Round table on functional linguistics,
1st April 1993, University of Vienna


Jan Firbas, Univ. of Brno
Henry G. Widdowson, Univ. of London & Univ. of Essex
Robert A. de Beaugrande, Univ. of Vienna

Introduction by the host, H. Schendl (Univ. of Vienna)

Ich möchte Sie sehr herzlich begrüßen, wir freuen uns sehr, daß trotz der sehr kurzen Verständigungszeit so viele von Ihnen gekommen sind. Ich werde sicher nicht lange reden, obwohl über unsere drei speakers sehr viel gesagt werden könnte. Ich möchte nur ein paar einführende Worte zu sagen:

Ich glaube, daß es bei diesem Thema wirklich überflüssig ist, Herrn Professor Firbas vorzustellen. Jeder, der von der Prager Schule oder vom Begriff der Funktionalen Satzperspektive je gehört hat, kann gar nicht umhin, immer wieder auf den Namen Jan Firbas zu stoßen; zuletzt 1992 als sein Buch Functional sentence perspective in written and spoken communication bei Cambridge University Press erschienen ist.

Kollege Widdowson, der vergangenes Semester hier in Wien lehrte und somit allen persönlich bekannt sein dürfte, hat gerade in den späten siebziger Jahren den funktionalen Ansatz ganz entscheidend geprägt.

Kollege Beaugrande arbeitet derzeit an einer kognitiv fundierten Functional Grammar of English.

Die gleichzeitige Anwesenheit dieser drei Kollegen in Wien ist wohl eine einmalige Gelegenheit für einen Round Table on Functional Linguistics. Dabei möchte ich nicht verschweigen, daß die Idee von Doktor Seidlhofer stammt, der ich für ihre Initiative danken möchte.

Well, the participants have decided on the following procedure: the Round Table will be in English, each of the speakers will start by giving a statement of about ten minutes first, starting with Professor Firbas, who will include some brief historical information on the development of Functionalism. Then Professor Beaugrande will, I think, enlarge on some of these aspects, followed by Professor Widdowson's presentation. Finally Professor Beaugrande will present his latest research in the field.

Thank you all very much for coming. May I ask Professor Firbas to start.

Prof. J. Firbas

I should like to say a few words about my approach to language. I approach a language, or languages, or language in general, both as a learner and as a scholar.

First a word on the learner. As a Czech, I speak a language that is spoken only by a comparatively small community. Speakers of Czech are heavily outnumbered by those who do not speak the language. This is certainly not disproved by the possibility that when an Austrian football team plays a Czech football team, roughly half the Austrians may have Czech names and roughly half the Czechs German names. Czech is simply not a world language. If a Czech wishes to communicate with a non-Czech, (s)he has to learn a foreign language or foreign languages. In this way, I have become a learner and of course realize that there is no end to learning. As a learner, I am frequently faced with the problem of finding linguistic means that would adequately satisfy my communicative needs; in other words, I am frequently faced with the problem of finding linguistic means that in the act of communication would adequately serve my communicative purpose. In order to be able to apply the linguistic means in an adequate way, I must know how they operate in the act of communication. This is certainly a problem worth investigating. This brings me to my endeavours as a scholar.

It is a commonplace nowadays that in studying a language, we cannot sever it from its function or functions. Nevertheless, it is a commonplace that must be taken very seriously. I should like to make two points in this connection. First, let me turn my attention to the very act of communication, or rather the moment at which a sentence is uttered and/or perceived. As early as 1884, Henri Weil wrote the following words: „Car dans la parole, ce qu'il y a de plus essentiel, c'est le moment de la conception et de l'énonciation; c'est dans ce moment que se trouve toute la vie de la parole, avant ce moment elle n'existait pas; après, elle est morte" (Weil 1884: 27). The English translation runs: „For in speech - or in the spoken language - the most important thing is the instant of conception and utterance. Into this instance is compressed all the life of speech: before it, speech had no existence: after it speech is dead" (Weil 1887: 30) . In a way, this dictum may sound a little exaggerated, but I wholeheartedly agree that the moment of utterance and/or perception of a sentence is a phenomenon of paramount importance. It is at the moment of utterance and/or perception that the sentence serves a particular communicative purpose and hence functions in a definite perspective. It is not without interest to note that Mathesius, who knew Weil's work, coined the felicitous term „aktuální c*lene*ní ve*tné" („aktual' noe chlenenie predlozheniya", „aktuelle Satzgliederung", „la division actuel le de la phrase"). As English „actual" is not an exact equivalent of Czech „aktuální", another term had to be found for English. I accepted Professor Josef Vachek's suggestion and started using the term „functional sentence perspective" (FSP; Firbas 1957). The term is based on Mathesius' term „Satzperspektive". Vachek's suggestion has added the qualification „functional". This is the way the term „functional sentence perspective" (FSP) has found its way into the literature. For years I have been endeavouring to study the factors and the signals which operate at the moment of utterance and/or perception and which orient, in other words, perspective, the sentence towards the element that conveys the high point of the message, i.e. the rheme proper of the sentence.

