clac 19


Frank Scheppers

Vrije Universiteit Brussel, Center for Linguistics (CLIN)


Some notions remain unquestioned throughout otherwise highly divergent schools of thought within the domain of linguistics and in neighboring disciplines. One such central complex of notions is the standard view of linguistic communication. Although many differences can be noted between competing approaches (for an -incomplete- overview, see Berge 1998), most models have, often implicitly, at least the following core features in common (for a remarkably explicit version of this ‘standard’ view, see Levelt 1989):

- communication is localized in two distinct psychological units (two subjects), the speaker and the addressee;

- these units interact by means of the transfer of information:

- the speaker codes (‘packages’, ...) a ‘prelinguistic intention’ of his in such a way that this content (‘message’) becomes available to the addressee;

- the addressee in his turn ‘decodes’ (‘interprets’, ...) the message;

- the message itself is conceived of as being an object with an independent status from both speaker and addressee.

Of course, lately, the psycholinguistic modelling of this coding-decoding process has evolved into a much more complex matter than the classic semiotic model: the attention has shifted away from the speaker towards the addressee, in that the coding process on behalf of the speaker is now conceived of as implying extensive monitoring of the addressee’s cognitive state (see e.g. Brown 1995). Still, the psychological subject is the ultimate locus of the ‘sense’ of discourse, in that it is defined in terms of the intentions of the speaker and the interpretation of these intentions by the addressee. Furthermore, the psychological subject is nowadays often understood as corresponding to an underlying biological organism, mostly only considered in neurological terms, i.e. mainly as a brain.

The purpose of this paper is a modest one. It is not intended to offer conclusive evidence against any of the views under scrutiny, and no systematic attempt will be made to offer a full-grown alternative for the prevailing paradigm. However, I will bring together various elements that point toward an alternative conception of the phenomena, and thus will try and show that the speaker-addressee doctrine is not a necessary one and depends on specific methodological/epistemological choices.


1. Coherence, sense and intentions

There are serious arguments for claiming that the ‘sense’ of behavior (as in ‘this behavior makes sense’ or in ‘sensible behavior’) can be interpreted in terms of its ‘pragmatic coherence’, i.e. in terms of the specific ways its component actions make up a more complex action.

The ‘point’ of a particular action (its pragmatic function) cannot be isolated from its position within a more global sequence of actions: it depends both on when it is executed with respect to actions that precede and follow it and on the qualitative relation it has with its neighbors and with respect to the more complex encompassing action. I have argued elsewhere (Scheppers 2003) that this feature of coherent behavior can be represented by assigning a hierarchical structure to behavior (insofar as it is coherent).

Consider the following example. Someone reaches for his pocket. The point of this action becomes apparent as he grasps for and takes out a box of matches. The point of this more complex action ‘taking a box of matches’ in its turn depends on the way it fits within a more global sequence of actions. If the agent has already fetched a kettle, filled it with water and now uses the matches to light a burner of the kitchen range to cook the water, the point of these actions and the encompassing action ‘heating water’ may get their point from the fact that he’s e.g. cooking tea. On the other hand, if he takes out his matches to light a cigarette, the pragmatic function of this action does not go beyond the encompassing action ‘smoking a cigarette’ and it would (normally) be fruitless to try and link it with any of the preceding or following actions.

Observe that exactly the same elements that enter into an analysis of the internal hierarchical structure of a complex action also serve as answers to the question as to why the agent does whatever he does, i.e. as an analysis of his intentions: he reaches for his pocket in order to take out a box of matches, because he wants to light a match, with the intention of lighting a cigarette, so as to smoke. Insofar as this line of argument is adequate, what an agent is doing is indistinguishable from why he is doing whatever he does. In order to understand behavior (of an observed agent or -for that matter- one’s very own), intentions do not enter the game as separate items, apart from what makes the behavior coherent and recognizable in the first place.[1]

Furthermore, the recognition of the ‘sense’ of certain sequences of actions -as e.g. reflected in the act of naming the actions- involves the recognition of the type of action this particular action token belongs to. Even if one chooses to formulate the sense of the action tokens of an agent in terms of his intentions, this formulation will necessarily imply reference to action types (making sense of ‘this agent cooking tea’ implies knowing what it is to ‘cook tea’).[2] The type-token distinction holds as well for ‘novel’ or ad hoc behavior: even if some aspects of observed behavior are ‘new’ to the observer or the agent (e.g. while trying to make sense of an object s/he has never seen before), in order to be perceived as sensible/coherent, its component actions and the relations that hold between them should be of recognizable general type (grasping, lifting, moving, ...). Obviously, these action types are not reducible to the particular agent’s intentions, rather the inverse is true: one cannot intend to perform an action without somehow knowing what it is to do so.

Thus, both the stereotyped and the ad hoc or novel aspects of an agent’s behavior -insofar as they can be interpreted as ‘making sense’- depend on structural features of the behavior itself in which no reference needs to be made to the particular intentional states of the agent. The same structuring mechanisms accounting for the coherence and sense of behavior can be repeated over and over again in different token actions, across different agents and different observers. Both from the point of view of the agent and from the point of view of the observer, ‘making sense’ is a matter of managing the coherence relations between the actions that make up a more complex action, and these coherence relations have an independent status with respect to the subjects (or organisms) involved and can accordingly be analyzed in an insightful way without making reference to the cognitive states of these subjects.[3]


2. Sense and communication

This way of analyzing the coherence of behavior applies to verbal behavior, i.e. to discourse, as well. Within the field of Discourse Analysis and (linguistic) Pragmatics the kind of coherence relations that can hold between discourse segments are an object of ongoing research (for references and discussion, see Scheppers 2003; also see e.g. Mann & Thompson 1988 and Roulet & al. 2001). This research involves more or less elaborate typologies including relations such as the one between a claim and evidence for that claim, between the two members of a contrast, between the setting in which an event takes place and the event itself, topic-comment relations, etc. Again the coherence relations constituting the internal structure of discourse can be interpreted so as to represent the point of its different segments and hence the intentionality of the discourse as a complex action.

