We have seen that linguists like to think of linguistic items as having fixed and stable properties, among other things inert, literal meanings. This way of looking at linguistic phenomena as objects is coupled with a very popular model of linguistic communication, i.e. the one which portrays the communication process as the conveyance of a message, or, in other words, some kind of transportation of a certain, fixed message from the speaker to the listener, a transfer of given thoughts and feelings as if these thoughts and feelings were independent of and prior to the “encoding” and “decoding” processes in communication. This view is associated with the metaphor that meanings are objects (VII.1):

“Meanings are objects. Linguistic expressions are objects. Linguistic expressions have meanings (in them). In communication, a speaker sends a fixed meaning to a hearer via the linguistic expression associated with that meaning. On this account it is possible to objectively say what you mean, and communication failures are matters of subjective errors: since the meanings are objectively right there in the words, either you didn’t use the right words to say what you meant or you were misunderstood.” (Lakoff & Johnson 1980:206)
The view that communication is basically a transfer or transportation of fixed messages shows up in many disguises, under several different but related metaphors. One such metaphor is involved in “the bucket theory” of meaning

“Words, like little buckets, are assumed to pick up their loads of meaning in one person’s mind, carry them across the intervening space, and dump them into the mind of another” (Osgood 1979:213)
Common to this and many other theories of linguistic communication is the view that words, sentences, and other expressions are containers loaded with meaning (content); words are, in other words, meaning-full (sic!). Reddy, in an interesting paper (1979), treats the matter under the name of the conduit metaphor, a frame of thinking which implies that:

“(1) language functions like a conduit, transferring thoughts bodily from one person to another; (2) in writing and speaking, people insert their thoughts and feelings in the words; (3) words accomplish the transfer by containing the thoughts or feelings and conveying them to others; and (4) in listening or reading, people extract the thoughts and feelings once again from the words.” (Reddy 1979:290)
Reddy points out that the conduit metaphor underlies many common English ways of talking about linguistic communication; it is deeply rooted in our (Standard Average European) culture and hence very difficult to free oneself from. Here are a few expressions that show how we have conventionally and inadvertently become accustomed to the conduit metaphor in our everyday language:2)

(1) Try to get your thoughts across better
(2) Whenever you have a good idea practice capturing it in words.
(3) Try to pack more thoughts into fewer words.
(4) The lines may rhyme, but they are empty of both meaning and feeling.
(5) Can you actually extract coherent ideas from that prose?
(6) He writes sentences in such a way as to seal up the meaning in them.
(7) His words carry little in the way of recognizable meaning.
(8) Please pay attention to what’s there in the words.
Another metaphor for basically the same perspective on linguistic communication is one which is particularly familiar to linguists, psycholinguists and communication theorists; it construes the communication process as a series or recodings, or mechanical translations, of the same message with a certain fixed meaning. Accordingly, this has been termed the translation theory of speech communication (Garrett 1975, Linel1 1982a). In short, the theory says this. For some reason, a speaker comes up with a certain “idea” or “thought” which he wants to communicate to someone else. This “message” is therefore encoded in the speaker’s brain into, say, patterns of neural activity which travel along nerve paths down to the various speech organs, where the message is translated or recoded into articulatory movements, which in turn give rise to a new form of representation, i.e. acoustic sound waves. These waves convey the message to the hearing organs of a listener, and there it is translated again into new codes. After a number of additional recodings, the message is finally regained in approximately its original form, when it reaches the listener’s brain or mind.

The theory just sketched is obviously inspired by Shannon and Weaver’s classical model of technical information transfer. It recurs in almost every introductory textbook on linguistics or speech communication (e.g. Denes & Pinson 1963). I have given it couched in basically physical terms, but essentially the same type of translation-theoretical approach permeates generative psycholinguistics, where, instead, scholars prefer to talk about abstract mental representations corresponding to the constructs of linguistic competence theory. Thus, the message is there automatically processed through a series of purportedly “linguistically significant” representations. For example, the production processes may be explained as a linear sequence of translations of the same utterance via representations such as the following:

Semantic Deep-structural Shallow- ~
´ representation representation structural

representation representation PHONETIC SIGNAL

Exactly what “stages” are included in this series of translations naturally depends on the particular theory espoused. But the general outline is the same; although early (“remote”) representations may be very different from late (“superficial”) representations, they still “represent” the same message unit (usually a sentence, Û VI.1) with the same fixed meaning. The process of perception and comprehension would, at least according to the most naive theories, be essentially a translation in the reverse direction.

