necessarily endorse. And the criterion is presently stated at a level of generality that lets it be applied to a polity when it should be applied to a stage, a forum, or a register, which the Indian case makes plain and which the European material had allowed the paper to overlook.

§7 Institutional Consequences, and the Cost

7.1 The Requirements the Candidate Would Impose

Taking the formulation of §5.7 provisionally, three requirements follow. Revisability: the arrangement must contain occasions at which its own terms, including its linguistic terms, can be raised as a matter for decision rather than settled by default. Translation as constitutive: the apparatus of interpretation belongs to the deliberative process rather than to its administration, which has the practical consequence that it must be present where drafting occurs and not only where voting occurs. Distribution of the burden of formulation: the question is not only whether all languages are available but who is required to formulate in a system not their own, and at which stage.

A fourth requirement is stronger than these and follows from the active clause of Claim 3. Generation of the countervailing capacity: it is not sufficient that revision be formally available, since availability without the resources to sustain contestation is availability to those who already have them. The arrangement must produce, as a byproduct of its own operation, the standing and the occasions by which the power its interpreting generates can be pressed against. What this looks like institutionally is not obvious and the account does not claim to have specified it; the equal authenticity of language versions is one instance, since it manufactures a standing to dispute a rendering that no party had to fight for, and the presence of interpretation at the drafting stage is another, since it creates occasions of restatement that would not otherwise exist. Both are cases of an arrangement generating what would otherwise have to be brought to it.

7.2 Existing Practice Read Against It

Some of the Union’s practice satisfies these and some does not, and the account is not improved by pretending otherwise. The equal authenticity of language versions in Union law is a genuine instance of the second requirement, and an unusual one: it builds the denial of a privileged original into the legal order itself. The rule that a Member State may address the institutions in its own language and receive a reply in it satisfies the letter of the first. Against this: the concentration of drafting in a working language, the sequencing by which documents appear in other languages after positions have formed, and the practical unavailability of interpretation at the committee and expert-group stages are, on the account here, failures of the second mode: conversion burdens imposed at the stage where the remainder still has somewhere to go, and rights honored at the stage where their exercise is least consequential. The status of regional and minority languages, which the Charter addresses (Council of Europe 1992) and which the Union’s own regime largely does not, is the point at which questions of conversion closure arise.

7.3 The Cost

The arrangement the candidate criterion would recommend is expensive, slow, and does not pay for itself by the measures normally applied. Interpretation and translation consume a large budget and a larger amount of time; multilingual drafting is slower than monolingual drafting and the difference is not marginal; and there is no demonstration that the resulting decisions are better by any independent standard. The efficiency objection is correct on its own terms, and the account does not dispute it. What it disputes is the sufficiency of those terms: the cost buys revisability, and revisability is not a deliberative efficiency but a condition on the arrangement being democratic at all. Stating the cost is preferable to letting a reader discover it, and a reader who judges the cost too high has understood the argument rather than missed it.

7.4 Answering the Standing Positions

To the no-demos thesis: the argument grants Grimm’s observation and denies his inference. What is absent in Europe is a pre-existing communicative commonality, and it is absent. The inference to the impossibility of European democracy holds only if commonality must precede democratic relation. On the account here it is generated by such relation, unevenly and slowly, which is what the trajectory of the European legal vocabulary in §4 exhibits. Whether it is being generated fast enough, or at all, in the political rather than the legal register is an empirical question this paper does not settle.

To the lingua franca solution: the disagreement is narrower than it appears and sharper for being narrow. Van Parijs’s redistribution addresses the cost of the asymmetry to those who bear it. It does not address whether the arrangement in which the asymmetry obtains remains revisable, and that is a prior question, since a compensated asymmetry that can no longer be raised as a question is conversion closure with a payment attached. The reply available to Van Parijs, that revisability is itself best secured by everyone’s being able to speak to everyone else, in whatever language, is a serious one, and the argument between the two positions turns on it. Nothing here refutes it.

To the pluralist solution: recognition and provision are necessary and are not sufficient, and parallelism is the demonstration.

Central proposition. Democracy does not require that citizens share a language. It requires arrangements under which a common meaning can continue to be generated across the languages they have, and under which the terms of that generation, including which language the generation occurs in, remain open to revision by those who must live under them. A polity that has achieved perfect linguistic commonality and can no longer put that commonality in question has satisfied the older criterion and failed this one. A polity that shares no language and whose arrangements keep the question open has failed the older criterion and may yet satisfy this one.

§8 Open Questions

The argument opens more than it settles, and setting the residue out is part of what a first discussion is for. The questions below are given in the order of their bearing on the account: the first three are ones the account requires an answer to and does not supply, the remainder are ones it makes askable and leaves standing. None is rhetorical, and for several the author’s inclination is recorded and should be read as an inclination.

8.1 The Authority to Certify a Conversion

§4.5 established that some party certifies which rendering counts as what was said, and that the certification is not clerical. It did not say who should hold that authority, and the obvious answers are each unsatisfactory. If the receiving institution certifies, the closure of §5.4 is built into the procedure. If the claimant certifies, no conversion can ever be completed and the institution cannot act, which is a different way of denying him a hearing. If a professional intermediary certifies, the question recurs one level up, since by what standard is the intermediary’s rendering assessed, and by whom.

A partial answer suggests itself and is not developed here: that the authority in question is not the authority to fix a rendering but the authority to reopen one, and that these may be held by different parties without contradiction. An institution may certify provisionally for the purpose of acting while the claimant retains the standing to reopen, which is close to what an appellate structure does with legal error. Whether that arrangement is stable, or whether the power to act simply absorbs the power to fix, is the open question.

8.2 The Distribution of the Interpretive Burden

Every multilingual interaction imposes a burden of formulation, explanation, and adaptation, and the account has said that its distribution is a matter of justice without saying what a just distribution would be. Symmetry is the wrong answer, since the burden cannot be symmetric where the numbers are not: a rule requiring each participant to meet every other halfway imposes vastly more on the member of a small language community than on the member of a large one, and the arithmetic is not a detail. Proportionality to numbers is the wrong answer in the other direction, since it reproduces the dominance it was meant to moderate.

What the account can say is that the burden is a cost of the arrangement rather than a private cost of being a minority speaker, and that arrangements differ in whether they treat it as such. What it cannot say is how much of it any party owes. This is the point at which the argument most obviously needs the resources of a distributive theory it has spent §2 declining to rely on, and the tension is real rather than apparent.

8.3 Revisability and the Cost of an Unsettled Question

The account recommends that linguistic arrangements remain open to revision. It should be recorded that minority-language communities have frequently and reasonably preferred the opposite: a fixed constitutional guarantee, a territorial settlement, an entrenchment that removes the question from ordinary politics. The preference is not confusion about their interests. A permanently revisable settlement is a permanently contestable one, and contestation is won by those with the resources to sustain it, which is not usually the minority.

This is the sharpest objection to the paper’s position and it has no answer here. Two lines are available and neither is developed. The first distinguishes the object of revision: the terms under which claims are heard may remain open while particular guarantees are entrenched, so that entrenchment is itself the outcome of a process that stays open. The active condition of Claim 3 bears on this and may answer part of it: what the minority community has reason to fear is not revisability but revisability without the resources to sustain contestation, which is exactly what the criterion refuses to count as sustained generation. An arrangement that generated the standing and the occasions for pressing a claim, rather than merely permitting them, would not present the community with the choice between entrenchment and exposure. Whether any arrangement does this reliably, and whether entrenchment is not simply the most dependable way of generating it, remains open. The second concedes the point for the case of a minority under pressure and restricts the account to arrangements among parties none of which is at risk of disappearing, a restriction that would sharply narrow its application, and one the author is reluctant to accept without seeing whether the first line can be made to work.

8.4 Fidelity, and Whether There Is an Original

Translation theory has organized itself around fidelity to an original. The present account, in treating meaning as re-specified by the receiving system’s conditions on valid continuation, sits uneasily with the notion of an original meaning standing behind the versions, and the Union’s own doctrine of equal authenticity sits uneasily with it too.

Two readings are available and the paper does not choose. On the first, there is no original, fidelity is a regulative fiction, and what looks like faithfulness is agreement among the parties that a rendering will be treated as what was said. On the second, the original is real but is a relation rather than a content, what the speaker was doing in his own system, and fidelity is possible in principle though never verifiable. The first is cleaner and risks making the claimant’s protest that his meaning was lost unintelligible, which would be a serious cost, since that protest is the phenomenon §5.4 was written to accommodate.

8.5 Whether Misunderstanding Is Always a Failure

The efficiency framing treats residual misunderstanding as loss. A different possibility is that the impossibility of complete conversion is among the conditions of generation rather than an obstacle to it: where no ready equivalent exists, the parties must make something, and what they make is new to both. The European legal vocabulary of §4.2 is at least consistent with this reading.

The account can distinguish the productive case from the destructive one only by the criterion it already has, whether the misunderstanding remains open to correction by the party who was misunderstood, and whether that distinction does enough work is unresolved. It plainly does not follow that misunderstanding is generally productive, and an account that celebrated it would be a bad account.

8.6 Transformation and Identity

Generative translation requires that the receiving system be altered by what it receives. But a system altered without limit is not the same system, and a minority language transformed by continuous accommodation to a dominant one is describing assimilation rather than generation. The account therefore requires a distinction between transformation that a system undergoes as itself and transformation that ends it, and does not have one.

This is not a question peculiar to language. The same problem arises for scientific traditions, institutions, and persons, and any general answer the framework gives will have to be given at that level rather than here. It is named as a question the present account presupposes an answer to and does not provide.

8.7 Shared Meaning or Shared Procedure

Procedural accounts hold that democracy needs fair rules and not agreement on meanings. The account here suggests that procedures are not independent of meanings, since a rule is applied under some interpretation, and that the question of which interpretation governs is not itself settled procedurally without regress.

Whether this is a serious difficulty for procedural accounts or a familiar one they have already absorbed is a question for those accounts’ defenders, and the author does not claim to have shown the former.

8.8 Untranslatability

If some concepts genuinely resist conversion, the question is whether this is an obstacle to democracy across difference or a condition of its being a relation at all. The inclination here is toward the second, a fully transparent polity would require no relation because there would be nothing to work out, but the inclination is not an argument, and stating it as though it were would be the kind of move the paper has criticized elsewhere. The philosophical literature on untranslatability has resources the account has not used (Cassin 2014).

