Pragmatics in Semiotics: Bakhtin/Volosinov

Bakhtin, M. (1981), The Dialogic Imagination. University of Texas Press.

Banfteld, A. (1982), Unspeakable Sentences. Routledge and Kegan Paul.

Brown, P. and Levinson S. (1978) 'Universals in Language Usage: Politeness phenomena'. In E. Goody, ed., Questions and Politeness, pp. 56-310. Cambridge University Press.

Biihler, K. (1934), Sprachtheorie. Jena.

Burge, T. (1978), 'Individualism and the Mental', in P. French et al., eds., Midwest Studies in Philosophy, vol IV, pp 73-121. University of Minnesota Press.

Clark, K. and Holquist, M. (1985), Mikhail Bakhtin. Harvard University Press.

Coates, R. (1981), English Armorial Syntax. Grazer Linguistische Studien, 15.

Davidson, D. (1984), 'Communication and Convention',Synthese, 59, pp. 3-17.

Hurston, Z. N. (1986), Their Eyes Were Watching God. Virago.

Levinson, S. (1979), 'Activity Types and Language'.Linguistics, 17, 5/6, pp. 35699.

Levinson, S. (1983), Pragmatics. Cambridge University Press.

Medevev, P./Bakhtin, M. (1985), The Formal Method in Literary Scholarship. Harvard University Press.

Morris, C. (1938), Foundations of the Theory of Signs. University of Chicago Press.

Morris, C. (1946), Signs, Language and Behavior.

Pascal, R. (1977), The Dual Voice. Manchester University Press.

Pateman, T. (1986), Review of Sperber and Wilson 1986. Poetics Today, vol 7, Nr. 4, pp. 745-754.

Pateman, T. (1987), Language in Mind and Language in Society. Clarendon Press.

Pêcheux, M. (1982), Language, Semantics and Ideology. Macmillan.

Pratt, M.L. (1977), Towards a Speech Act Theory of Literary Discourse. Indiana University Press.

Smith, N. (ed.), (1982), Mutual Knowledge. Academic Press.

Sperber, D. and Wilson, D. (1986), Relevance. Basil Blackwell.

Stich, S. (1983), Beyond Belief. Massachusetts Institute of Technology Press.

Tobler (1887), Zeitschrift fur romanische Philologie. XI, 437.

Todorov, T. (1981), Mikhail Bakhtin: le principe dialogique Seuil.

Volosinov, V. (1926), 'Discourse in Life and Discourse in Poetry: Questions of Sociological Poetics'. In Bakhtin School Papers ed. A. Shukman, pp. 5-30. (Russian Poetics in Translation, 10, 1983).

Volosinov, V. (1973), Marxism and the Philosophy of Language. Seminar Press.

Walker, A. (1983), The Color Purple. Women's Press.

Lightly revised from the article of the same title appearing in Journal of Literary Semantics, vol XVIII number 3 December 1989, pp 203 - 216. Copyright in the original appears to be with Julius Groos Verlag 1989

1. Introduction

In his book on the work of Mikhail Bakhtin, Tzvetan Todorov calls him the 'fondateur modeme' of pragmatics (Todorov 1981, p. 42), and in this paper I explore what Bakhtin's contribution to pragmatics was and is. Since the authorship of the works with which I shall be concerned - notably, Volosinov's Marxism and the Philosophy of Language (Volosinov 1973; Russian original, 1929 and 1930) - is still in some dispute, I shall for convenience refer to them and their authors by the names attached to the existing translations.

The use of the term 'pragmatics' raises certain immediate problems. Whereas Todorov writes of 'pragmatics', Katerina Clark and Michael Holquist in their biography of Bakhtin write of 'sociolinguistics' (Clark and Holquist 1985, e.g. p. 216). Now 'sociolinguistics' in English-language writings has come to be identified with the kind of positivistic correlational studies undertaken in America by William Labov and in Britain by researchers like Peter Trudgill. It is fairly clear that this is not the kind of study which Bakhtin would have been particularly concerned to promote. Bakhtin's concerns with the socially-situated utterance and with the structuring of linguistic form and meaning by context, and his specific views on these subjects, relate much more obviously to the contemporary literature of pragmatics, as surveyed, for example, by Stephen Levinson in his textbook Pragmatics (1983). Pragmatics thus understood includes all the work inspired by philosophers of language such as J. L. Austin, H. P. Grice and John Searle, and undertaken by linguists and ethnomethodologists such as Gerald Gazdar, Ruth Kempson, Levinson himself, Harvey Sacks, Emanuel Schegloff, Dan Sperber and Deirdre Wilson. Notably, there is nothing in pragmatics which excludes extension of its theories to studies of the written word. Indeed, there have been productive studies of the pragmatics of literature by such people as Richard Ohmann and, notably, Mary Louise Pratt in her book, Towards a Speech Act Theory of Literary Discourse (1977) - studies which merge into work going under such names as text comprehension and reception aesthetics. All such work can be subsumed under the general heading of 'pragmatics' when it is given the scope it has in C. W. Morris's 1938 definition of pragmatics as encompassing the study of the relations between signs and their users. This is because 'users' includes hearers and readers as well as speakers and writers. (Morris's 1946 work in behaviourist semiotics, Signs, Language and Behaviorcontains what to my knowledge is the first English-language reference to the Russian original of Volosinov's Marxism and the Philosophy of Language. I imagine that it could have been Roman Jakobson who supplied Morris with the reference.)