Proponents of such models of speech communication naturally admit that the whole process may be disturbed by “noise” in several ways. But the idealized version of the model clearly implies that one and the same fixed message gets across without distortion. If we assume that this model is reasonably adequate, then it follows that we should concentrate our efforts of linguistic analysis on studying the message at that particular stage, where it is most easily accessed, i.e. in the observable products (the phonetic behavior). It is thus assumed that the meaning of what is said can be gained simply by applying a linguistically correct analysis to these linguistic products. This is the view of the autonomous linguistic message once again.

Two examples of translation-theoretical models in psycholinguistics and phonetics.

(from Denes x Pinson 1963:4)
“A possible information-flow model of speech production” (from Cooper 1980:298)

The translation model is a thoroughly misleading and inadequate model as applied to speech communication in face-to-face interaction. It grossly underestimates the complexity of the social interplay between speaker and listener, and of their interaction with the surrounding situation:

“This model of communication objectifies meaning in a misleading and dehumanizing fashion. It influences us to talk and think about thoughts as if they had the same kind of external, intersubjective reality as lamps and tables.” (Reddy 1979:308)
Translation models are inadequate for all sorts of meaningful communication between human beings; contrary to what these models suggest, we always take into consideration the productive and interpretive activities of senders and receivers3) situated in social contexts. However, it seems that the conduit or translation metaphors are less inaccurate as applied to communication by written messages. Indeed, it is obvious that the whole frame of thinking is inspired by such communication; someone writes a letter or a book, sends it away to an addressee/receiver (or a group of receivers), who will read it. The written text has to be relatively explicit and relatively autonomous in various ways. Although the reader too must rely on various expectations and background knowledge, he is normally strongly guided by the text in his interpretive activities. Thus, the whole process reminds us to some extent of a mechanical transfer or transportation of a message.4)

The reasons why this model cannot be exploited in the explanation of spoken dialogues are many. Most of them have already been discussed. Let me sum up a few points.

First, the situational interpretations relevant to speaker and listener are never equal to the linguistic meaning associated with the utterance itself; any interpretation goes beyond the linguistic meaning which in itself is vague and allusive (VII.1,5).

Secondly, there is no complete linguistic meaning nor any fully developed intended interpretation in the mind of the speaker before the utterance has been compiled and its outer form has been determined; many aspects of meaning are the result of the verbalization process itself (Linell, forthcoming.), and some interpretations are clearly discovered only after the verbalization (VII.5).

Thirdly, there are no uniquely correct situational interpretations; both speaker and listener may, e.g. vary in their depths of intention and understanding (VII.5). Any interpretation is in principle inherently negotiable and extendible, i.e. there are no fixed meanings being transferred in communication.

9.2 The functions of language

Linguists tend to have rather definite preferences for certain theories of linguistic meaning and communication, as we have seen. Attached to these theories are certain implicit evaluations of which communicative functions are most important and most characteristic of language.

In this broad overview we may distinguish between functions along three different dimensions. First of all, one should single out the dialogic functions in social interaction from the monologic functions in thinking and other kinds of intraindividual communication. This point will be further discussed in IX.3-4. It may be recalled that the conditions on written communication are such that both communicating parties may be said to use language in a monologic fashion.

Secondly, it is possible to focus on the different interacting factors of the communication process and discern the following four aspects:

a) focussing-on the sender: the expressive functions. What is communicated is, under this aspect, seen as expressions of the sender’s beliefs, views, feelings, attitudes, volitions, needs etc. Naturally, some of these are unconscious to the sender and thus not intended by him.
b) focussing on the receiver: the evocative functions. Communication is here seen as directed towards evoking certain reactions on the receiver’s part. What is conveyed serves to make him perceive or understand something, have certain feelings and attitudes, or perform a certain action.
c) focussing on the subject matter, i.e. on the imaginary or objective reality that the message refers to: the referential functions. We are then concerned with how communication is used to refer and describe, to analyze, argue about, and explain things in the world.
d) focussing on the relation between sender and receiver: the social functions. From this point of view communication serves to establish and maintain social contact between the communicating parties. A great deal of oral discourse takes place simply because social situations and conventions require it, perhaps because one feels obliged to avoid an embarrassing silence. Some speak of the phatic function of language (Malinowski 1949:315).
Thirdly, we shall relate communication to different psychological dimensions of the communicating parties. There seem to be at least three different aspects:

