On Normative Discourse

If speaking of justice were the same thing as banging on the table, then normative discourse could not be taken seriously. The aim of this paper, however, is to vindicate the meaningfulness, and rationality, of normative discourse, and to outline its conditions of possibility. Normative discourse can be understood as if there were, in its structure, different “stages,” or layers. In the transition from one stage to the next, complexity increases. Thus, I shall depict the emergence of normative discourse as a kind of genetic process.


The First Stage: Functionality and Technique
The "phase" that paves the way for the emergence of normative discourse does not in itself have anything normative: It does not fall within the realm of normative discourse proper. It is the phase of functionality, or functional explanation.
By itself, the accomplishment of a function, be it a biological one or the function accomplished by an artifact, can be accounted for in purely descriptive terms. A tree's tropism can be described by referring to the model of the functioning of a complex machine. The logic of functionality is not in the least normative, yet it is highly relevant to normative discourse, as will be seen presently.
Since living individuals typically depend on their environment for their survival, they have a positive perception of favorable circumstances and a negative perception of unfavorable ones. A tree perceives water positively, thus extending its roots in that direction: In this case water is perceived prima facie as fundamentally good, useful, desirable (in a figurative sense, of course).
The starting point in the conceptual genesis of normative discourse is the domain of reflective understanding of the different ways of securing favourable conditions (and avoiding unfavourable ones) for one's survival and biological flourishing. This is the domain where we see the development of the kind of reflection leading to the formulation of technical rules-namely, rules of the form "If you want to attain goal G, you have to phy" (where phy-ing is, it is assumed, conducive to the attainment of G; see von Wright 1963).
The technical conceptually precedes all other elements of normative discourse. Technical rules instantiate an instrumental, means-to-end rationality: The purpose is taken as given, and the rule provides instructions as to the means conducive to its attainment.
Note that there is a profound difference between the functional and the technical. The logic of functional explanations (which may in turn rely, in the case of the biological, on evolutionary explanations) does not fully overlap with the logic of technical reasoning.
Specifically, there is an important difference between the performance of a function-e.g., by an organ in a living body (the case of the tree) or by an artifactwhere the end is determined by evolution, by design, or whatever, on the one hand, and "setting oneself an end," or "having something in view as an end," on the other. The key difference is that between biological or artificial ergon, on the one hand, and skopos, on the other. "Having something in view as an end"-the optative-is not the same as blindly working to fulfill a function: It is a kind of understanding that takes human and, somehow, nonhuman minds (animals, too, have desires) beyond the domain of mere functionality.
True, the adoption of an end by somebody is a matter of fact, and it can be stated and described as such. But the optative-the attitude of having an end in view-is not the descriptive. The proper expression of the adoption of an end is not by purely descriptive language. Technical rules, articulating means-to-end rationality, imply the optative (our adoption of ends). While the optative is not yet, as such, the normative, it shares a crucial feature with so much of practical thought and discourse: a world-to-mind direction of fit (it is the world that has to fit what is thought or said; in case of a mismatch, what is deemed inadequate and, as it were, blamed on this account, is the way things go: Anscombe 1972, par. 32;Searle 1975).
Technical rules typically apply to an individual: They find their application in the first place with individual action. In using these rules, groups replicate the instrumental reasoning of individuals. There is, on the one hand, a form of instrumental rationality-strategic rationality-that specifically deals with issues where a plurality of agents are involved, and their respective choices are interdependent (Elster 1983, chap. 1). How is one to choose among options (i.e., which choice can be said to be rational), where the result of each agent's choice depends on the choices of others-and everybody knows that (everybody knows, and so on) this is so?
On the other hand, there are ways of doing things together-jointly engaging in cooperative or collective activities-which cannot be reduced to the working out or the application of technical rules by individuals (Bratman 1992;Gilbert 1989;Hurley 1989;Searle 1995;Sugden 1993).
These two forms of agency cannot be fully captured as a simple replica of individual instrumental rationality. Yet individuals involved in a strategic interaction problem are engaged in trying to find out ways of satisfying their (previously given) desires. For this reason, strategic rationality does not pose a separate issue. On the other hand, most forms of plural agency are more or less inextricably interwoven with institutions and institutional practices.

The Second Stage: Institutions and Constitutive Rules
By virtue of this new set of properties and characterizations, behaviors and actions that may previously have been interpreted as merely functional or expedient take on a meaning that goes beyond functionality or expediency: The dueness of these behaviors (their being "right" or "due") is something different.
