Interpersonal Communication as Social Action

1. Language as a Social Institution

Since the publication of Austin’s (1962) How to Do Things with Words, speech act theory has been a very influential approach to the study of linguistic communication. Speech act theory comes in different versions, all of which share some basic assumptions: that language is used to perform certain specific types of actions, called “speech acts”; that by producing a single linguistic utterance a speaker typically performs several speech acts at different logical levels (in Austin’s view a locutionary, an illocutionary, and a perlocutionary act); that the main concern of the theory is the analysis of illocutionary acts (like, for example, informing, requesting, promising, greeting, etc.); and that the force of an illocutionary act derives from specific conventions shared by the members of a linguistic community.

The last point is the most problematic. Indeed, Austin did not clearly explain what he meant by a convention; most plausibly he wanted to clarify the distinction between illocutionary and perlocutionary acts: between informing and convincing, requesting and inducing, threatening and frightening. One possible interpretation of why Austin considered illocutionary acts to be conventional is that to inform, request, or threaten someone, it is crucial that the addressee understands the speech act as a case of informing, requesting, or threatening; on the contrary, understanding is not sufficient to convince, induce, or frighten: these are possible further effects on an addressee, which a speaker may fail to achieve even if their illocutionary act has been properly understood. This seems to be what Austin (1962, 138) had in mind when saying that to perform an illocutionary act, it is essential to “secure uptake.”

Today’s most popular version of speech act theory is the one offered by Searle (1969). In Searle’s (1969, 61) view, in performing a speech act, a speaker intends to produce a certain illocutionary effect in virtue of the fact that “the meaning of the item he utters conventionally associates it with producing that effect.” ¹ The conventions regulating meaning are regarded as the constitutive rules of a social institution, the institution of language. According to Searle (1995, chap. 3), language is the “basic social institution,” because it plays a crucial role in the creation and maintenance of all other human institutions.

What we can call institutional theories of communication, ² like Austin’s and Searle’s, have two notable advantages. The first is that they provide a unified treatment for all types of speech acts, whether they presuppose complex nonlinguistic institutions (like the speech acts involved in baptizing someone or firing someone from a job) or are simple, purely linguistic communicative acts (like informing or requesting). Moreover, institutional theories simply accommodate speech acts that have normative effects (like promises, invitations, agreements, etc.) because such effects are typical of institutional actions.

To summarize, the illocutionary acts of speech act theory are social in at least two distinct ways: first, they are social in that they are institutional actions, and as such presuppose shared acceptance of the constitutive rules of the institution of language; second, they are social because their performance can produce social consequences, like, for example, normative effects (which are regarded as institutional facts created by a speaker within the institution of language). Interpersonal communication, therefore, can be viewed as a large-scale social phenomenon, in that it essentially involves forms of group-level normativity.

This view of communication is not without problems. Possibly the main difficulty is that the actual reconstruction of illocutionary acts as conventional actions has proved impossible. Consider, for example, the utterance, “I’ll be at office tomorrow.” Depending on the circumstances, in producing this utterance the speaker may be trying to make an assertion, or to inform the addressee, to make a promise or a threat, to issue an invitation, and so on; however, it is clearly impossible to formulate a set of conventions able to account for the illocutionary act that is actually performed. In speech act theory, this problem is typically faced by assuming that in uttering a sentence a speaker can perform indirect illocutionary acts (e.g., a promise) different from the literal speech act (e.g., an assertion); but even Searle (1975) explains the meaning of indirect illocutionary acts in terms of nonconventional inferential processes.

Of course, that the illocutionary force of an utterance is (at least partially) nonconventional does not rule out the fact that the sentence uttered is itself conventional, in the sense that it follows the conventions of the lexicon and grammar of English. But in certain cases, even the “utterance” may be nonconventional: a fictional example is offered by Sperber and Wilson (1986/1995), who describe a situation in which two prisoners succeed in escaping by coordinating their actions thanks to the production by one of them of a spontaneous, nonconventional signal. Speech act theory cannot explain similar cases nor the communicative activities of preverbal children. Bruner (1983), however, has convincingly argued that children learn to deal with illocutionary forces before they acquire the conventional rules of their native tongue, and recent research by Tomasello and collaborators shows that preverbal children are capable of communicative interactions that could not be explained if one assumes that communication always presupposes the acquisition of social conventions.