Weil's dictum brings me to the second point I should like to make. As I see it, I find Weil's dictum applicable both to the written and to the spoken language. After all, Weil made his statement while comparing the word orders of ancient Latin and Greek with those of modern French and German. Accepting the dictum, I do not interpret the term „speech" in the sense of de Saussure's „parole". After all, Weil's book on word order was published about half a century before the appearance of de Saussure's Course. The factors determining the functional perspective of a sentence, as well as the signals yielded by these factors, operate in an interplay. A special place within this interplay is occupied by the contextual factor, which through the immediately relevant context, verbal and situational, plays the dominating role in effecting the embedding of the sentence into the flow of communication. Now, participating in fulfilling the communicative purpose, a linguistic element of any rank performs a specific function. This function cannot be determined if the element is examined in isolation and if its relations to the concurring elements are disregarded. This points to the systemic character of languge. I agree with Vachek and others that language is a system. In fact, I go the length of agreeing with Vachek that language is a system of systems. However, I also agree with him (and Danes*) that language is not a closed system, but a system that has its centre and its periphery.

It follows that the approach I subscribe to is functional and systemic, systemic in the sense of Prague linguistic structuralism.

Prof. R. de Beaugrande

I came into this area when I was facing a practical problem. I was working on a theory of translation and I didn't find American linguistics especially helpful. I began collecting Czechoslovakian research papers, and I noticed that Western linguists were very curious about FSP but knew little about it. Even though a good deal of the studies were written by scholars of English and Anglo-Saxon in very good English, they were published in sources that were extremely difficult for Westerners to get to for political reasons in those days, mostly just was Slavic Institutes or the occasional Ost-Europa Haus.

I did find the work extremely useful, when I finally got to Czechoslovakia in 1987, the researchers presented me with much of their work, back issues, some dating as far as the 1920s, say of Travaux Linguistiques de Prague, Brno Studies in English, and Sbornik praci filosoficke fakulty brnenske university. So I sat down to read them all, and I wrote a paper for the Academy of Sciences on 'The heritage of Functional Sentence Perspective' in Linguistica Pragiensa, (vol. 34/1-2, 1992, pp. 2-26 and 55-86), in which I attempted to show in retrospect how much we owe to this work.

In 1926, Mathesius said that we have the opportunity to approach language by starting with the form and going to the function, or start with the function and go to the form. In the West, given the way science was done in those days and is often still done today, the main idea has been to start with the form and maybe go to the function later. We have here two academic ideologies:

(Fig. 1)

In the one we could - unpolemically and descriptively - call 'formalism', the basic fact of language is form and function is one corner of that, e.g. when Chomsky put forth his formal grammar and leave out functions by claiming that 'the rules of stylistic reordering are very different from grammatical transformations', the latter being 'much more deeply embedded in the grammatical system' and using 'markers drawn from a fixed, universal language-independent set', while 'stylistic' ones are 'peripheral', apply to 'performance', and 'have no apparent bearing' 'on the theory of grammatical structure' (Aspects, pp. 127, 222f).

In the ideology we could call 'functionalism', the basic fact of language is function, which ought to be obvious in a commonsensical way, and there are only some formal aspects in the phenomena.