8.9 Citizenship and Its Formation

If democratic capacity resides in arrangements rather than endowments, what a citizen must be able to do is unclear. Language competence is the traditional answer and is not obviously the right one; a capacity to sustain relation across interpretive systems is closer to what the account requires, and is not a skill any curriculum currently teaches or any instrument measures.

The developmental question that follows, how such a capacity is formed, whether it is formed differently under multilingual and monolingual rearing, whether a capacity formed in childhood transfers to political participation, is not addressed in this paper and is not addressed by way of promise either. The supposition that multilingual childhood is the developmental foundation of multilingual democracy is attractive, has very little evidence behind it, and reproduces a developmental teleology this framework declines elsewhere. It is a question, and it is a different paper.

8.10 Machine Translation

Automatic translation now handles a large share of the conversion this paper has discussed, and the question is whether it changes the problem or relocates it. The account’s expectation is relocation: a system that renders reliably still renders into categories, those categories are inherited from the corpus on which it was trained, and the corpus is not neutral among interpretive systems. Certification does not disappear when it is automated; it becomes harder to identify and harder to contest, since there is no party who can be asked why this rendering rather than that one. Whether the criterion of §5 can be applied to an arrangement whose certifying party is a model is genuinely unclear, and it is the question on this list most likely to become urgent soonest.

8.11 The Unit to Which the Criterion Applies

§6.4 exposed a defect that the European material had concealed. The criterion has been stated as a condition on an arrangement, and applied as though a polity were one arrangement. It is not. A settlement may be open at the level of constitutional politics and closed in the register in which administrative decisions are actually taken; a language may be revisable in the legislature and fixed in the higher judiciary. Applied to a polity as a whole the criterion averages over conditions that differ, and an average is not what a criterion should return.

The repair is presumably to specify the unit: a forum, a stage of a procedure, a register. Which unit, and whether the criterion behaves consistently across units of different scale, is not worked out here, and until it is, the applications in §7 should be read as applying to the stages they name rather than to the Union.

8.12 The Minimal Condition of a Generative Relation

The most general question the paper raises. The framework accepts asymmetry as a permanent feature of relation and holds that asymmetry alone is not domination. The criterion of §5 suggests a candidate for the difference, that an asymmetry becomes degenerative when the weaker party can no longer modify the structure producing it, and if that candidate holds it would apply well beyond language, to educational, institutional, and international arrangements alike.

Whether it holds is not established here. The multilingual case is one case, chosen because it makes the constitution of the interpretive conditions unusually visible, and a criterion that works in the case where the phenomenon is visible may fail where it is not. Testing it elsewhere is the natural continuation of this line and is not undertaken in this paper.

§9 Limits and Further Work

The account rests on claims that could be wrong, and stating them as such belongs to setting it down.

The empirical base is thin. The comparative treatments of §6 are brief to the point of being schematic, each compressing a literature into a paragraph, and a specialist in any of the four will find the compression unsatisfying; they are offered to test whether a distinction discriminates, not as accounts of the cases. One deliberative finding, from a movement setting with self-selected participants and politically committed volunteer interpreters, is the principal evidence that a multilingual arrangement can distribute participation more equally than a monolingual one; it has not been reproduced in formal institutions, and there is reason to expect that it might not be. The reading of European legal vocabulary as generated rather than translated is a reading, and a legal scholar might describe the same material as convergence on a meaning that was there to be found.

The candidate criterion awaits an operational form, and no comparison has been made among it and the alternatives noted in §5.7. To say that an arrangement is democratic in the measure that an interpretation can still revise the arrangement that produced it is to state a condition, not a test. What would count as evidence that revision remains available, short of an actual revision occurring, is not settled here, and the distinction between a conversion whose remainder is admitted and one that has closed is at present a distinction a careful reader can apply and not one a procedure can decide.

The extension of the account beyond natural language is not attempted. The definition in §3 was framed so that other symbolic practices might satisfy it, and whether disciplinary, professional, or affective registers do so, and what would follow politically if they did, is left open. The temptation to assert the extension is considerable and has been resisted, since the argument here is supported by the European material and the extension would not be.

Several of the questions of §8 are not merely unanswered but bear on whether the account is correct, in particular the objection recorded there that revisability is itself costly to those least able to sustain contestation. That objection has no answer in this paper, and the two lines sketched against it are sketches.

---

A note on method. No formal apparatus has been used. The candidate of §5.7 could be written as a condition on the reachability of certain states of an arrangement from within it, and closure as the loss of a class of transition; the notation would be exact and would derive nothing that the prose does not already state, while borrowing a rigor the argument has not earned. What is claimed is claimed in prose, and stands or falls there.

References

Alcalde, Javier (2018). Linguistic Justice: An Interdisciplinary Overview of the Literature. In Michele Gazzola, Torsten Templin, and Bengt-Arne Wickström (eds.), Language Policy and Linguistic Justice. Cham: Springer, 65–149.

Cassin, Barbara (ed.) (2014). Dictionary of Untranslatables: A Philosophical Lexicon. Trans. ed. Emily Apter, Jacques Lezra, and Michael Wood. Princeton: Princeton University Press. [French orig. 2004.]

Council of Europe (1992). European Charter for Regional or Minority Languages. ETS No. 148. Strasbourg.

Council of Europe (2001). Common European Framework of Reference for Languages: Learning, Teaching, Assessment. Cambridge: Cambridge University Press.

Council of Europe (2020). Common European Framework of Reference for Languages: Companion Volume. Strasbourg: Council of Europe Publishing.

Doerr, Nicole (2012). Translating Democracy: How Activists in the European Social Forum Practice Multilingual Deliberation. European Political Science Review, 4(3), 361–384.

Doerr, Nicole (2018). Political Translation: How Social Movement Democracies Survive. Cambridge: Cambridge University Press.

Fricker, Miranda (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press.

Grimm, Dieter (1995). Does Europe Need a Constitution? European Law Journal, 1(3), 282–302.

Habermas, Jürgen (1995). Remarks on Dieter Grimm’s “Does Europe Need a Constitution?” European Law Journal, 1(3), 303–307.

Habermas, Jürgen (1996). Between Facts and Norms: Contributions to a Discourse Theory of Law and Democracy. Trans. William Rehg. Cambridge, MA: MIT Press. [German orig. 1992.]

Habermas, Jürgen (2006). Religion in the Public Sphere. European Journal of Philosophy, 14(1), 1–25.

Kraus, Peter A. (2008). A Union of Diversity: Language, Identity and Polity-Building in Europe. Cambridge: Cambridge University Press.

Patten, Alan (2003). Liberal Neutrality and Language Policy. Philosophy & Public Affairs, 31(4), 356–386.

Patten, Alan (2014). Equal Recognition: The Moral Foundations of Minority Rights. Princeton: Princeton University Press.

Réaume, Denise G. (2000). Official-Language Rights: Intrinsic Value and the Protection of Difference. In Will Kymlicka and Wayne Norman (eds.), Citizenship in Diverse Societies. Oxford: Oxford University Press, 245–272.

Van Parijs, Philippe (2011). Linguistic Justice for Europe and for the World. Oxford: Oxford University Press.

中文

超越共同语言的民主

一种生成性关系的多语民主理论的初步讨论

黄万宏 · huangwanhong@serendip.ngo

---

摘要

民主理论一般把一个共享的公共领域当作正当性的一个条件，并把一个共享的公共领域当作要求一门共同语言。欧盟移除了民族国家赋予那一假设的自然性外观。现存的诸回应分为两派：一种带补偿性再分配的通用语，与一种受保护的语言权利的多元主义；本文论证两者都依据一个它们共享的前提，即民主能力是一种分布于诸个体之上的性质，而它们之间的僵局随那个前提而来、而非随任何关于它的分歧而来。它转而提议民主能力驻留在参与者之间的安排之中，从而意义的共同性是民主关系的一个产物、而非它的先决条件，而翻译是一个共同意义借之被生成的机制、而非意义跨一个边界的转移。无论是权力的在场，还是一个主张被异化进非其作者之术语，都无法充当失败的判据，因为一个指控任一者的标准会指控每一个安排。本文转而辨认的，是闭合：一个安排停止生成其自身修正之手段的那一点。一条候选判据被提供，连同被指出而未被比较的诸替代方案，以及一组论证所开启而不了结的问题。

关键词： 多语民主；语言正义；翻译；关系性生成；诠释系统；欧盟；协商。

关于本文定位的一则说明。 这是一次初步讨论，且作为一次初步讨论被提供。论证被放下在它已变得可陈述、而非在它已被确立的阶段：它的核心概念，诠释系统、转换闭合、翻译的生成性形式与同化性形式，被定义得足以被反驳，而不足以由一个程序来应用，而§8颇为详尽地铺陈论证所开启而不回答的诸问题。在这一阶段发表是刻意的，其看法是一条早早陈述并被争议的判据，比一条私下打磨的更有价值，但读者应以适合一次初次陈述的标准来衡量整体。它也是一部政治理论的著作，而非欧洲政策分析的著作。欧盟通篇作为一个民主理论的假设变得可见并能被检验的情境而出现，而非作为要被评估的对象；无意对欧盟的语言体制作任何裁断，而在描述其实践之处，该描述意在对研究它的人是可辨认的、而非对他们是新奇的。阐述的次序追随论证的次序：现存的诸立场先被铺陈、并以其最强的形式，而关系性的说明只作为它们所留下未加处理者被引入。倘若能证明关于多语能力与民主翻译的现存文献已然供出此处所寻的判据，后面的诸节便失去它们的立足之处。

关于对他人工作的概述的一个词。§2中所铺陈的诸立场被压缩，在某些情况下从整本书压缩而来，而压缩会扭曲。此处一个立场被概述时，是为论证所转于其上的那单一特征而被概述的，而那不是它的作者会置于前景的特征；读者应把诸引用当作指向那些作者所论证之物的指针，而不把诸概述当作它的替代品。误归是作者最会懊悔的错误，且是最迫切被想要的更正。