Consider e.g. a discourse segment like ‘Yesterday I was at the bank, and there was a guy, and ...’, after which follows a story with the ‘guy’ as the main character. It is quite obvious that producing this stretch of discourse gets its point from being the introduction to the ensuing story (introducing its setting and its main character): it is normally only interpretable as the opening of such a story (and only at its opening!), and it would not normally be produced at all if not as the beginning of that story. Likewise, discourse segments presenting the different actions that make up the plot of the story (e.g. ‘and suddenly he pulled a gun’ or ‘and then he ran away’) get their respective points from the position they take within that plot. The intention to produce this particular segment and the intention to produce that particular story are interdependent.

As in the case of non-verbal behavior, the type-token distinction can be applied. As is shown by the analyses in the fields of rhetoric, narratology and discourse analysis, different genres of discourse (an anecdote, a joke, a fairy-tale, a forensic speech ...) imply specific ways of sequencing and articulating the segments and sub-segments the genre implies. Each segment gets its point from the position it has within the overall structure of the discourse (e.g. the punch-line with respect to the joke, the narration with respect to the forensic speech, “Once upon a time ...” and “They lived happily ever after” with respect to the fairy-tale, etc.), and, conversely, the discourse is coherent and makes sense insofar as the necessary structural relations between its segments are maintained.

Understanding (‘making sense of’) discourse is then to a large extent analogous to understanding (observed or experienced) non-verbal behavior, in that both fundamentally imply the ability to construct the coherence relations between the action segments that make up the behavior, and hence their ‘point’ or ‘sense’. In both cases, the explicit reconstruction of the agent’s intentions (incl. one’s own) as something separate from the structurally inherent ‘sense’ of the behavior itself is not necessary for an insightful analysis.

Levelt (1989, 58) states: “It is generally assumed, and it seems to be supported by introspection, that speakers produce utterances in order to realize certain communicative intentions”. It is clear from the above that I simply deny the existence of ‘communicative intentions’ as specific to and necessary for verbal behavior, as well as the purported evidence from introspection. Consider what actually happens while watching someone cooking tea or listening to someone telling a story about what happened at the bank the other day. Phenomenologically, understanding what other people are doing/saying is not normally experienced as if it implies the decoding of a message and the subsequent interpretation of its underlying intentions (what would be the message of cooking tea?). Rather, what the observer sees/hears presents itself to him as a matter of ‘following’ what the observed agent is doing/saying: he recognizes the observed actions as being a coherent whole of a familiar type and he experiences their ‘sense’ or ‘point’ in the same way as if he were performing the actions himself. Understanding others is -so I claim- largely homologous to self-understanding, but the converse is also true: understanding one’s own actions involves the very same mechanisms as understanding observed behavior.[4]

The viability of the notion of ‘communicative intention’ depends on the viability of the notion of ‘intention’ in general: if one denies that the sense of behavior depends on pre-behavioral intentions, this goes for communicative behavior as well. If self-understanding on behalf of an agent/speaker is not fundamentally different from what it takes for an observer/addressee to understand what is going on, the specificity of communication looses a lot of its intuitive charm.

Thus, if one chooses to stress the structural features of coherent behavior that are constitutive of its sense as an alternative to the (rather foggy) concept of ‘(an agent’s) intention’, the sense of communicative behavior (i.c. discourse) as it is experienced appears to be largely analogous to the sense of non-communicative behavior as it is experienced.[5]


3. Cross-speaker coherence

The structural interpretation of the coherence/intentionality of behavior in terms of its internal coherence can be extended to multi-agent (‘co-operative’, including antagonistic) behavior. Think e.g. of the coordination of the actions of different agents implied by dancing, or of boxers in a boxing match or the players in a football game (both within a team and between adversaries).

For the present purposes the most important type of co-operative multi-agent behavior is conversational discourse. In this section I will review a few structural features of conversation that are not limited to one speaker.


(a) adjacency and sequentiality

The internal structure of conversation has from the outset been one of the major issues in Conversation Analysis as founded by Harvey Sacks’s classic analyses (Sacks 1992/1995a, 1992/1995b). Sacks’s analysis of the sequentiality of utterances in conversation gave rise to the familiar notion of ‘adjacency pair’ (e.g. Greeting-Greeting, Summons-Compliance, Question-Answer, etc.), but included also more complex forms of sequentiality.

Furthermore, some of the cross-speaker coherence relations bear a close resemblance to regular one-speaker counterparts. Thus e.g. the relation between a wh-question and the answer to it functions almost exactly like a topic-comment relation. Conversely, many typical dialogic structures can be used within a one-speaker discourse: e.g. the rhetorical use of questions in a monologue as a means of marking topic-comment relations (compare “Now, what about the theoretical implications of this model? ...”, and “As far as the theoretical implications of this model are concerned, ...”).

Genre-specific patterns also determine the different conversational roles which the participants assume; for instance, the roles of patient and doctor are inherent features of the discourse genre(s) which are typical in a medical context: these roles are not features of the participants in the interaction in se but of the type of interaction (‘genre’) itself (the very same persons may engage in quite different interactions as well, e.g. as friends); entire branches of discourse analysis are concerned with the investigation of the functions of discourse within wider interactional contexts (cf. various contributions to Schiffrin, Tannen & Hamilton 2001), i.a. focusing on the functions of verbal behavior as part of essentially non-verbal behavior (e.g. on the workfloor, or in the context of the sociological features of different groups of teenagers, etc.). For the present purposes it is interesting to note that the genre-specific roles of the participants often directly reflect into structure of the conversation. Thus, the different roles in a conversation may be generally ‘symmetrical’: the participants have a more or less equal status and power in the group, the conversational genre in which they engage does not imply a specifically different role for the different participants, etc. In these cases, the overall pragmatic structure of the conversation itself will often be made up of structurally symmetrical cross-speaker patterns, e.g. some dialogic genres require antagonism between the participants (e.g. debates, quarrels, bluffing sessions ...); this role-assignment will reflect into the discourse structure under the form of highly frequent contrast-like adjacency pairs. Conversely, an asymmetrical relation between the different roles typically reflects into structurally asymmetrical patterns in the discourse, as in the following cases: (i) some dialogic genres (‘interviews’ in the broad sense) imply that one of the participants (the ‘elicitor’) does all the questioning, whereas the other participant(s) do(es) all the answering; in some cases this asymmetrical distribution is obviously related to a role-distribution which is pre-existing with respect to the local discourse context (e.g. journalist/researcher-interviewee; police-detainee; doctor-patient); (ii) meetings in a professional context are often presided over by a ‘chairman’, which (often but not always) is hierarchically superior in the organization, and whose role implies for him to regulate the interactions; likewise, the contributions of the host at a party reflect his special status in that e.g. it typically is his role to introduce new participants and to generally regulate the overall development of the interactions.

