The basic idea of our project consists in a generalization of speech act theory to an analysis of communicative and cooperative actions in general, of linguistic acts in particular. This generalization takes, however, a distinctive form.

In order to distinguish ‘classical’ speech act theory in the tradition of Austin and Searle from our generalized approach, we would like to call the latter a "pragmatic analysis of the form(s) of communicative and joint actions" or, shorter, "form analysis of communication". Classical speech act theory presupposes propositional content and conventional linguistic meaning as already understood. It believes in a method of making some implicit rules of human actions explicit. Pragmatic form analysis, on the other hand, does not start with believes in a concept of preestablished meaning. It does not presuppose contents of expressions or of speech acts that are only actualized in a particular utterance. Nor do we take the distinction between types and token as already understood. It is only a relative distinction: a particular attempt to perform a communicative act is even as a particular token always to be ‘understood’ as a type. The hearer’s reactions must be of the ‘right type’, and this concept of correctness refers to a certain way of attributing a type to what the speaker S is doing. The attribution of types and the judgements of correctness take place in a realm of cooperation, joint actions and joint normative judgements.

Moreover, we have to distinguish between explicit attribution of types by naming them and implicit re-actions of the hearer H that fit to the (type of) action of S. That is, the implicit or explicit attribution of types to singular actions takes place in a common and open realm. The implicit or explicit judgement of the correctness of such an‘understanding’, as it is shown in H’s re-action, takes place in a kind of joint jury. Neither the speaker’s judgements nor the hearer’s nor any mere appeal to ‘usual’ or ‘normal’ meanings and understandings alone will be sufficient. But, of course, conventional attributions of ‘types’, normal rules of ‘inferences’, conventional rules of language use, for instance, help a lot and are, therefore, accepted as implicit norms of correctness of understanding a communicative action. They are used also as explicit appeal courts in disputes about the ‘correctness’ (and about the ‘entitlements’) of H’s understanding as it is shown in his (verbal or non-verbal) reactions.

In order to describe ‘norms’ and ‘rules’ in joint interactions, it is indeed a good idea to distinguish with Robert Brandom very generally between norms and rules of commitment and of entitlement . The distinction of these two kinds of ‘rules’ are quite well explained in the context of Lorenzen’s dialogical logic . Rules of commitment of a ‘proponent’ P tell us what an ‘opponent’ O may ask P to do (later) or what O is entitled to do – given the commitments of P. Rules of entitlement say what P or O is allowed to do. Of course, O and P change places in the course of a dialogue such as speakers and hearers do. And, of course, there is a kind of jury or score-keeping presupposed by which ‘we’ judge if the players played the dialogical game of commitments and entitlements ‘correctly’.

By liberalizing the metalevel ‘rules’ of judgements about the correctness of dialogical games of giving and asking for reasons in particular, and of acting and reacting in appropriate ways in general interactions, our approach accepts certain criticisms against the‘conventionalism’ or ‘regularism’ brought forward by intentional semantics in the tradition of Paul Grice. On the other hand, we try to name and avoid shortcomings in an intentionalistic ‘theory’ of communicative actions, as it was formally worked out, for example, by Georg Meggle. These shortcomings consist, we believe, in a "speaker-centered" approach to the concept of understanding and the corresponding implicit mentalism. This mentalism consists, we claim, in a lack of reflection on or analysis of the basic concepts of intending something and believing something. These concepts are presupposed in Meggle’s definitional reconstruction of the concept of communicative actions.