Pragmatics has developed quite independently of any Marxist concerns, and this is true also of Volosinov's misleadingly titled book. The occasional references to 'class' and 'dialectics' don't disguise or alter the fact that Volosinov's book assumes and defends a general sociology, indifferent to the specific tendencies of Marxism. When it was published a year before Marxism and the Philosophy of Language, Medevev/Bakhtin'sThe Formal Method in Literary Scholarship was subtitled, 'A Critical Introduction to Sociological Poetics'.Marxism and the Philosophy of Language might have been more aptly titled, A Critical Introduction to Sociological Linguistics for that - in the absence of a use for 'pragmatics' - is more or less what it is. Just as Durkheim is said to have sociologised French neo-Kantianism, one might well say that Bakhtin's achievement in the 1920's was to sociologize the German neo-Kantianism in which he was steeped.

I turn now to a series of illustrations of some of Volosinov's leading ideas.

2. 'Scissors!'

I'll begin with some analysis of a socially situated utterance of the linguistic form, 'Scissors!', uttered by some particular surgeon during some particular operation with a view to getting some particular assistant to pass a particular pair of scissors.

According to Volosinov:

An interlocking organic unity joins the form of communication (for example, on-the-job communication of the strictly technical kind), the form of the utterance (the concise, businesslike statement) and its theme [what it is about - here the thought of some definite person passing a particular pair of scissors - what Sperber and Wilson 1986 call the 'thought']. Therefore, classification of the forms of utterance must rely upon classification of the forms of verbal communication. The latter are entirely determined by production relations and the sociopolitical order (1973, pp 20-1).

If anything is going to exemplify this argument, 'Scissors!' contextualised as I have outlined, is surely going to do so.

However, there are problems. Working backwards through the quotation, we can dispose of the last sentence rapidly enough. The forms of verbal communication found in operating theatres are not so much determined by production relations or the sociopolitical order as they are part of them. To speak of determination we would have to be able to imagine wordless operations. This may be conceivable (though not in terms of Volosinov's arguments, which insist on the omnipresence of the word in all conscious acts), but it is not conceivable when we consider other situations which the general claim purports to cover. There is no such thing as the wordless lecture, which somehowdetermines the form of lecturing.

The second sentence of the quotation, despite the flourish of a 'Therefore' is obviously a non sequitur. Why classification of the form of utterance, 'Scissors!', must rely on the classification of the form of verbal communication, performing a surgical operation, has not been made clear. But perhaps it can be. 'Scissors!' is uttered as a way of getting someone to pass a pair of scissors. The precise classification of what speech act theorists call the illocutionary force of the utterance, as arequest or order, for example, does depend on the participants' classification of the situation in which it is uttered. What makes 'Scissors!' an order, on this occasion of its utterance, is the fact that, for good reasons, if the assistant heard it, understood it, had no good grounds to dispute its legitimacy, but still declined to pass the scissors, that would be grounds for disciplinary action. In another situation, 'Scissors!' would not have the force of an order: my two year old might say 'Scissors!', but I can decline or prevaricate. She can request me to pass a pair of scissors, but not order me. So it emerges that one link between form of communication and form of utterance is that reference to the former is necessary to making a classification of the illocutionary force of the latter. What Volosinov calls 'form of communication' actually reappears fifty years later in contemporary pragmatics as Stephen Levinson's concept of activity type, which he defines as:

a fuzzy category whose focal members are goal-defined, socially constituted, bounded events with constraints on participants, setting and so on, but above all on the kinds of allowable contributions. Paradigm examples would be teaching, a job interview, a jural investigation, a football game, a task in a workshop, a dinner party, and so on. (Levinson 1979)

