If the fundamentalist goal of a theory of sense is given up, the possibility of an explicative theory of sense follows, but so does, at the same time, the necessity of having to accept several language-action systems. This corresponds to Wittgenstein's step from the Tractatus to the Philosophical Investigations, in the sense of a radical withdrawal of the foundations that are claimed in explaining the possibility of the significant proposition.¹⁰ In a certain respect one thereby goes from the frying pan into the fire. If one admits only a single language system as the possibility of language, then speaking about it falls outside its possibilities (as the Tractatus shows) and would be nonsensical. If one admits several, the theory of sense seems to become pointless (as the Philosophical Investigations suggest here and there), because the many language-games, as Wittgenstein puts it categorically and yet, as a philosopher, visibly frustrated, have "no one thing in common" at all (PI § 65).

How, then, can the functioning of a language-game be explained? How can the (self-referential) connection between meaning (possibility of sense) and sense be explained? In the Philosophical Investigations this question is taken up as the problem of correct rule-following: how can the connection between a rule and its application be explained? In several examples and counter-examples Wittgenstein shows that rules cannot be fixed in their content by binding instructions for action, since every action can somehow be brought into accord with one and the same rule, even an "obviously" incorrect one. Wittgenstein's decisive insight is that the application of a rule is not its interpretation (which can be arbitrary) but action within the framework of a particular practice of a form of life, and that this action takes place (unlike interpretation) without choosing between alternatives: "When I follow the rule, I do not choose. I follow the rule blindly." (PI § 219) Whereas in the Tractatus the universal language system guaranteed the self-referential connection between possibility of sense (rule) and sense (application of the rule), in the Philosophical Investigations this is achieved by the respective form of life.

Once again: the fact that the interpretation of a rule can bring every action into accord with it shows "that there is a way of grasping a rule which is not an interpretation" (PI § 201).

"And hence also 'obeying a rule' is a practice. And to think one is obeying a rule is not to obey a rule. Hence it is not possible to obey a rule 'privately': otherwise thinking one was obeying a rule would be the same thing as obeying it." (PI § 202)

"It is not possible that there should have been only one occasion on which someone obeyed a rule." (PI § 199)

This raises the requirement that rule-following be mutually checkable: rules and language-action games would have to be understandable, legitimable and criticizable. This in turn raises the question of the ethical dimension: if the "heterogeneity of language-games" (Lyotard 1986) is to be assumed, what becomes of ethics, which after all claims universal validity? This is the culmination point of the current postmodernism debate. The heterogeneity of language-games makes intersubjectivity, legitimation and critique urgently necessary, and at the same time impossible: for a language-game that strives for these "does violence to the heterogeneity of language-games" (Lyotard 1986, p. 16).

Although, from a fundamentalist point of view, language-action games have nothing in common (although, that is, not a single possibility of sense occurs in every one of them), the following can nevertheless be established for all of them together. (1) Each can function smoothly as a self-referential game without the regulation of law and order.¹¹ (2) Beyond this, each language-action game points to an alternative, to openness, at the moment when its functioning becomes a problem and the language-action game at the same time becomes tangible as a restriction, i.e. as a system. The language-action game then no longer functions, as in (1), as a closed self-referential system; the horizon within which self-regulation now takes place widens to alternatives, to other language-action games. But since these are not of the same kind but show "family resemblances" (PI § 67), they do not permit a universal language-game of consensus, but rather a dynamic form of life between different language-games. A language-action game holds without alternative (a priori) for a communicative situation, but a theory-of-sense explication is in principle open with regard to alternative language-action games. (3) The idea of this openness is the language-action ideal. A language-action game can only ever be brought into a system, evaluated as such, and so on, by another language-action game against the background of the language-action ideal. But the ideal as a game (the ideal language-action game) does not exist.¹²

What perspectives follow from this for a theory of sense?

(1) A theory of sense cannot single out and monopolize one language-action game: for example, one type of science (the error of Kantian transcendental philosophy), or one language-game (the error of the Tractatus), or one world-view (the error of every ethnocentric ethnology). What is required is the opening of the theory of sense with regard to the variety of all possible language-action systems: from the systems of the sciences "down" to the systems of everyday life, here from the social to the psycho(patho)logical dimension, and "across" to the systems of foreign cultures.