Note that the above remarks apply equally well to the structure of antagonistic behavior. The antagonism implied in e.g. a debate is a structural feature of such a discourse genre, which implies recurrent Contrast patterns opposing two contrasted claims.


Beside and beyond the structural relations that express the coherence of the contents of a conversation, some other cross-speaker regularities can be observed.


(b) echoism, cross-speaker utterance completion

In spontaneous conversation, it is a quite common phenomenon that the addressee echoes the last word(s) of the previous speaker’s utterance, or finishes the previous speaker’s utterance before he can (see Sacks 1992a, 647-655, on what he calls “appendors”). Obviously, the addressee (the second speaker) does not take up a role different from the one the first speaker occupies; the execution of the same role seems to be divided between both speakers.


(c) “uh huh” and other other-speaker fillers

In his May 24 1971 lecture, Sacks starts from the remark that “uh huh”, like “uh”, ‘fills a pause’, but -unlike “uh”- fills a pause in the other person’s talk. Sacks then adds a few remarks (Sacks 1992/1995b, 410-412): “uh huh” does not occur after the other person has paused, but typically occurs in such a way that their is ‘no gap and no overlap’ between speakers and thus seems to anticipate the other speaker’s pause; “uh huh” does then not simply mean “go on”, but anticipates the other’s intention to go on (as well as his intention to pause). What is interesting about this phenomenon is that, although in some sense different speakers contribute to the discourse, intuitively, these interventions can hardly be said to constitute a separate turn in any real sense at all. The discourse of the current main speaker is not interrupted by these interventions, no separate pragmatic point is carried by ‘uh huh’, (arguably) not even a regulatory one, and accordingly no separate ‘intention’ can be attributed to the speaker in any real sense of the term. This seems to point in the direction of an interpretation in which the different participants do not really each contribute their share to the development of the discourse, but rather seem to ‘embody’ the very same discourse. More lexicalized fillers, like “yeah”, “yes”, “o.k.”, etc., and echoic phenomena as the ones mentioned above can be observed to have comparable functions.


(d) no gap, no overlap

The very fact that different speakers’ utterances in fluent conversation often follow each other without there being any gap between them and quite often (though not always) without there being any overlap, has been considered as a feature of verbal communication that deserves attention. As has been noted by Sacks (Sacks 1992b, passim), contributions of speakers overlapping each other is all in all rarer than could be expected, and this ability for speakers to time their utterances in a very meticulous manner (if they want to) should be accounted for. Again, this feature a priori seems to fit in nicely with our notion that pragmatic structure somehow transcends the individual participant.


These features of conversation, which taken together are not marginal at all, obviously do not fit very comfortably with the standard speaker-message-addressee picture of linguistic communication. The notion of ‘communicative intention’ becomes rather unappealing if applied to the ‘uh huh’ phenomenon: an analysis in terms of highly complex (and very fast) monitoring procedures of the other participant’s intentions would seem inelegant and uneconomical for an apparently ‘basic’ and reflex-like behavior. On the other hand, from the point of view adopted here above, the coherence between more obviously contentful structural features of conversation is not qualitatively different from (fits in the same picture as) the “uh huh” phenomenon. According to this picture, these structures cannot be located exclusively in either one of the participants. They suggest structuration mechanisms that are not reducible to the intentions of the participants as autonomous psychological units, but rather as patterns beyond the individual unit in which these units take part.

These observations, again, can be extended to non-verbal interaction: think e.g. of the motor coordination involved in structurally coupled behaviors such as dancing or physical combat, in which the major task of the participants is to ‘blend in’ with the other’s actions.


4. ‘Communication’ beyond the psychological subject or the biological organism

In the previous sections I have followed two different lines of thought that converge with respect to the implications they have for the locus of the ‘sense’ of action in general and discourse in particular:

(1) in section 1 I have argued that the sense of the behavior/discourse of an agent/speaker depends on structural features of that behavior which have a status that is independent of the cognitive state of both the agent/speaker and the observer/addressee; in section 2 I have tried to show that, as a consequence of this, the sense of so-called communicative behavior (i.c. monologic discourse) need not be fundamentally different from the sense of non-communicative behavior;

(2) in section 3 I have argued that the sense of multi-agent verbal behavior (i.e. conversation) can convincingly be analyzed in terms of structural patterns that account for the coherence of the segments that make up the conversation, and that the actualization of these patterns in conversation is not reducible to the properties (intentions, cognitive states, etc.) of the participants.


The notion that the proper level of analysis for communication is not the speaker and/or addressee is not completely absent from the literature. I will now briefly review a few approaches in which this notion is implied.[6]


(a) ethnology and ethnolinguistics

Within the ‘ethnological’ tradition out of which emerged Conversation Analysis, it has been claimed that verbal interactions (‘speech events’) have a proper status which is independent of the participants which engage in it (cf. Erving Goffman’s aphorism “Not men and their moments, rather moments and their men”). Although extreme and too simple formulations of this claim have been replaced by more complex models, the ‘structural’ bias has remained a fundamental feature of ethno-linguistics, conversation analysis and related disciplines (for a historical review, see Heritage 2001).


(b) system-theoretical cognitive biology

As a starting point for their approach to biology, Maturana and Varela defined the biological unit (or unity) as an autonomous item characterized by an ‘autopoietic’ (self-creating) organization, i.e. a set of relations between its components that allows for the maintenance of its autonomy and identity through the structural transformations it may undergo: even if all the actual components change, the unit remains recognizable as being ‘the same’ for an observer. The most simple living unit in this sense is the cell. Recurrent interactions between cells may give rise to a second-order unit that again has an autopoietic organization, i.e. that in its turn retains its identity for the observer throughout the structural changes it undergoes (‘structural coupling’); this is the case for meta-cellular organisms like humans. Cognition is then a matter of changes within the internal structure of the organism, just like growth or action.