Basic problems of conventionalism and formal and mental intentionalism

1. Communicative or cooperative actions cannot be understood just as instantiations of regular action types or as mere rule following behaviour.
2. Even if we assume a principle of expressibility according to which anything we can mean can be expressed in principle in a possible language, we still have to take the paradox of analysis into account: Verbal explications post hoc that use, for example, (re)constructions of types and rules and conventions of behaviour do not say that what was done praeter hoc was ‘really’ done in view of what is said about it later. Or rather, it is not too clear what it means to accept such reconstructions and what it means to accept Searle’s principle of expressibility.
3. The concept of communicative actions cannot be explained sufficiently, either, if we just presuppose a concept of definite intention on the side of the speaker S and of definite understanding on the side of the hearer H. If we presuppose "intending" and "understanding" as basic, we presuppose the most crucial terms and explain by them the easy terms "communicative action".
4. In a way, the usual procedure is similar to that of axiomatic arithmetic. One believes that an axiomatic system could ‘define’ at least the relative use of the words or ‘concepts’ occurring in the axioms in an exact, formal, way. But the Hilbertian and Carnapian belief that the (Peano-)axioms ‘define’ and ‘ground’ arithmetics puts things upside down. In reality, the choices of axioms are justified by the theorems we wish to deduce, not the other way round. This was already seen by Frege. It is, however, not easily understood why it is an age-old myth to believe that the theorems are ‘justified’ by the axioms and that the axioms are just a kind of shorthand for describing some true statements. (Of course, we can use axiomatic and algebraic systems in order to talk about many ‘models’ or ‘individual structures’ at once. Nevertheless, the axioms are justified by the ‘models’ we are interested in. Things get blurred linguistically by a use of the word "structure" in the sense of a family of individual structures.)
5. In our case, too, we would put things upside down if we presupposed a vague concept of intention and understanding. Precisely these concepts, more precisely: the family resemblance of their different uses, are to be laid out. For this it is not enough to put the words into a system of axioms. We should distinguish, for example, between a vague concept of being in an intentional state on one side form a definite concept of intending or meaning X rather than Y or Z. The question is, as Robert Brandom puts it: What makes meaning and content definite?
6. We should also take into account that any explication of such an X aims at being ‘correctly’ understood by a speaker S as well as by a hearer H. In other words, definite meanings not only of words and sentences but of utterances and communicative actions must already be types and they must not be ‘finer’ than what ‘can be understood’ by more than one person alike.
7. There is another idea which in part goes back to Robert Brandoms approach: Judgements about a correct understanding of a communicative act are made from some neutral perspective. Neither has the speaker an absolute knowledge or authority with respect to what he means or intends to do in his (attempt of a) communicative act, nor with respect to what he judges as a correct reaction on the side of the hearer. The latter ‘correctness’ refers to a judgement of being understood well, i.e. to the satisfaction condition that turns a mere attempt of a communicative act into a successful one.
8. Such a success does not entail that all ‘perlocutionary’ intentions of S must be fulfilled. In fact, it is not too clear how to judge if the satisfaction conditions of ‘understanding’ an attempt of a communicative act are fulfilled not just insofar as H knows that S has made an attempt of communication, but that it was an attempt with definite content X and not Y or Z. The question concerns the status, the definiteness and openness of such satisfaction conditions. How are they connected with and how do they differ from the mere feeling of S or H that the communicative act was successful as such?
9. Our proposal is to say that any judgement about a success of an attempt of a communicative act refers on some kind of third perspective, to a view of umpires or score keepers that can be S and H themselves, but only if they change their ‘immediate’ attitude and cooperate or interact in playing the role reflecting referees about their own intentions and understandings. In other words, the very concept of a successful communicative act of content X rather than Y or Z and the very concept of understanding an act as an attempt to communicate the content X refers to a ‘shared’ point of view and to a concept of joint actions that goes conceptually beyond what can be described in a framework of intentionalism as a branch of methodical individualism.
10. In fact, a very basic feature of human behaviour is the establishment of a joint perspective – at first on external things, but then, also, on ones own behaviour.
11. It is in our view a petitio principii to describe such a joint perspective, as in the proposals of David Lewis, as nested intentions and believes about each others intentions and believes. The very concept of a definite intention or belief already presupposes a joint perspective of judgements about the identity of its contents.
12. On the other hand, regularism can be defended only so far as it expresses doubts against a reconstruction of conventions in a mere intentionalistic framework (as proposed by David Lewis and supported by Meggle). It cannot be defended if it is read as the belief that any intention to communicate a content X presupposes a possible expression x in a possible language L that represents the content X in an invariant and context-independent way.
13. Hence, regularism is to be replaced by an analysis of how joint understandings of forms or types or contents X of (attempts of) actions are constituted even when no invariant expression x of these forms or contents are already available.
14. The idea is to describe what it means to understand a particular content or type or form X by describing a general form of interaction in which we determine jointly what it is to understand X ‘as it was meant’ or what it is ‘to have an intention’ Y and not Z. This form is, roughly, speaking, the form of judgement of a jury of referee who judges between what a speaker does and what he claims to do (in an explicit self-description of his intentions) and what a hearer H does in a kind of ‘consequence’ of the speaker’s act and what he claims to have understood about the intentions (‘meanings’) of S.
15. It is important to add, in distinction to what can be found in Brandoms book, that we play multiple roles in communicative interactions ‘at the same time’. We are speakers and hearers, actors and spectators, and we ourselves are scorekeepers and referees with respect to normative judgements of right or wrong. Such judgements say, for example, if the act A of S fits to his possible description of his intentions or if H’s re-action to A draws the right conclusion to S’s doing A. This re-action can be a verbally expressed partial ‘interpretation’ of A or just a behaviour. Understanding is a normative insofar as there is a distinction and judgement between right or wrong understanding. It would be wrong to cut off some of these of roles. If we do so, the analysis of intending and understanding (communicative) acts will not be sufficient.
16. Any formal talk about intentions and meanings, beliefs and expectations is highly ideal . In using variables for definite intentions and definite beliefs, we often neglect the real constitution of the idealised objects the variables refer to. In doing so, we talk from a God’s perspective. From such a high perspective, the real constitution and ‘ real concept’ of intending and understanding cannot be fully understood, just as the concept of a geometrical form cannot be fully understood from the ideal perspective of pure mathematics that transcends all question about impure, but real, figures, gestalts, and measurements.
17. Our analysis wants to be more down to earth than any formal and hence ideal analysis can be. Hence, we must be sceptical against any formalist approach. Such an approach cannot but presuppose ideal meanings and intentions and beliefs and expectations as given and allegedly already well defined values of variables that occur in formal expressions.
18. Hence, our leading questions are the following: What does it mean to speak of the content of an attempt of a communicative act? What do we refer to if we talk about speakers intention or ‘what he wants to say’? What has an addressee really to do in order to ‘understand’ what a speaker means?
19. Many approaches to an analysis of communication take a (psychological) concept of subjective intention and a concept of subjective understanding for granted. We want to confront these analyses with a really pragmatic reconstruction of what we do when we talk about intention and understanding.