The Cambridge Handbook of Systemic Functional Linguistics 1107116988, 9781107116986

Table of contents :
Introduction 1
Wendy L. Bowcher, Lise Fontaine and David Schoenthal
Part I. SFL: The Model:
1. Firth and the origins of systemic functional linguistics: process, pragma, and
polysystem 11
David G. Butt
2. Key concepts and the architecture of language in the SFL model 35
Jonathan J. Webster
3. Semantics 55
Miriam Taverniers
4. The clause: an overview of the lexicogrammar 92
Margaret Berry
5. The rooms of the house: grammar at group rank 118
Lise Fontaine and David Schoenthal
6. Context and register 142
Wendy L. Bowcher
7. Intonation 171
Wendy L. Bowcher and Meena Debashish
8. Continuing issues in SFL 204
Mick O'Donnell
9. The Cardiff model of functional syntax 230
Anke Schulz and Lise Fontaine
10. SFL in context 259
Christopher S. Butler
Part II. Discourse Analysis within SFL:
11. Models of discourse in systemic functional linguistics 285
Tom Bartlett
12. Cohesion and conjunction 311
Maite Taboada
13. Semantic networks 333
Andy Fung and Francis Robert Low
14. Discourse semantics 358
J. R. Martin
15. Appraisal 382
Susan Hood
16. SFL and diachronic studies 410
David Banks
17. SFL and multimodal discourse analysis 433
Kay L. O'Halloran, Sabine Tan and Peter Wignell
18. SFL and critical discourse analysis 462
Gerard O'Grady
Part III. SFL in Application:
19. Language development 487
Geoff Williams
20. Applying SFL for understanding and fostering instructed second language
development 512
Heidi Byrnes
21. Language and education: learning to mean 537
Peter Mickan
22. Systemic functional linguistics and computation: new directions, new challenges
John Bateman, Daniel McDonald, Tuomo Hiippala, Daniel Couto-Vale and Eugeniu
Costetchi 561
23. Clinical linguistics 587
Elissa Asp and Jessica de Villiers
24. Language and science, language in science, and linguistics as science 620
M. A. K. Halliday and David G. Butt
25. Language and medicine 651
Alison Rotha Moore
26. Language and literature 690
Donna R. Miller
27. Language and social media: enacting identity through ambient affiliation 715
Michele Zappavigna
28. Theorizing and modeling translation 739
Erich Steiner
29. Language typology 767
Abhishek Kumar Kashyap
Index 793

Citation preview

The Cambridge Handbook of Systemic Functional Linguistics Introduction ​1 Wendy L. Bowcher, Lise Fontaine and David Schoenthal Part I. SFL: The Model: 1. Firth and the origins of systemic functional linguistics: process, pragma, and polysystem ​11 David G. Butt 2. Key concepts and the architecture of language in the SFL model ​35 Jonathan J. Webster 3. Semantics ​55 Miriam Taverniers 4. The clause: an overview of the lexicogrammar ​92 Margaret Berry 5. The rooms of the house: grammar at group rank ​118 Lise Fontaine and David Schoenthal 6. Context and register ​142 Wendy L. Bowcher 7. Intonation ​171 Wendy L. Bowcher and Meena Debashish 8. Continuing issues in SFL ​204 Mick O'Donnell 9. The Cardiff model of functional syntax ​230 Anke Schulz and Lise Fontaine 10. SFL in context ​259 Christopher S. Butler Part II. Discourse Analysis within SFL: 11. Models of discourse in systemic functional linguistics ​285 Tom Bartlett 12. Cohesion and conjunction ​311 Maite Taboada 13. Semantic networks ​333 Andy Fung and Francis Robert Low 14. Discourse semantics ​358 J. R. Martin 15. Appraisal ​382 Susan Hood 16. SFL and diachronic studies ​410 David Banks

17. SFL and multimodal discourse analysis ​433 Kay L. O'Halloran, Sabine Tan and Peter Wignell 18. SFL and critical discourse analysis ​462 Gerard O'Grady Part III. SFL in Application: 19. Language development ​487 Geoff Williams 20. Applying SFL for understanding and fostering instructed second language development ​512 Heidi Byrnes 21. Language and education: learning to mean ​537 Peter Mickan 22. Systemic functional linguistics and computation: new directions, new challenges John Bateman, Daniel McDonald, Tuomo Hiippala, Daniel Couto-Vale and Eugeniu Costetchi ​561 23. Clinical linguistics ​587 Elissa Asp and Jessica de Villiers 24. Language and science, language in science, and linguistics as science ​620 M. A. K. Halliday and David G. Butt 25. Language and medicine ​651 Alison Rotha Moore 26. Language and literature ​690 Donna R. Miller 27. Language and social media: enacting identity through ambient affiliation ​715 Michele Zappavigna 28. Theorizing and modeling translation ​739 Erich Steiner 29. Language typology ​767 Abhishek Kumar Kashyap Index 793