Within this framework, communicative behavior (linguistic or not) cannot be described in terms of input and output of information by both units, since they are defined as autonomous closed systems; the interactions of these systems with their environment are fully determined by their own internal organization and the domain of perturbations that this organization allows for without loss of identity (i.e. by their ‘plasticity’). The regularities that an observer perceives between the behaviors of both units and their common environment, which he interprets as ‘communication’, cannot be ascribed to a flux of information between these items, but has to be formulated as the creation of a third-order structural coupling, by virtue of which both organisms participate in a higher-order organism with a common domain of perturbations (a common ‘world’).[7]

Attributing intentions to communicating organisms is then always the work of an observer seeing both organisms and their environment at the same time, without taking into account the fundamental internal organization of the organisms.[8]


Although communication or social interaction in general are not basic issues in the Anglo-Saxon tradition of Philosophy of Mind (characterized by a certain preoccupation with empirical data from neuro-psychology as well as from Artificial Intelligence and related disciplines), some recent developments in that field are pertinent to the present discussion in that they seem to undermine the status of the psychological subject (as coinciding with a neuro-biological mind) as the fundamental locus of cognition (and hence of communication). Interestingly, the post-phenomenological philosophy of Derrida (characterized by a close reading of the classics of the continental philosophical tradition) has come to very similar results.


(c) Derrida

In La voix et le phénomène (Derrida 1967) Derrida formulates a detailed critique on Husserl’s notion of ‘pure (i.e. pre-expressive) intentionality’. According to Husserl, intentionality (in the sense of ‘sense’) has the ‘Cartesian’ subject as its ultimate source: something is meaningful if it proceeds from an intentional act by a consciousness. Derrida shows that meaning crucially involves an intrinsically iterable procedure for making differences, which is incompatible with a unitary source of meaning: a difference implies a boundary between at least two items; the making of such a difference implies an (iterable) procedure. In Derrida’s work these notions are connected to a reflection on the notion of ‘writing’: many of the attributes which are traditionally ascribed to writing (iterability, potential decontextualization, potential absence of the writing subject, mediateness ...) can in fact be extended to ‘live speech’ and -for that matter- soliloquy.[9]


(d) Philosophy of Mind

In his best-seller Consciousness Explained (Dennett 1991/1993), Daniel Dennett assumes an empirical stance in his investigation of how the phenomena which constitute so-called consciousness relate to the functions of the brain as a neuro-biological item, adducing arguments from neuro-psychology (incl. the classic pathological cases), artificial intelligence and traditional philosophical-analytical argumentation. For the present purposes the following of his findings are interesting:

- neurobiologically speaking, nothing in the functional architecture of the brain corresponds to the psychological notion of ‘self’ or ‘psychological subject’ as the locus of ‘consciousness’, supposedly the central level of cognition where all the specialized types of information are brought together; rather, the functioning of the brain consists of a ‘pandemonium’ of parallel processes, shaped by biological evolution and performing different functions directly related to the survival of the organism;

- the apparent stream-like character and the apparent unity of consciousness are largely post hoc effects of the way people (‘we’) represent ourselves in discourse; this self-representation crucially involves behavioral patterns which have evolved through interaction and are transmitted by culture, and thus cannot be defined in terms of the individual brain alone (see section 5 below).

In the same broad research tradition, Susan Hurley’s excellent book Consciousness in Action (Hurley 1998) deals with issues in neuro-psychology and philosophy of mind (the mind-body problem; the input-output picture of the distinction between perception and action, etc.) but also with philosophical issues in the interpretation of Kant and Wittgenstein. Amongst many other things, Hurley shows the following:

- the distinction between perception and action (conceived of as an input-output relation) does not account for the phenomenology of human cognition and is not tenable from a (neuro-)psychological point of view, in that the well-attested interdependence between action and perception in everyday settings proves to be not just instrumental but intrinsic;

- the causal paths involved in cognition cannot be construed in terms of a strict inside-outside distinction with respect to the brain or the central nervous system, but crucially involve loops through the context in which the brain functions (both inside and outside the organism as a whole), which is particularly clear in the case of proprioceptive/kinesthetic perception; thus, neurologically speaking, there is no need to maintain the brain (or the central nervous system) as the sole substratum for cognition;

- the interwovenness between action and perception suggests that the neurological correlates of behavioral phenomena may be reformulated in terms of a “horizontal modularity”, i.e. in terms of task-based patterns, shared by different cognitive modes, rather than “vertical” modules based on representation-mode; this cross-modal task-based modularity essentially involves context-dependent factors.


These quite different lines of argument all bear upon the standard view of communication under scrutiny in that they all seem to indicate that the psychological subject or biological organism need not be the relevant substratum to which a scientific explanation of ‘sensible’ behavior can be reduced; in other words, there need not be any similarity at the level of the organism between different cases of what is understood as being ‘the same (type of)’ behavior.


5. Mimesis

Thus, the problem arises of how to integrate two notions which both have an intuitive appeal: (i) the structure of the discourse (or any other action) as an iterable pattern as the locus of sense (‘intentionality’, ‘meaning’, etc.), and (ii) the ‘person’ (‘subject’, ‘organism’) as the basic unit for psychology and (some types of) biology, as well as for jurisprudence, politics and ethics. In other words: how can we integrate the concept of ‘the individual’ (person, subject, organism) with the higher-order (‘supra-individual’) forms of structuration?

In this context, the notion of mimesis can be introduced. In Ancient Greek, the word m€mhsiw, often translated as ‘imitation’ (or ‘representation’), is connected with theatrical performance, and does not necessarily imply an explicit model which is imitated. Thus, the notion can be linked with the metaphor of ‘performing a role’, well-established in sociology, but should be extended beyond the sociological usage, so as to include the psychologically and biologically important notion of imitation.