There is an essential aspect to this contrast. Normative discourse, as opposed to a purely technical stance, does involve a consideration of ends. Practical reasoning is also about ends, which can be chosen, validated, exposed as vulnerable and fragile, criticized.
As a result, although the connection with functional explanation and technical rationality remains active and is often evident, it does not explain this new horizon of the human condition. The domain of technical rules is normatively blind insofar as it does not enable us to consider and evaluate ends, but only to work with means directed at already specified ends, which are taken as given.
Normative discourse appears to break free of the grip of the technical. Technique makes it possible to kill, but not to execute (to carry out a death sentence); it makes it possible to copulate or rape, but not to enter into matrimony. When it comes to doing these kinds of things-things and actions otherwise impossible, expanding our freedom-what comes into play is a new kind of rules, over and beyond technical ones.
Constitutive rules (rules of the form "X counts as Y in context C") make possible new forms of behavior-institutional behaviors (Searle 1969(Searle , 1995-thereby widening the range of human freedom. This way, we can do things we could not do before. Specifically, we act in normatively qualified ways by interpreting our behaviors as (institutionally) legitimate. It is constitutive rules that make possible, and populate, this stage of normative discourse. Moreover, the world of institutions requires us to choose from among several possible ends.
In the institutional world, however, our choices are more or less firmly constrained by the roles we happen to play, by our positions on the social and institutional chessboard. It is "as an X" (as a father, a teacher, and so on) that, in institutional settings, we face a choice between ends. True, we can, at least sometimes, choose whether to take part in a given institution or not, whether to play a given institutional role or not. From what has been said so far, however, nothing shows that these choices also fall within the scope of normative discourse. There appears to be a distinction to be traced between infra-institutional "oughts" or reasons, on the one hand, and extra-institutional "oughts" or reasons, on the other hand.
In this account of full normative discourse, however, there is no stark distinction between these two "oughts," these two kinds of reasons. There are no fully extrainstitutional "oughts," no wholly decontextualized reasons for action: Fully normative "oughts," good reasons for action, develop critically from within institutional settings, partially transcending them. There is no way to give an exhaustive account of the institutional world in terms of means-end rationality and functionality alone. Constitutive rules are different from technical rules.
True, constitutive rules make room for new technical rules. In institutional reality, new functions come into existence. Both points are unproblematic. First, as hinted above, institutional practices, constituted by sets of constitutive rules, identify more or less specific sets of (institutional) goals, or ends. Awareness of the (institutional) conditions conducive to the attainment of these ends generates new sets of technical rules ("If you want to get married, you have to do this or that" and so on). Second, together with these ends, and ways of attaining them, new functions develop: Institutional roles and characters set out functions which can be more or less effectively accomplished by people occupying those roles or playing those characters. Words like teacher, president, and so on, are functional words. This, once again, shows that the connection with the previous stage remains open.
There is a necessary connection between constitutive rules, on the one hand, and (institutional) prescriptions, on the other. Constitutive rules, in Searle's (1969, 34) phrase, make possible, "and also regulate," new forms of behavior. Despite the logical-genetic primacy of constitutive rules-their primacy in the process by which normative discourse emerges, as well as in the institutional world at large, a specifically human world-regulative rules, their necessary side effect, are also there from the start. How do regulative rules "follow" from constitutive ones, i.e., how exactly is the normative "ought" of regulative rules "entailed" by sets of (shared) constitutive rules?
There is no single way in which normative concepts and arguments emerge in institutional settings. Institutional practices and normative discourse are interwoven in several ways. The simple way, as shown in the previous section, is as follows: Institutions are made up of constitutive rules; they could not work, however, without regulative rules. Constitutive rules create and also regulate the relevant kinds of behavior.
Prescriptions arising in this way are, on the face of it, fully infra-institutional, contextual rules. But there are more complex issues involved, and different connections. Obligations, prohibitions, and permissions-the mark of the normativemay also come into existence (institutional existence) as a component part of a peculiar kind of institutions, namely, institutions which, although existing (like any other institution) by virtue of agreement and shared beliefs and attitudes, are in fact necessary conditions for the viability of any human society: basic institutions which secure the minimal conditions for the existence of a relatively stable, peaceful, ordered social group.
