A retrospective view of Systemic Functional Linguistics, with notes from a parallel perspective

Abstract

This paper gives a brief overview of the development of Michael Halliday’s work in Systemic Functional Linguistics over the last sixty years and recalls some anecdotes relating to the research projects he directed at University College London in the 1960s. It locates the author’s approach in relation to Systemic Functional Linguistics and raises some questions to do with possible avenues of future research and discussion in that model.

Background

The fortieth annual meeting of the International Systemic Functional Congress, held at Sun Yat-sen University in Guangzhou, China in July 2013 (henceforward, ‘ISFC40’), provided a context in which it seemed appropriate to look back on the beginnings of this functional approach to language. The present paper recalls some of the early days of work in this model of linguistics from the perspective of someone who was privileged to be able to take part in some of them and attempts to trace the trajectory of developments in the thinking of Michael Halliday, who initiated the model and who has been at the heart of its development ever since. In the first part of the paper there is an outline and ordering of a selection of Halliday’s publications, mixed with some anecdotal reminiscences; and towards the end some questions are raised to do with the direction of future developments of work in Systemic Functional Linguistics.

Early days

My recollections of working in this model of linguistics go back a long way, to the early days in London. At ISFC40, it was exactly fifty years since, as a research assistant working under Basil Bernstein in his Sociological Research Unit at the London University Institute of Education, I first became a student of Michael Halliday’s at University College London; and it was forty-nine years since I became, with Ruqaiya Hasan, a linguistics research assistant working under Halliday in the Nuffield Programme in Linguistics and English Teaching at University College London. I am therefore in a position to look back on the early years of what began as ‘Scale and Category Grammar’ and has now become ‘Systemic Functional Linguistics’ (henceforward, ‘SFL’). I would like to do so both anecdotally and also by considering more seriously some aspects of that span of time and achievement, which has largely overlapped with the establishment of Chomsky’s Transformational Grammar (henceforward, ‘T-G’) and related approaches.

To begin anecdotally, work in those early days seemed to be housed either in basements or attics. I remember watching the London Post Office Tower growing to its full height from a skylight window in the Tavistock Square attic in which Bernstein’s project began; and Halliday’s two projects lived in a basement in Malet Place, just opposite what was then Dillon’s Bookshop.

I remember Chomsky’s first visit to London. He gave a lecture that we all crowded in to attend and at the end of it Michael Halliday asked him a question, the implication of which was that one of his claims did not hold water. Chomsky’s dismissal of this very pertinent challenge was magisterial: ‘that is merely a putative counter-example, which does not affect the validity of the argument in any way’. The phrase ‘putative counter-example’ has stayed with me ever since as a useful summary of the T-G attitude to data. It also led me to see Chomsky himself more in the light of a powerful (and bruising!) rhetorician than the heavy weight logician that he was often presented as being. This is a view that has not diminished with time.

Development from the ‘Scale and Category’ model to SFL

In thinking about the development of SFL from Scale and Category Grammar I believe it is relevant to take into account the contemporary influence of work and ideas in T-G. Two aspects of this strike me as particularly interesting. One is the consistent emphasis that Halliday, and SFL generally, have placed on the analysis of real texts, from a wide range of different genres and registers. As a rich, and always growing, source of ‘putative counter-examples’, this approach challenges, head on, the T-G restriction of focus to the ‘native-speaker’s intuitions’. This has remained one of the major contrasts between the two schools of linguistic thought over the years, though SFL could be said to occupy the middle ground between T-G and some varieties of ‘corpus linguistics’ in this respect. It has not rejected testable intuitions that could be supported from textual analysis but it is also by no means an exclusively data driven approach that derives its categories from the statistical analysis of corpora. In as far as there is some movement towards a more functionally-oriented approach in some T-G based studies, such as the analysis of English questions by Robert Fiengo (2007: 146, 170–1), there might now be something more of a move towards developing a consensus that both data and intuitions are needed.