(a) mimesis as imitation

First, it is important to stress the fundamental importance of ‘imitation’ as a basic reflex-like feature of human behavior (cf. Hurley 1998, 409-412, 416-417):

- the ability to imitate implies the ability to recognize the perceived behavior and the ability to relate this perceived behavior to motor-patterns of one’s own (e.g. infants are able to imitate ‘sticking out one’s tongue’ very early on); thus, this ability must be innate, because it seems to be a prerequisite for acquisition and apparently is a reflex-like reaction, even in infants; recently, a neurological basis for this reflex-like imitating behavior has been discovered by the discovery of the so-called ‘mirror neurons’ (for references and discussion, see e.g. Hurley 1998, 411-418);

- the acquisition of culturally transmitted patterns implies the ability to imitate the perceived behavior of elders, i.e. acquisition is a process of ‘becoming more like the others’, or to be able to assume a recognizable role in the community in which one lives;

- pure imitation plays a similar role in quite simple and basic forms of group behaviour, such as singing and dancing at parties, contagious laughing, or crowd behavior in mass events (applauding, chanting, ...); see also what has been said about echoism in conversation; likewise, imitation is constitutive of more complex phenomena like the spread of cultural patterns, as e.g. involved in language change;

- psycho-pathological cases suggest that ‘rational’ or more generally ‘normal’ behavior seems to imply the inhibition of directly and simply imitative behavior (Hurley 1998, 410), i.e. the ability to assume one’s role in a more complex fashion; this shows -again- that imitation as a simple reflex is one of the primary factors in human behavior.


(b) mimesis as the basic communicative mechanism

In section 2 above I have argued that understanding the other’s behavior (say: as an observer) is not a matter of decoding either (nor just empathy), but of the active ‘following’ of this behavior. Likewise, multi-agent actions such a conversation (or football) can be described insightfully without reference to the internal states of the participants: it suffices to indicate the roles they take in the action (or ‘game’) at hand.

In any case, sensible behavior involves access to the action-type of which the observed behavior is a token, but need not involve ‘representations’ in the usual ‘informational’ sense of the word.[10] Furthermore, the patterns which constitute such action-types intrinsically involve structural ‘roles’ which cannot be reduced to the individuals who assume them in any particular action-token. In this sense, mimesis can be construed as the basic communicative mechanism - not only in acquisition, but also in ‘mature’ social behavior


(c) mimesis as constitutive of person-hood

It follows from the above that the psychological subject (as the locus of consciousness) does not coincide with the biological organism (or its neurological system) as such. It is impossible to describe person-hood (incl. consciousness, intentionality etc.) on the level of the bio-chemistry of the organism:[11] these notions essentially involve the ‘normative’ notions which enter into the interpretation of behavior as sensible action. For there to be consciousness, there needs to be mimesis: the construction of subject-hood implies the recognition that one is ‘like the other one’ (cf. notions such as the ‘mirror stage’ in the psycho-analytical literature; cf. e.g. Kristeva 1977, 377-379 et passim) and the stuff which makes up the contents of consciousness crucially involve supra-individual patterns.


(d) mimesis as constitutive of culture (memes)

In the context of trying to account for human behavior within a broader biological frame-work, the problem of the phylogenetic origin as well as the propagation and survival of recurring and historically determined patterns as invoked in the above has been addressed in terms of the notion of ‘meme’, which has been coined on the basis of the root of ‘mimesis’: what genes are for the biological organism, is called memes for the higher order structural couplings which constitute social life. Action types (in the sense of sections 1, 2 and 3 above) can be considered as memes; language in general and a language in particular can be viewed as a prototypical instance of a system of memes (for references amd discussion, see e.g. Dennett 1991/1993, 199-226).[12]


Thus, the notion of ‘mimesis’ may be construed so as to cover all relations between the subjects and as the fundamental mechanism underlying both society and person-hood:

- mimesis is the basic reflex-like behavioral pattern underlying social behavior;

- mimesis constitutes the actual structure of communicative behaviour;

- mimesis is the ontogenetical and phylogenetical basis for the emergence of subject/person-hood;

- mimesis accounts for the phylogenetic emergence and transmission of supra-individual patterns.


6. Apparent arguments in favor of the subject-based view

In the previous sections, I have pointed out various lines of thought which converge in that they yield a conception of communication which does not start from the speaker-message-addressee picture. Still, the fact remains that this picture has prevailed for quite some time and still seems to function properly as a basis for many branches of scientific research. Thus, the question has to be addressed as to when the speaker-addressee picture does seem to be attractive. Three recurrent types of situations immediately come to mind as at first sight incompatible with the alternative views:

(a) meta-behavior (explaining one’s intentions, ...);

(b) faked communication (deceit, fiction, ...);

(c) failed communication (misunderstandings, ...).[13]

These cases seem to support the notion of the speaker’s intention as the source of the meaning of communication rather than rather than the superordinate practice in which both participants participate, in that they appear to involve a distinction between the overt form of the message and the underlying intention of the speaker. Furthermore, the last two cases apparently show a clear asymmetry between speaker’s meaning and addressee’s meaning.


(a) meta-behavior

In many cases (though by no means in all cases) people are able to give some account for why they do whatever they do. In some cases (but by no means in prototypical ones) they also perform some explicit planning of what they are going to do, before actually performing the intended action. These phenomena suggest that intentions are pre-existent with respect to the behavior itself, which in its turn may suggest that the sense of actual actions resides in this kind of pre-behavioral intentions.

However, the link between explicitly formulated intentions and the actual behavior is a complicated one, as has been extensively pointed out by Wittgenstein (see footnote 1 above). It should be noted that expressing one’s intentions (whether aloud or while thinking about what one wants to do) is a behavior that is of a different type (‘meta’-behavior) as the explained behavior itself and is by no means an essential part of it. Note that in many cases one finds it difficult to adequately express one’s intentions in any detail, beyond rather dubious stock explanations (cf. also notions like ‘sub-conscious’ or ‘unconscious intentions’, etc.). These particular kinds of meta-behavior (‘planning one’s actions’, ‘accounting for one’s actions’ or ‘expressing one’s intentions’) and the relation they have with the ‘actual’ actions intended can be compared with other kinds of meta-behavior, such as praying or performing magical rituals before embarking on an action: though they are felt to have a particular relation to or even to be an integral part of the ‘real’ action they are connected with, they actually constitute an independent action and nothing garanties that these actions are direct reflections of how the action they are combined with actually function.