As Ales Svoboda pointed out to me, a large number of the FSP people were Anglo-Saxon scholars; when they looked at Anglo-Saxon word order, it had for Czechs a rather familiar feel to it, much more than modern English. And to study word order in Anglo-Saxon, they enquired how Anglo-Saxon speakers decide what sort of patterns they want, and they found a principle similar to Czech, in which impossible to formulate a sentence without taking functional sentence perspective into consideration. In Modern English, which, due to by the introduction of widespread literacy, particularly in the 18th century, was 'frozen' - this is my own hypothesis - in the transition from being an inflected language, as was Middle English still, and of course Anglo-Saxon, over to being a genuinely isolating language in the sense that Mandarin Chinese is. It got frozen in the middle and ended up with formal substrate, (what's called in physics a 'frozen island' in an otherwise fluid system).

These 'frozen islands' of syntactic patterns offered the Western formalists the main encouragement for their formalist approach. But the longer you look at the older forms of English the more you're convinced that English is a highly functional language; it would not have occurred to an Anglo-Saxon to write a 'formalist grammar'. One paper in particular went to the heart of the matter; when I went back and read it again, I realized it said a lot more than I had thought, namely Danes 'Three-Level Approach to Syntax' (Travaux Linguistiques de Prague 1, 1964, 225-40). To bring home the point, I drew a contrasting chart here.

(Fig. 2)

intonation - prosody
(a) Formalist levels
(b) Functionalist levels

In the dominant Western scheme (you'll find this in Bloomfield's Language), phonemes are the smallest units of sound, and morphemes are combinations of phonemes, the lexemes are the words, combinations of morphemes, and then you have the phrases which were called „syntagmemes". In the formalist scheme, the layers are related in terms of size and constituency, each one is a piece of the other. In the functionalist scheme, the front-end of language is not phonemes but intonation, recalling here that most of the important early functional work was done by people like Bolinger, Danes, Firth, Halliday, and the Sinclair group, who were scholars of intonation or prosody: they had to adopt a functionalist approach. The grammar does the main work of organizing the language, so, it includes morphology, syntax, semantics, and many functional issue that would go under stylistics. The discourse, rather than being the largest unit, is the entire system, where the contributions of the subsystems are all put together. In contrast to the formalist system, the relation is not in size; intonation is a factor of the whole text, grammar is a factor of the whole text, so you're not getting bigger pieces as you go on, but to at aspects that are functionally integrated. Danes pointed out in his paper on 'The three levels' that the one level is a means for the other level, which seem to me to lie the heart of the functionalist, the Prague approach Danes' system (see Figure) covers the 'organization of the utterance', the 'semantic structure of the utterance', and the 'grammatical structure of the utterance', and what he meant by 'grammatical structure' turns out to be this rich kind of 'functional grammar' that is coming into style today.

(Figs. 3 & 4)

organization of utterance
individual utterance event
semantic structure of sentence
minimal communicative unit of text
grammatical structure of sentence
abstract structure within the grammatical system
Fig. 3: Danes' 3 levels
Fig. 4: Danes' levels of organization

If you look at the sentence from a functional viewpoint, as an 'individual utterance event', where pragmatics, communicative semantics, structures syntax they're all put together, and it's their interaction that's interesting. In a formalist scheme you have simple components with complex interactions, e.g., once you've got the whole phonology set up, you ask how it might interact with morphology. But that picture makes the interactions impossibly complex. Say, you have a whole syntax set up and then you put the semantics on top of it, and if you've tried to write all the rules you know how difficult that is. A functional approach has the components be more complex by introducing functions, and has the interactions then be more direct and simple.

We are at a watershed now. The reason why modern linguistics has become functional is because we have discovered we don't have the constraints to write a description of any language based on form alone. There is no formal grammar of any natural language because there cannot be; all you can attempt to do is rewrite functional information in a formalist way. In the late seventies people would ask me: 'if this is FSP, how do we formalize it?' How do you write the transformational rules to capture all this, but without speakers, without contexts, in a sense, to 'de-functionalize' it? Since the 1980s, ironically, some linguists, particularly in America, are now applying the name 'functional' to formalist models to sell them better, to get on the new wave, giving us half-hearted, nebulous formalizations of the functional approach. And we must take care to distinguish these from the central functionalist approach as it came to us from the Prague school.