The second aspect relates to Chomsky’s concept of ‘deep structure’. This had an impact of a more technical kind on Scale and Category: namely, an increasing emphasis on systems, as opposed to structures. Halliday’s paper, ‘Some notes on “deep” grammar’ (1966) is seminal in this respect. This shift is of course reflected in the change of name to Systemic Functional Linguistics. But it is worth noting that when the first edition of Halliday’s book, Introduction to Functional Grammar (henceforward, ‘IFG’), appeared in 1985, the description of English which it gave was expressed almost exclusively in terms of structures, even if a systemic analysis under-pinned it. This is particularly interesting because the book represented his teaching over the intervening years. As he says in the ‘Foreword’ (ix), ‘This book grew out of seventeen pages of class notes’.

University College London

Those early days in the UCL Malet Place basement, from 1964 onwards, were very intellectually stimulating. It was during this period, up to the beginning of the seventies, that many of the ideas that have proved lasting in Systemic Linguistics were first established. This is the period in which Halliday’s three papers on transitivity and theme in English appeared in the Journal of Linguistics (1967–8) and the later paper on modality and mood in English appeared in Foundations of Language 6 (Halliday 1970b). Taken together, these still form the foundation of a great part of the description in the third (revised) edition of IFG in 2004. It was also the period in which Halliday first put forward his concept of the four meta-functions and the associated four components of the grammar, beginning with his paper, ‘Options and Functions in the English Clause’ for Brno Studies in English 8 (Halliday 1969) and expanded in the following year in his ‘mood’ paper and also in the chapter, ‘Language Structure and Language Function’ , in a collection edited by John Lyons (Halliday 1970a). This is one of the key foundations of the model as we know it now. I would also place the main work of Halliday and Hasan’s collaboration on their book, Cohesion in English, essentially in this period, even though it was completed and published later (Halliday and Hasan 1976).

During this time, again, John Sinclair began implementing Firth’s approach to the vocabulary in terms of collocations, by a large scale computational study at the University of Birmingham; and Halliday developed his Edinburgh work on the intonation of English in relation to the grammar, for teaching purposes. The initial recordings of the illustrations for A Course in Spoken English: Intonation (Halliday 1970c) were made then (though later replaced), using a group of native speakers from UCL of whom I was one. There could in my opinion be no clearer indication of the contrast between Halliday’s approach and that of Chomsky than the mere existence of this body of work on intonation. As late as 2007 (38–43), Robert Fiengo, a relatively functionally inclined scholar who works within T- G, could claim that intonation is largely a matter of ‘affect’ and feel free to ignore it in a study of English questions.

Stages in Halliday’s thought and the development of the model

I believe it is not unreasonable to summarize the development of Michael Halliday’s work and thought in four main stages, with a possible fifth and sixth (although there is also considerable overlap between them):

Beginnings of the model

The first stage, ‘Scale and Category’ as set out in his Word article (Halliday 1961) goes back to his (1955) Cambridge PhD study of a late fourteenth century Chinese text, published as a Special Publication of the Philological Society as, The Language of the Chinese “Secret History of the Mongols” (Halliday 1959). In this stage we have the overall theoretical framework, including the four categories of ‘unit’ , ‘structure’ , ‘class’ and ‘system’ , and the three scales of, ‘rank’ (associated with size of unit); ‘delicacy’ (associated with progress through a system path and so with increasingly finer distinctions and a greater degree of detail in the description); and ‘exponence’ (later termed ‘realization’ , following Lamb, and associated with degree of abstract/concrete). It was also at this time that the foundations of Halliday’s work on English Intonation were laid. His two pioneering articles, ‘The Tones of English’ and ‘Intonation in English Grammar’ , belong here and were both first published in 1963 (Halliday 1963a 1963b) Geographically, this stage is associated with a wide range of different places: with The School of Oriental and African Studies in London, where Halliday studied under J.R. Firth; with China, in Beijing, and at Sun Yat-sen University in Guangzhou; with Cambridge, and subsequently with Scotland, at the University of Edinburgh.

Stage 1 is essentially what Halliday brought with him to London in 1963.

Systemic functional linguistics: the lexico-grammar

The second stage, at University College London, is the one that I have been outlining earlier. It sees a greater emphasis on systems, somewhat at the expense of structures. The three (1967–8) Journal of Linguistics articles on transitivity and theme are expressed and organized chiefly in terms of system networks. This is the stage that sees the establishment of Halliday’s multifunctional approach, which has remained central to the model ever since. It is also the period in which Halliday’s two major papers on English intonation were brought together and made more generally available in his (Halliday 1967b) book, Intonation and Grammar in British English.