(b) fake communication

Similarly, phenomena like lying and deceit but also fiction and irony seem to be more easily compatible with a subject-based notion of sense, in that they imply an opposition between the prima facie sense of the behavior on the one hand and the ‘real’ (but concealed) intentions which underlie it on the other hand. Thus, deceit implies an asymmetry between speaker/manipulator and addressee/victim which seems to support the notion that the speaker’s intention is the main locus of the sense of his behavior.

But then again, these actions are structural patterns themselves, implying the embedding of a ‘basic’ first degree structure in an encompassing second degree structure. In order to be able to understand deceptive (or playful) behavior as such, it takes the construction of a superordinate pattern in which the prima facie pattern is embedded. Obviously, the resulting complex patterns are in se iterable as well. In fact, it is always possible to continue adding superordinate interpretative frames on top each other (as is exemplified by the practice of paranoiacs but also by hermeneutic practices such as psycho-analysis).

Consider the following analogy. The functioning of an animated cartoon is in its effects dependent/parasitic on the way actual people/creatures move, and its success as a ‘representation’ depends on how well it mimics the movements it is intended to represent, but the actual procedures of making such a cartoon do not teach us anything on how actual creatures move. Likewise, deceit (or faked communication in general, including theatrical fiction), if it is to function, depends on the fact that it can be perceived as ‘regular’ communication, i.e. on the fact that it retains the relevant ‘surface’ features of regular communication.

A description of deceptive communication that does not take into account the deceptive aspects of it is in a way still a correct, though incomplete, description of it; as a matter of fact, a correct description should even start from its non-deceptive surface. Again, mimesis can be invoked as the relevant mechanism: the speaker who tries to deceive actually takes on a role that is understandable as such; in order to understand the deceit as such one has to understand that taking up this role in its turn part of a more complex meta-role. Thus, the problem only arises if one chooses to equate a role with a psychological subject and a biological organism.


(c) failed communication

Misunderstanding, infelicitous communication or even overt differences in the presuppositions underlying different interacting speakers’ utterances are intuitively best understood in terms of a divergence in the representations that these participants have of the situation at hand. In the same vein as the comparisons between verbal and non-verbal behavior made in the above, one can compare this situation to dancing off-beat or to the knock-out punch which ends the coordinated behaviors between boxers. In all these cases, accounting for the sense of the behavior implies a supra-individual pattern, but at the same time a failure for that pattern to establish itself or the breakdown of such a pattern.

In terms of the system-theoretic biology of Maturana and Varela (see above), this could be formulated as follows: the behaviors of two or more second-order organisms enter into a structural coupling which may give rise to the establishment of a third-order organism; in some cases however the structural coupling seems to be less complete (?) than in other comparable cases, or the pattern may be short-lived. These situations may be compared to phenomena which are called ‘death’ or ‘procreation’ in the case of second-order organisms: death is a limit to the persistence of biologic units as units for an observer; similarly, procreation is a matter of shifting the boundaries of such units: what was part of a certain biological unit, the next moment is better analyzed as a biologic unit of its own. Thus, the problems concerning the proposed ‘third-order organsisms’ have exact parallel at the level of the familiar second-order organisms.[14]

Note that this problem has a single-speaker counterpart as well: many phenomena of monologic speech which are perceived as performance dysfluencies (afterthoughts, digressions, grammatical dysfleuncies, ...) crucially involve the recognition of the ‘normative’ pattern which constituets the sense of the performance, but also the fact that other (non-normative? individual? cognitive?) factors have disturbed the ‘normal’ realization of that pattern (cf. the notion of ‘paracoherence’ discussed in Scheppers 2003). Again, these phenomena illustrate the problem of how to relate the supra-individual with what seems to be individual cognitive factors.

It is obvious that some of these problems are genuine problems which seriously challenge the type of approaches suggested in the above: the common sense notion of the individual substratum for cognition is hard to do without, and it is obviously not my purpose to argue that one should try to do without it at all cost.[15] Still, I hope to have shown that the notion of a supra-individual level of patterning is indispensable and that a reductionist stance with respect to the relationship between the individual and the supra-individual is highly problematic.


7. Conclusions

By way of conclusion, I will point out a few of the consequences which the lines of thought sketched in the above may have for various rather general issues in linguistics and neighboring disciplines.


(a) psycholinguistics, psychology and neuro-psychology

The alternative views to communication suggested in the above seem to be in direct contradiction not only with ‘common sense’ psychology but also with the most fundamental presuppositions of the standard neuro-approaches to cognition. However, we have pointed out that at least some authors working in that tradition have argued precisely against the traditional views, starting from neuro-psychological data. Furthermore, with the discovery of the so-called ‘mirror neurons’, interesting neurological data have come into view which seem to support the viability of the present mimesis-based approach, while at the same time bridging somewhat the seemingly enormous gap between this account and the neuro-sciences.

Of course, I do not mean to suggest that technical research issues in psycho-linguistics and (neuro-)psychology in general would immediately benefit from the lines of thought sketched in the above. Thus, for instance, the important problem of language acquisition (and acquisition in general) obviously necessarily involves the learner as an individual of which several consecutive states are compared; on the other hand, it also involves the ‘normative’ supra-individual behavioral patterns which s/he is acquiring (or not). It should be noted that the available frameworks for addressing technical issues in language acquisition do not address any of the issues touched upon in the above. Although -of course- no clear picture as to how to tackle the more technical issues concerning acquisition emerges from the above a priori considerations, it seems worthwhile to try and broaden the debate in this field by taking into consideration contemporary critiques on the input-output picture of cognition and the more pragmatically oriented view on cognition as well as e.g. the accounts of acquisition and growth in terms of plasticity and mimesis, as briefly sketched in the above.