反对、更正、反证与扩展全都热忱欢迎，且是本文以这一形式流通的缘由。事实的更正，关于欧洲诸机构、关于诸比较案例、关于论证所倚靠的诸文献，是特别被想要的，关于那个中心区分不成立的诸反对、关于§8的诸开放问题之一容许一个作者错过的已定答案的诸论证，以及该说明对它未曾考虑之诸情形的扩展，亦然。来信请寄 huangwanhong@serendip.ngo。

§1 引言

民主理论一般主张正当性要求一个公民能在其中彼此致意的公共领域，并一般假定彼此致意要求一门共同持有的语言。这一假定鲜少被辩护，而它鲜少被辩护的缘由是它鲜少可见。在该理论历史的大部分中，视野中的政体是民族国家，而民族国家在该理论到来描述其结果之前，便已通过义务教育、行政实践、兵役，以及政治成员资格被扩展所依据的诸条款，产生了一门支配性的公共语言。语言的共同性因而向理论呈现为政治生活的一个背景条件、而非它的诸产物之一。就这一历史主张精确是值得的，因为一个被夸大的版本招致一个正确的反对：没有任何民族国家在事实上曾是语言上同质的，而法国、意大利与西班牙在其巩固之时的内部语言多样性是可观的。该主张不是同质性曾成立，而是一门支配性的公共语言曾被制度性地产生、并此后显得自然。那一显得，正是使该假设不可见的东西，而那一显得本身是一项治理成就。

欧盟移除了那一外观。一个有二十四门官方语言、数百门区域与移民语言、一种严重倚赖英语的工作实践，以及一套繁复的口译与笔译装置的政体，无法把语言的共同性呈现为一个背景给定物。它把它呈现为一个问题，而这个问题有一个被公认的政治表达：在被汇集于民主赤字这一标题下的诸抱怨之中，语言的抱怨反复出现，即一个以大多数公民并非在其中长大之语言进行的公共领域不是他们的公共领域，而一个以二十四门语言进行的公共领域根本不是一个公共领域。

对这一问题的现存回应分派，且分得尖锐。一种主张一个共同的载体是一个人民所要求的，接受英语担此角色，并把由此给予其母语者的优势当作一桩呼唤以转移支付、以及以其他语言在其自身领域中的属地保护为形式加以补救的不义。另一种主张不应强加任何这样的载体，主张语言承载权利，而正义的安排是一种承认、通融与制度性多语的安排。第三种主张这个问题如所提出的那样无解：缺一个共同的公共领域便没有欧洲人民，因而没有欧洲民主，只有成员国中的民主。

本文论证前两个立场，尽管有种种分开它们的东西，共享一个前提，而第三个从同一前提汲取其力量。该前提是民主能力是一种分布于诸个体之上的性质，一门人所占有的语言、一项人所持有的权利、一种人所习得的能力，从而政治问题是这样的诸性质应如何被分配。允准该前提而分歧随之而来：人可以高效地分配单一一门语言并补偿失利者，或分配对许多门语言的诸权利并保护持有者，而没有第三个选项，因为在该前提上没有别的可分配之物。倘若这是对的，这一困难便不是立场之间的进一步论证会解决的困难，因为它随它们所一致之物、而非随它们所争议之物而来。

两批现存工作已然开始拒绝它，而本文对它们的亏欠应在开篇被陈述、而非在后来被发现。欧洲委员会的框架文件以一个说者能跨一个部分的、不均的、既是接受性的又是产出性的库存做什么的术语、而非以对数门语言的平行掌握的术语，来描述多语能力。关于民主翻译的工作把民主能力定位在协商安排本身之中，在口译的在场之中，以及在口译对说话之分配所做之事之中。这两个动作在此都不被声称为新的。所被提示的，带着两个文献都很大、而作者不会已读遍任一者的限定，是两者都尚未说出什么使这样一个安排成为民主的，也就是说，什么把一个意义跨差异被生成的安排、同一个意义跨差异被强加的安排区分开来，两者都是安排、且两者都涉及翻译。那一判据正是当下的说明所承担去供出的。

该说明可被陈述为单一的一个主张。

中心主张。 民主能力驻留在参与者之间的安排之中、而非驻留在他们各自带给它的一份禀赋之中。一个政体在其诸安排使诸诠释跨语言差异保持对修正开放的程度上是民主的，而在这样的修正被封闭的程度上失败。因而意义的共同性不是民主关系的先决条件、而是它的诸产物之一：一门共享的公共语言，在有一门存在之处，是被维系的民主关系可能生成之物、而非它所预设之物。翻译，在这一说明上，不是一个意义跨一个边界的转移、而是一个共同意义借之被生成的一个机制，而它的民主意义在于它对那个安排所做之事、而非在于它所达成的忠实。

该主张分四步展开。第一步以其最强的形式铺陈现存的辩论，并定位诸立场所共享的前提。第二步给出论证所要求的诠释系统这一概念，且限制性地这样做，因为该概念正是论证的一个过度延伸的版本会触礁其上的那个。第三步把翻译当作一个生成的机制、而非一个转移的机制来处理，把它的生成性形式同它的同化性形式区分开来，并问谁持有认证一次转换的权威。第四步，即该说明立足或倒塌其上的那一节，把关系的诸永久条件同一个安排围绕它们闭合的那一点分开，因为一套只能描述成功的理论不是任何东西的理论，而一套处处发现失败的理论也不是一套理论。

其余部分如下推进。§2铺陈无人民论题、通用语解决方案与多元主义解决方案，并论证它们之间的分派是它们所共享之一个前提的一个后果。§3定义一个诠释系统，并陈述默认语言的选取所安装的不对称。§4把翻译当作生成来处理。§5把关系的诸不可消除条件同闭合之点区分开来，并提供一条候选判据。§6对照欧洲之外的四个情境检验这一区分。§7引出制度后果，对照它们读现存的实践，并陈述代价。§8铺陈论证所开启而不了结的诸问题。一个结尾节标记该说明的界限。

§2 现存的辩论及它所共享的前提

三个立场占据这一场域。它们在此以其最强的形式被铺陈，因为一个借削弱其对手而进入的说明未曾挣得它的进入。

2.1 无人民论题

第一个立场主张民主正当性的诸先决条件当前在欧洲层面不到位。迪特·格林的是它的一个审慎陈述（格林 1995）。在他写作之时反对欧盟的宪法化，格林主张一个民主政体要求一个意见在其中被形成并被中介到政治机构的公共领域，这样一个领域依赖于跨政体运作的媒体、政党与社团，而这些当时在欧洲层面不在场；在他所辨认的诸障碍之中，是一个共享的传播基础设施的缺席，语言是它的一个组成部分。有两件事应被保持在视野中，因为这一立场常被以一个强于其作者所给的形式报告。格林的论证指向一个特定的提议、而非指向欧洲民主本身的可能性，而他允许他所发现缺席的诸条件可能随时间发展。语言是他说明中的一个要素、而非它的全部；此处的概述孤立那个语言要素，因为那正是当下这篇论文所关乎的，而这一孤立是本文的所为、而非他的。这一论题已被长篇争议（哈贝马斯 1995），而值得登记的是，它被绕开比被驳斥更频繁：回应它的若干提议，一个通过联网的诸国家公共领域而浮现的欧洲公共领域，或一个政治地而非文化地被构成的人民，接受格林关于什么曾缺席之描述的相当一部分，并论证别的某样东西可代它服务。

2.2 通用语解决方案

第二个立场接受一个共同载体是必需的，并问一旦一个被采纳正义要求什么。菲利普·范·帕里斯的是那个发达的陈述（范·帕里斯 2011）。在他的说明上，英语正通过一个无人设计的动态作为一门欧洲通用语扩散，而它的扩散总的说来是要被欢迎的，部分因为一门共享的语言给较差者一种彼此协调的手段。他辨认伴随它的两种不公平，一种由其第一语言他人必须付费学习的人所享的搭便车，以及对其他语言所占领域的压力，并提议直接处理它们、而非借抵抗那一扩散：一项针对第一种的语言税，以及一条属地原则，在其之下每一门语言在其自身区域内被保护，针对第二种。这一概述是被压缩的，而论证在那本书中被长篇发展。

应被登记的是，这一立场是当下这篇论文最有理由认真对待的立场，而非一个陪衬。范·帕里斯不把普遍化的多语当作那个补救，而他的理由不是一个对多样性之价值的诉求所能对付的：在他的说明上，期望每个人习得数门语言是有代价的，而语言学习的代价不均匀地落下，从而一项一般多语的政策会最沉重地压在拥有最少者身上。任何一个民主能力更好地被定位在多语库存中的论证都必须回答这一点，而借断言语言多样性的内在价值来回答它并不对付那个反对，而那个反对不是关于价值的。

2.3 多元主义解决方案

第三个立场谢绝那个共同载体，并详述正义对语言及其说者所亏欠之物。它是一族看法、而非一个，而它的成员们有别。艾伦·帕滕论证过一条平等承认的原则，在其之上国家对其公民的诸语言给予可比较的支持，并铺陈过诸替代方案的诸困难，包括按宗教政教分离之模型的语言中立（帕滕 2003，2014）；丹尼丝·雷奥姆论证过一门语言可能对其说者有一种纯粹工具性的说明所错过的价值，而官方语言权利回应那一价值（雷奥姆 2000）。他们在诸根据上不一致，而此处的概述只关乎这一族所共享之物：正义所推荐的诸安排是制度性多语的诸安排，以官方地位、以数门语言的提供，以及对区域与少数语言的保护为形式。一个想要这些作者之间诸分歧的读者不会在此找到它们。

这一立场特征性的困难不是不义、而是平行。承认与提供可能被完全地扩展到每一个群体，而使诸群体彼此并列、而非处在关系之中，在一个共同的制度屋顶下共存，各自进行它自己的公共生活，而没有一个共同意义在它们之间被生成的过程。不显然这样一个结果会如何仅以承认与提供的术语被命名为一个失败，因为在那些术语上没有任何东西失败了：每一项权利都被尊重、每一门语言都被提供。这是一个关于一条判据的评述、而非关于这些作者中任何一个所相信之物；所引诸著作中没有任何东西使他们承诺于对那一结果的漠不关心。它尽管如此看起来像一个民主的失败、而非一个正义的失败，是这一立场，在此处所提供的读法上，没有一个显然位置容纳的一项给定的事实；它的一个辩护者很可能回答说，民主从来不是该理论所承担去供出之物。