Preface

This volume has been several years in the making. It was first conceived of in 2013, when Cambridge University Press approached Lise Fontaine with the possibility of including a Handbook on Systemic Functional Linguistics as part of its series of Handbooks on Language and Linguistics. Recognizing this as a wonderful opportunity, Lise, Geoff Thompson, and Wendy Bowcher discussed the possibility of co-editing the volume. It was decided that Geoff would take the lead, and in consultation with various scholars, including Michael Halliday, he developed the conceptual framework for the book – a volume with a comprehensive, somewhat historical but also forward-looking overview of Systemic Functional Linguistics. Later, after Geoff’s untimely death, David Schönthal was invited to join the editorial team. As editors, we encouraged contributors to include both theoretical and practical details where possible – the latter noted by Halliday in personal correspondence as being an important part of the character of a ‘handbook’. Finding contributors to this volume was difficult, but in a positive way, as there are so many scholars around the world with expertise in the various areas covered who could have been approached. The final line-up, we feel, offers a wide scope of perspectives from a range of established and emerging scholars, some expert in more than the field of research which they have written about in this volume. We would like to take a moment here to thank all our contributors for the effort and expertise they have brought to this collection. Readers will notice that at the beginning of some of the chapters there is a note of tribute to several scholars who have passed away since the volume’s inception: Chapter 4 pays tribute to Geoff Thompson (see also the tribute to Geoff at the beginning of this volume), Chapter 7 to Bill Greaves, Chapter 23 to Johnathan Fine, and Chapter 26 to Ruqaiya Hasan. We felt it was important to include these tributes – to Geoff himself as the person who really got this project off the ground, and to Geoff and all the other scholars who have been such an important influence not only in the development of the

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:30:31, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.001

xxx

PREFACE

theory and practice of Systemic Functional Linguistics evident throughout the book, but in their commitment to furthering the field through their encouragement, generosity, and dedication to mentoring SFL scholars around the world. In the final stages of preparing the manuscript, we became aware that Emeritus Professor Michael Halliday, the founder of Systemic Functional Linguistics did not have long to live. It was not long after we had submitted the manuscript that he passed away, on 15 April 2018. It is a great privilege and honour to have included in the volume Chapter 24 ‘Language and Science, Language in Science, and Linguistics as Science’, which he co-authored with David Butt. After enquiring as to whether this was the last work that Michael Halliday penned, David Butt kindly offered to write a brief note on the nature of the co-authorship of this chapter. This note is presented at the end of Chapter 24.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:30:31, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.001

Introduction Wendy L. Bowcher, Lise Fontaine, and David Schönthal

I.1

The General Scope of This Volume

The Cambridge Handbook of Systemic Functional Linguistics presents a clear and comprehensive overview of one of the most appliable and socially progressive linguistic theories available today. As a social semiotic theory of language, Systemic Functional Linguistics (SFL) prioritizes language from the perspective of systems of meaning – how these systems are shaped, and, at the same time, how they play a role in shaping human social systems and how they relate to other systems of meanings within society. With its origins firmly grounded in a functional approach to language, SFL brings into theoretical rigour the concept of function in relation to language in a unique and robust way. That is, in SFL theory, function is ‘meaning in context’. It is not simply equated with ‘use’, but is considered a property of language at every level of description (cf. Halliday 1985: 17). Moreover, SFL places importance on its ‘appliability’. M. A. K. Halliday, the founder of SFL theory, has described a linguistic theory which is appliable as one which tackles problems and tries to answer questions – but questions that are asked, and problems that are raised, not by professional linguists so much as by other people who are in some way concerned with language, whether professionally or otherwise. There are large numbers of such people: educators, translators, legal and medical specialists, computer scientists, students of literature and drama, . . .; and it is their ‘take’ on language that is being addressed, at least to the point of clarifying what sorts of questions can usefully expect to be asked, and whether or not there is any hope of coming up with an answer. (Halliday 2013:128)