In addition to these difficulties, we have a more general objection to Searle’s idea that illocutionary acts are institutional actions defined by constitutive rules. In Searle’s (1969, 36) words, “constitutive rules, such as those for games, provide the basis for specifications of behavior which could not be given in the absence of the rule.” Now consider, for example, a promise. It is fair to define an act of promising as a speaker’s attempt to create an obligation to an addressee to perform an action that is in the addressee’s interest. However, creating an obligation has to be part of the capacities of human beings before promise can be defined: promising does not create the concept of an obligation in the same way as, for example, the game of chess creates the concept of a checkmate; and even when language is needed to specify the action that an agent promises to perform, it is not the use of language by itself that creates the obligation, nor is the fact that specific obligation-making conventions are followed.

2. Minds in Interaction

A sharply different approach to communication is based on Grice’s pioneering article on nonnatural meaning. As is well known, Grice (1957, 385) defined nonnatural meaning, or meaning_NN, as follows:

“A meant_NN something by x” is (roughly) equivalent to “A intended the utterance of x to produce some effect in an audience by means of the recognition of this intention.”

Even if Grice does not explicitly mention communication, Strawson (1964) remarks that Grice’s definition can be interpreted as an analysis of interpersonal communication. In the same article, Strawson criticizes Austin’s idea that the force of all illocutionary acts is conventional. He distinguishes between two categories of illocutionary acts: the first category includes those speech acts that do appear to derive their illocutionary force from a convention, like bringing in a guilty verdict in court, pronouncing a batsman out in a cricket match, introducing someone by saying “This is Mr. Smith,” or surrendering by throwing up one’s arms; the second category includes those speech acts, which we call communicative acts, whose illocutionary force does not appear to be fully determined by any convention. The example we gave in the previous section (saying “I’ll be at office tomorrow”) is an instance of this category.

To explain how a communicative act acquires its illocutionary force, Strawson interprets Austin’s concept of uptake in terms of (a revised version of) Gricean meaning. In the appropriate circumstances, for example, the utterance “The ice over there is very thin” will have the force of a warning (rather than, for example, the force of an assertion) not because it conforms to a specific convention regulating warnings, but because the addressee, considering the total situation in which the utterance is produced, recognizes the speaker’s intention to issue a warning, at least in part by having this very intention recognized.

Since Strawson’s article, several theories of communication based on the Gricean definition of meaning have been put forward. Most often, Grice’s original definition is somewhat modified. A first frequent amendment concerns the reference of the expression “this intention” in the original definition. As Grice (1969) clarified in later writings, he meant it to refer to the speaker’s intention “to produce some effect in an audience” through the utterance. The definition therefore involves two stratified intentions: a first-level intention to produce an effect in an audience and a second-level intention that the first-level intention be recognized by the audience. ³ However, there are good reasons to prefer a version of the definition involving a single reflexive intention, where the expression “this intention” is taken to refer to the whole intention “to produce some effect in an audience by means of the recognition of this intention” (see Carassa and Colombetti 2009 for a justification of this interpretation); following common practice, an intention of this kind we shall call a communicative intention. A second modification of the original definition concerns the concept of recognition by an audience. As argued by Schiffer (1972), the one-sided recognition of the speaker’s intention by the audience does not suffice: for genuine communication, the speaker’s intention must be shared (i.e., mutually recognized ⁴ ) by the speaker and the audience. As a third amendment, we propose to drop the qualification “in an audience” from the expression “to produce some effect in an audience.” The reason is that, as we shall argue later, the effect of a communicative act is not necessarily limited to the audience, as it may involve also the speaker. Considering these amendments, we define a communicative intention as follows:

A communicative intention is [the intention to achieve some effect, at least in part by means of sharing this intention with some other agent],

where “this intention” refers to the whole intention described within the brackets. A communicative act is then defined as an action performed with a communicative intention.