2.4 隐藏的前提

那两个建设性的立场在什么应被分配上彼此对立。它们在什么种类之物处于攸关这个在先的问题上一致。在两者中，民主能力是一种分布于诸个体之上的性质：对通用语解决方案而言，一种在那个共享载体中的能力，习得起来不均地有代价、因而呼唤补偿；对多元主义解决方案而言，一项就一个人的语言所持有的权利，不均地被尊重、因而呼唤承认。两者都是关于个体地被持有之诸好处的分配理论，而它们的分歧是一个关于被分配项的分歧。

这可能就是这一交锋一直持久的缘由。允准民主能力是个体地被持有的，而诸选项被这两个穷尽：共同持有一门语言并补救那代价，或持有许多门并保护每一门。没有第三个安排可为之论证，因为该前提已然固定了一个安排能是什么，即一个分配。无人民论题从同一来源汲取其力量：它观察到个体地被持有的诸好处加不成一个公共领域，这是正确的，并推断出没有公共领域是可得的，而这只在一个公共领域必须由个体地被持有的诸好处构成的情况下才随之而来。

主张 1。 通用语解决方案与多元主义解决方案之间的分派是两者所共享之一个前提的一个后果，即民主能力是一种分布于诸个体之上的性质，而非一个关于那个前提的分歧。只要该前提成立，诸立场便穷尽这一场域，而无人民论题随观察到两个分配都不构成一个公共领域而来。倘若那是对的，出路不是一个更好的分配、而是民主能力的一个不同定位，在参与者之间的安排之中、而非在他们各自所持有之物之中。这一重新定位是否成功，正是本文其余部分所为之目的，而此处所提供的诊断可能在那个前提上是错的，而任一立场都不在任何别的东西上是错的。

2.5 那个前提的现存移置

两批现存工作已作出这一动作的一部分，而随之而来的说明是它们的一个延续、而非一个背离。欧洲委员会的框架文件以一个说者所汲取的单一的、相互关联的库存的术语来描述多语能力，其中诸能力可能是部分的与不均的，且其中在一门人不会说之语言中的接受性技能算作能力（欧洲委员会 2001，2020）。那一描述把能力定位在一个说者能跨一个库存做什么之中、而非在一个分别被掌握之诸语言的储备之中。此处所提供的读法，即这在形式上是一个关系性的、而非一个承载者形状的概念，是当下作者的，而不是那些文件中所作的一个主张，那些文件是语言教育的工具，且不是为了了结政治理论中的诸问题而写的。彼得·克劳斯已更长篇地发展了欧洲多语的政治蕴含（克劳斯 2008）。关于多语协商的工作把民主品质定位在那个安排本身之中：妮可·德尔关于跨国社会论坛的诸研究考察当口译在场时协商发生什么（德尔 2012，2018）。一个更宽的关于语言正义的文献，由阿尔卡尔德所综述（阿尔卡尔德 2018），从数个学科方向接手那些制度问题。

然而两者都不陈述什么使这样一个安排民主。一个多语库存可能在支配之下、一如在民主之下同样容易地被投入使用；流利于宗主国语言的殖民地臣民有一个库存并跨它运作。翻译在帝国行政中一如在协商中同样在场。一个安排是多语的、且它的参与者跨诸库存运作，这本身并不把它同一个意义跨差异被强加的安排区分开来。供出那一区分是§5的工作，而它之前的两节准备它可被陈述于其中的术语。

§3 语言作为一个诠释系统

3.1 反对工具性看法

工具性看法把一门语言当作一个通道：一个思想被形成、被编码、被传输、被解码，而诸通道之间的选择是诸效率之间的一个选择。在这一看法上，一门公共语言的选取是一个带一个分配性残余的协调问题，而这恰恰是通用语解决方案处理它的方式，而这一处理是融贯的。

工具性看法无法登记的，是这一选取决定哪些经验以低代价可轻易表达、哪些概念已可供公共使用，以及哪些参与者承担以非其自身之术语表述其所指之物的负担。这些不是对一个讯息之传输的诸效果。它们是关于在一个给定的公共情境中根本能说什么的诸条件，而它们由那一选取本身、而非由参与者事后所做的任何事不均地分配。

3.2 一个诠释系统的定义

论证要求一个比”语言”更精确的概念，而该概念在此被限制性地陈述，因为那个过度延伸的版本正是一个这种论证失败之处。

定义。 一个诠释系统是一个对什么算作一个有效延续施加诸条件的实践，以致一个有能力的参与者能在它之内犯错，且按该实践自己的诸标准、而非按一个观察者的诸标准犯错。

该定义的两个特征做工。第一是规范性：一个诠释系统不仅仅是一套习惯或一种风格、而是一个带诸标准的实践，从而它之内的一个动作可能被更正，且被无需在别的任何事上一致的其他参与者更正。第二是这些标准是内部的：使一个延续有效的东西由该实践固定、而非由它在别处的诸后果固定。

自然语言直截了当地满足这一点。其他象征性实践是否满足它，学科语域、法律论证、数学记号、专业习语、身体与情感表达的被编码语域，是一个真实的问题，且在此被留作悬而未决。有些显然满足；另一些可能只在特定条件下满足；而自然语言之间的相互不可理解，同在一个法律语域与一个外行语域之间移动的困难之间的差别，是一个种类的差别，一个仓促的一般化会将它抹去。该定义被提供，以使论证后来的诸主张可查核，且以使任何超出自然语言的扩展都必须逐案挣得。随之而来的东西中没有任何东西取决于那个扩展。

3.3 不对称及其吸收

一门默认语言的选取安装一个常在的不对称：一些参与者在一个是他们的系统中以低代价表述，另一些在一个不是他们的系统中以高代价表述。这一不对称不由善意可消除，也不由任何特定语言从默认位置被移除可消除，因为某个默认将成立，哪怕只是通过消耗。

该框架关于不对称的立场是，不对称总生成权力，而问题从来不是权力是否在场、而是那个安排是否继续通过生成吸收它、还是让它巩固为一个固定的语法。应用于此：一门支配性工作语言的在场本身不是那个失败。那个失败是这样一个安排：其中支配停止可修正，其中谁承担表述之负担这个问题本身再也无法在其语言处于攸关之公共领域中被提起。什么把这两种情形区分开来是§5的主题。

§4 翻译作为生成

4.1 反对转移模型

转移模型把翻译当作一个意义从一门语言到另一门语言的传送，以忠实为其标准、以损失为其特征性的失败。作为一个关于一位译者在专业上有义务尝试什么的说明，这无可反对。作为一个关于政治概念在一个多语政体中发生什么的说明，它误描。

一个进入另一个诠释系统的概念受制于那个系统关于有效延续的诸条件。它获得那个系统所许可的诸推断，并失去它不许可的那些；它获得它先前不曾有的诸邻居、诸内涵，以及一段历史。其结果不是同一个意义换了一身新衣，也不仅仅是原件的一个退化副本。它是一个新系统中的一个新项，同第一个处在某种关系之中，而该关系是参与者们必须弄清、而非读出的一个。

4.2 欧盟自己的证据

欧洲的语汇供出诸实例，而它们是文献性的、而非例证性的。solidarity、solidarité 与 Solidarität 承载不同的制度与教义负载；subsidiarity 在其德语与意大利语接受中有一个天主教社会训导的谱系，而它的英语用法不承载这一谱系；sovereignty 在一个法国宪法语域中不是同一个项、一如在一个英国议会语域中。这些术语如今有欧洲的诸意涵。那些意涵不是由更好的翻译找到的，而它们不是诸国家意涵的交集。它们是通过制度实践，即诉讼、谈判、起草、对平等地存在于所有版本中之诸文本的累积诠释，历数十年被产生的，而它们仍在移动。

法律的情形是最锐利的。欧盟法不知有原件：每一个语言版本都平等地真确，而在诸版本分歧之处，法院参照该文书的目的与总体方案来诠释。在形式上，诸版本背后没有一个它们是其呈现的意义。意义是跨诸版本的诠释实践所产生之物。无论人怎么看待这一教义，它是一个建立在对转移模型之否认上的制度。

4.3 经验结果

一个经验发现直接关乎该主张，而它应连同其界限被给出。德尔关于跨国社会论坛中多语协商的诸研究，在对那些情境的延展观察的基础上论证，以志愿口译进行的协商在某些方面能够比单以英语进行的协商更具包容性：口译放慢那一交换、打断流利者与自信者的优势、要求一个贡献被重述，并创造一个位置，即口译者，一个贡献可从其被澄清、然后被回应（德尔 2012，2018）。

这些界限是可观的，且不应被最小化。这一情境是一个参与者由其对这一过程之投入而被选取的社会运动论坛，而非一个立法机关或一个选民团；那些口译者是政治上投入的志愿者、而非中立的专业人员，而那是那个机制的一部分、而非它其中的一个缺陷；这一发现未曾在正式机构的规模上被复现，而在那里激励是不同的。她的工作所支撑的比本文的论题更狭窄，而尽管如此是那个正确的形状：一个安排中翻译的在场能改变有效参与的分配，而且能以同效率论证所预测者相反的方向改变它。效率论证主张翻译是摩擦、而摩擦是代价。在这里那摩擦正在做民主的工作。

主张 2。 一个多语政体中的翻译不是一个意义跨一个边界的传送、而是一个共同意义借之被生成的一个机制，因为一个进入另一个诠释系统的概念被那个系统关于有效延续的诸条件重新界定。因而它的民主意义在于它对那个安排所做之事，在于对表述之负担的分配以及在于一个贡献可被重述并被修正的诸场合，而非在于它所达成的忠实。

有两样东西这一主张不许可。它不消解不可翻译性：某些项抗拒重新界定，而那一抗拒本身是有信息量的，这与该说明相容、并可能对它是必需的，而此处不对使不可翻译性成为其主题的哲学文献采取任何立场（卡桑 2014）。而它不使翻译成为充分的。翻译在无人会称为民主的诸安排中在场，而帝国的诸行政，除别的之外，曾是可畏的翻译机构。什么把这些情形区分开来是接下来两小节的主题。

