Systemic functional linguistics: advances and applications

In the early 1960s, the dominant linguistic theory was that of transformational (later generative) grammar, initiated by Noam Chomsky; this theory posited that the only aspect of language worth studying was its form (Chomsky 1957, 1965), and to do this it had to be reduced to an idealized structure, that is, a form which does not occur in natural human communication. In contrast to this, a small group of linguists in London argued against this prevailing orthodoxy, arguing in favour of a study of natural human language, which involves consideration of much more than the form of language. They developed the approach known as Scale and Category Grammar, which would later evolve into Systemic Functional Linguistics (SFL). I suspect that at that time, no one on the ground could have predicted that the movement initiated by this group would develop into a theory that would be taught and practiced throughout the world, in universities and institutes on every continent. Without in any way wishing to belittle the immense achievements of other members of this group, the name of one man has become indelibly associated with SFL, of which he is generally now considered the founding father. That man is, of course, Michael A. K. Halliday.

Michael A. K. Hallilday was born in Leeds in 1925. He studied Chinese, and from 1944 to 1947 he taught that language to British army personnel in India and subsequently in the UK. In 1948 he went to China where he worked, studied and taught, returning to Great Britain in 1950 to prepare his PhD, which he presented at the University of Cambridge in 1954. By the early 1960s he was in London, and the saga of SFL was under way (Steiner 2018; Webster 2005). In his 1961 article “Categories of the theory of grammar”, Halliday (2002 [1961]) posits, for example, that the structure of language is based on a number of ranks which form a scale. For English these would be sentence (later replaced by “clause complex”), clause, word, group, and morpheme. Each rank can be downwardly shifted, or rankshifted to function at a lower rank. Thus, for example, a clause can be rankshifted to function as a word, or a group. However, rankshift cannot operate in the other direction.

Scale and Category Grammar morphed into SFL because of the increasing importance of function and system. Function relates to the way language works both externally and internally. In Fontaine’s words “[t]he functions of language include both the use that language serves (i.e. how and why people use language) and linguistic functions (i.e. the grammatical and semantic roles assigned to parts of language)” (Fontaine 2013: 5). The first of these leads to the importance of language as a social phenomenon, a “social-semiotic” in Halliday’s terms (Halliday 1978), and hence of context as a causal feature in the creation of language; the second leads to the structuring of the clause as three interrelated metafunctions dealing with the representation of external or internal reality (the ideational metafunction), the relationships between the speakers and their addressees or with their messages (the interpersonal metafunction), and the way the message is organized (the textual metafunction). System is the network of choices that the language provides for the speaker. To give a simple example: if a clause is our starting point, this provides the choice between indicative and imperative; selecting indicative gives the choice between interrogative and declarative; selecting interrogative provides the choice between Wh-question and polar question. Theoretically the network can continue up to the point of lexical choice. At the same time Halliday (1967) recognized the importance of spoken language and hence the need to integrate phonology into this framework. Taking over the basic ideas of the Prague School, he then added the important distinction between thematic and information structure.

Many other theories seem to founder on the fact that their frameworks are fairly rigid, and once established provide little room for development. If SFL has been able in the intervening 60 years or so to continue developing and to branch out into new areas that probably could not have been foreseen in the 1960s, it is at least in part due to an openness of spirit, espoused by Halliday himself, whereby debate, divergence, and discussion are welcomed, where alternative points of view are given due weight, and where there is no official orthodoxy which must be adhered to.

The objective of this special issue is to attempt to show something of the breadth and depth of current investigation inspired by SFL. The development of the approach has branched out in so many directions that any attempt to be exhaustive would be impossible. Nevertheless, the contributions assembled here show something of the range, diversity, breadth, and spread of work in this field, while remaining true to the principles of SFL.

As mentioned above, system networks were an essential element in the early elaboration of SFL. In 1963, Halliday described system as “the ‘choice’ axis”, and went on to say that

a system is a limited (“closed”) set of terms in choice relation, such as the English system of “number” (for example boy/boys). While (in this instance) the ultimate exponents of the terms in the system are boy and boys, the direct exponents of these terms are the class “singular nominal group” and the class “plural nominal group”. (Halliday 2002 [1963]: 97, italics original)

In recent years, however, I feel that system networks have taken something of a back-seat role, with some notable exceptions, such as Matthiessen (1995).

Language learning was also a question of importance in the early years, both as learning one’s mother tongue (L1), evidenced by Halliday’s seminal work on the language development of his own son (Halliday 1975), and in the learning of other languages (L2) (cf. for example Chapter 9 of Halliday et al. 1964). It is the link between system networks and L2 language learning that is of interest to Jorge Arús-Hita, Christian M. I. M. Matthiessen, and Winfred Wenhui Xuan (“Learning how to mean in a second language: uses of system networks in L2 education”). They chart the ways in which system networks have been used in L2 language teaching and learning over the years, and suggest ways in which these can be adapted and extended. This includes a wide range of applications such as:

1. tracking language development as it was first used in L1 language development in Halliday (1975).