Prof. H.G. Widdowson

I think my perspective on sentences will be slightly different, functionally, from that of my colleagues. What I would like to do is to distinguish a number of different senses in which one can talk about functions, because I suppose I'm thinking of this particularly from the point of view of language pedagogy, and there's a good deal of confusion as to what a function is. And I think what we've been hearing about is one way in which one might define the idea of function. I'd like to distinguish three ways, and I think each of them has its legitimacy, and it's really a matter of recognizing the nature of descriptions in relation to these three kinds of functions, and what these descriptions can legitimately claim and what they cannot claim. so I think it's really not so much that one approach to the description of language replaces another, but that we need to be clear as to what the legitimacy of the claims are that are made. All descriptions of language tend to claim to be comprehensive and this in a sense is one of the difficulties.

One sense of function, let me call it the systemic, is what I take it Jan Firbas was talking about. The Prague School and his own work brings out a relationship, a combination indeed, of the functional and the systemic, and I think that sense of function is indeed a systemic sense. And that sense is still alive and well and living in descriptive linguistics. By it I mean that one is talking about the code-internal relations which linguistic forms as items contract with each other. So here we are talking about system, or systems, as Jan said, systems within which one finds terms, which by virtue of the fact that they are in systems, function in contrast with each other. And that is, if you like, a systemic function. So one could look at systemic relations in grammar. Indeed in Halliday's original conceptions grammar was a systemic grammar in precisely that sense. And one can talk about sense relations in lexis, and here we are talking about the internal functioning of forms in relation to each other. One of the disadvantages, and this I think is what Robert is pointing out, is that this kind of description is closed off so to speak from the outside world. You can talk about these formal relations as abstract objects, or abstract items, and apart from the fact that you make a neat model of them, there is no motivation for them in respect to what the language is supposed to do in the world outside.

So then we get a second, I think really quite distinctive notion of function, and this let me call for the nonce, a semiotic function. And this is where Halliday, for example, moves from a systemic grammar to a functional grammar. The name changes because the notion of function alters. Here we are talking not so much about linguistic forms as items, but about linguistic forms as signs, signs of something. Of what? Signs of something in the world OUTSIDE language, and the concern here I think is how the abstract systems that might be described, are in some way motivated by the social needs for language. We are talking here I think, about the way in which language formalizes or formulates social semiotic meaning. Halliday, famously, starts his linguistic descriptions almost always by asking the question „Why is language as it is?". And his answer is that language is as it is, is formed as it is, its systems are as they are, because they are developed to meet the communicative requirements of language. Hence, we have these three types of system in a Hallidayan grammar, in a functional grammar in this sense, each system reflecting a basic communicative or semiotic function. So you have the ideational function, and there you have systems which he calls transitivity systems of the grammar. We have the interpersonal function. If you're going to have a language, you are going to use it to talk to other people, interpersonally, so then you have systems which reflect that function, and these systems are the mood systems as he calls them. And then you have the textual function, and that is particularly of course what the Prague School has been most concerned with, which is as he described it, the function of language to make links with itself, and there are systems for that also and these are the theme systems. One of the advantages of looking at language in this way is that you provide a semiotic function outside language itself a motivation for the formal systems which you then define. The description is motivated by the need for an explanation. Instead of saying, language, that's how it is, these forms which you will study in isolation and separated from what purpose they might have, you provide some sort of an explanation for these systems in reference to the functions that they've evolved to serve. It's a sociological explanation, if you like. A semiotic explanation. Halliday is not the only one whose deals in explanations, the much-maligned Chomsky also has an explanation. And its important to recognize that he too, in this sense, sees the grammar in functional terms, but very different terms, because for him, grammar is as it is because for him, it reflects universal cognitive processes or innate endowment or whatever. He, too, is looking for an explanation, but his explanation of form is not a sociological but a psychological one, not a communicative one but a cognitive one. So, I would like to suggest as a second notion, or second way of thinking of function: as an explanation of form in terms of the social semiotic, and this is very much Halliday's functional grammar.