4.4 生成性翻译与同化性翻译

有两个操作应答翻译这个词，并同本文的判据处在不同的关系之中。

在第一个中，两个系统都移动。一个进入另一个诠释系统的概念被它重新界定，而那个接收系统反过来被它所接收之物所改变：一个范畴被拉伸、一个区分被引入、一个术语获得一个它先前不曾有的应用。§4.2的欧洲法律语汇不均地展现这一点：subsidiarity 进入英语法律用法，对那个接收用法的改变，一如那个英语用法对那个术语的改变。把这称为生成性翻译：这一关系在这一特定意义上是对称的，即两个系统都易受那一相遇的修正。

在第二个中，一个系统移动而另一个不移动。一个在一个诠释系统中被表述的主张被呈现进另一个的诸范畴，在这一过程中变得政治上可读，而那些接收范畴不受影响。该主张进入公共领域；它进入所依据的诸条款不随它进入。把这称为同化性翻译。它的标志不是不准确，一个同化性的呈现可能按每一个专业标准都是一丝不苟的，而是那一易受修正性的方向，它只朝一个方向而行。

这一区分不是一个好翻译与坏翻译之间的区分，而它不由译者的用心来决定。它是那一翻译发生于其中之安排的一条性质，而具体地是那个接收系统是否有待被它所接收之物所改变的一条性质。这正是使对整个说明的那个显然的反对，即欧盟已然有翻译、且以巨大的代价，能被回答、而非被偏转的东西。翻译基础设施是必需的，且不了结这个问题。了结它的，是那个基础设施是否许可外围系统改变那个中心系统、还是只许可它们进入它。

4.5 诠释权威

在这一区分之后立着一个转移模型不提起的问题：谁执行那一转换，且在谁的有效性诸条件之下。

在一个平等真确的体制中，没有版本是原件，而在形式上无人占有意义。在实践中那一呈现被认证：由一个依据既立诸对等物工作的翻译服务、由一个以一门工作语言运作的起草委员会、由一个诠释分歧诸版本的法院、由一个双语的阶层，其在两个系统中的能力使它成为它们之间的通路。认证不是一个文书功能。无论谁认证，都决定哪个表述算作所说之物，因而诸可能呈现中的哪一个成为那个机构所回应的主张。

这对那个主张者有一个应被直白陈述的后果，因为它是转换借之成为一桩错事、而非一个代价的机制。该主张以一个别处所撰之形式返回到它的作者，而他被要求把它接受为他自己的，才能被回应。他作为说了它的那个人的地位，已同他对所说之物的权威分开。在那个安排承认那一分开之处，在他可以说这不是我的意思、并被听为在作出一个主张、而非在阻碍一个程序之处，那一转换仍是一个他为其一方的操作。在那个安排不承认它之处，他已被弄成了一个他不曾作出之陈述的作者。该框架关于主体生产的一般说明在此无需调整地适用：那个安排已产生它所要求的那个参与者，而问题是它所产生之物是否仍能修正它。

§5 关系的诸条件与闭合之点

一个只描述成功情形的关于民主安排的说明什么也不描述，因为那个成功情形被预先辨认。本节辨认什么出错。然而它以什么不出错开始：一个批判性说明会受诱惑去指控之物的一大部分，是关系的一个寻常条件，而一个指控它的判据会什么也分不出。

因而本节从诸条件跑向那一闭合。第一小节确立无论是权力的在场还是异化的在场都无法充当那个判据。接下来两个描述展现两者、且保持敞开的诸安排。第四个辨认那个安排闭合的那一点，即该说明所命名的失败。然后两个进一步的形式被区分，而一条候选判据被提供。

5.1 不对称的不可消除

每一个关系在某个方面都是不对称的，而不对称生成权力：这是该框架的常在立场，而它不是此处勉强作出的一个让步。两方不处在关系之中而其中一方不在某个维度上被更好地安置：更流利、资源更好、更熟练，或更靠近那个将决定的机构。诸维度很多，而没有一方在它们全部中都占优势，但其总和从不为零，而一个消除了它的安排会是一个其中什么也没在被做的安排。

异化同样是象征之为象征的一个后果、而非特定诸安排的一个缺陷。陈述一个主张就是把它交付给一个不是该主张者私人占有物的符号系统，其诸术语承载他不曾撰写的诸历史与他不曾意图的诸推断，而它将被他人在他不控制的诸条件下理解。这是根本对任何人说任何东西的条件。它在单语的情形中一如在多语的情形中同样完全地成立；多语的情形所做的，是使它可见，借给那个交付之时刻一个名字，即翻译，以及一个所在。

对那个判据的后果是决定性的。倘若权力的在场使一个安排不义，每一个安排都会是不义的；倘若异化的在场使它如此，每一个言说的行动都会是。一个谴责一切的判据在任何东西之间都不辨别，而在实践中许可应用它的人已然倾向于谴责的无论什么。因而此处的说明不问权力是否在一次转换中在场，也不问该主张者的某样东西是否在其中被丢失。两者都总是真的。

它所问的，是那权力一经被生成便成为什么。一个不对称可能被吸收，被一个继续生成、并由此继续产生那一不对称能被抵着施压之诸场合的安排采纳，或它可能巩固，安顿进一个那个安排不再修正的固定语法。这一差别不是权力之数量的一个差别。它是生成是否继续的一个差别。

有一个进一步的点属于此处，因为它移除本节其余部分否则会招致的一个诱惑。一次转换被评估所依据的诸规范，即什么算作一个充分的呈现、哪些范畴可用、什么可被施压以及被谁，不是从它们所治理的关系系统之外被引入的。它们在它之内被生成，由那同一个过程，其诸产物它们随后评估。因而没有一个中立的立足点，一次转换能从其被认证正确，而对一个这样的立足点的寻找产生本文一直在抵抗的那两个幻想：一个应答一个在所有版本背后之意义的忠实翻译，以及一个无人承担表述之负担的语言中立的公共领域。这些规范是产物。问题只是它们是否保持对它们所治理之物负责、还是成为它们所治理之物再也够不到的诸前提。

5.2 排除

在第一种模式中，一个参与者根本无法进入：那门公共语言是一门他不具有的，没有口译被提供，而那个安排不给他任何供他从其说话的位置。这是现存文献处理得好的情形，而§2的两个现存立场都处理它，一个借普遍化在那个共享载体中的能力，另一个借扩展官方地位与提供。它在此为完整性被引入而不被论证，因为一个其贡献在于注意到排除的说明会没有贡献。

5.3 带一个被承认之余项的转换

在第二种模式中，进入是可得的，且要求转换。一个在一个诠释系统中被表述的主张必须被呈现进决定于其中被作出之系统的诸范畴，而那一呈现不承载一切：原件所许可的某个推断在那些接收范畴中不可得，它所转于其上的某个区分是那些范畴不划的一个。把未能承载之物称为余项。

这里的负担是真实的，且它是不对称的。无论谁必须转换，都承担一个无论谁母语地表述都不承担的代价：找到一个呈现的劳作、它所引入的延迟、那个呈现将被回应、而非那个主张被回应的风险。那一负担的分配是一件正义的事，且它是一个可观文献的主题。

然而它尚不是本文所关乎那种的一桩不义，而如此说是必要的，倘若这个概念要做任何工作。一次其余项被承认的转换，在那个主张者保有说这不是我所说的、并让它的这一说被当作一个关于此事的主张、而非一个对那个程序的反对之地位之处，是一个他仍为其一方的操作。它是有代价的、它是不平等的，而它使那个安排保持敞开。转换本身不是那个错事。倘若它是，每一个政治翻译的行动都会是一个，而这个概念会什么也分不出。

5.4 转换闭合

在第三种模式中，那一转换被当作完成的。那个余项不被承认为一个余项：未能承载之物不被认为存在，那个呈现无残余地算作所说之物，而那个被转换的主张无法修正接收了它的诸范畴。该主张者可能重复他自己，而那一重复被同一个操作再次转换。那个安排已围绕那一转换闭合。

这就是那个闭合，而它应被准确地描述：不是一个强的一方冤枉一个弱的一方、而是一个接收了某样东西、并从接收了它中生成了无物的安排。支配可能在场、且常常在场；它不是那个判据所辨认之物，因为它在敞开的诸情形中也在场。最干净地显示它的实例是一个原住民主张同一个殖民者法律系统之间的相遇。一个涉及世系、义务与集体地位的对土地的关系，只在被问它是诸可得范畴中的哪一个之后才被准入那个诉讼程序：财产，在那种情况下它必须有一个所有者和一个边界；环境利益，在那种情况下它必须对着开发被权衡；或宗教自由，在那种情况下它是诸个体所持有、并作为此被保护的一个信念。每一个呈现使那个主张可读。每一个也预先决定那个主张所提起的问题，因为处于攸关的恰恰是那个关系是否是那种有所有者之物、还是一个信念。那个主张被听闻、被转换、被回应，而回应了它的诸范畴不因这样做了而被改变。那个诉讼程序也不生成任何场合，带来那个主张的一方能在其上使它们改变。没有任何东西被排除。那个诉讼程序是一丝不苟的。而那个决定在其中被作出的诠释系统，恰如它进入那样从那一相遇中出来。

同第二种模式的区分是那一改变的在场或缺席，而诸补救相应地有别。带一个被承认之余项的转换在那个安排之内是可补救的：那个负担能被重新分配、那个排序被改变、那个起草阶段被纳入那个多语体制、那个余项被给予一个程序性的位置。转换闭合不由那些手段可补救，因为人会用来为那个补救论证的手段已被弄得受制于同一个转换。

同弗里克的诠释学不义的关系应被陈述、而非被留给一个读者去注意（弗里克 2007）。弗里克把一桩错事定位在集体诠释资源中的一个空缺之中，从而一个人的经验不可理解，因为会呈现它的那个概念不存在，而这桩错事可能在无人有过错的情况下成立。此处所描述的错事不是一个空缺。那些接收范畴是完全成形的、并对它们自己的诸目的完美地充分；那个主张被无困难地呈现进它们。被闭合的，是那个安排对它所接收之物的易受修正性，而那种闭合被某人维持，在一个法院的解释、一个起草惯例，或一个既立对等物被维持的寻常意义上。这两个说明是邻居而不是同一个，而那一差别是一个命名一个缺席、而另一个命名一个闭合。