While particular institutions, such as games or academic rituals, define particular, restricted contexts, which can more or less dramatically be transcended (participants can more or less easily step down and quit), such basic institutions are, in a sense, nontranscendible. The contexts they define are, for empirical or conceptual reasons, a necessary framework for human life in society as such (as Hart shows, at a suitable level of abstraction, the two kinds of reasons cannot, in this case, be separated from each other: Hart 1958, 79-80;1961, 188). The obligations, permissions, powers, and responsibilities associated with these institutions may take various forms in different temporal and spatial settings. However, they recur as recognizable instances of a single recognizable abstract type.
Nevertheless, in the case of these (nontranscendible) institutions, reasoning as to the means for achieving the ends embodied in the institution may generate nontranscendible technical "oughts"-nontranscendible technical reasons for action. Ways of achieving the goals laid down by the institution may simply be means that human life in society cannot do without.
Thus, in many ways the institutional world implies normative discourse. Some regulative rules have an instrumental underpinning. They are more or less firmly grounded in the goals embodied in institutions, or in the goals we may, or in some cases must, pursue by engaging in institutional practices. However, a remarkable fact about our institutional world is that it also takes shape through prescriptions that go well beyond functionality and instrumental reason. Some prohibitions take on a stern appearance that allows no scrutiny of their conduciveness to any identifiable ends or functions, and admits of no exceptions of any kind.
The phenomenon is two-sided. On the one hand, we are facing here the transcending of both functionality and instrumentality. On the other, we find that some rules seem to allow no exceptions. They are, apparently, neither prima facie or pro tanto reasons nor defeasible ones. Or, in other words, some prescriptive generalizations become, in Frederick Schauer's (1991) phrase, "entrenched." Nontechnical "oughts" seem to arise as "absolute" (neither defeasible nor merely contributory, or pro tanto) requirements. As we shall see, it is only through further development-i.e., at a later stage in the genesis of normative discourse-and through critical self-reflection, that such requirements, or at least some of them, finally become defeasible, pro tanto, reasons for action.
In light of these considerations the "certain" should be understood as the historical datum, the effectively given (pretended) authority-or authoritative dictumthat prescribes without necessarily explaining, and with an attitude that seems brutal. Some rules-legal rules, moral rules, and so on-exhibit a characteristic lack of rationality, a feature widely known since Aristotle: These rules present themselves as arbitrary. Such arbitrariness can even run contrary to functionality or to instrumental rationality, and brings in what appears to be a purely normative dueness. In short, the regulative rules framing the institutional articulation of normative discourse are characterized by a possible but constitutive lack of rationality. The relative lack of rationality characterizing lists of commandments, as well as social taboos and the like, is handed down as authoritative tradition, ingrained habits, dispositions, embodied ways of thinking and acting: a huge part of our "second nature."

The Third Stage: From Criticism of the "Certain" to Principles
Fully normative discourse is critical reflection on what Vico (1999) called certo (certain). Institutional normative practice stands in need of justification. When asking for a justification of the certain, what form of rationality are we bringing into play? In order for critical reflection to take off, we need standards of correctness: standards of right and wrong, reasons for treating institutional prescriptions as pro tanto reasons or as defeasible generalizations. In other words, what is the meaning of "ought" within our critical exercise of reason, in our quest for a justification of the certo?
I submit that a (qualified) holistic and coherentist approach provides an answer to these questions. In a nutshell, we avail ourselves of pieces of institutions, in order to criticize other pieces of (different) institutions. It is by appeal to (favored segments of) traditions that we ask for justification of different (segments of other) traditions.
Normative discourse, at its critical stage, is holistic and coherentist. First and foremost, it does not leave behind instrumental rationality, consigning it to the past. Rather, it calls into question the distinction between instrumental and noninstrumental rationality, i.e., it shows the distinction to be provisional and of limited import.
While we may certainly ask ourselves what the causally efficient means to a given well-defined and circumscribed end are (the domain of technique), when it comes to open-ended goals (where it is possible to devise many conceptions of these goals), and to noncausal means (ways of achieving such goals) things take a different turn. Deliberation about means and deliberation about ends can and do interact (Dewey 1929). 1 Deliberating about the appropriate way of pursuing the end may amount to scrutinizing which among the different conceptions of the end is to be pursued. Furthermore, it calls into play further ends, as well as ways of conceiving of and pursuing them ("the appropriate way"). This gives rise to a net of interacting ends, ways of conceiving them (specifications, determinations), and ways of pursuing them. There is no need for a different form of rationality (different, that is, from instrumental rationality). The point where full-fledged, critical normative discourse emerges is where instrumental rationality becomes holistic and coherentist: It is here that we part company with functionality and technique.