There is however a third sense of function, and I think it is frequently confused with the second, and this brings in what Jan Firbas has referred to as the „contextual factor". Here I'm referring to the function of language in the context of its occurrence, the pragmatic function. And - in a sense - is not a function of language, but a function of people's use of language - it is how people realize its communicative potential. Halliday refers to descriptions of language as descriptions of meaning potential, but how this potential is actually realized in the achievement of meaning in context is another matter. This seems to me to be not a matter of simply seeing how features of a language occur in context, one is looking here at the way in which code and context interrelate, how each one, each side so to speak, modifies the other - it's a kind of reciprocal interaction. And I think that in semiotic terms, one is thinking of the functioning of the linguistic sign, not as a semantic symbol, but as a pragmatic index, that is to say, one is looking at what the contextual constraints are on how people mean, how people use the resources of their language to engage with each other in particular occurrences of use. It's a matter not of correlating the linguistic forms with the features of the context, but how the context leads people to convert and in some sense to subvert the linguistic system in order to achieve the meanings they want.

Now having distinguished these three, quite distinctive functions, I should like to suggest that the real problem in all language description is how these three can be meaningfully interrelated. Of course, one would like to include them all in a vast comprehensive scheme. But what in effect happens, I think, is that we go for one rather than the other, and you get movements whereby one kind of function is emphasized at the expense of the others, and another kind of function is emphasized at the expense of the others and so on, which I think is in many ways regrettable. I think one has to recognize that there are these three and it is a perfectly legitimate and honest endeavor to concentrate your attention on one rather than the other two. The problems arise when you make claims that in dealing with one function you're in some sense including all the others.

I'd like to make just three brief observations about the third of these functions, the pragmatic. They are deliberately controversial.

The first observation is that context and code operate on each other. Context works in very mysterious ways. In the way we make language function pragmatically, we operate on a least effort principle, so that there are occasions where the amount of language we need to achieve our contextual meaning is minimal. Although, as Robert has said, there are no strict levels, so to speak, in pragmatic functioning, nevertheless, one does see, in discussions of discourse analysis, a description of discourse as above the sentence. And the FSP of course, is concerned with sentence perspective. But a great deal of language of course operates below the sentence and one has, for instance, one word texts, very commonly. They are texts, they are single word texts. We don't communicate by sentences. In fact, I would argue that the sentence is not a unit of communication at all. Utterances are what we use in communication, written or spoken and utterances can be very minimal. Why? Because they are compensated for or they act upon the context in all kinds of strange and mysterious ways. This is why, it seems to me, we are concerned not with how you identify linguistic forms occurring in context but how you work out the relationship between code and context in the achievement of meaning. We are talking about the realization of meaning and not simply the replication of it as if it were ready-made in the language system itself.

The second point I'd like to make is I think that the pragmatic function in languages calls for a really quite radical reconsideration of the relationship between lexis and grammar. With other functions of language, systemic and semiotic, lexis is seen to be at the service of syntax. And this also true pedagogically, that an approach to language which is modelled on formal descriptions or systemic functions, tends to simply use lexis to embody syntactic structures, and the lexis itself is thought to be at the service therefore of the syntax, to demonstrate, to illustrate it, to show how it works. It seems to me that in the pragmatic functions language use there is an argument for reversing that. To say that what in effect happens is that one uses more or less ready-made lexical expressions and one then regulates these by reference to grammar so that they key-in to a particular context. That is to say, it is not a matter of generating sentences from scratch, but of invoking more or less ready-made idiomatic stretches. I think there is a good deal of language that is memorized in this sense and that the syntax services these chunks and fine tunes them for contextual use. If one is really thinking about description of language as communication, or the teaching of language as communication, there is therefore it seems to me a case for starting with lexis and demonstrating how grammar is used to mediate the relationship between lexis, not just individual words but lexical chunks, idiomatic stretches, how grammar is used to mediate between the lexis and the context. That of course means that grammar is at the service of lexis, and that if one is really interested in the description of pragmatic functions, that is where you start, and not the other way around.

The third observation I'd like to make is a specifically pedagogic one and it has to do with functions, as they appear in notions and functions in communicative language teaching. Now I mention this perhaps simply as a tailpiece because I find the confusion between ideas of function very apparent here. If you look at a notional functional syllabus, you'll see that notions are defined in terms of semiotic function and functions in terms of pragmatic function. So that what you tend to see is people assuming that notions were handled, as it were, apparently by the grammar and that functions were pragmatic things that you had to achieve in context. In fact, what I would like to suggest is that notions in the sense of achieving referential meaning is as much a pragmatic function as is the achievement of illocutionary meaning. So in reference to a notional functional syllabus, I think we need to define notions and functions as both of a semiotic type as both of a pragmatic type, but we can't have it both ways, or at least if we do, then we are likely to be led into confusion.