5.5 象征性闭合

一个相关的形式值得命名，因为它在一个单一系统之内、一如在两个之间同样运作。一个术语可能停止可修正，并开始作为一个归属的检验来运作，不再是一个其应用可论辩的概念、而是一个其接受示意成员资格、其被质疑示意外部性的标记。ever closer union，在某些用法上；European values，在某些用法上；以及，在另一个方向上，sovereignty 如它在某些国家语域中所运作的那样。闭合的标志是那个术语的应用再也无法被论辩，而它的使用转而是一个位置的信号，从而一个质疑它的参与者被听为在宣告一个效忠、而非在作出一个主张。

这种闭合不是一个语言的现象。它在一个单一语言之内、一如跨数门同样容易地发生，而如此指出对该说明的诚实是重要的：多语既不是此处所描述之诸失败的一个充分原因、也不是对它们的一个保护。

5.6 平行

多元主义安排特征性的结果，即诸群体被完全提供、并彼此并列而立，应作为一个不同的失败、而非作为一个温和的成功被引入。所缺席的不是提供、而是接触：没有一个在一个群体中被形成的诠释能被另一个回应的过程，因而没有一个共同意义被生成的过程。平行在这一方面是转换闭合的镜像：在那里，一个主张被转换、而那些接收范畴不被改变；在这里，没有转换被尝试，而没有任何东西供任何人改变。在下面的判据上，平行失败，不是因为任何人被排除、而是因为那个安排没有修正可发生的场址。

5.7 通向一条判据

前面诸小节所支撑的比一条判据更狭窄，且作为此被提供。它们确立那个失败不位于何处，不在权力的在场之中，不在一个主张被异化进非其作者之术语之中，而它们辨认某样确实出错之物的一个点。它们所不确立的，是随之而来的表述是陈述那个点的正确方式、或唯一方式。它是一个候选，而它的诸替代方案在其后被铺陈。

主张 3（一条候选判据）。 无论是一个安排中权力的在场，还是一个主张被异化进非其作者之术语，都不是那个失败；两者都是关系的诸条件，而一条指控它们的判据会指控每一个安排。可在它们的位置上被提议的，是闭合：一个安排停止生成其自身修正之手段、从而它所产生的诸诠释规范不再对它们所治理之物负责的那一点。一个安排在其中的诠释之行动本身产生抵着那诠释所生成之权力施压之能力的程度上维系生成，从而所要求的不仅仅是修正在形式上不被封闭、而是修正的诸场合、诸地位与诸资源在那个安排自己的诸产物之中。在这一表述上，一个民主之失败，不在某些人在转换中处于不利之处，那总是如此，而在那个安排接收一个被转换的主张、并从接收了它中生成无物之处。这一表述是为审查而被推进的、而非作为一个定义。

有一个限定在这条判据同它的诸邻居被比较之前是应有的，因为没有它，该说明便获得一个它不想要的承诺。被维系的生成不是诸安排应当追求的一个目的。此处没有任何东西说一个政体应以生成性为目标、说更多的生成比更少的好，或说那个生成性的安排是那个好的；一个那种形状的判据会安装一个目的论，并会为什么人应当想要它负责。该主张是否定性的、且更狭窄：生成是一个安排自己的诸诠释规范保持对它们所治理之物负责所处的条件，而在它停止之处，那些规范已成为诸前提、而非诸产物。一个安排可在那一根据上被批评，而生成性不被命名为它的目的。

若干其他表述会适合本节的材料，而本文未曾比较它们。

术语说明。 generative（生成性）这个词已然在一个既立文献中在工作。罗恩·埃格拉什的生成性正义问一个过程中由参与者所产生的价值如何被他们保有、而非被并未产生它的诸系统所攫取。此处的问题是一个安排的诸诠释规范是否保持对它们所治理之物负责。上面所抛出的判据被留作未命名，以使两者保持分开。

人可以把这条判据建立在诸诠释规范的可修正性之上，这接近上面所说的，但把那个条件定位在那些规范之中、而非在产生它们的那个安排之中。人可以把它建立在诠释负担的分配之上，这有可度量的优点、以及是一条分配性判据、带§2所引起之诸困难的缺点。人可以把它建立在重开一个已定呈现的能力之上，这更狭窄，且可能以一种当下表述所不能的方式可操作化。人可以把它建立在生成性本身的保存之上，这更一般、并相应地更难应用。这些不是等价的，它们会以不同方式分出某些情形，而在它们之间选择是本文未曾做的工作。所被主张的，是那个失败在于闭合之点、而不在于先于它的诸条件；究竟在哪里围绕那一点安置那条判据是悬而未决的。

这个候选尽管如此应同一条它将与之混淆的邻近判据区分开来。哈贝马斯路线中的协商性诸说明问参与者能否达成一个理性地被推动的一致，并界定诸条件，其中包括包容、非强制，以及更好论证之力量，在其之下这样一个一致会是可达成的（哈贝马斯 1996）。此处所划的区分不是这样的诸说明忽略了把一个主张呈现进一个公众能接收之术语的问题。哈贝马斯后来关于公共领域中之宗教的文章把那一呈现当作由宗教的与世俗的公民同样承担的一个相互负担（哈贝马斯 2006）；它的主题是宗教的、而非语言的差异，而它在此被引用，只因它所描述之要求的结构，即一个主张被翻译进公开可及的术语、且这样做的负担不单落在一方身上，接近一个多语的批评者会为之施压之物。那一类比是否成立在此不被论证。那个区分在别处。协商性判据探问一个理由空间之内的一致；此处的判据问那个理由空间本身是否对在其之内协商的那些人开放、可被修正。一个安排能满足第一个而失败于第二个：一致可能是可达成的、非强制的与包容的，在参与者中没有一个能再争议的诸术语之内。正是第二个失败是那个多语情形使之可见的，因为在一个多语政体中，那个理由空间的构成不是隐藏的：它在公开中被做，借选择那讨论发生于其中的哪些语言。

§6 欧洲之外的机制

该说明迄今依据两个情境：欧洲诸机构，以及一个从殖民者-拓殖裁决中取来的转换闭合的单一实做案例。那对一个关于诸安排一般的主张而言是一个狭窄的基础，而它招致一个特定的反对，即§5.4中所描述那种的闭合是拓殖者法律系统的一个人为产物，那些系统有一种接收原住民主张而不被它们所改变的独特利益，而非多语诸安排的一个一般特征。本节接手四个进一步的情境，看那个敞开的转换与闭合的转换之间的区分在它们之中是否做任何工作。这些处理是简短的，且由一个在这些案例中的任何一个上都不是专家的人所写。它们是为检验一个区分是否辨别而被提供的，而非作为对所关涉诸政体的刻画；在它们同研究这些案例的人的诸说明冲突之处，后者应被优先。

6.1 殖民语言行政

殖民行政是闭合的干净案例，而陈述缘由是值得的，因为那缘由不是那个显然的。殖民政权不缺翻译。它们曾是可畏的翻译机构：它们委托编纂语法与词典、训练口译者、以方言产生法律法典，并配备诸法院，在其中当地主张被长篇听闻。那个失败不是主张无法进入。它是它们所进入的诸范畴不易受它们所接收之物的改变。一个土地关系被接收为租佃、一项习惯义务被接收为契约或被接收为那个法典指派给它的技术意义上的习俗、一个宗教权威被接收为一个职位。每一个呈现在行政上是可用的，而每一个预先了结那个主张所提起的问题。治理那些呈现的诸诠释规范不是在诸方之间的关系之内被生成的；它们随其中一方一道到来。

这是§5的区分最不可争议的案例，而正因这一缘由它是证明得最少的案例。它显示那个机制不限于拓殖者裁决。它不显示那个机制在并非显然不义的诸安排之间辨别，而那是一条判据为了值得拥有而必须做之事。

6.2 后苏联语言立法

苏联的继承国呈现那个更难的案例，而该说明应对着它、而非绕着它被检验。它们中的数个采纳了确立那个名义语言为公共行政、教育与公共生活之语言的语言法，在俄语曾占据那个位置、且大量说俄语的人口留存的境况中。这些安排已在关于少数保护的文献中被长篇考察，而此处的概述不作为对它的一个贡献被提供。

对这些安排的两个描述是可得的，而两者都有些道理。在第一个上，那立法回应一个该政体无份于确立的在先定案：一门从公共生活被移置的语言被恢复到它，而在这一描述上，那个较早的安排是那个其诸规范不在它们所治理的政体之内被生成的。在第二个上，那立法把如今形成一个少数的说者置于把他们的诸主张转换进一门他们不控制的语言与一套范畴的位置，而在这一描述上，相关的问题是那些说者在当前安排下面对什么、而非在那个较早的之下什么曾成立。

此处的说明不在这些之间了结，而说出缘由、而非显得已忽略了这个问题是重要的。§5.7的候选判据问一个安排是否生成其自身修正之手段。它不问、也无法问，一个今天被强加的闭合是否被一个昨天被强加的闭合所担保，因为那是一个关于一个定案之辩护、而非关于那个定案是否保持敞开的问题。一个想要第一个问题之答案的读者不会在此找到一个，而诚实的报告是这是那条判据的一个局限、而非它的一个特征。那条判据关于这些案例所能说的是狭窄的：一个其中那个语言定案之诸条款能被生活在它之下的那些人提起、争议并修正的安排，同一个其中它们不能的安排立得不同，无论产生了那个定案的历史为何，而这是一个比上面两个描述中的任何一个都更狭窄的观察，且不在它们之间选择。

6.3 魁北克与加泰罗尼亚

两个其中一个较大国家之内的少数语言已被弄成其自身领土中公共生活之语言的案例。两者都频繁地被多元主义立场引用为一个属地定案能保护一门语言而不排除任何人的证明，而两者都在那个领土之内被那个定案置于不利者所争议。

这些案例对当下说明的旨趣在于一个它们共享、且那条判据挑出的特征。在每一个中，那个语言定案本身是一个常在的政治争执项：它被立法、被诉讼、被修订、被作为竞选议题，而这些安排在活着的记忆之内已不止一次被大幅修正。无论关于它们别的什么为真，那些条款不是被固定到超出生活在它们之下者的够及范围之外。那正是那条判据所探问的，而值得注意的是，那条判据相应地比一个分配性说明可能被这些安排困扰得更少，因为在两个案例中表述之负担都不均地、并出于设计地落下。无意对任一定案作任何裁断，而这一观察是关于那条判据所留意之物、而非关于这些安排是否被辩护。这是那条判据的一个强处、还是它不够苛刻的一个指示，是一个作者不视作已定的问题。