In using (segments of) traditions in order to seek justification for (other segments of other) traditions, we move at various levels of generality (Rawls 1993, 8). We try to frame principles that could selectively explain segments of our institutions, and we appeal to considered judgements that provisionally hold as fixed points in our deliberative and evaluating practice. The process is intended to achieve reflective equilibrium between principles and judgments, at all levels of generality.
There thus emerges in the institutional world a fundamental characteristic of normative discourse tout court, in that the institutional world necessarily (and historically) takes in an element of arbitrariness. Technical reasoning, as such and unqualified, explains itself without residue. Normative discourse, on the other hand, pronounces actions, attitudes, and so on to be right or due, without-or at least without necessarily-referring to a technical rule, or set thereof, by which to explain, exhaustively, such dueness. In institutional settings, actions and behaviors will be described as right, good, worthy, called for, pure (or else wrong, bad, unworthy, uncalled for, impure) regardless of their direct or indirect effects, as assessed from a purely technical standpoint, and even if these effects should work against functionality. Duties and obligations will come to the fore, and they will do so even when lacking functionality or any technical underpinning. This is the philosophical error most apt to defeat, at this stage, a rational reflection on normative discourse. One might come to think that because rules necessarily present themselves as carrying an element of arbitrariness, when this element becomes excessively burdensome, because it is counter-functional or otherwise technically ineffective ("times change"), a critical revision will then be in order. This process often occurs, of course, but by underscoring this particular source of the need for critical revision, we are sure to miss a fundamental characteristic of normative discourse, namely, the fact that it is not by accident nor episodically that normative discourse becomes subject to criticism. Rather, it does so constitutively (though sometimes unwittingly), precisely because it starts out on a potentially arbitrary basis.
This poses two interrelated problems. First, in normative discourse at its (conceptual) birth, in its transcending of functionality and technique, there is an element of arbitrariness (the "certain"). But, it may be asked, is there also, in normative discourse at its (conceptual) origins, a non-arbitrary component (i.e., one which cannot be reduced to functionality and technique)? Second, let us grant that the possibility of criticism is a structural, constitutive feature of normative discourse, this feat being explained by its-at least partial-arbitrariness. What are the criteria with respect to which this possibility is to be exercised?
An answer to the first question, albeit a partial one, is provided by arguments purporting to show that the world of human institutions is rooted in a framework of highly abstract and indeterminate needs and necessities that work as a nontranscendible context for the development of viable, relatively stable societies. The arbitrariness of normative discourse, at its origin, lies (in part) in the determination and specification of highly abstract, generic, and largely indeterminate duties, prohibitions, powers, and responsibilities arising from this framework.
As to the second question, it is clear that the criteria for criticism cannot be provided by this abstract framework of needs and necessities, taken by itself. So, where do they come from? The third aspect, or "stage," of normative discourse is that of critique. The institutional is criticized by "forcing" it to provide its own justification: Rules need to be argued for. From this perspective, every institution must be justified, irrespective of the contents of that institution, and irrespective of the actors who have issued the relevant prescriptions or who represent the institution. The criteria by which regulative rules are to be criticized and argued for we call principles.
The language required by normative discourse at its third stage of development (critique of institutional reality) is the language of argumentation. Argumentation is carried forward on the basis of principles. Principle-based argumentation revolves around justification, the giving and taking of reasons; it requires arguments and explanations. Principles make it possible to effectively criticize the arbitrariness of rules and their irrationality (their argumentative irrationality).
Argumentation presupposes a minimal mutual Anerkennung by which the parties acknowledge each other as acknowledging each other, and so on; i.e., they see each other as "self-authenticating sources of valid claims" (to borrow Rawls's phrase: 1993, 32). In other words, where, and to the extent that, arguments are offered or required the parties recognize each other as capable of giving and taking reasons: of following, as it were, the thread of reasons. This is the practical, morally significant side of argumentation. Argumentation itself is a practice, partly an institutionalized one, with its own constitutive rules, roles, and so on: It is governed by an ethic of its own (engaging in dialectics requires specific education). Recognizing each other as selfauthenticating sources of valid claims, the parties to an argument do, at least within the argument, credit each other with the equal dignity of persons. It is here that normative discourse shows its inner drive, leading it beyond the institutional stage. Crucial to the ethic of argumentation is an attitude of "equal concern and respect" that implies an attempt to overcome the very arbitrariness of institutions. The parties "must" offer reasons which present themselves as grounded in standards which lie beyond the raw authority embedded in the institutions. Putting forward such reasons can be defined as "argument," and the relevant standards are "principles." What is peculiar to principles is that resort to them is governed by the mutual acknowledgment of the parties to a discourse, by the "equal concern and respect" attitude, which is constitutive of argumentative practice. Principles, then, exhibit a characteristic double-sided openness. On the one hand, they do not have the rigidity of rules-or, rather, the rigidity that rules pretend to have. On the other hand, they are addressed to, and claim to be appealing to, parties to a common argument not sharing our-disputed-views. This is where their strength-their inconclusive strength, as we shall see-lies.