(2) To understand a foreign system of sense means to search self-referentially, within the framework of one's own system, for possibilities of sense that make it possible to reconstruct and respect (as far as possible) the self-reference of the other system. Playing or joining in (understanding) a language-action game rests on implicit knowledge of the rules of the system in question. If understanding functions, the game cannot be got behind; it is closed, self-referentially secured, autarkic. What functions as reality in this game has the integrity of a reality in itself. Only when doubts arise about what is to be understood (what someone means) does communication rise to another level. A theory of sense always starts from the difference between two systems of sense, which arises when a linguistic act is not understood.¹³ This means that theory-of-sense insights into understanding are grounded in the fact of not-understanding. A theory of sense arrives via negationis (through alternatives, demarcations and counter-examples) at an understanding of the system of sense. I fail to understand as long as I do not know the possibilities of sense of this system; I "experiment" with possibilities of sense until I can reconstruct the self-reference between linguistic act and possibility of sense.¹⁴

If, on the other hand, a system of sense is described within the horizon of foreign possibilities of sense, it is treated as other-referential. It is alienated into an object of scientific explanations, and the linguistic acts of this system are degraded to forms of behaviour to be explained causally.¹⁵ But: "We must do away with all explanation [in the sense of a scientific explanation – K.N.], and description alone must take its place." (PI § 109) Description aims at the systematization of the language-action game in the sense of a "perspicuous representation" (PI § 122) of what "lies open to view" anyway (PI § 126) when the game is understood, and what therefore does not need to be explained (ibid.). Language-games whose possibilities of sense are familiar and systematically available, whose "grammar", whose "use of words" we "command a clear view of", since we may "invent" them ourselves (PI § 122), function here as "objects of comparison" (PI § 130), but without any normative claim:

"For we can avoid unfairness or vacuity in our assertions only by presenting the model as what it is, as an object of comparison – as, so to speak, a measuring-rod; not as a preconceived idea to which reality must correspond. (The dogmatism into which we fall so easily in doing philosophy.)" (PI § 131)

Every system must be regarded as rational. Rationality, seen in a value-neutral (i.e. system-neutral) way, is in the first instance nothing other than the connection between a possibility of sense and a linguistic act, between system and game. In this respect rationality is to be respected as a category immanent to the system. Criteria of rationality of one system cannot be applied to another. The same holds for criteria of objectivity: what is real in one system need not be real in another, and vice versa. Reality and objectivity show themselves in the linguistic acts of a system: they are the epitome of the possibilities of sense available in a system. The existence of an object is the possibility of being able to speak about it significantly within a language-game; the object is a possibility of sense. "What looks as if it had to exist, is part of the language" (PI § 50) of a given language-game.

(3) The plurality of systems, including its consequences (the equal value of all systems; the relativization of criteria of rationality and objectivity), leaves every evaluative discipline, above all ethics, a traditionally highly developed philosophical discipline, few chances of development,¹⁶ insofar as it pursues traditional goals. Presumably a turn is under way in ethics similar to the pragmatic turn of the Philosophical Investigations, which detached the explanation of language from the assumption of (possibly universal) mentalistic acts conferring meaning and grounded it in linguistic practice.

In the field of tension between the heterogeneity of language-games and the corresponding urgency of mediating between them, one thing is in any case clear. If a theory of sense avoided those principles that are tacitly in play when one (e.g. one's own) language-action system is singled out, then ethics, as an idealizing opening of the horizon of a language-action ideal, would have the appropriate space of freedom.¹⁷ Perhaps an ethical intervention could start more from communicative practice and understand conflicts as conflicts between different action games, not between an (ideal-typified) front of good and evil. Perhaps a normative theory of this kind, with a view to the relativizability in principle of games within the horizon of the language-action ideal, can offer new action games. More generally, it can bring innovative strategies into play with which one can step out of games and with which, in cases of conflict, threatening system fixations and boundaries that clash heterogeneously can be overcome.