(b) ethico-political consequences

On different occasions on which I presented my arguments about coherence and intention (as summarized in section 1 and 2 here above; also see Scheppers 2003) to different audiences with different backgrounds (philosophy, cognitive psychology, linguistics), different interlocutors have formulated ‘humanistic’ objections against my views on action and discourse and especially against the views on the status of the speaker/agent elaborated in the above; these objections invariably argued that the theoretical ‘abolition of the subject’ implies de facto an infringement of the irreducible status of the individual human as the basis of the Western political and ethical values. I agree that the choice of basic assumptions is not politically innocent. But from the point of view of the approach presented here, I argue that the notion of ‘individual’ is -precisely- a strictly political or juridical one: it occurs as an irreducible ground only in contexts in which the responsibility for this or that effect is at issue; in most human activities, outside these ‘forensic’ contexts the notion hardly plays a role; e.g. for the scientific description of human behavior, I have extensively argued that it is not a viable notion. But it should be noted that science is a different genre of action from politics or jurisprudence, and hence can be expected not to imply the same cognitive content or basic categories. Thus, the views suggested in the above are not only fully compatible with insisting on the importance of defending the irreducibility of the individual human in political and ethical matters, but could even be the starting point for arguing against the mitigation of this principle on the basis of ‘scientific’ arguments, e.g. the role psychiatry/psychology plays in forensic matters and the role ‘sociological’ notions such as ethnicity increasingly play in politics.


(c) status of the message/text

On a more theoretical level and in a closer connection with linguistics proper, the alternative views on communication have consequences for the status of the ‘message’ or ‘text’ as well. These aspects cannot be dealt with here and can only be mentioned.

First, the blurring of the difference between content (locution) and action (illocution) as described in sections 1, 2 and 3 above has for a consequence that the ‘message’ has to be described as less independent of the agents’ intentions. >From the point of view of the structure of the message (‘text’) itself (the point of view adopted in the above), the text qua discourse itself constitutes the speaker as such:[16] the speaker becomes one of the structural positions within the overall iterable pattern which constitutes the text as a discourse; the text thus cannot be viewed as a pure object, independent from its ‘realization’ (interpretation, utterance ...).[17]

Likewise, as a consequence of the more holistic view of discourse, the difference between ‘text’ and ‘context’ becomes a highly problematic one: if discourse is viewed as a structural pattern which intrinsically involves the speaker and addressee as part of its structure, the ‘outside’ of discourse becomes a very relative notion.[18]




In this paper, the convergence has been noted of a few different lines of thought which all point in the direction of a view on communication in which speaker and addressee qua psychological subjects or biological organisms are not the main locus for the sense. It has been argued that the intentional/sensible behavior (incl. communicative behavior) can be described in terms of the structural features of the behavior itself and need not involve the level of the individual organism/subject as such. On the other hand, a few of the many puzzles which -at least apparently- involve both the supra-individual and the individual level have been summarily described.




8. References

Anscombe, G.E.M. 1976 (2nd edition). Intention. Oxford, Blackwell.

Bennington, Geoffrey & Derrida, Jacques. 1991. Jacques Derrida. Paris, Seuil [Coll. Les contemporains].

Berge, K.L. 1998. “Communication”. Mey (ed.) 1998, 140-147.

Brown, Gillian. 1995. Speakers, listeners and communication. Explorations in discourse analysis. Cambridge, Cambridge University Press.

Denyer, Nicholas. 1991/1993. Language, Thought and Falsehood in Ancient Greek Philosophy. London-New York, Routledge.

Dennett, Daniel C. 1991/1993. Consciousness Explained. London, Penguin.

Derrida, Jacques. 1967. La voix et le phénomène. Introduction au problème du signe dans la phénoménologie de Husserl. Paris, Presses Universitaires de France.

Derrida, Jacques. 1990. Limited Inc. Paris, Galilée.

Heritage, John. 2001. “Ethno-sciences and their significance for conversation linguistics”. Brinker, Klaus & Antos, Gerd & Heinemann, Wolfgang & Sager, Sven F. (eds.), Text- und Gesprächslinguistik. Linguistics of Text and Conversation. Volume 2, 908-919. Berlin, de Gruyter.

Hoekstra, Marieke & Scheppers, Frank. 2003. “ÖOnoma, =∞ma et lÒgow dans le Cratyle et le Sophiste de Platon. Analyse du lexique et analyse du discours”. Antiquité Classique 72, 55-73.

Hurley, Susan L. 1998. Consciousness in Action. Cambridge (Mass.), Harvard University Press.

Kristeva, Julia. 1977. Polylogue. Paris, Seuil.

Levelt, Willem. J. M. 1989. Speaking: from intention to articulation. Cambridge (Mass.), MIT press.

Mann, William C. & Thompson, Sandra A. 1988. “Rhetorical Structure Theory: Toward a functional theory of text organization”. Text 8, 243-281.

Maturana, Humberto R. & Varela, Francisco J. 1980. Autopoiesis and Cognition. The Realization of the Living. Dordrecht-Boston-London, Reidel.

Maturana, Humberto R. & Varela, Francisco J. 1992. The tree of knowledge: the biological roots of human understanding. Boston, Shambhala.

Mey, Jacob L. 1993. Pragmatics: an introduction. Oxford, Blackwell.

Mey, Jacob L. (ed.) 1998. Concise encyclopedia of pragmatics. Amsterdam, Elsevier.

Mey, Jacob L. 1998a. “Pragmatic acts”. Mey (ed.) 1998, 701-703.

Mey, Jacob L. 2001 (2nd edition). Pragmatics: an introduction. Oxford, Blackwell

Roulet, Eddy & Fillietaz, Laurent & Grobet, Anne & Burger, Marcel. 2001. Un modèle et un instrument d’analyse de l’organisation du discours. Bern, Lang.

Sacks, Harvey. 1992/1995a. Lectures on Conversation. Volume I. [edited by Gail Jefferson; with an introduction by Emanuel A. Schegloff]. Oxford, Blackwell.

Sacks, Harvey. 1992/1995b. Lectures on Conversation. Volume II. [edited by Gail Jefferson; with an introduction by Emanuel A. Schegloff]. Oxford, Blackwell.

Schank, Roger C. & Abelson, Robert P. 1977. Scripts, Plans, Goals, and Understanding. An Inquiry into Human Knowledge Structures. Hillsdale, Erlbaum.

Scheppers, Frank. 2003. “P(ragmatic)-trees. Coherence, intentionality and cognitive content in discourse and non-verbal behavior”. Journal of Pragmatics 35, 665-694.