Let us now compare mainstream speech act theory with Gricean theories of communication. As far as communicative acts are concerned, the main difference is that Gricean theories do not rely on a concept of convention or constitutive rule. ⁵ But this involves a very different idea of what it means for a communicative act to be a form of social action: rather than stressing the role of language as a large-scale social institution, Gricean theories stress the importance of the small-scale interaction between individuals. What is at stake, in Gricean approaches, is human intersubjectivity, understood as the capacity to share with others certain types of mental states, inclusive of communicative intentions.

It should be remarked that a certain level of intersubjectivity is presupposed also by the institutional theories of communication. Such theories usually insist that the conventional rules of language apply to utterances that are used literally and seriously, and this rules out those uses that are either metaphoric or ironic. But how is an addressee supposed to understand, for example, that an utterance is used ironically and standard conventions are therefore suspended? Should we assume some form of metaconvention to signal metaphors and ironies? But then, we should be able to signal that the metaconventions too are used literally and seriously, which would lead to infinite regress. It seems that the only viable solution is to assume that speakers are able to share with their addressees their intention to use an utterance literally and seriously: but this inevitably brings intersubjectivity into institutional theories of communication.

Gricean theories view interlocutors as “minds in interaction,” and treat communication as a small-scale social phenomenon, with a number of advantages. Nonconventional forms of communication, like the ones discussed in the previous section, are no longer problematic. Moreover, these theories can be extended, with due care, to the communicative processes of preverbal children. It will therefore come as no surprise that the Gricean structure of communication plays a prominent role in Tomasello’s cooperative model (Tomasello 2008), to which we devote the next section.

However, also Gricean theories of communication face some difficulties. First, if minds in interaction take center stage, it is not possible to marginalize the psychological dimension of communication; rather, it is necessary to investigate what mental capacities are put to use in communicative interactions, how it is concretely possible to share mental states with others, and so on. We shall come back to this important point in the next section. Second, if communication is no longer assumed to take place in the context of the institution of language, it is not easy to explain how communicative acts can have normative consequences, as is obviously the case with such acts as promises, proposals, and agreements. Proposing a solution to this problem is a major concern of our own research, which will be sketched in the fourth section.

3. Communication for Cooperation

The last few decades have witnessed remarkable progress in the developmental and comparative psychology of communication. Michael Tomasello and colleagues, in particular, have carried out a vast amount of experimental research on the communication processes in nonhuman primates and children. On the basis of the results of this research activity, Tomasello (2008) has recently formulated a comprehensive theory, the cooperative model of communication, in which human communication is viewed in the wider context of cooperation.

In Tomasello’s view, human communication is radically different from the communication processes of other animal species. This is due to the fact that the communicative capacities of humans are part and parcel of their wider ability to cooperate, which is a distinctive feature of the species. A deep comprehension of the psychological infrastructure involved in cooperation is therefore necessary to understand human communication, and this infrastructure can be investigated by carrying out systematic empirical research in developmental and comparative psychology.

According to the cooperative model, the main elements of the psychological infrastructure of human communication include the capacities for shared intentionality, certain prosocial motivations, and an understanding of social norms. As far as shared intentionality is concerned, Tomasello places center stage Clark’s (1996) concept of common ground, together with the capacity to form joint intentions and goals: for these concepts, Tomasello mainly refers to the works of Searle (1990, 1995), Bratman (1992), and Gilbert (1989). In particular, Tomasello draws on Searle (1990) to remark that cooperation is mutualistic, in the sense that to cooperate one has to regard others as possible candidates of cooperative action.

The infrastructure for shared intentionality, however, would make no sense if humans did not have a repertoire of motivations that drive them to engage in cooperative behavior. In Tomasello’s view, such prosocial motivations concern helping others and sharing feelings and attitudes. Coherently with the idea that we see each other as possible candidates for cooperation, these motivations are both active in each subject and assumed to be active in the others. For this reason, the motivation to help underlies both the acts of informing others in their interest and requesting from others something that is in our interest. Requesting, informing, and sharing are therefore assumed to be the three fundamental types of motivations underlying communication.