But this link between form of communication and form of utterance is not the only one. Volosinov rightly thinks that reference to form of communication/activity type will be involved in the explanation of why, to continue with my example, the surgeon used the form 'Scissors!' rather than any other linguistic form by means of which she could have communicated the same theme or thought - that is, successfully referred to a pair of scissors and expressed her desire that a particular person pass them to her. Why didn't the surgeon say 'Scissors, please!' or 'Would you kindly pass the scissors, Mr Smith?', and so on? This question can be answered by referring to the form of communication/activity type and claiming that power relations and social distance between addresser and addressee, together with the presumed importance of the action commanded, all work - in this instance - in the same direction. They tend to exclude the use of indirect or polite forms and work in favour of a minimally elaborated and maximally efficient form. How these various causal agents work, singly or jointly, to select the form of utterance can only be determined by considering a range of different cases. If surgeons say 'Scissors!' in operations even when the addressee is another equal status surgeon rather than a nurse, we might - for example - conclude that the importance of the action commanded is what accounts for the fact that polite forms are - if they are - dispensed with in operating theatres.

I should add here that power, social distance and importance are the variables identified by Penelope Brown and Stephen Levinson as explanatory of choice between different utterance types. Brown and Levinson's 1978 paper on politeness phenomena is a major contribution in contemporary pragmatics, and one which ought also to be attended to by anyone concerned to develop a specifically Bakhtinian pragmatics.

Utterances other than 'Scissors!' could have accomplished the passing of the scissors ('Scissors, please' would presumably get the scissors passed, even if the recipient was puzzled by the 'please'. Pointing to the scissors might also succeed). So also could 'Scissors!' be used outside of an operating theatre to get a pair of scissors passed. In consequence, one very strong claim apparently made in Marxism and the Philosophy of Language (p. 20) cannot be sustained. This is the claim that the relationship between form of communication and form of utterance is one to one. That is to say, it is a relationship which defines a set of forms of utterance with one or more members and which would be exclusive to a form of communication and obligatory for it - so that 'Scissors!' could only occur in the context of an operation and in that context would be the obligatory way of getting scissors passed.

If such relations between form and context did hold, the job of the critical linguist, for one, would be greatly simplified. For we would be able to read off form of communication from form of utterance (if you like, context from text) and vice versa. It is occasionally the case that we can do this, but it is, in my view, not a reflection of a necessary constraint on language use. I can, however, produce one entertainingly obscure example of a stricter one-to-one co-relation between context and text. The example is taken from a paper by Richard Coates (1981).

In the heraldic activity of blazoning the coats of arms of jousting competitors, the herald performs one kind of speech act only, blazoning, and does so using utterances constructed largely according to a single syntactic pattern to describe coats of arms yielding uttterances such as:

Gules, on a bend or between two boars heads, a cross sable

From the first-order use of such a sentence one can infer heraldic blazoning with much more reliability than one can infer a surgical operation from the occurrence of a 'Scissors!'. Furthermore, from heraldic blazoning (the activity type) you can infer the utterance of such a sentence as the one above, given an appropriate state of affairs in the world, namely, the presence on the jousting field of a coat of arms satisfying the truth-conditions of the sentence. One only has to exclude external contingencies, such as the herald losing his voice or misreading the coat of arms.

But such a case is relatively exceptional. In general, I think that the correct position on relation between context and text or, equally, function and form, is that articulated by Donald Davidson when he writes that:

it is not an accidental feature of language that the ulterior purpose [aim, goal] of an utterance and its literal meaning are independent, in the sense that the latter cannot be derived from the former [and vice versa]: it is of the essence of language. I call this feature of language the principle of the autonomy of meaning (1984, p. 12)

This position is, I think, in many ways irresistible but in Volosinov you will find attempts to resist it, attempts which basically seek to dissolve meaning into context. Yet in the end, I think Volosinov (and Bakhtin) hold back from an ultra-Wittgensteinian contextualism ('meaning is use') in favour of something closer to an orthodox distinction between semantic and pragmatic aspects of meaning, distinguishing what a sentence means from what a sentence-in-an-utterance means and does.

3. 'Well!'