6.4 印度

印度是最抵抗本文之定框的案例，且正因这一缘由被纳入。宪法定案承认大量语言，那三语公式以一种没有欧洲安排所接近的方式跨学校教育分配表述之负担，而这一安排已被反复地、并在相当的压力下修正，包括通过重塑了它的反印地语鼓动。按§5.7的判据，这一安排看起来不寻常地敞开。

它在其他量度上看起来不那么敞开。英语在高层行政、高等教育与高层司法中所占据的位置已被证明持久，而那些被承认语言同那些在被承认集合之外者之间的关系，已被以类似于§5.4所描述之物的术语讨论。两个观察都能是真的，而它们的共存是那个有用的教训：一个安排能在一个层面上大幅敞开、而在另一个上闭合，而一条被应用于一个政体整体的判据将返回一个在应被分别评估之诸条件上取平均的答案。这是那条判据在此被陈述之方式中的一个缺陷、而非一个关于那个政体的观察，本文没有立场评估那个政体；它相应地被记录在§8中。

6.5 这些案例所显示的

三样东西，而第三样是最不令人安适的。那个机制不限于拓殖者裁决：殖民行政清楚地展现它，而且在广泛翻译的同时如此，而那正是该说明最需要一个例子的点。那个敞开的定案与闭合的定案之间的区分确实在并非显然不义的诸案例之间辨别，而魁北克与加泰罗尼亚的案例显示这一点，而且以一个分配性说明未必会认可的方向如此。而那条判据当前被陈述在一个让它能被应用于一个政体、而它应被应用于一个阶段、一个论坛或一个语域的一般性层面，而印度的案例使这一点清楚，而欧洲的材料曾容许本文忽略它。

§7 制度后果，以及代价

7.1 该候选会施加的诸要求

临时地采取§5.7的表述，三个要求随之而来。可修正性：那个安排必须包含它自己的诸条款，包括它的诸语言条款，能作为一件供决定之事被提起、而非被默认地定下的诸场合。翻译作为构成性的：口译的装置属于那个协商过程、而非属于它的行政，这有一个实践后果，即它必须在起草发生之处、而非仅在投票发生之处在场。表述之负担的分配：问题不仅是所有语言是否可得、而是谁被要求在一个不是他们自己的系统中表述，且在哪个阶段。

一个第四要求强于这些，且随主张 3 的主动子句而来。那一制衡能力的生成：修正在形式上可得是不充分的，因为没有维系争执之诸资源的可得性，是对那些已然拥有它们者的可得性。那个安排必须产生，作为它自己运作的一个副产品，抵着它的诠释所生成之权力能被施压所凭的诸地位与诸场合。这在制度上看起来像什么并不显然，而该说明不声称已界定它；诸语言版本的平等真确是一个实例，因为它制造一个争议一个呈现的地位，而无一方必须为之奋争，而口译在起草阶段的在场是另一个，因为它创造否则不会存在的重述之诸场合。两者都是一个安排生成否则会必须被带给它之物的情形。

7.2 对照它读现存的实践

欧盟的一些实践满足这些，而一些不满足，而该说明不因假装并非如此而被改善。欧盟法中诸语言版本的平等真确是第二个要求的一个真正实例，且是一个不寻常的：它把对一个享有特权之原件的否认建进那个法律秩序本身。一个成员国可以以它自己的语言致意诸机构、并以它收到一个回复这条规则满足第一个的字面。对着这一点：起草在一门工作语言中的集中、文件在诸立场已形成之后以其他语言出现所依据的排序，以及口译在委员会与专家组阶段的实践上不可得，在此处的说明上，是第二种模式的诸失败：在那个余项仍有某处可去的阶段被强加的转换负担，以及在其行使最无后果的阶段被尊重的诸权利。区域与少数语言的地位，宪章处理它（欧洲委员会 1992）而欧盟自己的体制大体上不处理它，是转换闭合的诸问题生起之点。

7.3 代价

那条候选判据会推荐的安排是昂贵的、缓慢的，且不以通常被应用的诸量度自己回本。口译与笔译消耗一笔大预算和一个更大量的时间；多语起草比单语起草更慢，而这一差别不是边缘的；而没有一个证明，即由此产生的诸决定按任何独立的标准是更好的。效率反对以它自己的术语是正确的，而该说明不争议它。它所争议的，是那些术语的充分性：那代价买来可修正性，而可修正性不是一种协商效率、而是那个安排根本成为民主的一个条件。陈述那代价比让一个读者发现它更可取，而一个判断那代价过高的读者已理解了那个论证、而非错过了它。

7.4 回答那些现存立场

对无人民论题：该论证允准格林的观察并否认他的推断。在欧洲缺席的，是一个先已存在的传播共同性，而它是缺席的。到欧洲民主之不可能的推断，只在共同性必须先于民主关系的情况下成立。在此处的说明上它由这样的关系所生成，不均地并缓慢地，而这正是§4中欧洲法律语汇的轨迹所展现的。它是否在政治的、而非法律的语域中被生成得足够快、或根本被生成，是一个本文不了结的经验问题。

对通用语解决方案：这一分歧比它显得的更狭窄，而由于狭窄而更锐利。范·帕里斯的再分配处理那一不对称对承担它者的代价。它不处理那一不对称成立于其中之安排是否保持可修正，而那是一个在先的问题，因为一个再也无法作为一个问题被提起的被补偿的不对称，是带一笔付款附着的转换闭合。范·帕里斯可用的回复，即可修正性本身最好由每个人能对每个别人说话、以无论什么语言来确保，是一个严肃的回复，而两个立场之间的论证转在它之上。此处没有任何东西驳斥它。

对多元主义解决方案：承认与提供是必需的、且不是充分的，而平行是那个证明。

中心命题。 民主不要求公民共享一门语言。它要求这样的诸安排：在其之下一个共同意义能继续跨他们所拥有的诸语言被生成，且在其之下那一生成的诸条款，包括那生成发生于其中的哪门语言，对必须生活在它们之下的那些人保持对修正开放。一个已达成完美语言共同性、并再也无法把那一共同性付诸质疑的政体，已满足那个较旧的判据、并失败于这一个。一个不共享任何语言、且其诸安排使那个问题保持敞开的政体，已失败于那个较旧的判据、并可能尚且满足这一个。

§8 开放的问题

论证所开启的多于它所了结的，而把那残余铺陈出来是一次初步讨论所为之目的的一部分。下面的诸问题按它们对该说明的关涉次序给出：前三个是该说明要求一个答案、而不供出的，其余的是它使之可问、并留其悬而未决的。无一是修辞性的，而对其中数个，作者的倾向被记录，且应被读作一个倾向。

8.1 认证一次转换的权威

§4.5确立某一方认证哪个呈现算作所说之物，且那一认证不是文书性的。它未曾说谁应持有那一权威，而那些显然的答案各不令人满意。倘若那个接收机构认证，§5.4的闭合便被建进那个程序。倘若那个主张者认证，没有任何转换能被完成，而那个机构无法行动，这是拒绝给他一次听闻的一个不同方式。倘若一个专业的中介认证，那个问题在上一个层面复现，因为那个中介的呈现按什么标准、由谁被评估。

一个部分的答案提示自身，且在此不被发展：即成问题的那一权威不是固定一个呈现的权威、而是重开一个的权威，而这些可被不同的诸方持有而无矛盾。一个机构可为行动之目的临时地认证，而那个主张者保有重开的地位，这接近一个上诉结构对法律错误所做之事。那一安排是否稳定、还是行动之权力仅仅吸收固定之权力，是那个开放的问题。

8.2 诠释负担的分配

每一次多语互动都强加一个表述、解释与调适的负担，而该说明已说它的分配是一件正义的事、而没有说一个正义的分配会是什么。对称是那个错误的答案，因为在数目不对称之处那个负担无法是对称的：一条要求每一个参与者对每一个别的迎半的规则，对一个小语言共同体的成员强加的远多于对一个大的成员，而这算术不是一个细节。同数目成比例是另一个方向上的错误答案，因为它再生产它本意要缓和的那个支配。

该说明所能说的，是那个负担是那个安排的一个代价、而非作为一个少数说者的一个私人代价，而诸安排在它们是否如此对待它上有别。它所不能说的，是任何一方欠它的多少。这是论证最显然地需要一套它已花§2谢绝去倚赖之分配理论之诸资源的点，而这一张力是真实的、而非表面的。

8.3 可修正性与一个未定问题的代价

该说明推荐诸语言安排保持对修正开放。应被登记的是，少数语言诸共同体已频繁地并合理地偏好相反：一个固定的宪法保障、一个属地定案、一个把那个问题从寻常政治中移除的固化。这一偏好不是关于他们利益的混淆。一个永久可修正的定案是一个永久可争议的，而争执由拥有维系它之诸资源者赢得，而那通常不是那个少数。

这是对本文之立场最锐利的反对，而它在此没有答案。两条路线是可得的，而两者都不被发展。第一条区分修正的对象：诸主张被听闻所依据的诸条款可能保持敞开，而特定的诸保障被固化，从而固化本身是一个保持敞开之过程的结果。主张 3 的主动条件关乎这一点、并可能回答它的一部分：那个少数共同体有理由畏惧的不是可修正性、而是没有维系争执之诸资源的可修正性，而那恰恰是那条判据拒绝算作被维系之生成的东西。一个生成了施压一个主张之诸地位与诸场合、而非仅仅许可它们的安排，不会向那个共同体呈现固化与暴露之间的选择。任何安排是否可靠地做这一点、以及固化是否不仅仅是生成它的最可依赖之方式，仍是悬而未决的。第二条为一个处在压力下之少数的情形让步那个点，并把该说明限制到诸方中没有一个处在消失之风险中的诸安排，一个会尖锐地收窄它的应用的限制，且一个作者不情愿去接受、而不先看第一条路线能否被弄得起作用的限制。

8.4 忠实，以及是否有一个原件

翻译理论已围绕对一个原件的忠实组织它自己。当下的说明，在把意义当作被那个接收系统关于有效延续之诸条件重新界定时，同一个立于诸版本背后之原初意义的概念不安地相处，而欧盟自己的平等真确教义也同它不安地相处。