The cooperative model posits different kinds of intentions that are relevant to communication. Social intentions concern the actual social interactions that speakers try to achieve by performing communicative acts (helping, getting help, and sharing feelings or attitudes). Relying on the mutual assumptions of cooperativeness, speakers try to achieve the intended effects by intentionally sharing their social intention with their interlocutors, and therefore perform communicative acts with a Gricean communicative intention, whose main (but not only) function is to create expectations of cooperation.

Interestingly, there is much empirical evidence that while cooperative communication of this kind is precluded to nonhuman primates, it is already within the capacities of preverbal children of about 1 year of age. Later on (from 3 to 4 years of age), children start to appreciate that communicative acts also have normative effects. In Tomasello’s approach, the social normativity of communication is strictly related to the fact that communication is functional to cooperation and is related to the logical structure of communicative intentions: as already noticed by Strawson (1964), communicative intentions make what we say overt and avowable, and “the fact that I have communicated to you overtly, publicly, actually creates not just expectations of cooperation but actual social norms, whose violation is unacceptable” (Tomasello 2008, 91-92). Such normativity arises at two distinct levels. The first level concerns the performance of the communicative act itself: “if I attempt to communicate with you . . . , you cannot just ignore me as though I did not attempt to communicate” (Tomasello 2008, 92). The second level of normativity concerns the addressee’s compliance to the speaker’s social intention: “if I make a small request . . . , you cannot really reply with ‘No’—unless you make some excuse for why you cannot comply in this circumstance” (Tomasello 2008, 92).

In the cooperative model, the idea of sociality that underlies human communication is coherent with the role of communicative interactions for cooperation and presents both small-scale and large-scale aspects. As far as small-scale sociality is concerned, young children of about 1 year already demonstrate prosocial motivations and basic capacities of nonnormative shared intentionality, including the ability to take into account common ground and to form and pursue common goals. Later on, between 3 and 4 years of age, children become sensitive to the large-scale social norms regulating communicative acts. Importantly, the normative effects of communication crucially depend on the logical structure of communicative intentions. The role attributed to such intentions is to make the communicative act “public”; as a consequence, speakers fall under relevant social norms. Tomasello (2008, 214-15) clarifies this with an example concerning a guest who would like to have more wine at a dinner:

If I place my empty wine glass in a conspicuous position, hoping my host will see it and fill it, but (for reasons of politeness) making sure he does not see me do it and so does not view it as an overt request, no norms apply. . . . But if I signal him overtly by brandishing the empty glass in his direction, in most cases this would trigger a norm—we know together that he has seen the empty glass and presumably has inferred from the brandishing act toward him that I want a full glass, and so he must deal with it in some way, or pretend that he did not in fact see my act.

The global picture, therefore, is that communicative interactions do not create normative relationships from scratch, but rather instantiate preexisting social norms, recognized at group level.

At present, Tomasello’s cooperative model is the most remarkable attempt to embed a theory of interpersonal communication in a larger theory of human cooperation. What makes it especially valuable is the vast empirical evidence on which it is based, gathered in observational and experimental settings and concerning both the phylogeny and the ontogeny of communication. However, the cooperative model leaves a number of crucial aspects unclarified. An important point concerns social normativity: is all normativity a large-scale phenomenon, so that specific normative relationships are always produced by instantiating preexisting group-level norms, or can normative relationships directly arise from small-scale intersubjective interactions? And if so, what features of such interactions are involved in bringing about normative relationships?

In the next section, we shall briefly present our approach, which specifically aims to clarify how small-scale normativity can be directly created by the shared recognition of communicative intentions, rather than by instantiating group-level norms.

4. Normativity from Interaction

Let us go back to our previous example utterance, “I’ll be at office tomorrow.” Let us also suppose that in producing this utterance, the speaker intends to make a promise, and that the addressee correctly understands this. How can we account for such a communicative interaction? A promise is an attempt to create an obligation of the speaker to the addressee to perform an action in the addressee’s interest; therefore, we have to explain how such an obligation can be brought about through a communicative interaction. The answer given by speech act theory is that promises are part of the institution of language and are performed by carrying out a conventional procedure, whose institutional effect is to create a certain type of obligation. But in our example (as in most everyday cases of promises), it appears that no specific conventional procedure for promising is followed. Therefore, a different explanation is needed.