Unlike theories of language that separate ‘langue’ from ‘parole’ and which consider parole as a somehow ‘flawed’ version of language, SFL recognizes the ‘symbiotic relation’ between langue and parole:

Downloaded from https://www.cambridge.org/core. Stockholm University Library, on 30 Oct 2019 at 11:18:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.002

2

WENDY L. BOWCHER, LISE FONTAINE, AND DAVID SCHÖNTHAL

language as an already produced and an evolving system is predicated on the symbiotic relation between parole and langue – between process and its product – so we may claim that the system enables the efficacy of process while process fashions the very rules by which it attains this efficacy. (Hasan 2009:310)

This symbiotic relation is also seen in the relation between language and society, and is explained through the social semiotic perspective on language that SFL holds: The universal characteristics of parole – its orderly variation, its flexible regularities – are functional (Halliday 1970): they have their origin in the relations of parole to the community’s living of life, while at the same time, the various dimensions of a community’s social contexts of living depend on parole for their creation, maintenance, and evolution. Language as a social semiotic is predicated on this mutual relation between parole and social contexts. (Hasan 2009:309–10)

These principle perspectives of SFL theory underpin the contributions in the present volume, with these contributions covering the theory’s origins, architecture, key concepts, levels of analysis, and areas of application. Key terms are defined within the chapters, and key concepts are crossreferenced where relevant. Such cross-references to chapters within this handbook are given by reference to the author of the relevant chapter – for example, ‘see Butt, this volume’ to guide the reader (in this case) to Chapter 1. While the volume could be read from beginning to end, it is not necessary for readers to do so. Rather, readers wishing to understand a specific area of the theory or its application can refer to those chapters most relevant to their area of interest.

I.2

The Structure and Contents of This Volume

The chapters in this volume cover a comprehensive range of theoretical perspectives and applications of Systemic Functional Linguistics written by some of the world’s foremost SFL scholars, including M. A. K. Halliday, the founder of SFL theory. As editors, we have endeavoured to compile a volume that can be used primarily as a reference tool with descriptions and explanations of theoretical concepts and discourse analytical tools along with some exemplar analyses. There are also ample citations pointing readers to further literature wherein more detailed information and analyses can be obtained. Many of the chapters include a brief discussion of possible future directions in which research might be conducted or issues be further investigated and resolved. Ultimately, we hope that readers will not only become better informed about the various features of SFL theory and the value it can bring to solving societal problems, questions, and ambiguities in which language features, but that they may be inspired to

Downloaded from https://www.cambridge.org/core. Stockholm University Library, on 30 Oct 2019 at 11:18:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.002

Introduction

3

pursue some of the challenges and issues raised within the volume – be they theoretical or practical in nature. The volume is divided into three parts: Part I, ‘SFL: The Model’; Part II, ‘Discourse Analysis within SFL’; and Part III, ‘SFL in Application’. The following paragraphs summarize the contents of the chapters in each of these sections.

I.3

Part I – SFL: The Model

As the introductory part of the volume, Part I, ‘SFL: The Model’, has been designed in such a way as to cover the core features and terminology of the SFL framework. The organization of this part is designed in part to reflect the key perspectives on the theory. Opening the part, in Chapter 1, David G. Butt lays out in considerable detail the origins and history of how the SFL approach evolved. Halliday’s interest in a language-based linguistics is shown to derive directly from J. R. Firth, and, by better understanding Firth’s concerns, the reader gains valuable insight into Halliday’s development of SFL. A description of key terms in the SFL model is then given in Chapter 2, by Jonathan J. Webster. These two chapters form a necessary background for the more specific chapters that follow. In Chapter 3, Miriam Taverniers takes up the central concept of semantics and explores how it is conceptualized and modelled in SFL theory. In particular, she teases out the different conceptions of semantics within SFL. Importantly, she relates the key concepts of abstraction, patterning, and actualization to stratification and metaredundancy. Chapters 4 through 7, then, combine to provide a detailed discussion of four key approaches or perspectives on language in the SFL framework: the clause, units of the clause, context, and sound patterns. The multifunctional view of the clause is detailed by Margaret Berry in Chapter 4 on the lexicogrammar. Berry presents a concise analytical overview of the clause from the experiential, interpersonal, and textual metafunction. In Chapter 5, Lise Fontaine and David Schönthal present a critically engaged description of the units of ‘group’ and ‘phrase’. After reviewing the different units below the clause, they go on to challenge the distinction between the units of ‘group’ and ‘clause’. Context and its relation to text type is examined by Wendy L. Bowcher in Chapter 6, as she details the concepts of context and register within the model. Specifically, Bowcher discusses the history of these two concepts and their relation between one another, and reviews seminal SFL research on context and register. Chapter 7, by Wendy L. Bowcher and Meena Debashish, details how intonation and English tone groups are situated within the SFL framework. Not only do the authors offer a usable description of English intonation, they also raise important issues related to topics currently under debate and possible areas of future research.