2. diagnosing problems in L2 education, where a network can supply the target to be achieved, but networks can also be used to chart a learner’s choice, thus providing the transitional state of their learning curve, which in other approaches would probably be called their interlanguage (Selinker 1972).

3. supporting sequencing in the curriculum, where the system network will give the L2 resources that the learner needs.

4. designing exercises, which can be informed by system networks.

5. guiding L2 learners (and perhaps teachers too), where system networks can operate as cartographic tools.

6. contrasting the resources of the L1 and the L2, to bring out the differences between them.

The well-known Systemic Functional analysis of discourse into three interrelated strata, or metafunctions, of meaning, ideational, interpersonal, and textual, was already present in early accounts of the theory (Berry 1977; Halliday 1973, 1978). In those early accounts the interpersonal metafunction was defined as that part of meaning which established the relationships set up by the speaker, both with his addressees and with his discourse, and was said to be realized by mood and modality, in Halliday’s words the speaker’s “determination of the choice of roles for the addressee (mood), and the expression of his judgements and predictions (modality)” (Halliday 1973: 361). Towards the end of the twentieth century the introduction of Appraisal (Martin 2000; Martin and White 2005) would vastly extend the field covered by the interpersonal metafunction. This is the area which concerns Boitshwarelo Rantsudu and Tom Bartlett (“The role of deontic modality in the construction and mitigation of evaluation in hard news reporting”) in this issue. They are particularly interested in the use of deontic modality (or modulation) in press reports. They point out that previous studies using Appraisal have tended to concentrate on epistemic modality (or modalization). On the basis of a corpus of newspaper articles in the Botswana press dealing with the 2021 public sector strike, the authors wish to see how the journalists deal with interpersonal features in a genre which is supposedly objective, even if the definition of objectivity remains to some extent debatable. Deontic modality, it appears is used as a mitigation strategy, reducing or overriding evaluating language.

Right from the earliest days, Halliday (1967, 1970) integrated questions of phonetics and phonology into the model he was developing. Notable in this is the relationship between phonology and lexicogrammar, and how the two interrelate. The distinction made by Halliday between thematic structure and information structure is primordial, and marks off his approach from that of the Prague School, whose ideas had provided the initial impetus. This shows that the units of written and spoken language are different, that of the former being the clause, while that of the latter is the tone group. Although the literature on phonology in SFL is much less abundant than that on lexicogrammar, important updates have been provided by Tench (1996) and Halliday and Greaves (2008). In this issue, Gerard O’Grady (“Theme in spoken language: when a tone group is not a clause”) pushes forwards the boundaries of study in this area. In an intriguing and detailed analysis of a data set of spontaneous speech by nine respondents, he argues against two commonly held assumptions: (1) that clauses and tone groups are coterminous, and (2) that theme and given coincide, and that the new occurs in the rheme. The first of these is consistent with the observation that the majority of tone units seem to have no more than two or three feet, and that the maximum would seem to be of the order of seven or eight, given that we all need to take a breath fairly regularly, which engenders a tone group boundary (Banks 2004). O’Grady’s analysis teases out the details of the complex relationships between lexicogrammatical features and prosodic choices, including some tendencies which are not those which might have been predicted.

In 1981, Halliday said

I have hoped […] to exemplify my earlier point that the features of a description, and therefore of the model that lies behind it, are relatable to the aims of the model and through these to particular applications of linguistics. (Halliday 1981: 27)

This would develop into the notion of appliable, as opposed to applicable, linguistics. In 2005, he was asked in an interview about the distinction.

AB: Is it applicable linguistics? Is it appliable linguistics? Are there differences in your view?

MAKH: ‘Applicable’ to me has a difference. If I say applicable then it’s with a sense of applicable to something. I’m not making a big issue of this! But there is a significant difference. If I say something is ‘applicable’ then it refers specifically to some task or at least some particular sphere of action. I want a more general sense, that of ‘capable of and designed for being applied’. (Halliday and Burns 2006: 124)

A few lines later he mentions the clinical area, as a possible area of appliability. And systemic functional analysis has indeed been used to great effect in the clinical domain, including that of mental disorders. De Villiers (2005), for example, shows how systemic functional analysis can be used to make sense of the apparently impenetrable discourse of autistic patients.

In their contribution (“Grammar, cohesion and the co-ordination of the ‘self’ in a current psychotherapeutic technique”), David G. Butt, Alison R. Moore, Caroline Henderson-Brooks, and Kristin Khoo home in on the psychological condition known as Borderline Personality Disorder, and the use of the Conversational Model of Psychotherapy in treating it. Their detailed analysis of exchanges between a patient and her therapist show the validation of the depth of her suffering, the root of the problem in repetition, and the reconstrual of the situation as a context of “agonizing beauty”. The grammar and cohesion of the discourse help the therapist sort out the language of the patient. This is a prime example of the appliablility of linguistics at the grassroots of everyday life.