We believe that the main fault of the institutionalist explanation of a promise is that the addressee is supposed, first, to recognize that the speaker is carrying out a conventional procedure for promising, and second, to derive from this (by instantiating a general constitutive rule of promising) that the speaker is thereby trying to create a certain type of obligation. If no convention is involved, however, the explanation must necessarily go the other way around: first, the addressee understands that in performing her utterance, the speaker is attempting to create a certain type of obligation; and it is only as a consequence of this that the addressee is entitled to describe the speaker’s communicative act as a promise. But then we are left with the problem of explaining how people can directly create obligations, and similar normative relationships, without relying on specific conventions.

Very few scholars have devoted their attention to what we call interpersonal normativity (Carassa and Colombetti 2014), that is, the normative relationships that two or more agents can create intersubjectively, without relying on preexisting group-level norms. Interpersonal normativity has the distinctive feature that it binds precisely those agents who jointly create it. To our knowledge, the only extensive attempt to account for this phenomenon is Gilbert’s (1989) plural subject theory. A number of agents constitute a plural subject when they create a joint commitment of doing something “as a body,” which in turn entails directed obligations (and the correlative rights) between the agents (Gilbert 2013). It is remarkable that creating a joint commitment does not presuppose preexisting institutions: according to Gilbert, it is only required that the relevant agents express, in conditions of common knowledge, their readiness to do so.

Gilbert (2009) submits that joint commitments are the essence of shared intentionality. Her position, however, is highly controversial. According to Bratman (1993, 112), for example, shared intentions are “primarily a psychological—rather than primarily a normative—phenomenon.” A similar position is put forward by Tomasello (2014), who argues that the ability to collaborate appeared in an evolutionary stage in which humans still lacked the psychological infrastructure necessary for normativity. Tomasello’s conception of normativity, however, encompasses large-scale, group-level norms and does not take into account interpersonal normativity.

We agree with Tomasello that simple forms of collaboration, like group hunting, do not necessarily involve normative relationships: such collaborative efforts may be driven by the fact that every member of the group appreciates that their contribution is critical for the success of the current endeavor and therefore for the achievement of their personal goal. However, it seems plausible to assume that human normativity did not appear at once in the fully fledged form of group-level norms. At an intermediate stage, humans may exploit interpersonal normativity to support more complex forms of collaboration; for example, some kind of interpersonal normativity seems necessary to support delayed exchanges (I do this for you now, you will do that for me tomorrow). But what do we mean, exactly, by “interpersonal normativity”?

In our view, interpersonal normativity consists in the fact that between two agents, ⁶ a relationship holds such that (a) it makes one of them responsible or answerable to the other for doing certain things and (b) it has been autonomously brought about by the same agents who participate in it. Following Darwall (2006), we conceive of responsibility as a second-personal relationship, in the sense that it assigns correlative deontic positions to two agents: A is responsible to B to do X if, and only if, B has a legitimate claim against A that A do X. It must be remarked, however, that not every relationship of responsibility, although second-personal in the sense just specified, is interpersonal. Indeed, we reserve the term interpersonal responsibility to those relationships of responsibility that are autonomously created by two agents (that we call the parties of the interpersonal responsibility), rather than deriving from the instantiation of a group-level norm. To clarify the difference, consider, for example, Ann’s duty to take care of her old father Bruce. This is a case of second-personal responsibility, because it makes Ann responsible to Bruce to take care of him and grants Bruce a legitimate claim against Ann that she so does. However, Ann’s responsibility is not interpersonal, because it has not been autonomously created by Ann and Bruce, but rather it instantiates the general norm, recognized at society level, that everybody ought to take care of their old parents. On the contrary, if Ann promises to Bruce to bring him a bottle of wine, and Bruce accepts the promise, then Ann accrues a responsibility to Bruce that is genuinely interpersonal, because the two of them jointly created it by mutual agreement.