SFL and social semiotics

I would place the beginning of a third stage, in which Halliday published complex ideas to do with social semiotics, essentially at the end of the UCL period and afterwards, from 1971 on. However, the inception of this work can be seen much earlier, in his Inaugural Lecture, as the first Professor of General Linguistics at University College London, later published as, ‘Grammar, Society and the Noun’ (Halliday 1967a). In that, he set out his own position in relation to Whorf’s work and ideas of ‘Linguistic relativity’ . This third stage shows the influence of Bernstein’s work on the relationship between language and social class. Halliday’s (1971b) essay, ‘Language in a social perspective’ in the Educational Review and his 1972 paper, ‘Towards a sociological semantics’ , both reprinted in his Explorations in the Functions of Language (1973), illustrate this, as does his later book, Language as social semiotic (1978). Halliday’s work at this time also shows the influence of Sydney Lamb’s development of Stratificational Grammar in relation to the concept of ‘realization’ and different levels (strata) of abstraction, which represented a development of the scale of exponence. It is the period in which Halliday substantively enters the field of child language development with his analyses of the language of Nigel at 19 months and in which he generally works to expand the perceived relevance of the model. His best known article on literary language, ‘Linguistic function and literary style: an enquiry into the language of William Golding’s The Inheritors’, belongs here in terms of publication (1971a), although its origin was as a 1965 conference paper in London. The development of a computational approach to the grammar, using systems, in the work of Bill Mann and Christian Matthiessen at the University of Southern California (1983), which began following guest lectures by Halliday at Stanford University in 1980, is associated with the latter part of this period. Geographically, this stage includes time spent in North America but is also, from 1975, the first of those associated with Australia, at the University of Sydney.

Return to the lexico-grammar; grammatical metaphors

I would suggest a fourth stage in Halliday’s work, beginning with the publication of Introduction to Functional Grammar in 1985 and including its second edition in 1994, which incorporated the concept of ‘grammatical metaphors’. This represents a return to an emphasis on descriptive linguistics as such, as distinct from social semiotics. The first two editions of IFG are organized and expressed very largely in terms of structures. In harmony with this return to a structural analysis of texts, there is much further work by Halliday in the description of particular texts and types of texts in different registers, including his (1985b/1989) book, Spoken and Written Language, and several important papers on the language of science, including, ‘The Language of Physical Science’ (Halliday 1988).

Expansion of systems

Perhaps it would be in order to suggest a fifth stage, of which one major achievement would be the publication of the third edition of IFG in 2004, in collaboration with Christian Matthiessen. If so, this would represent something of a re-visiting and further development of the stage 2 approach, in the sense that there is much more of an explicit account in terms of systems in the third edition of IFG than in either of the two earlier versions. This aspect of the revision and extension is attributed largely to Matthiessen (IFG 3: ix), who draws on some of his specialized computational expertise in this area (cf. Matthiessen, 1995). I believe that the forthcoming fourth edition of IFG continues and expands this direction. Another major achievement here is the revision of Halliday’s work on intonation in collaboration with Bill Greaves in their (Halliday and Greaves 2008), Intonation in the Grammar of English.

Return to Chinese linguistics

Finally, in view of his opening plenary lecture at the Fortieth International Systemic Functional Linguistics Congress, entitled, ‘That Certain Cut: Towards a Characterology of Mandarin Chinese’ and his current position as advisor to the Functional Linguistics Institute at Sun Yat-Sen University, it may not be fanciful to suggest a sixth stage in Halliday’s work, representing a return to his first love in linguistics: the study of the Chinese languages.

Narrow limits of this account

In this grotesquely brief summary I have just sketched the trajectory of some of the theoretical developments made by Michael Halliday himself, and I have purposely left out of account almost everything done by anyone other than him, except in as far as they have been his collaborators in particular publications. This is especially unjust to Ruqaiya Hasan, who has done so much to consolidate and extend the model and three of whose seven volume series of collected works, edited by Jonathan J. Webster and published by Equinox, have now appeared (2005 2009 2011). It is also unjust to Matthiessen, with, for example, his (1995) monumental work on English systems, and Martin, with his (1992) and later work on discourse and genre, and on ‘Appraisal Theory’ with White (2005).