Relevance and the "Situation Potential"
What has been said so far suggests that the conclusiveness requirement may not be the only fundamental requirement at work in the construction of normative arguments. Jullien (2004) offers an interesting insight in his treatment of efficacy: Instead of constructing an ideal Form that we then project on to things, we could try to detect the factors whose configuration is favorable to the task at hand; [. . .]. In short, instead of imposing our plan upon the world, we could rely on the potential inherent in the situation. (ibid., 16) We find here instructive pages devoted to bringing out the distinctive features of Chinese strategy, as discussed in the "ancient treatises," the key idea throughout being that of leveraging the "situation potential." We rule out the idea of fixing the course of events on the basis of a plan worked out in advance, and more or less final. We therefore work our way out of the logic of the model (the eidos as a project-and-idea tasked with prefiguring the shape of things to come) and take up what for Jullien is the Chinese logic of things in process (the logic of development).
In other words, the situation potential is circumstantial and circumstances make up the potential. Thus, what becomes fundamental is to look at the various factors at play, having their exploitation in sight: strategic exploitation in the Chinese doctrine of efficacy, argumentative exploitation in the context of a normative critique of institutions. The heedful study of circumstances is of crucial importance. In military strategy and diplomacy alike, an evaluation of the factors at play is always a must, for the elements that combine to make up a given situation always consist of that wide constellation of circumstances on whose potential we must somehow operate.
To proceed from the circumstances, to pay heed to the specific situational datum, is to embrace a different perspective-different from the technical-poietical attitude aimed at the construction of a normative model to be brought into effect. To proceed with a model is to accept a principle of conclusiveness: An argument bound up with a model will be more or less conclusive, and here the perfection of a normative argument would ideally coincide with the closure of discussion. If an argument presents itself as perfectly conclusive, argued discussion will thus appear to have come to an end. The various arguments conceived from within the logic of an ideal project to be carried through can be more or less easily compared with one another by the measure of their greater or lesser conclusive force.
If, on the contrary, we start out from the existing institutional and normative horizon and we interpret it in the first place as a situation whose potential, in the sense specified above, is what matters normatively, with a view to developing principled arguments, we shall no longer evaluate our argument by the criterion of its conclusiveness alone. We may have left out of the account a certain aspect of the situation, a detail that carries decisive implications; we may have failed to consider the potential of the initial situation in its irreducible specificity. The import of a critical normative argument will be that of proving more or less relevant as far as the potential of the situation is concerned.
Relevant means "pertinent": There is expressed the idea that an argument is particularly suited to criticizing a given institution (or in general any given situation), and therefore there is expressed the force or value of an argument. The force of an argument will depend on whether the argument proves satisfactorily suited to the circumstances. Importance depends on pertinence.
At this stage, a word of caution is in order. First and foremost, we must take seriously the possibility that the fragility of normative argument may also show itself in its inconclusiveness-meaning, by this, not that it goes on without end when relevant considerations are put forward (this is something I have already granted), but rather that it can end in a stalemate. We have not been promised that a solution necessarily exists for all our problems. This is the case with genuine dilemmas.
Furthermore, the very notion of a "relevance principle" can be questioned. The principle, if spelt out ("Argue relevantly!") would be in the best scenario quite trivial, and unfaithful to what relevance is about, namely, the potential of specific situations. The point about relevance is, at bottom, that there are no settled rules governing it, and there never can be.
Finally, normative discourse seems after all unable to reject other kinds of arguments that aspire to be conclusive. The relevance principle is itself more or less relevant: It is not autonomously and unilaterally conclusive. The logic expressed by the principle applies to the principle itself.