My contribution is an attempt perhaps to oversimplify, by taking a number of perhaps simple-minded bearings on the whole notion of function. I think we need to distinguish between these three kinds of function, we need to accept that each has its legitimacy, that the problem is the relationship between them, and I don't think myself it is helpful to attempt to be comprehensive in relation to all of them, within one, unitary model, though we shall of course keep on trying.

Prof. R. de Beaugrande

I found what Henry had to say quite congenial. My only impulse to see if we can't broaden it to reconcile his terminology with mine with respect to context and code working together. This has of course been a neglected topic in linguistics since its early decision to distinguish between language and language use as a necessary precondition for becoming a science. A great deal of linguistic theory since then including Saussure's and Chomsky's is an enormous construction taking the hypothesis as proven that you can in fact describe a language without looking at its use. However, after 70 years or so no such description has ever been produced and the results are in fact moving steadily further from consensus. It would therefore be reasonable to regard the hypothesis as refuted in that the announced intent to study the code was always an implicit and uncontrolled study of contexts.

Formalism has been popular because the forms of the language seem to be all that is left when you take out the social and the semiotic aspect. The former repertories look like the langue or the competence which is subject to scientific treatment by rewriting into formal notations. This also coincides with rather shallow and premature views of how science works. In addition, enormous problems arise if you describe syntax, morphology and phonology separately, you need impossibly complicated mechanisms to put them back together in communication. The fact that nobody has convincingly been able to reassemble them is eloquent evidence that they do not function as independent components or as the fashionable term is now nearly decomposable systems.

The current shift toward functionalism reflects my desire to look not merely at language in use but at language functioning as a whole. An unfortunate consequence is that it is very hard to study anything without studying everything. Functionalism does not allow you very well to break out a nice tidy issue like adjectives, adverbs of a domain of a complete theory. It is like going to the ocean, and seeing a fishnet and when you take hold of it you find out it is connected to the entire sea bottom and you can keep pulling it up endlessly until you give up or you wilfully cut off a small piece. We have here the basis of arbitrariness and what Henry has called partiality Taking language away from use and breaking it down into components artificially heightens the arbitrariness of the results, simply because a considerable extent of motivation has been removed.

Current results in the emerging science of complexity suggest that the issue of motivation is most productively approached from a rather different standpoint than has previously been attempted. This motivation would be based on functional principles upon which all complex systems are being found to opeate, from the evolution of life organisms out of the „primordial soup" of chemicals all the way up to the stock market, and the international economics of buying, selling, and exchange. Despite the rather imposing name „complexity theory" the principles are in fact much simpler and more unifying than had been anticipated. The interactions of relatively simple and local constraints often suffice to generate quite sophisticated global constraints; the term „self-organising systems" has been proposed. I am currently attempting to work out a model of language that functions along comparable lines. A language description in this sense would be a model of an evolutionary system that is able to learn from the environment to perform a range of complex operations. The essence of the model is that every organism is an informational field with a data substrate interacting with the material substrate. In lower order organisms, the data substrate is fairly hard-coupled to the material substrate, so that behaviour is simple and little, if any, communication is possible. In higher order organisms, the data substrate is soft-coupled to the material substrate and a variety of behaviours considerable as is the communicative potential. In human language, a very small investment of matter and energy suffices for an enormously rich information transfer. In addition, language serves as a modality for open-ended higher amplification through cognition and communication. So we can envision language as a complex system in a highly dynamic sense so that it gains or reduces complexity and determinacy, thereby regulating the relationship between context and code. It would follow that the activity of producing discourse does not tap the knowledge of English as a whole your knowledge of the lexis, but a reduced version of that which is still an operational system adequate for the purpose - in Firthian terms, the „context of situation". The idiomatic stretches and collocations are „islands of complexity" (which is to say they have undergone self-organisation and can be managed more simply) If the lexis really operated word by word, putting them together would be an impossibly complex task as transformational grammar unintentionally demonstrated. If you try to write all the rules at the level of the morpheme and the lexeme you would get what we might call the „infinity effect", so that a complete formal grammar would either be infinitely long, or else you would need an infinite number of grammars, neither of which could be reasonably proposed as a research project, a more realistic operational description would provide for standing rules or constraints which would cover the more or less „frozen islands" of the language that have been formalised fairly successfully plus indefinitely large sets of emergent rules or constraints which are generated to fit the context and which have, so far, stubbornly resisted formalisation. The downfall of all formal linguistics descriptions has lain in assuming that the entire language functions on these standing constraints or frozen islands, whereas in fact these constraints very soon shade out into the area of emergence as soon as we leave behind the relatively well-behaved and hand-picked examples that have been used as linguistic data in the past. 'The same factor was responsible for the stagnation of early formal text linguistics, which made the same general formality supposition for texts had previously proven intractable for sentences, and of course, it was a disaster.