I have used an example of my own in discussing Volosinov's distinctions between form of communication, form of utterance, and theme, and in order to relate those distinctions to work in contemporary pragmatics. Now I'll take a passage from Volosinov's 1926 paper, 'Discourse in Life and Discourse in Poetry' to introduce some specifically Bakhtinian themes. Here is the passage:

A couple are sitting in a room. They are silent. One says, 'Well!'. The other says nothing in reply. For us who were not present in the room at the time of the exchange, this 'conversation' is completely inexplicable. Taken in isolation the utterance 'well', is void and quite meaningless. Nevertheless the couple's peculiar exchange, consisting of only one word, though one to be sure which is expressively inflected, is full of meaning and significance and quite complete.

In order to discover the sense and significance of this exchange, we must analyse it. But what, strictly speaking, can we subject to analysis here? However much we fiddle with the purely verbal part of the utterance, however finely we define the phonetic, morphological and semantic features of the word 'well', we will not be a step closer to understanding the integral sense of the exchange.

Let us suppose that we also know the intonation with which our word was articulated, and that it was indignantly reproachful, but softened with a touch of humour. Though this somewhat fills out for us the semantic void of the adverb 'well', it does not reveal the meaning of the whole.

So what are we missing? That 'non-verbal context' in which the word 'well' sounded intelligibly for the listener. This non-verbal context of the utterance is formed out of three factors: 1) a spatial purview common to the speakers (the unity of what is, the room, the window and so on), 2) the couple's commonknowledge and understanding of the circumstances, and finally 3) their commonevaluation of these circumstances.

At the moment of the exchange bothindividuals glanced at the window and saw that it was snowing. Both knew that it was already May and long since time for spring, and finally, that they both were sick of the protracted winter. Both were waiting for spring and wereannoyed by the late snowfall. The utterance depends directly on all this - on what was 'visible to both' (the snowflakes beyond the window), what was 'known to both'(the date was May) and what was 'similarly evaluated' (boredom with winter, longing for spring); and all this was grasped in the actual meaning of the utterance, all this soaked into it yet remained verbally unmarked, unuttered. The snowflakes stay beyond the window, the date on a page of the calendar, the evaluation in the mind of the speaker, but all this is implied in the word 'well'. (1926, pp. 10-11)

This is a remarkable passage. I do not think there was any serious development of similar ideas in any country for 30 or 40 years - not until H. P. Grice began to develop the line of thinking (in part originating in a critique of behaviourist semiotics) which culminated in his 1967 William James lectures at Harvard - lectures from which much of contemporary linguistic pragmatics stems directly. It is with the benefit of this kind of hindsight that I would now like to look more closely at the passage quoted above. Volosinov clearly realises that what we might call the total occasion meaning of this utterance of 'Well!' is, in some sense, the joint product of its context and its own form as an exclamatory utterance of 'Well!' with a particular intonation. (It would be a Stanislavskian exercise to realise 'Well!' in an indefinitely large number of different ways.) And Volosinov sets out to classify context in such a way that from the conjunction of context and form we can see how to derive (and, by extension, theorise how the actual addressee derived) the total occasion meaning of the utterance, as made up of what is said and what is contextually implied. The key idea is that context is to be defined in terms of assumptions which the speaker makes about what the hearer knows or believes, including assumptions about the hearer's assumptions about the speaker's knowledge and beliefs. So defined, the speaker's utterance is shaped by and responds to context - in other words, to thehearer who, in this sense, is present (Bakhtin would later say, dialogically present) in the utterance. If all utterances are context bound in this specific kind of way, then all speech (and all written text) is dialogic in the sense that it responds if not to an actual hearer then to assumptions about an actual or potential hearer, addressee or audience.

In current linguistic pragmatics Volosinov's approach in his 1926 paper has been independently developed in theories which place mutual knowledge and mutual belief at the centre of a theory of speech production and understanding. The volume edited by Neil Smith, Mutual Knowedge (1982) contains several examples of the approach, for example that of Herbert Clark. And in literary theory, what goes under the name of 'reception aesthetics' is, effectively, a range of accounts of the presence of the reader in the text.

There are alternative, or would-be alternative approaches, I should add. Notably, in recent work brought together in their book Relevance (1986), Dan Sperber and Deirdre Wilson have sought to offer an approach to understanding which places much greater emphasis on our inferential abilities and much less on the idea of recourse to a stock of assumptions about the speaker's knowledge and beliefs about us and about our assumptions about such assumptions. I cannot pursue this here; I have discussed Sperber and Wilson's book at some length elsewhere (Pateman 1986). What I want to do now is to pick up again Volosinov's distinctive idea that utterances are shaped by an implied audience and their anticipated response, And I want to consider how this idea developed into the more radical realisation that hearers or audiences - in general, others - actually get inside the form of utterances. What I take to be my words are, often enough, the words of another. Volosinov chooses to exemplify the general idea through a study of a phenomenon in literature, the phenomenon of free indirect discourse, and to this I turn.