Purely for reasons of space, I have mentioned little of the work of the Birmingham ‘school’ and none of that at Cardiff, for example, Fawcett (2000/2010), both of which have extended the range of SFL. I have also not referred to the debates concerning theme and the nature of the register/genre distinction, or indeed the huge amount of work carried out by scholars all over the world in applying SFL to different kinds of texts, for a wide range of particular purposes. The theme of ISFC40 was, ‘Broadening the Path’; but it is immediately clear, even within the confines that I have set myself, how ‘broad’ the path of SFL already is.

Discussion: suggestions for further debate in SFL

Standpoint and background of author in relation to SFL model

The remarks that follow are geared to some core theoretical aspects of SFL, as opposed to its applications. They should be seen as made from the standpoint of a friendly fellow-searcher after understanding: someone whose research began within SFL but has become less centrally related to it over time, though I have taught both undergraduate and postgraduate courses using SFL approaches and materials, including different editions of IFG. Let me briefly sketch the nature of my research connections with SFL. My original involvement in the UCL research programme led me to an early exploration of comment adjuncts (Davies 1967) along lines that were not then standard, although later partly adopted in SFL, and to some attempts to set out what was then the standard account of elements of clause structure (Davies 1968b) and the relation of the model to other linguistic theories (Davies 1968a). From 1967, when I left London for Cardiff University, I was working on an interactional role theory approach to the deep grammar of mood and condition in English. This issued in a dissertation (1976) and a later book (Davies 1979) which made some fairly substantial theoretical departures from the SFL model and proposed some different components in the grammar, notably one of ‘telling’ , but still used systems to set out some of the surface grammar complexities of English interrogatives.

Perhaps on this basis it was included in the bibliography of the first edition of IFG in 1985; and some papers of mine are cited under ‘Further Reading’ for the chapter on the interpersonal component in the second edition in 1994, including one on modal verbs (Davies 1988a) in which I again used systems. But subsequently I have developed an approach in terms of set theory and operators, first sketched in my 1979 book, and have used it to develop the different components proposed there: in particular those of ‘knowledge’ (Davies 2001) and ‘telling’ (Davies 2006). In a later paper (Davies 2012: 237–238,240–245, 249–250) I have moved towards proposing a different ‘entry’, in terms of a model of interaction with factors in the existing Common Ground, to an area that is close to the textual meta-function in SFL and again involves telling. Telling is seen as what is done with reality (events and states of affairs, knowledge of them and decisions and wishes about them) to construct and present language for use between people in interaction. It is the only purely linguistic, and the only essential, meta-function and operates on those of decision and knowledge (which partly correspond to the interactional and experiential meta-functions in SFL). This is the ‘parallel perspective’ of my title.

Three suggested topics

I would like to suggest that some of the major theoretical issues which have been raised by what I regard as Halliday’s seminal insights should remain open to further debate and investigation. I believe that one respect in which SFL could be broadened now would be to explore links with other approaches and I want to indicate briefly three major topics on which I believe the theoretical discussion should be seen as still open.

Grammar and discourse

The first of these concerns the linguistics/pragmatics interface and the relationship between grammar and discourse. Put briefly, this latter is presented as a seamless transition in Halliday’s account, which uses categories, such as ‘exchange’ , derived from discourse as the foundation of his analysis of the mood component in the grammar. He earlier maintained that discourse was not merely a ‘larger form of grammar’ but involved different kinds of organization; but it might be claimed that his treatment of the mood system has some tendency to treat this aspect of grammar as a ‘smaller form of discourse’. In one sense, Sinclair and Coulthard (1975) could be said to have done something comparable (though in the opposite direction) by extending the use of categories designed originally for grammatical analysis (such as the rank scale) to the analysis of discourse.

An alternative to both these approaches would be to see the interface between grammar and discourse as a location for ‘changing gear’: moving from one type of category to another related type, which is differently constituted. This is more the approach taken in formal grammars, with their division between syntax and pragmatics, and it is also something that I have tried to do (Davies 1979 1985 1988b) with the notion of categories of ‘significance’ , seen as derived from the combination of linguistic and extra-linguistic meanings. It seems to me that this is an area with which SFL might re-engage.