a) the cognitive functions which have to do with knowledge, beliefs, and intellectual understanding. From the expressive point of view communication may be seen as expressing the views and beliefs of the sender, and in an evocative perspective communication is viewed as directed towards arousing beliefs, conveying information and bringing about understanding. Since cognitive activities always have some “intentional objects” – they always “are about” something in the world that the parties believe or want each other to believe – the cognitive and referential functions are heavily intertwined and are seldom kept properly distinct.
b) the emotive functions; in communication the sender ex presses his feelings, attitudes, emotions and desires, and this may also evoke the corresponding states and activities in the receiver.
c) the practical functions; much communication is used for guiding the behavior of the receiver, i.e. the messages are conveyed in the hope of arousing a readiness in the receiver to act in certain ways. For obvious reasons, practical and evocative functions often go together. But the sender may also use language for the purpose of planning and guiding his own actions; in that case language is used monologically, and the sender simultaneously plays the part of receiver.
Most linguistic messages are multifunctional, although many are specialized in various ways so as to stress certain particular aspects. However, the actual state of affairs is rather inadequately reflected in most theories of linguistic meaning, which are very much focussed upon the cognitive and referential aspects. These are the properties which are most independent of particular senders and receivers, and the medium of written communication very strongly emphasizes precisely these aspects. The aim behind certain types of texts is very much to describe reality as explicitly as possible and to display the lines of argumentation as clearly as possible (III.1). The ideal is often to pursue description, explanation and argumentation in such a way that they appear to be independent of the views of the author (cf. legal and scientific texts).

In spite of the fact that such objectives are typical of certain forms of written communication, there is also a wide-spread belief that language in general is primarily directed towards the cognitive and referential functions of communication.

All this means that little attention is paid to those functions which often dominate in face-to-face interaction. That is, the expressive, evocative and social functions, i.e. those which are directly related to the communicating subjects, tend to be neglected, and the same holds for the emotive and practical aspects, as opposed to the cognitive side. Notice that these functions are quite dependent on the context at large, on nonverbal communication and prosody (cf. VI.2).

We only need to take a short look at contemporary work in linguistic semantics to see that only cognitive and referential functions are really considered to be of fundamental linguistic importance. This is, as was just pointed out, in full agreement with the written language bias. It also squares well with the emphasis on individuals and monologues rather than social interactions and dialogues that is typical of much of American psychology and linguistics, and perhaps of Western culture in general).

American academic psychology has always had a distinct bias towards individual psychology. This is also true of psycholinguistics, both the variant inspired by generative linguistics and the more recent information processing theories in “cognitive science”. One may also recall Chomsky’s view that language is primarily a means for the expression of thought. The utilization of language in communicative dialogues is to Chomsky more or less an accidental phenomenon (Chomsky 1975:56 ff.). Thus, language is consistently seen as a means for storing, representing, transmitting (transporting) knowledge, not as an ingredient in people’s social interaction. Since writing is monological rather than dialogical, this is exactly what we should expect.

Among those scholars who have thought deeply about the written language perspective in our Western culture is Walter Ong. Here are two pertinent passages:

“Because our concept of what words are is so tied up with a feeling for words as written or printed, a basic difficulty in thinking about words today is our tendency to regard them largely or chiefly or ideally as records.”
“Once we can get over our chirographic-typographic squint here, we can see that the word in its original habitat of sound, which is still its native habitat, is not a record at all. The word is something that happens, an event in the world of sounds through which the mind is enabled to relate actuality to itself. To understand more fully what this implies, we must examine in some detail what an oral-aural culture in general is like.” (Ong 1974:167-8)
The functions of words in oral cultures are of course very much emotive and practical. As an additional point, which is not merely a curiosity, one might note here that “primitive” cultures often regard speech as magic. Words have a magical function; one can bring about things by the spell of words; nature and other people may be changed (Ong op.cit.:168-9).