4. 'John said he would come tomorrow'

Suppose John says:

(1) 'I'll come tomorrow'.

This can traditionally be rendered in writing as either reported speech (2) or indirect speech (3):

(2) John said, 'I'll come tomorrow'.

(3) John said that he would come on the following day.

But (1) can also be rendered, in undated but posterior writing as (4), though (4) can only be spoken with difficulty to mean what it means in undated writing:

(4) John said he would come tomorrow.

What happens in (4) is that a term the reference of which is relative to the time of utterance, that is, the temporal deictic 'tomorrow', is retained in representing at a different, future time what John said, though the tense of the verb changes to indicate the anteriority of John's speaking. In this way, a word - 'tomorrow' - which could only be John's own has found a place in the writer's words but without quotation marks, though in one sense it must be and is recognisable as a quotation, for otherwise 'tomorrow' would have the wrong or no assignable temporal reference. To use a term of Karl Biihler's (from his Sprachtheorie (1934)), the I -Origo of 'tomorrow' is John.

In (4) we have a simple example of what is generally called free indirect discourse, though it rejoices in a variety of other names in English, French and German. In Volosinov's Marxism and the Philosophy of Languageit is called quasi-direct discourse, and the whole of the third section is given over to a study of the phenomenon.

The phenomenon is pervasive in literature and has been an object of attention for a hundred years since it was first identified in an article by Tobler (1887). There are recent excellent book-length treatments in The Dual Voice by Roy Pascal (1977) and in Unspeakable Sentences by Ann Banfield (1982). I'll give an example of free indirect discourse (FID) from a novel I have recently read, an example which will, I hope, indicate some of the wider significance of the phenomenon.

In Zora Neale Hurston's Their Eyes Were Watching God(1986: originally 1937), we find the narrator reporting some thoughts of the male character, Tea Cake, in the following words:

Tea Cake didn't say anything against it and Janie herself hurried off. This sickness to her was worse than the storm. As soon as she was well out of sight, Tea Cake got up and dumped the water bucket and washed it clean. Then he struggled to the irrigation pump and filled it again. He was not accusing Janie of malice and design. He was accusing her of carelessness. She ought to realise that water buckets needed washing like everything else. He'd tell her about it good and proper when she got back. What was she thinking about nohow? He found himself very angry about it. (pp. 259-60)

In this passage, reporting the inner thoughts of a character, the character's inner words are taken up into the narrator's, though not completely. Tea Cake is a speaker of a Black English Vernacular (BEV); the narrator is not. Yet BEV 'nohow' appears where a Standard American English 'anyhow' would be the narrator's word. And, less obviously perhaps, 'good and proper' is the character's rather than the narrator's expression ('good and proper' would not be allowed in a school exercise indirect speech rendering of Tea Cake's thoughts). Elsewhere, the narrator translates the character's thoughts, as in 'She ought to realize' where BEV would have 'She oughta realize'. (In another passage, in contrast, I noted the narrator retaining BEV 'hisself')

What is at issue here is the relation between narrator and character or, differently put, between representing and represented speech. In some literary styles, a rigid separation between the languages of these two levels is insisted upon and Volosinov and Bakhtin describe such styles as being dogmatic and authoritarian. (In Bakhtin's 'Discourse in the Novel' and the other essays in The Dialogic Imagination (1981) the concern with local features of text is transformed into large-scale historical typologies). In other styles, the narrator disappears as a separate voice. Alice Walker achieves this in The Color Purple (1983) by using the device of the epistolary novel. She thereby avoids the problem which Zora Neale Hurston has of having as represented speech a language which, if it is not used for representing, will appear somehow inadequate and the object of patronage. By mingling represented and representing speech in FID, Hurston goes some way towards avoiding this, but by no means as far as other authors have gone.

Volosinov and Bakhtin add to the comparative philologist's account of FID the significant claim that when represented speech is taken up into representing speech, it becomes reaccented by the voice which speaks it. The incorporation of the speech of the other necessarily transforms it by giving it an intonation or an evaluation which it did not previously possess. So, for example, in the passage quoted from Their Eyes Were Watching God, Hurston's use of FID both brings the reader closer to the plight of Tea Cake, mortally ill with rabies, and implicitly places Tea Cake's words in a context of what we might call tragic irony. (I note in passing that for Sperber and Wilson irony is necessarilyechoic) What Tea Cake doesn't know, but the narrator and we do, is that Tea Cake soon won't be able to tell anyone anything 'good and proper'. The use of FID emphasises the unreal, delirious character of his thinking.