Varela, Francisco J. 1979. Principles of biological autonomy. New York, North Holland.

Wittgenstein, Ludwig. 1989. Werkausgabe Band 1. Tractatus Logico-Philosophicus. Tagebücher 1914-1916. Philosophische Untersuchungen. Frankfurt am Main, Suhrkamp.




© Frank Scheppers. Círculo de Linguística Aplicada a la Comunicación 19, September 2004. ISSN 1576-4737.


clac 19



[1] This remark obviously reminds one of the analyses of the notion ‘intention’ (‘Absicht’) in Wittgenstein’s Philosophische Untersuchungen (Wittgenstein 1989, 343 et passim; cf. also Anscombe 1976). See section 6 below.

[2] The classical approach to action types in the cognitive sciences is of course Schank and Abelson’s ‘script’ model (Schank & Abelson 1977).

[3] Cf. the analyses of the so-called ‘private language’ argument by both Wittgenstein and Derrida. See Bennington & Derrida 1991. 

[4] This has an important epsitemological corollary: the sense (the coherence) of an action can only be observed if the observer and the observed agent are sufficiently alike.

[5] Of course the present line of argument is entirely parallel to a quite traditional line of thought in linguistics, as exemplified by Saussure’s langue-parole dichotomy. Obviously, e.g. word meaning is not inherently “private”, can be repeated indefinitely and retains its type-identity across an indefinite number of reiterations, within or across speakers. Here I propose to generalize the non-private/reiterable/etc. character of language to the whole of sensible behavior. It is interesting to note that the lines of argument in sections 1 and 2 remind one of some of the developments in -at first sight- quite different philosophical traditions. The notion that sense depends on repeatability within or across speakers/agents (or even in the absence of the speaker himself) and the denial of any privileged position for the psychological subject with respect to the sense of his own acts are, on the one hand, central to the early works of Derrida, especially in his careful analyses of Husserl. Interestingly, quite analogous considerations are, on the other hand, present in recent developments in the Anglo-Saxon, more empirically and epistemologically oriented tradition (e.g. Dennett 1990/1993 and Hurley 1998); see below sections 4 and 5. Some of the issues I am addressing here closely remind of the rather fierce discussion (?) between Derrida and Searle (as related in Derrida 1990), which concerned precisely such issues as the ‘iterability’ of contents, the status of the status of the speaker, as well as some related issues such as the status of the notion of ‘context’ in general and the ‘normality’ of some realizations of a text as opposed to other ‘abnormal’ realizations (see also sections 4, 6 and 7 below).

[6] It should be noted, for that matter, that is significant that linguistics in its traditional form of ‘grammar’ has been alive for centuries without there being any particular doctrine on how it operates within the language user.

[7] Cf. Maturana & Varela 1992, 195: “We call social phenomena those phenomena associated with the participation of organisms in constituting third-order couplings”; and: “As observers we designate as communicative these behaviors which occur in social coupling, and as communication that behavioral co-ordination which we observe as the result of it”. For a critique of the notion ‘transmitted information’, cf. Maturana & Varela 1992, 196.

[8] This system-theoretical approach presents interesting epistemological aspects as well, in that it takes into account the fundamental role which the observer, an autopoietic unit himself, plays in determining what the relevant facts are. Furthermore, this theory implies a self-referring epistemology: the theory is fully applicable to itself; something similar goes for the pragmatic approach to ‘sense’, as sketched in Scheppers 2003 and sections 1 and 2 above: scientific practice can be analyzed as a ‘genre’ of human behavior on a par with other types of activity.

[9] For an accessible introduction to Derrida’s work, see Bennington & Derrida 1991.

[10] One may construe the notion of ‘representation’ in terms of the plasticity of the organism as a whole, but then ‘breathing’ is a representation of the air and walking a representation of the floor.

[11] Note that the etymology of the word ‘person’ belongs to the same semantic field as ‘mimesis’: (Latin ‘persona’ means ‘mask’, ‘role’, ‘character in a play’)

[12] It should be noted that some of the approaches and ideas described in the above (esp. the system-theoretical biology of Maturana and Varela, but also the notion of ‘meme’) emerged from or were influenced by work in the field of cybernetics. These cybernetic lines of thought continue developing as well and some arguments parallel to the present ones can be found in that field; for references, cf. e.g. the Principia Cybernetica Web website (by F. Heylighen, C. Joslyn, V. Turchin),

[13] An interesting epistemological parallel can be noted. The prevailing conception of semantics in logic and linguistics in terms of truth-value, and with this the problem of reference/denotation as the main issue, emerged quite late in the history of mankind (to be exact: with Plato’s Sophist, see Hoekstra & Scheppers 2003), whereas previously the possibility of falsehood (errors or lies) was the main conceptual problem (see Denyer 1991/1993). Likewise, the conception of communication in terms of speaker’s and addressee’s intentions seems to be mainly motivated by infelicitous or deceptive cases.

[14] Other conceptualizations in system-theoretical biological terms may be possible, e.g. the competition between third order couplings to impose themselves.

[15] Note that one can regard this epistemological problem as an avatar of the old mind-body problem as well as of the old langue-parole or competence-performance problems.

[16] Cf. the article “La fonction prédicative et le sujet parlant”, in Kristeva 1977, 323-356 et passim.

[17] Note that if one chooses to look at discourse from the point of view of the speaker as a biological organism or a psychological subject, the discursive process should be construed as a process within this organism/subject, which -if one pursues this point of view in a systematic way- makes the message as an independent object (and hence as an iterable and understandable/sensible one) disappear as well: the ‘classical’ view of communication (as described at the beginning of this paper) thus is necessarily à cheval between two points of view, the compatibility of which is highly questionable.

[18] Cf. also Jacob Mey’s theory of ‘Pragmatic acts’ (Mey 1998a; Mey 2001, 206-235), which stresses the notion that textual and contextual aspects involved in any adequate analysis of discourse should be viewed as an integral part of a complex structure which he calls a ‘pragmatic act’ (or ‘pract’), which -as its name indicates- is of a primarily pragmatic nature, and which is to be conceived of as an iterable pattern, involving all the pragmatically relevant elements (including speaker and addressee, referents, etc.).