Henry said, and I wrote it down here, that „context and code interact in strange and mysterious ways". Perhaps, some of the newer models that are coming out of complexity theory to suggest how language acquired the capacities for context and code to interact may make the operation seem if not less strange at least as mysterious.

Prof. H. Schendl

Basically that ends the first part of the round table, I think Prof. Firbas would like to answer immediately.

Prof. J. Firbas

I should like to add three notes on the three types of function dealt with by Henry Widdowson. [At the round table I only presented Note 2 in full. Notes 1 and 3 have been expanded on here.]

As I see it, the system of language reckons with and responds to the tree types of function discussed by Henry Widdowson.

Note 1. Function and structure

Language is a tool of communication. Like any other tool, it is shaped in such a manner as to fulfil the communicative purposes it is to serve, i.e. to function in the acts of communication. Linguistic elements are interrelated to one another in order to form a structure capable of serving communicative purposes of various types. For instance, in regard to the relationship of semantics to syntax, it holds that syntactic structuration effects a semantic connection, i.e. a connection of meanings (Reichling 1961: 1, Danes* 1968: 55). Seen in this light, the syntactic structuration that takes place when a sentence is produced is undoubtedly functionally motivated. Generally speaking, the structure of language is not a haphazard cluster of linguistic elements; the elements constituting this structure are interrelated to one another in a hierarchical systemic way.

Note 2. General functions of language

Halliday's general functions of language can be traced back to those established by the Viennese psychologist Karl Bühler, who speaks of „Darstellung", „Kundgabe" and „Appell". These functions are served by various means offered by the structure of language. For instance, under the heading of „Kundgabe", and also that of „Appell", for that matter, come the language user's emotional attitudes to the message conveyed. These attitudes can, for example, be conveyed (signalled) by word order. As for the way in which word order can serve as a vehicle of emotion, and the extent to which it can do so, languages or different historical stages of one and the same language may differ. The differences are due to the differences in the structures of the languages or their different historical stages. The structure of a language determines the extent to which the various word order principles may assert themselves. Word order constitutes a system determined by the interrelations between word order principles (Mathesius 1942). Let us compare the operation of the system of word order in Modern English with that of the system of word order in Old English. In both systems, the emotive (marked) word order is caused by the deviations from the requirements of the leading word order principle. Whereas Modern English, emotive (marked ) word order is due to deviations from the requirements of the grammatical principle, Old English emotive (marked) word order is due to deviations from the requirements of the FSP (functional sentence perspective) linearity principle. Roughly speaking, the basic requirement of the FSP linearity principle is the placement of the element conveying the high point of the message (rheme proper) in end position. The basic requirement of the Modern English grammatical principle is the placement of the subject before the verb, which in its turn is to precede the object or the subject complement. Let us comment on the following example sentences taken from the Old English version, and several Modern English versions, of the New Testament.



... butan intingan hig me weorðiaþ and læraþ manna lara.