The situation potential to which critical argument governed by the relevance principle refers is not an atomic unit, a shell of normative potential: it unfolds as a network of related factors. The situation potential does not present itself as a singular, given, atomic context capsule; on the contrary, the institutional realities upon which the situation potential depends, and from which normative discourse properly speaking emerges, are nested realities, meaning that they exist in a system of relations and interdependencies and are closely bound up with a social and institutional context to be understood as intensely plural, especially after the demise of compact, homogeneous societies that thrive on sets of shared values. An institution's original self-interpretation, then, is not to be taken as decisive. The point is delicate and apt to produce misunderstandings. Imagine three circular blots of color, say, red, yellow, and blue. Now, keep the red circle red and replace the other two with scarlet and crimson circles. The red circle has not changed one bit, yet its role in the system in which it is now nested (quite a simple system in truth) has altogether metamorphosed: Red initially existed as one of the three primary colors; now we have red as a chromatic variation on a common base. Similarly, the institutional practice of arranged marriages may have one meaning within its original context and take on an entirely different meaning when nested in a different, pluralistic environment. In such a context, to arrange a marriage may mean to oppress or exploit women, to exert undue pressure on individuals, to exploit the younger members of the household, and so on.
The relevance principle, then, does not operate upon situations given as capsules, as circumscribed elements free of context. Further, the principle does not operate independently of a given doctrinal tradition: Normative arguments are produced in a thick web of erudite references, footnotes, citations, conceptual loans, options drawn more or less consciously from a reservoir of more or less shared concepts. Even resorting to Aristotelian endoxa (the weight of authoritative opinion) has an operative rationality of its own. It is not in a void that arguments are devised but within a doctrinal and material horizon, a horizon that needs to be understood and interpreted.

The Technical Use of Institutional Facts
It is now time to say something more about the institutionalist assumptions underlying this account of normative discourse. The starting point is an obvious remark: Terms that are used to refer to institutions, to institutional facts, entities or properties, do not refer to anything that, as it were, is directly observable. Or, rather, the way they do so is profoundly different from the way in which terms referring to natural entities refer to directly observable entities.
A Quercus pedunculata is directly observable, just as a citizen is. True, the systematic typology which makes what we see a Quercus pedunculata (versus, say, a Quercus sessiliflora), and the complex, if not explicit, legal philosophy implied in the term "citizen" are the theoretical background underpinning the meaning of these terms. Trees and human beings are of course conceivable only in the frame of a broader typology.
So far, so good. Citizens and quercus differ in one crucial respect, however. According to Searle (1995) brute facts need some kind of linguistic convention in order to be described: Described facts are still brute facts, though, while the fact that we describe them is already institutional. Citizens, contracts, or money, on the other hand, would not even be there unless they were described-understood, and used-as such. Furthermore, their conceptual structure is given by sets of deontic powers, positive and negative (i.e., obligations and prohibitions; permissions, rights, and responsibilities; and so on). Thus, a different kind of institutional background underpins such terms. What such terms refer to would not even exist were it not for sets of constitutive rules. Institutional terms, like citizen, unlike quercus, presuppose a normative frame both for their use and for the very existence of the kind of entity they refer to.
Institutional terms are thus doubly connected with rules. Apart from their use depending on linguistic rules-a feature they share with noninstitutional termswhat they express is the concept of a kind of thing whose very possibility is itself constituted by (both constitutive and regulative) nonlinguistic rules (meaning rules which govern not only linguistic behavior but also behavior tout court). We can understand them only as far as we understand that (sets of) constitutive rules are in place. We cannot understand what "signing a contract," "cashing a cheque," or "marrying" are without being aware of a preexisting normative world.
As a result, we often need to conjure up the inherently normative nature of some actions, in order to understand the meaning of the verb which describes them. The same holds for objects, properties, facts, and so on. Such terms are utterly meaningless without a presupposed system of rules (even minimal). Without this background, what they refer to loses its reality-if it has any at all.
All this does not rule out the possibility of manipulating institutional facts. Can I use money as a weapon, can I use a weapon as money? The answer is (a qualified) yes. Nothing is against that. In fact, it is always possible to step out of the shared circle of relevant beliefs and construe any constitutive rule which underlies a legal or social institution as a technical rule which informs us about the right way of proceeding in order to pursue our own advantage (Trasymachus' injustice argument; Plato 1982, 47-9). If we take such a step, we may try, if we feel confident enough, not to obey, but to steal, kill, pursue power and pleasure; if we feel ambitious we can also aspire to enjoy the prerogatives of dictatorship and tyranny. It is, in other words, always possible to use law and norms in a technical way, i.e., as a means to an end-an end quite different from the ends of law and norms when we accept them in their constitutive force.