两个读法是可得的，而本文不选择。在第一个上，没有原件，忠实是一个调节性的虚构，而看起来像忠实之物是诸方之间一个呈现将被当作所说之物的一致。在第二个上，那个原件是真实的、但是一个关系、而非一个内容，即那个说者在他自己的系统中曾在做什么，而忠实在原则上是可能的、尽管从不可验证。第一个更干净，且冒使那个主张者关于他的意义被丢失的抗议不可理解的风险，而那会是一个严重的代价，因为那一抗议正是§5.4被写来容纳的现象。

8.5 误解是否总是一个失败

效率定框把残余的误解当作损失。一个不同的可能性是完全转换之不可能是生成的诸条件之一、而非对它的一个障碍：在没有现成对等物存在之处，诸方必须造出某样东西，而它们所造之物对两者都是新的。§4.2的欧洲法律语汇至少同这一读法一致。

该说明只能借它已有的判据、即那误解是否对被误解的一方保持对更正开放，来把那个生产性的情形同那个破坏性的区分开来，而那一区分是否做足够的工作是未解决的。它显然不随之而来即误解一般是生产性的，而一个颂扬它的说明会是一个糟糕的说明。

8.6 转变与身份

生成性翻译要求那个接收系统被它所接收之物所改变。但一个被无限地改变的系统不是同一个系统，而一门被向一门支配语言的持续通融所转变的少数语言，是在描述同化、而非生成。因而该说明要求一个在一个系统作为它自己所经历的转变、同结束它的转变之间的区分，而没有一个。

这不是一个语言所特有的问题。同一个问题对科学传统、诸机构与诸人生起，而该框架给出的任何一般答案将必须在那个层面、而非在此处被给出。它作为一个当下的说明预设一个答案、而不提供一个的问题被命名。

8.7 共享意义还是共享程序

程序性诸说明主张民主需要公平的规则、而非关于诸意义的一致。此处的说明提示诸程序不独立于诸意义，因为一条规则在某个诠释之下被应用，而哪个诠释治理这个问题本身不无回归地程序性地被了结。

这是对程序性诸说明的一个严重困难、还是它们已然吸收的一个熟悉的困难，是一个供那些说明之辩护者的问题，而作者不声称已显示前者。

8.8 不可翻译性

倘若某些概念真正地抗拒转换，问题是这是对跨差异之民主的一个障碍、还是它根本成为一个关系的一个条件。此处的倾向朝向第二个，一个完全透明的政体会不要求任何关系，因为会没有任何东西可弄清，但那个倾向不是一个论证，而把它陈述得仿佛它是一个论证，会是本文在别处所批评的那种动作。关于不可翻译性的哲学文献有该说明未曾使用的诸资源（卡桑 2014）。

8.9 公民身份及其形成

倘若民主能力驻留在诸安排、而非诸禀赋之中，一个公民必须能做什么是不清楚的。语言能力是那个传统的答案，且不显然是那个正确的；一种跨诸诠释系统维系关系的能力更接近该说明所要求之物，而它不是任何课程当前教授、或任何工具度量的一种技能。

随之而来的发展性问题，即这样一种能力如何被形成、它是否在多语与单语养育之下被不同地形成、一种在童年被形成的能力是否转移到政治参与，在本文中不被处理，也不以许诺的方式被处理。多语童年是多语民主之发展性基础这一假定是有吸引力的、其背后有很少的证据，且再生产一个这个框架在别处谢绝的发展性目的论。它是一个问题，而它是一篇不同的论文。

8.10 机器翻译

自动翻译如今处理本文所讨论之转换的一大份额，而问题是它改变那个问题、还是移置它。该说明的预期是移置：一个可靠地呈现的系统仍呈现进诸范畴，那些范畴从它被训练于其上的语料库被继承，而那个语料库在诸诠释系统之间不是中立的。认证不在它被自动化时消失；它变得更难辨认并更难争议，因为没有一方能被问为什么是这个呈现、而非那个。§5的判据能否被应用于一个其认证方是一个模型的安排是真正不清楚的，而它是这份名单上最可能最快变得紧迫的问题。

8.11 那条判据所应用到的单位

§6.4暴露了一个欧洲材料曾遮蔽的缺陷。那条判据已被陈述为一个安排上的一个条件，并被应用得仿佛一个政体是一个安排。它不是。一个定案可能在宪法政治的层面上敞开、而在行政决定实际被作出于其中之语域中闭合；一门语言可能在立法机关中可修正、而在高层司法中被固定。被应用于一个政体整体，那条判据在有别的诸条件上取平均，而一个平均不是一条判据应返回之物。

那修复大概是界定那个单位：一个论坛、一个程序的一个阶段、一个语域。哪个单位，以及那条判据是否跨不同尺度的诸单位一贯地行为，在此不被弄清，而在它被弄清之前，§7中的诸应用应被读作应用于它们所命名的那些阶段、而非应用于欧盟。

8.12 一个生成性关系的最小条件

本文所引起的最一般的问题。该框架接受不对称为关系的一个永久特征，并主张不对称本身不是支配。§5的判据提示那个差别的一个候选，即一个不对称在那个较弱的一方再也无法修改产生它的那个结构时成为退化性的，而倘若那个候选成立，它会应用得远超语言，同样地应用于教育的、制度的与国际的诸安排。

它是否成立在此不被确立。那个多语情形是一个情形，被选取是因为它使那些诠释条件的构成不寻常地可见，而一条在那个现象可见的情形中起作用的判据，可能在它不可见之处失败。在别处检验它是这一路线的自然延续，且在本文中不被承担。

§9 界限与进一步的工作

该说明依据一些可能是错的主张，而如此地陈述它们属于把它放下。

那个经验基础是薄的。§6的诸比较处理简短到近乎图式的程度，各把一个文献压缩进一个段落，而四个中任何一个的一个专家会发现这一压缩不令人满意；它们是为检验一个区分是否辨别、而非作为对这些案例的诸说明被提供的。一个协商发现，来自一个带自我选取之参与者与政治上投入之志愿口译者的运动情境，是一个多语安排能比一个单语安排更平等地分配参与的主要证据；它未曾在正式机构中被复现，而有理由预期它可能不会。把欧洲法律语汇读作被生成、而非被翻译，是一个读法，而一个法律学者可能把同一材料描述为对一个曾在那里有待被找到之意义的收敛。

那条候选判据等待一个可操作的形式，而未曾在它与§5.7中所指出之诸替代方案之间作出任何比较。说一个安排在一个诠释仍能修正产生了它的那个安排的程度上是民主的，是陈述一个条件、而非一个检验。什么会算作修正保持可得的证据，短于一个实际的修正发生，在此不被了结，而一次其余项被承认的转换与一次已闭合的转换之间的区分，当前是一个仔细的读者能应用、而非一个程序能决定的区分。

该说明对自然语言之外的扩展不被尝试。§3中的定义被这样定框，以使其他象征性实践可能满足它，而学科的、专业的或情感的诸语域是否如此、以及倘若它们如此政治上会有什么随之而来，被留作悬而未决。断言那个扩展的诱惑是可观的，且已被抵抗，因为此处的论证由欧洲材料支撑，而那个扩展不会。

§8的诸问题中的数个不仅仅未被回答、而是关乎该说明是否正确，尤其是那里所记录的反对，即可修正性本身对最不能维系争执者是有代价的。那个反对在本文中没有答案，而针对它所勾勒的两条路线是勾勒。

---

关于方法的一则说明。 没有形式装置被使用。§5.7的候选可被写作一个安排的某些状态从它之内的可达性上的一个条件，而闭合被写作一类过渡的丧失；那记号会是精确的，且会推导无物、非散文已然陈述之物，同时借来一份该论证未曾挣得的严格。所被主张的，以散文被主张，且在那里立足或倒塌。

参考文献

Alcalde, Javier (2018). Linguistic Justice: An Interdisciplinary Overview of the Literature. In Michele Gazzola, Torsten Templin, and Bengt-Arne Wickström (eds.), Language Policy and Linguistic Justice. Cham: Springer, 65–149.

Cassin, Barbara (ed.) (2014). Dictionary of Untranslatables: A Philosophical Lexicon. Trans. ed. Emily Apter, Jacques Lezra, and Michael Wood. Princeton: Princeton University Press. [French orig. 2004.]

Council of Europe (1992). European Charter for Regional or Minority Languages. ETS No. 148. Strasbourg.

Council of Europe (2001). Common European Framework of Reference for Languages: Learning, Teaching, Assessment. Cambridge: Cambridge University Press.

Council of Europe (2020). Common European Framework of Reference for Languages: Companion Volume. Strasbourg: Council of Europe Publishing.

Doerr, Nicole (2012). Translating Democracy: How Activists in the European Social Forum Practice Multilingual Deliberation. European Political Science Review, 4(3), 361–384.

Doerr, Nicole (2018). Political Translation: How Social Movement Democracies Survive. Cambridge: Cambridge University Press.

Fricker, Miranda (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press.

Grimm, Dieter (1995). Does Europe Need a Constitution? European Law Journal, 1(3), 282–302.

Habermas, Jürgen (1995). Remarks on Dieter Grimm’s “Does Europe Need a Constitution?” European Law Journal, 1(3), 303–307.

Habermas, Jürgen (1996). Between Facts and Norms: Contributions to a Discourse Theory of Law and Democracy. Trans. William Rehg. Cambridge, MA: MIT Press. [German orig. 1992.]

Habermas, Jürgen (2006). Religion in the Public Sphere. European Journal of Philosophy, 14(1), 1–25.

Kraus, Peter A. (2008). A Union of Diversity: Language, Identity and Polity-Building in Europe. Cambridge: Cambridge University Press.

Patten, Alan (2003). Liberal Neutrality and Language Policy. Philosophy & Public Affairs, 31(4), 356–386.

Patten, Alan (2014). Equal Recognition: The Moral Foundations of Minority Rights. Princeton: Princeton University Press.

Réaume, Denise G. (2000). Official-Language Rights: Intrinsic Value and the Protection of Difference. In Will Kymlicka and Wayne Norman (eds.), Citizenship in Diverse Societies. Oxford: Oxford University Press, 245–272.

Van Parijs, Philippe (2011). Linguistic Justice for Europe and for the World. Oxford: Oxford University Press.