Identity and number of the meta-functions

A second major issue, in my view, is the question of the identity and the number of the meta- functions, and their associated components in the grammar. Here, not only ‘theme’ but also the ‘logical component’ have perhaps not yet reached their final versions; and the question of whether the latter should be separately established at all may be open to further debate. There are also issues involving the status of the ‘Subject’ element. In the mood component, this is treated (together with the Finite) as a constituent of the Mood element; but under theme, it is the question of whether or not the Subject, as a separately established element, combines with the topical Theme that gives marked/unmarked theme selection status to the (declarative) clause (Halliday & Matthiessen, 2004: 73–74, 80). In this way, the Subject element of structure has more of an independent existence in the theme component of the grammar than in that of mood; and the same is true of transitivity, where the Subject may combine with a wide range of different roles, varying, for example, from Actor to Carrier. The Subject seems to be treated as being at a different level of abstraction in the mood component as compared with theme and transitivity. If a full description of the clause consists in bringing together its description in terms of all the components of the grammar, how does a part of an element of one kind (the Subject as part of the Mood element) combine with the whole of an element of a different kind, such as Theme or Actor? This leads in to questions to do with the relationships between the different grammatical components and their contrastive delineations. The development of the concept of a multiplicity of functions, operating simultaneously and relating to different areas of lexico-grammatical organization, has always seemed to me to be one of Halliday’s most fruitful contributions to linguistic theory. But belief in multi-functionalism, which I share, need not mean that the currently established components should always remain exactly as they now stand, in every respect.

The role of systems in the model

There is a third, very major, topic that I believe would benefit from more open discussion. This is the role of systems in the model. In Halliday’s ‘deep’ grammar paper, systems and structures are initially presented as being at the same level of abstraction: equally ‘deep’ as compared, respectively, with the more surface categories of paradigm and syntagm (Halliday 1966: 60). However, as the paper develops, ‘some possible consequences of regarding systemic description as the underlying form of representation’ are considered (62–63), one of which would be that, ‘that part of the grammar which is … “closest to” the semantics may be represented in terms of systemic features’. ‘Structure would then appear as the realization of complexes of systemic features’ (63).

This second position is adopted in ‘Notes on transitivity and theme in English’ (Halliday 1967c: 37) and thereafter. In IFG 3, structural configurations are specifically presented as less ‘deep’ than terms in systems, in the sense that the former are shown as realizing the latter. In particular, the presence/absence of a given element of structure may be shown as a systemic choice, as it is in the mood system where the presence or absence of the Mood element (consisting in a combination of Subject and Finite) is shown to distinguish indicative from unmarked imperative clauses (Halliday & Matthiessen 2004: 111, 135, 138). In the relational clause network of systems under transitivity, we have not only the presence/absence of particular elements of structure shown as realizing terms in a system, but also detailed specification of what these elements of structure are themselves realized by in terms of the class of unit of the rank next below. For example, in the system ‘MODE OF RELATION’ the term ‘Attributive’ is shown as realized by the presence of the two elements of structure, ‘Carrier’ and ‘Attribute’ , with the further feature that the element ‘Carrier’ is realized by a group of the class ‘nominal’ (Halliday & Matthiessen 2004: 215–222).

The mere citation of these examples shows one major and obvious respect in which the model has evolved from the ‘deep’ structure paper of 1966: namely, by the introduction of the multiple meta- functions and their associated components in the grammar, yielding different kinds of elements of structure. That is, whereas in 1966 there were just Subject, Predicator, Complement, and Adjunct as elements of clause structure, we now also have Mood and Residue, Theme and Rheme, and a long list of different transitivity roles, including Actor, and Carrier as above, all treated as (different kinds of) elements of clause structure. This in turn leads to the possibility of distinctions to do with different combinations of these different kinds of elements, such as that of the Theme and Subject combinations mentioned above. This development represents a change in the model which could be said to affect the role of systems, in the sense that it greatly extends the range of the realizations of their terms, and, in fact, of their own number and identity, thereby substantially enriching the description.