“Primitive man commonly feels that one can use words to hurt people as one can use an arrow or a spear; hence various magic formulas.” (Ong 1974:168)
Since literate culture so strongly de-emphasizes the magical aspects (excepts for certain obvious cases such as rituals and other ‘performative’ uses), we seem to have become blind to these aspects of oral communication. After all, people still frequently exploit the magic of words, in attempts at enticing, flattering, offending, hurting etc. There is always an element of magic in (true, full-blown) communication, especially in face-to-face interaction, when a speaker – by means of words and non-verbal signals – makes things happen with the listener, when he induces thoughts, images, feelings, moods, and volitions in the listener’s mind. Moreover, if we turn to children’s play with words, we can see the indissoluble relation between language and social and physical reality. In (symbolic) play children pretend that things happen, they make things happen in the concrete world of play, and this they bring about by the mere use of words (Strömqvist 1980).

9.3. Thought and expression: Content and form

Translation theories of linguistic communication suggest a separation between the content and the expression of the linguistic message. Early in the process of utterance production we find the thoughts and “representations” which directly encode or are even equal to the content of the message, and only later the whole message gets transformed into phonetic behavior; the message, and its meaning, is given an outer form. The comprehension process is essentially the reverse; the listener starts with the expression and ends up with the content. These theories are legion in the overwhelming majority of books and articles dealing with utterance production and perception within modern psycholinguistics, but the general ways of thinking are old; the theory of utterance production just sketched may be associated with, among others, Wilhelm Wundt (Gardiner 1932, Blumenthal 1970).

There are three elements in the philosophy underlying translation theories of utterance production (and comprehension) that are worth discussing in this context:

a) the thesis that speech, or linguistic products in general, are nothing but expressions of thoughts and ideas,
b) the thesis that content and form may be ripped apart as two independent phenomena,
c) the thesis that messages and their meanings or interpretations are mental entities, i.e. things existing in the inner world of our minds (though they presumably have physical substrates in our brains).
Although these theses are interrelated, we shall here deal with them separately.

It is a popular conviction that the process of saying something meaningful involves two steps first the speaker comes up with and elaborates an idea or thought, and then he expresses this idea linguistically in a verbal utterance. As a general theory of discourse production, this is wrong, however, because speakers elaborate their messages through the verbalization process itself, i.e. form and content are created simultaneously. On the other hand, there are of course cases when speakers have prepared their utterances carefully and when they in fact express ideas that have been conceived in advance. An academic lecture or a solemn speech are good examples. In fact, the “translation theory” (IX.1) of discourse production is part and parcel of classical rhetoric, which analyzes the whole process into the discovery and arrangement of ideas (“invention, disposition), the discovery of appropriate expressions (“elocution”), and then the memorization and actual delivery of the speech (e.g. Beaugrande & Dressler 1981:15). But the most obvious application of the theory under consideration concerns the expression of messages in writing. Many final written products are very well prepared, and they may appear to be quite careful and explicit expressions of consciously intended messages. (Yet, not even these meticulously compiled texts “contain” their interpretations (cf. VII.6)). The study of such written products may lead the scholar’s thinking onto wrong paths. He may forget that he is only analyzing the final product of a long series of trials and errors, in which the author only gradually managed to elaborate his points adequately.

A more serious error is committed by the linguist, if he believes that the mere analysis of final written products could provide a theory of how thought and language interact in spoken discourse. It is simply misleading to regard speech only as “the use-of articulate sound-symbols for the expression of thought” (Gardiner 1975:17). This theory was criticized earlier in VI:1, when I discussed the traditional definition of the sentence as “the expression of a complete thought. In face-to-face interactions we do not find so much of the expression of complete thoughts (cf. quotation from Gardiner in VI.1). On the one hand, thoughts are elaborated and accomplished through, not before, the interaction. On the other hand, we must not put emphasis only on the intellectual side of language; in addition to the referential and cognitive aspects, we have the expressive, evocative, social, emotive and practical aspects which often dominate in spoken dialogues (9.2).