[ in vain they me worship and teach men's lore] (Matt 15.9)


(a) But in vain do they worship me, ... (Phillips)

(b) ... vain is their worship of me, ... (Moffatt)

(c) They worship me in vain; ... (NIV)

(d) But their worship of me is all in vain, ... (Goodspeed)

(e) ... ; their worship of me is in vain ... (NEB)

(f) Their worship of me is in vain, ... (Knox )

(g) But their worship is to no purpose, ... (BBE)

(h) They do me empty reverence, ... (NAB)

(i) It is in vain that they keep worshipping me, ... (NW )

(j) Uselessly, they worship Me with their teaching of human commands (MLB)

It follows that whereas the word order of the first sentence of OE 1 is rendered emotive (marked) due to the placement of rheme proper in front position, the word orders of ModE 1a and 1b are rendered emotive (marked) owing to the deviation from the S-V-O/C [subject Complement] order. The majority of the ModE versions, however, do not show this deviation. It is not without interest to note that, owing to the fact that the grammatical principle has established itself as the leading principle in the ModE word order system, ModE word order shows a decrease in emotiveness (markedness). (A certain degree of markedness is shown by the word orders of ModE 1i and 1j. In 1i, in vain is thrown into relief by means of the cleft construction. In 1j, Uselessly becomes conspicous on account of its unusual front position, for adverbs of manner in -ly do not normally occur before the subject. Let me note that Uselessly is rhematic, but does not serve as rheme proper, the sentence being perspectived to with their teaching of human commands. Unlike its OE conterpart ModE 1j is not a compound, but a simple sentence. (For a discussion of the relationship between FSP and word order, see Firbas 1992: 117-48.)

Needless to say, emotive (marked) word order is not the only means that serves the function of „Kundgabe", or that of „Appell", for that matter. Any deviation causing markedness serves either or both functions.

Note 3. Function and context

At the moment of communication, the sentence becomes embedded in context, verbal (written or spoken) and situational. Language does not operate outside context. It is constantly affected by it. As I see it, context co-effects the perspective in which the sentence functions at the moment of utterance (written or spoken) and/or perception. It does so through the operation in language of the contextual factor, which plays the leading role in the interplay of factors determining the functional perspective of the sentence (FSP).

The signals yielded by this factor are (a) the occurrences (actual presence) of pieces of information in what has been delimited as the immediately relevant preceding verbal context, and (b) the re-expression(s) of such information at the moment of utterance and/or perception; or (a) the actual presence of a referent in what has been delimited as the immediately relevant situational context, and (b) the expression of this referent at the moment of utterance and/or perception. Expressions conveying information retrievable from the immediatly relevant context do not express the high point of the message (rheme proper), towards which the sentence is perspectived. They constitute or co-constitute the foundation (theme), upon which the core of the message (non-theme, consisting of transition and rheme) is built up. In this way, they perform an important role in co-determining the perspective in which the sentence functions, and consequently participate in revealing the language user's communicative purpose. (This is indeed indicated by perspectiving the sentence to the element conveying the high point of the message, i.e. the rheme proper. (For a more detailed discussion of the immediately relevant context and the operation of the contextual factor in FSP, see Firbas 1992: 21-40).

My notes could be summed up as follows. Henry Widdowson has drawn our attention to various aspects of function and cautioned against integrating them prematurely into one system. His warnings must certainly be taken seriously. But I only wish to point out that, if function is understood as the operation of linguistic means in fulfilling the communicative purposes of the language user, the three types of functions discused in Notes 1, 2 and 3 do not operate outside the system of language and are therefore to be looked upon as integrated into this system.


°The contributions represent edited versions of the talks given by the participants. The differences in format reflect the divergent editorial practices by the contributors themselves.


Danes*, F. 1968. „Some thoughts on the semantic structure of the sentence", Lingua 21: 55-69.

Firbas, J. 1957. „On the problem of non-thematic subjects in contemporary English", C*asopis pro moderni filologii 39:171-3. English summary of „Kotázce nezákladových podme*tú v souc*asné‚ anglic*tine*", ib.: 22-42, 165-73.

---. 1992. Functional sentence perspective in written and spoken communication, Cambridge: Cambridge University Press.

Mathesius, V. 1942. „Ze srovnávacích studií slovosledných" [From comparative word order studies], C*asopis pro moderni filologii 28: 181-90, 302-7.

Reichling, A. 1961. „Principles and methods of syntax: cryptoanalytical formalism", Lingua 10: 1-17.

Weil, H. 1844. De l'ordre des mots dans les langues anciennes compar‚es aux langues modernes, Paris: Joubert.

---. 1887. The order of words in the ancient languages compared with that of the modern languages, Boston: Ginn & Co.