Another respect in which the model has developed with respect to systems is that any apparent connection between rank and realization (‘exponence’) in the old Scale and Category model has been largely severed. That is, whereas in the latter a term in a system operating in the environment of an element of structure in a unit of clause rank was realized by a class of unit of the rank next below (group), we now have, mainly, a term in a system operating at clause rank realized by a type of clause, defined in terms of its own internal structural configuration. For example, the presence/absence of a given element of clause structure yields two different structural types of clause: no change of rank is involved. This development is not yet entirely consistently applied, however, as in the instance mentioned above, which allows a mixture. There the term in system, ‘Attributive’ , is shown to be realized by the presence of two elements of clause structure (‘Attribute’ and ‘Carrier’) and also by a particular form of further realization of one of them. The Carrier element must itself be realized by a particular class of the unit next below on the rank scale, the nominal class of group.

This part of the realization statement for ‘Attributive’ appears to revert to the earlier Scale and Category model.

Both these kinds of development in the way systems are used in the model are significant in terms of the different descriptions that can be generated and probably deserve to be argued through in the literature more fully and explicitly.

Directions of future research in SFL

The theme of ISFC40 was ‘broadening the path’, and this is certainly what has come about in SFL in the last half century. In the course of this happening, the nature of the approach has changed quite substantially: from an approach to language that centered essentially on the analysis of the lexico- grammar, to one which is very much wider and includes the study of non-linguistic phenomena. In this way, SFL is en route to becoming what physicists call a ‘toe’ (a theory of everything) - something which I believe can bring dangers of its own.

This leads me to the issue of where the greater theoretical effort in terms of the SFL model should be directed from now on: should it be in the direction of greater ‘explanatory adequacy’ to adopt Chomsky’s term; or should it be in the direction of attempting to extend the model to account for further different types of phenomena? This is probably the greatest general question about future theoretical developments in the model.

One of the greatest strengths of SFL has been, and is, what Chomsky called ‘descriptive adequacy’. The large, and growing, body of work involving the application of SFL in the analysis of texts in different languages, registers and styles may well exceed, both in amount and variety, that done in any other linguistic model. Some may feel that this demonstrated descriptive adequacy of the current model is enough, and that further theoretical debate is not necessary; but I am not sure that I would agree. I believe there are remaining challenges to do with explanatory adequacy; and that this matters.

SFL has to be seen both in relation to other functional grammars and in relation to the body of formalist approaches that are related in some degree to Chomsky’s work. We are fortunate in having an account of the former in Chris Butler’s (2003) two volume comparative study of three functional grammars. But, in my view, the latter, more vexed, topic of the relation of SFL to formalist approaches is the area in which theoretical advances in this new millennium most need to be made.

Conclusion

I will end with three questions which I believe should be discussed within SFL in the coming years, if only ultimately to give a full defence of the current status quo. The first is the nature of the interface between the lexico-grammar and discourse. If we want neither discourse categories imposed on the grammar, nor grammatical categories imposed on discourse, what is the nature of this ‘change of gear’? The second is the number and identity of the meta-functions and associated components of the grammar. For example, is a ‘logical component’ either justified or necessary? Thirdly, I would like to see SFL discussing rigorously how systems relate to structures and making fully explicit just what it is of value that systems add to the model. Chomsky dismissed them as merely a taxonomic device. In exactly what terms can SFL theory refute this criticism and conclusively justify their use?

With these enquiries for the future I will end this outline of some of the very considerable amount of work in the past. But whatever remains for further discussion and development, no-one can be in any doubt of the towering, and continuing, achievements of the man who began this model and has sustained it through nearly sixty years: Michael Halliday. To him, all of us who have worked within or alongside this framework owe a very substantial debt; and that is something which I personally find pleasure in acknowledging with gratitude.

Authors’ information

Eirian Davies has spent most of her academic life in the University of London, UK, with five years at Cardiff University and one, with the help of a Fulbright award, at the University of Minnesota, USA. In the early days of Systemic Functional Linguistics, she worked first under the direction of Basil Bernstein at the London University Institute of Education and then under the direction of Michael Halliday as a research assistant at University College London. She was one of Professor Halliday’s post-graduate research students there, later obtaining her PhD in Linguistics under his supervision. Now retired, she has subsequently held posts in departments of English in London University, at Bedford College and Royal Holloway, and is currently writing a book for De Gruyter (‘Elements of English Functional Grammar’) developing the theoretical approach mentioned under ‘standpoint of author’, above.