Our second point is one which has already been touched upon in IX.1; thoughts and meanings are often assumed to exist prior to, as it were independent of, the linguistic expressions that encode them in communication. In other words, thought and language are considered to be autonomous and mutually independent in a sense; meanings can be discussed in isolation from their corresponding expressions, the same meanings can be expressed in many ways and in entirely different languages, and, conversely, the same linguistic expression can be used with entirely different meanings. Language, in the sense of an expression system, is assumed to be an autonomous, uncontroversial “outer form”; linguistic expressions are seen as containers, empty by themselves but in various communication situations filled with content to be transmitted. Thus, this fits the translation or transportation metaphors in communication theory (9.1).

It is quite obvious that the theory just sketched is extreme and unacceptable also in the eyes of most linguists, who do, after all, a & it a relation of n solidarity” between the form and content of linguistic signs. Nevertheless, it lurks behind many popular descriptions in linguistic textbooks; we have that, e.g., translation theories of communication are legion there. It is fascinating to speculate over the historical origins of these ideas. It seems to me that one possible source may in fact be found in linguists’ traditional work with translations between various languages. The ideal here is of course to keep the content constant over different languages; linguistic expressions would then seem to be only the outer appearance. However, translators and interpreters of course know that, in practice, translations are never semantically identical. Translation is a creative reformulation with clear effects on the content. Yet, for various reasons the assumptions of autonomy are more motivated as far as written language is concerned. If we consider spoken dialogues with its bewildering richness of signals, impulses, stimuli and responses, intentions and expectations that are partly private and partly mutually shared and mutually ascribed to one’s interlocutor, then it seems that the idea of semantically equivalent expressions is very far-fetched indeed. Meanings and interpretations are created through the activities of verbalization and social interaction themselves.

I have argued that the written language bias in linguistics has promoted the idea of an autonomy of language. However, we have seen that this idea occurs in two rather different forms which seem to lead up to a serious contradiction. On the one hand, we have just seen that arguments have been made for the autonomy of linguistic expression with respect to linguistic meaning (and vice versa). On the other hand, we found in Û VIII.10 that the idea of language as an autonomous and primary force in mental life has engendered the hypothesis of linguistic determinism, which in fact assumes the link between expression and meaning to be very stable and rigid indeed. Thus, these ideas seem to be entirely inconsistent. However, attempts have been made, e.g. in generative linguistics, to reconcile the two opposing views. Thus, this theory assumes, on the one hand, that linguistic expressions as surface structures are very indirectly related, to thought and meaning (the autonomy of linguistic expressions), while, on the other hand, deep structures, or semantic representations, are very closely related to thinking (which amounts to a specific variant of linguistic determinism). The thesis of the autonomy and independence of language is therefore taken to an extreme in Chomskyan linguistics (cf. also V:5-6, X).

The third main point of this section is the idea that meanings are mental things in the minds of language users. In VIII.3 I suggested how such a conclusion may be arrived at. That is, the argument there was concerned with images that had to do with expressions (sound images), but similar arguments can be made as regards meaning. The philosophical problems would also be much the same.

The “intentional objects” of thoughts and verbal utterances, i.e. that which we think or talk about, are, after all, usually extralinguistic phenomena in the physical and/or social reality around us. For example, if someone thinks of or talks about Big Ben in London, then it is the “real” Big Ben which is the intentional object, not an image of Big Ben or any other mental object in the speaker’s mind. It is another matter that any given speaker at any given occasion must talk about Big Ben under a certain aspect, which means that the speaker always “contributes” something to the way he thinks or talks about it6). On the face of it, the most puzzling phenomenon about thinking and talking may be the fact that we can use non-existing things as intentional objects. Suppose we think about a unicorn; where is the intentional object of such a cognitive act? Some may be tempted to answer that the unicorn is a mental thing that exists in the subject’s mind. However, in my view this leads to unsolvable problems. The only reasonable answer is that, to the best of our knowledge, no real counterpart of such an intentional object exists. In other words, our question was wrongly put. This is not to deny that the act of thinking is real enough, or that we have a subjective experience of seeing a unicorn in our “inner eye n . The important thing is to distinguish between acts of thinking and speaking (and conditions, schemas, background knowledge etc involved in such acts), and the intentional objects of these activities; the latter may be objectively non-existing just as an image according to optical theory may be only virtual.