SFL is a determinately empirical theory. Unlike some other more Cartesian, theory-driven approaches, SFL is based on observation of real language, and real language means texts, either written or spoken. Written texts tend to be monologic, while spoken texts are often dialogic. In the former the addressee is not usually present and in a position to give an immediate reply, whereas this is frequently the case in the in spoken language. Nevertheless, monologic spoken language may well be as old as language itself. The origins of a speech delivered by an individual with authority to a listening crowd are lost in the proverbial mists of time, and this type of, often political, speech is still current today. Technological advances during the late twentieth and early twenty-first centuries have multiplied the possibilities of monologic spoken discourse, producing a plethora of new forms, such as TED talks, 3-minute theses, and various forms of social media texts (Rossette-Crake 2019). In this special issue, Fiona Rossette-Crake takes up this issue (“‘Spoken & monologic’: modelling oratory, past and present, through the framework of SFL”). She begins by considering examples of what might be considered the traditional form of oratory with examples by Abraham Lincoln, Nelson Mandela, John Kennedy, and Barrack Obama. This provides a template against which she can measure the changes brought about by the creation of new digital monologic forms. This includes increased use of interpersonal resources and additive syntax, giving the discourse a more conversational feel.

SFL has always had sociological leanings, and thus sees language as a social-semiotic (Halliday 1978). Hence language cannot be divorced from its context, and the context is important for the understanding of a text. The metafunctions (ideational, interpersonal, and textual) can be related to the three variables of the immediate context, field, the ongoing activity of which the language is a part, tenor, those taking part in the language activity, and mode, the means by which the language is communicated (Halliday 1989). Beyond register, the wider context, or genre (though the terminology has at times wobbled) is described as a goal-oriented process (Martin and Rose 2008). This can perhaps be extended to include historical, social, and economic features of the wider situation. Banks (2017), for example, shows how the differing historical situations of England and France in the seventeenth century, and associated social and economic factors, influenced the editorial decisions of the early editors of the Philosophical Transactions and the Journal des Sçavans, and thus, ultimately the language of the items they published. In this issue, Y. J. Doran, J. R. Martin, and Michele Herrington (“Rethinking context: realisation, instantiation and individuation in systemic functional linguistics”) take another look at field, tenor, and mode pushing the boundaries of the categories of these resources to new levels of delicacy. The system network that they produce is illustrated with extensive extracts from two sources dealing with cell biology: a transcript of a classroom lesson and a popularized history of the science of cell biology. Their discussion leads to a much more detailed and delicate understanding of what register, in terms of field, tenor, and mode, entails.

SFL was in its conception a theory concerned with language. Language, however, while being the prime means by which humans communicate with each other, is not the only possible channel of communication. Non-verbal communication can be achieved by visual, aural, and gestural means. Systemic functional analyses have been applied to numerous forms of non-verbal communication, or multimodality. O’Toole (1994) applied it to the visual arts; van Leeuwen (1999) did so to music, and Kress and van Leeuwen (1996, 2001 have applied it to a wide range of non-verbal forms of communication. Starting from studies such as these, work on multimodality has become a major part of Systemic Functional studies, even taking in such unlikely forms of communication as that offered by children’s pram toys (van Leeuwen and Caldas-Coultard 2004). However, it is only fairly recently that an attempt has been made to apply the principles of SFL to dance. This is what Ariana Maiorani and her collaborators have done over the last few years. She has elaborated a Functional Grammar of Dance, based on the model of SFL, and with her collaborators developed the ELAN system of annotation. In “The Functional Grammar of Dance applied to ELAN annotation: meaning beyond the naked eye”, Arianna Maiorani and Chun Liu describe the Functional Grammar of Dance model, and apply it to a number of case studies showing not only how it works in practice, but also how it can be adapted to experimental forms of dance which may not fit into a classical model. Through this the systemic functional analysis of dance can be seen as a fascinating and integral part of multimodal studies.

Any attempt to cover the whole gamut of ways in which SFL is moving forward in the twenty-first century would have produced something of encyclopedic proportions, and probably still have been incomplete. Nevertheless, each of the seven contributions to this special issue shows one of the ways in which SFL is developing and is reaching out into new areas of enquiry. These contributions have covered the uses of systems in language learning, the use of Appraisal in the analysis of deontic modality in news reports, the relationship between phonology and thematic structure, the use of grammar and cohesion in the therapeutic treatment of Borderline Personality Disorder, the analysis of new monological oral genres, the production of a more detailed network for register going far beyond field, tenor and mode, and the production of a model for adapting SFL to dance. More than 60 years after Halliday’s seminal paper “Categories of a theory of grammar” (Halliday 2002 [1961]), these papers are evidence of the breadth and depth of current study in this model, and show its vivacity in moving forward in both theoretical elaboration and appliability. I hope that readers will not only find these papers of interest in themselves, but that they will stimulate and inspire them to continue or join this endeavour to attain ever greater understanding of human communication.