We are now left with the problem of explaining how two individuals can autonomously create normativity, in the form of interpersonal responsibility. We have argued elsewhere (Carassa and Colombetti 2014) that interpersonal responsibilities are created by performing communicative acts, thanks to the distinctive logical structure of communicative intentions. To clarify this important point, let us assume that A intends to become responsible to B to do X. ⁷ How can A try to implement her intention? First note that, due to the second-personal nature of responsibility, A cannot be responsible to B to do X unless B has a legitimate claim against A that A do X. But as we are concerned with interpersonal responsibility, it must become shared by the two agents (in the sense of “shared” clarified in footnote 4) that each of them autonomously accepts that the responsibility holds. Clearly, A cannot achieve this effect by her alone: she can only give her individual contribution to the creation of the responsibility, in the hope that B will produce a matching contribution. More precisely, A’s contribution will consist in seeing to it that her intention to become responsible to B is shared by her and B; and B’s matching contribution, if it is produced, will consist in seeing to it that his intention to acquire the correlative legitimate claim against A is shared by him and A.

From what we have said so far, it follows that A’s intention to become interpersonally responsible to B implies A’s intention to share such an intention with B. But this means that A intends to achieve an effect, at least in part, by sharing her intention with B, which exactly corresponds to our definition of a communicative intention. Note that the qualification “at least in part” is crucial, because A will not accrue the intended responsibility until B communicates back to A his matching intention to acquire the correlative legitimate claim against A.

To make a concrete example, consider again Ann’s promise to bring Bruce a bottle of wine. In making her promise, Ann intends to become responsible to Bruce to bring him a bottle of wine, at least in part, by sharing her intention with Bruce. Now suppose that Ann’s promise to Bruce is correctly understood, in the sense that Ann’s communicative intention becomes shared by the two of them. This is still insufficient to create the intended responsibility, because understanding a promise and accepting it are two different things. To accept Ann’s promise, Bruce will have to perform a suitable communicative act, expressing his intention to acquire a legitimate claim against Ann that Ann brings him a bottle of wine. Only when both Ann’s and Bruce’s communicative intentions become shared by the two of them will the interpersonal responsibility be brought about. ⁸

Our analysis shows that, at least in the case of promises, it is by performing certain communicative acts that an interpersonal responsibility is created. We believe that the analysis can be extended to other types of communicative acts, but with a caveat. Promises (and commissive acts in general) are special, in that bringing about interpersonal normativity is the very point of performing them. Other types of communicative acts also bring about normative effects, but this is not necessarily their point. Consider, for example, an act of informing. We suggest that “A informs B that P” means that A makes P epistemically available to B, taking responsibility to B that P is actually the case. Most of the times, B will accept A’s responsibility, and this will give B a reason to believe that P. In other cases, B will accept A’s responsibility even if he is not really convinced that P is the case. But there are also marginal cases in which B will not accept A’s responsibility, for example, if A is dead drunk or otherwise mentally incapacitated.

In our approach, the sociality of communicative acts is essentially the sociality of minds in interaction. We believe we have shown that such small-scale communicative interactions are capable of bringing about normative relationships between the interacting agents; such normative relationships do not derive from the large-scale normativity of the institution of language but are a direct consequence of the logical structure of communicative intentions. Of course, we do not deny the crucial role of linguistic conventions: no complex propositional content could be communicated without language; moreover, language provides us with a large choice of tools to specify or at least constrain the illocutionary force of an utterance (e.g., in normal conditions, an utterance like “I might be at office tomorrow” cannot be interpreted as a promise). However, the role that we assign to language is different from the one advocated by institutional theories: in our view, language affects interpersonal normativity not because producing an utterance counts as a normative undertaking within the institution of language, but because it allows speakers to share complex communicative intentions with their addressees.

5. Conclusion

As we have seen, different approaches put forward different views of what it means for interpersonal communication to be a form of social action. Speech act theory regards communicative acts as part of the large-scale social normativity underlying the institution of language; Gricean theories focus on the small-scale sociality of minds in interaction; and Tomasello’s model regards communication as functional to human cooperation and concentrates on the phylogenesis and ontogenesis of the relevant psychological infrastructure.

In our view, all these approaches fail to account for the human capacity to create interpersonal normative relationships, which we regard as crucial for cooperation. We propose to analyze such normativity in terms of relationships of interpersonal responsibility that people create by sharing communicative intentions. While much work is needed to work out the details, we believe that our proposal allows for a better comprehension of the social structure of communication and of the normative aspects of interpersonal relationships.