9.4. Social and individual aspects of language.

It is characteristic of Wundt and a long tradition in linguistics (IX.t) that the direct associations between thinking and meaning, on the one hand, and linguistic form on the other are cut off. Thoughts are assumed to exist independent of and prior to language; first we think and organize meaning, then we speak and transfer the message! But this is not the only unfortunate separation in traditional linguistic philosophy; another separation, inherent in e.g. Saussure’s distinction between language and parole, is that between the social and individual aspects of language and its use:

“En separant la langue de la parole, on separe du meme coup: 1 ce qui est social de ce qui est individuel; 2Á ce qui est essentiel de ce qui est accessoire et plus ou moins accidentel.” (Saussure 1964:30)
In this theory, “language stands in opposition to utterance in the same way as does that which is social to that which is individual” (Volosinov 1973:60). On the one hand, we have the social system of stable signs, i.e. la langue, on the other hand we have the individual’s use of this language, which is then conceived of as a more less accidental use of ready-made units, i.e. words and phrases with fixed meanings.

Such a theory is entirely inadequate as a background for understanding spoken dialogues. Communication in normal speech situations is an inherently social enterprise, a complex interaction in which the various moves of the agents involved can never be understood from a purely individual perspective. Nor can the social aspects be entirely relegated to the domain of prefabricated social rules (la langue and the like). Speakers and listeners participate in social interaction, and so do writers and readers, although in a much less conspicuous way. In fact, if we consider written communication, Saussure’s attempt to derive the social/individual distinction from the underlying differences between language and the use of language, becomes much more natural. Both writer and reader normally work alone, thus seemingly performing individual (monologic) activities, and in doing so, they apply the rules of a language system that is socially shared and normatively standardized to a greater extent than is the case in speech communication. Thus, Saussure’s abstract objectivism7) (as regards his conception of la langue) and his view of linguistic performance reflect a bias towards written language, in spite of his own ardent defense of the primacy of spoken language (see V.2).

9.5. Semiotics

Semiotics is usually defined as the study of signs. Communication is necessarily based on various forms of signing and signaling, and semiotics is therefore, at least potentially, of vital interest for communication theory.

It is customary and appropriate to distinguish between the European tradition of semiotics founded by Saussure and an American variant going back to Ch. S. Peirce. Saussurean semiotics (or semiology, as his term was) has been deliberately based on structuralist models of language structure. It is therefore no wonder that this kind of approach to communication is strongly ly biased by the written language perspective in linguistics.

What is perhaps most characteristic of Saussurean semiotics is the conception of a sign as an association of two thing-like phenomena, expression and content. The linguistic sign, we recall, was construed as a more or less stable combination of a mental representation of the sound signal (image acoustique) and a “concept”, and the latter is also a mental representation of some kind. This seems to be a generalization at a somewhat more abstract level of how linguistic expression and meaning are sometimes represented in the written medium; the expression is given as a graphic word (or a combination of such words), and the meaning is portrayed, if possible, as a picture of the thing(s) meant.

The philosophical difficulties inherent in the theory of mental representations and images have already been alluded to. But there are other features of Saussurean semiotics which are rather unsatisfactory, especially when applied to communication through face-to-face interaction. The most obvious flaw is the neglect of the communicators, the sender and the listener. Whereas American semiotics does comprise also the pragmatic dimension, Saussurean semiotics does not really tackle the problems of dynamic interaction in communication. It is an approach which entirely focuses on the underlying system of signs, and this is never sufficient if we want to understand communication, i.e. the use of signs-for various purposes in different situations. Accordingly, classical European semiotics tends to describe communication as an exchange of various stable signs (cf. communication as a transportation of fixed messages).

To be sure, there are semiotic approaches which are less tied to extreme structuralism. As we have already noted, Peirce’s pragmatist semiotics seems better equipped to tackle the interplay of signs and symbols in face-to-face communication. Unlike Saussurean semiotics it is not an extension of linguistic modes of analysis, but is based on more behaviorally oriented traditions in American philosophy.

There are also some modern semioticians in Europe who have realized that utterances, texts or human acts in general cannot be ascribed unique and stable meanings. Thus, Julia Kristeva (1969) stresses the manifoldness of interpretations; she prefers to analyze texts in terms of acts and processes, or, more precisely, in terms of meaning-creating activities. The production and assignment of meanings must be seen as praxis; the communicating parties work with the creation of meanings and interpretations in the production of situated, (partially) shared understandings, they do not simply function as passive sources and goals for a transportation of messages.