Constitutive rules can therefore be read as technical rules; and the same holds for the prescriptions which, as we claimed above, are part and parcel of institutional practices. Soccer is constituted by those rules which describe and set the normative existence of the game-based on the skill of mastering a ball without using the hands or deliberately hurting the other players. Now, the regulative rules (e.g., "Do not hurt members of the other team") which derive from soccer's constitutive rules can always be read as a technical rule ("Do not hurt members of the other team if you can foresee that this will mean trouble for you or for your team; do not hesitate to act this way if you are sure it will be useful-because nobody can see you or because it is absolutely necessary to stop that dangerous action"). Such a technical rule does not belong to the normative core of the game: It is, rather, parasitic on it.
A distinction is called for here. Any cluster of constitutive rules, defining an institutional practice, will imply a number of technical rules which play a role as guidelines for those who want to engage in that practice. This "proper technical use" of constitutive rules, and of the prescriptions deriving from them, is different from the parasitic one just illustrated. The distinction, however, is a tricky one.
If every institution had a precise end or set of ends of its own, it could be claimed that a parasitic use of a constitutive rule-one of the rules defining the institution-is any use that does not have in sight the proper end of that institution. This, however, is not always the case. Institutional practices do not always serve a precise end or set of ends. Moreover, given an institutional practice, we may often distinguish between internal ends and external ones. Now, between the internal ends of an institution (to score at a soccer game), proper external ends (winning), and improper external ends (to earn a huge sum of money out of it, no matter what), there is often a continuum.
A rough-and-ready criterion will be the following: When the ends you have in view in taking part in such and such a practice could not be explicitly stated without undermining your participation in the practice, then you are a "parasite." But this criterion probably does not cover all possible cases. Generally speaking, the parasitic use of an institution has to do with cheating, a manipulative or dissimulative strategy revolving around some kind of hoax. It may look like you are playing soccer, but, truth be said, you are doing something else.
A sheer technico-political rule of government would thus read: "Pass laws to your advantage, in order to remain the stronger party and enjoy the prerogatives of dictatorship." Trasymachus nevertheless warns us that even such selfish, unscrupulous laws must "chastise as a law-breaker and a wrongdoer" (Plato 1982, 49) the man who deviates from them. In other words, the stronger actor has to "proclaim" the just in order to equate it with her or his own advantage. Not even the cruel dictator who rules in reckless disregard of any moral scruple can afford to set aside the concept of justice. Each government must conjure up that idea, but for an instrumental use. "Italy is un unjust republic" cannot be a constitutional provision: This would involve us in a performative contradiction (Alexy 1992).
According to Trasymachus this is the one principle of justice that identically obtains in all states. Specific rules, technically arising out of the tyrant's individual belief (concerning her or his advantage), must always lend themselves to being construed as norms underlying genuine institutional facts.
A government can be unjust, but it cannot proclaim itself to be unjust. The claim of justice seems to be a constitutive element of its very nature. Authority is de facto authority (it qualifies as such) only if it claims de jure authority (Raz 1986)-even if such a claim is unfulfilled, or mere pretense, an attempt at manipulation.
Even according to Trasymachus' conception (law as a technical device used by an individual in order to master brute facts), law must generate a set of collective shared beliefs supporting institutional facts. The law, a mere weapon in the hands of an unjust individual, must claim to be "moral" and "just": It must claim to have a justification. By this very claim, it starts to be processed as an institutional fact by the collectivity which lives under it. Its claim to justified authority is, at once, the first step of normative criticism. This is not meant as a sociological remark: Even if conceived, or wielded, as a weapon, law has a structure that implies liability to normative critique. It cannot help but claim authority, and it therefore lends itself to an inquiry on the normative soundness, or legitimacy, of such a claim.
Law as a simple weapon, or mere technico-political rule, enjoys no conceptual autonomy.
There is, then, an everlasting possibility that, on the one hand, technical rules, which are supposed to master brute facts, may have institutional consequences and, on the other, that institutional facts and norms, which are supposed to be the result of constitutive rules and common intentionality, as well as full-fledged normative argument, prove vulnerable to a merely technical exploitation. This twosided challenge to every conception which tries to rely on institutional facts to explain social reality is unavoidable.