In Bakhtin's work the idea of FID is the springboard to the more encompassing idea of heteroglossia. I don't propose to discuss that key concept here. I want to change tack, quite radically but, I hope, interestingly.

5. 'Doctor, I have arthritis in my thigh'

In this section, I want to relate, briefly, Bakhtin's sociological semantics, which from several angles sees meaning as a social rather than an individual phenomenon, to a similar line of argument in contemporary philosophy of language, represented most forcefully perhaps by Tyler Burge in an essay, 'Individualism and the Mental' (Burge 1978).

Burge invites us to contemplate a man who has and believes he has arthritis in the ankle, wrist, etc. waking up one day with a pain in his thigh and going off to the doctor to report that he believes his arthritis has spread to his thigh. No, it hasn't, says the doctor, arthritis is a disease of the joints and your thigh isn't a joint. Compare this, says Burge, with a community where everything is the same except that doctors etc. believe that you can have arthritis of the thigh - in other words, a community which has a different concept of arthritis from our one. Burge's central idea is that although the second community has a different concept of arthritis from us, which we might call tharthritis, the individual in the first community does not have a different concept from that of his own community. If he did, his belief that he has arthritis in his thigh would not be mistaken, but simply show him to be operating with a different concept of arthritis than that of his community. It would actually be wrong, says Burge, to reconstruct the content of the man's thought in such a way that he comes out believing something other than that he has arthritis in his thigh.

If Burge is right, then it follows that the meaning of 'arthritis' is not in this man's head, since what is in his head is inconsistent with the meaning of 'arthritis', though he doesn't realise it. Further, the content of what is in his head is partly constituted by something outside his head in the community. We say that he believes he has arthritis in his thigh (rather than saying he believes he has tharthritis) and we do this in virtue of facts about the community to which he belongs. Equally, by parity of reasoning, the meaning of 'arthritis' is not in any other individual's head; the meaning of 'arthritis' exists between members of the community. And this, Burge says, is a perfectly general claim about the nature of meaning - and one which, I would add, is entirely consonant with Bakhtin's position.

Now there are objections to Burge's position - those developed in Stich 1983, for instance (and see discussion in Pateman 1987, ch. 5). The point I would like to make here is that Burge's and Bakhtin's positions make the use of words, in effect, citational - that is, radically heteroglossic. When the man says, 'Doctor, I have arthritis in my thigh', he does not really use the word 'arthritis'; he mentions it. What he is saying might be glossed as something like this: 'I have a pain in my thigh, which I think you would call "arthritis" and since I want to call it what you would call it, I am calling it "arthritis".' The doctor's response is then to be understood as saying that the man is not in fact calling, his pain by the name the doctor would give it. From this citational or mentioning view of language use it is a short step to seeing the contribution of the individual to their utterance as residing in the manner or tone in which they utter the words they use - L'homme, c'est le style meme Language use takes the form of a comment or gloss or in some other way, a relation to what is cited - what Pêcheux would call the pre-constructed (Pêcheux 1982). But how the individual is able to establish a relation to mentioned words is as unclear in Bakhtin as in Pêcheux; their individuals appear to enjoy neither rationality or, what amounts to the same thing, creativity. Hence, they lack all autonomy.

A further problem concerns how far the citational view of language can be extended: Burge simply says that 'much of our vocabulary is taken over from others' (1978, p. 80). For example, can the citational view be extended to a word like 'pain'? If the man says, 'Doctor, I have a pain in my thigh', does he mean 'I have something in my thigh which I think you would call a pain and I want to call it what you would call it'; thus allowing, in principle, for the possibility that the doctor might reply, 'Oh no, what you have in your thigh is pins and needles'. The example, of course, immediately recalls postWittgensteinian discussion of pain language and pain behaviour. On the one hand, the incorrigibility of claims to be in pain, low-level linguistic mistake excepted, suggests that the claim to be in pain cannot involve a citational use of 'pain'. On the other hand, Wittgenstein's private language argument might be used to mobilise an opposing line of thought, that 'pain' can be just as much citational as 'arthritis'. Unfortunately, I must leave such fascinating questions for discussion on other occasions.