The Paradox of Normative Discourse
The endpoint of normative discourse could be described as positions that everyone should accept, but that someone could, by argument, reject. This is a crucial, and delicate, issue. First, "positions that everyone should accept" means validly drawn conclusions, i.e., conclusions that, given the premises, should, in fact, be accepted. This does not yet differentiate critical normative arguments from arguments oriented by the conclusiveness requirement. On the other hand, claiming that critical normative argument, if successful, leads to conclusions that everyone must accept would simply be too strong: Such a claim would put it on a par with, for example, the demonstration of theorems in Euclidean geometry.
So, "positions that everyone should accept" should be read, here, as a weaker claim. It does not mean that everyone is rationally compelled to accept them. It means, rather, that from the point of view of the speaker-the one who puts forward the argument-it is reasonable to claim that everyone should reasonably accept them. Such a claim puts pressure on each member of the audience to which the argument is addressed, because the argument is conceived as relevant and must therefore be either accepted or dealt with through counterargument.
The additional clause "but that someone could, by argument, reject" is meant both to prevent, once again, the assimilation to "conclusive" arguments, and to keep open the structure of normative discourse. If an argument is, so to speak, too strong, it defies the logic of normative discourse.
As long as we keep within the domain of normative discourse, the path of a possible counterargument must be clear of obstacles. So, for instance, some of the arguments provided by the New Natural Law Theory (Finnis 1991) are certainly interesting: Any action that is directed against a fundamental value is simply ruled out, as such, from the range of legitimate options. No exception is allowed, in principle. These arguments, which deserve respect from a scientific point of view, are nevertheless too strong to qualify, according to the here described notion of normative discourse, as principled arguments. Were they accepted as successful, the discussion would be concluded, finished.
A standard normative argument, then, is so structured that it necessarily leads to a paradox. On the one hand, if used to force others into belief, it serves a technical function: Wielded as a weapon, it merely aims at generating persuasion as a desired strategic result. The point, however, is that not only does a normative argument cease to be such the moment it becomes merely a rhetorical tool, if it convinces us in the manner of a fox or threatens us in the manner of a lion; it also ceases to be such the moment it aspires, in its structure, to be conclusive (and therefore to bring normative discourse to conclusion).
On the other hand (the other side of the paradox), too gentle an argument will not be an argument at all: It will rather be an articulated plea. If an argument does not have us cornered-if it fails to somehow force us to accept the normative direction provided by the argument itself-it will too closely resemble a petition, a complex request for a favor, an argued entreaty.
Only an argument aspiring to relevance and principled universality can demand this or that policy "by normative power," rather than by force and concession. A normative conclusion cannot present itself as a privilege granted, or as the outcome of a standard bargaining process: An articulated argument is neither pleading nor deceptive.
An argument will be all the more relevant if it meets certain standards. Thus, we do not want an argument to exhibit logical or pragmatic fallacies, or be inadequate to the problem at hand, or fail to make an articulate connection between the critique offered by the principles it relies on and the situation potential considered in its full complexity. One error that typically weakens an argument is oversimplification of the problem under scrutiny. Conversely, an argument will be the stronger for showing articulation, complexity, and richness. These are obvious requirements. It may not be trivial, however, to dwell on some of their different aspects.
A good normative argument will not only take into account the complexity of the institutional reality it is operating on, but will also help define this reality. A relevant argument, first, teaches us something about the institutional world it refers to, and is in this sense informative. Second, it contributes-by reinterpreting both words and deeds, by shifting accents, by projecting analogies, by making the implicit explicit, and so on-to modifying the intentionality that shapes the structure and point of the institution. Thus, an argument is found to be convincing to the extent that, in describing and reshaping the institutional reality we live in or would like to live in, it meets both a requirement of adequacy and a requirement of appeal.
Relevance both dictates respect for the situation potential, considered in its specificity and requires appropriateness, as manifested in the principles implied in the arguments produced. Rejecting articulation along these two dimensions is tantamount to embracing the technical world of weapons.
Finally, it should be emphasized that, as in the case of any other human enterprise, normative discourse can (and often does) fail. There can be several different explanations for this, most notably disattentiveness, lack of creativity, inaccurate assessment of the circumstances, poor listening skills, intractable disagreement, or sheer bad luck (rules and principles, facts and values are so arranged that no discursive path is open). We are dealing here with a fully human activity: There is no promise of success (neither by God nor by a Leibnizian belief in a good faith Calculemus), no natural law determinate enough to unerringly lead the way, and certainly no liberal shortcuts.

Department of Jurisprudence University of Modena and Reggio Emilia
Via San Geminiano, 3 41100 Modena Italy E-mail: gianfrancesco.zanetti@unimore.it