Downloaded from https://www.cambridge.org/core. Stockholm University Library, on 30 Oct 2019 at 11:18:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.002

4

WENDY L. BOWCHER, LISE FONTAINE, AND DAVID SCHÖNTHAL

The final three chapters of Part I each step outside the standard SFL model in different ways. In Chapter 8, Mick O’Donnell explores aspects of the model which are under debate within the field. He outlines the main points of interest, showing where there is scope for significant contributions from researchers. Chapter 9, written by Anke Schulz and Lise Fontaine, presents the model of functional syntax as developed within the Cardiff Grammar. As a model with its roots in SFL theory, the Cardiff Grammar shares many of the same principles as outlined in the other chapters in this part. However, there are some important differences illustrated in this chapter. In the final chapter of Part I, Chapter 10, Christopher S. Butler situates SFL in its theoretical context in relation to other functional approaches, or what he refers to as ‘functional-cognitive space’. This chapter, based on a detailed comparison of sixteen different models, shows that while some differences are highlighted, there are also some interesting points of shared concerns.

I.4

Part II – Discourse Analysis within SFL

The second part of the volume contains eight chapters which present various discourse analytical tools developed within the framework of SFL theory. This part begins with Chapter 11, by Tom Bartlett, who first describes SFL from a discourse analytical perspective and then discusses some of the main approaches to discourse analysis within SFL. This chapter includes critical comments on some of the analytical approaches and effectively sets the background for the chapters which follow. Chapter 12, by Maite Taboada, focuses on cohesion and conjunction. This chapter describes Halliday and Hasan’s (1976) early work on cohesion, and includes a brief description of cohesive harmony (Hasan 1985; Khoo 2016), as well as work on rhetorical structure theory (RST) (Mann et al. 1992). In her discussion of conjunction, Taboada briefly points out some of the differences between Martin’s (1992) and Hasan’s (1985) descriptions. The chapter also includes a brief description of some areas in which Halliday and Hasan’s model of cohesion and coherence has been applied, such as in computational studies and foreign language teaching and learning. Chapter 13, by Andy Fung and Francis Robert Low, focuses on semantic networks as developed by Hasan (1996), but the chapter first situates this discourse analytic framework in relation to other semantic-level approaches within SFL. The chapter describes the basic unit of analysis, the ‘message’, and demonstrates the utility of this framework through an analysis of a mother–child interaction. The latter part of the chapter illustrates how semantic networks have been applied, with a focus on pedagogic and journalistic discourse. Chapters 14 (J. R. Martin) and 15 (Susan Hood) describe two related analytical frameworks: discourse semantics and the system of appraisal. Martin

Downloaded from https://www.cambridge.org/core. Stockholm University Library, on 30 Oct 2019 at 11:18:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.002

Introduction

5

briefly explains how his ‘discourse semantics’ differs from Halliday and Hasan’s (1976) concept of coherence and cohesion before elaborating on various systems within his framework. The second part of the chapter presents a text analysis demonstrating the application of the different discourse semantic systems and the value of this analytical approach in highlighting threads of ideational, interpersonal, and textual meanings throughout a text, and how these threads relate to each other. Towards the end of the chapter, Martin discusses the interpersonal system of appraisal, which provides a natural segue to Hood’s chapter on Appraisal. Hood first situates the appraisal system within SFL theory and then introduces the three main appraisal sub-systems: attitude, graduation, and engagement. The latter part of her chapter discusses some of the ways that the system of appraisal has been applied, ending with a discussion of the current trends in Appraisal research, including multimodal studies, research into legal language, and studies of identity and affiliation. Chapter 16, Diachronic Studies by David Banks, while not technically about a discourse analytic approach, is about a discourse analytic perspective – diachrony – and Banks illustrates some of the discoursal features that are focused on by SFL researchers whose data and analytical perspectives would fall within the domain of diachronic research. Chapter 17, by Kay L. O’Halloran, Sabine Tan, and Peter Wignell, covers multimodal discourse analysis, a particularly fruitful and growing area within the SFL theoretical framework, abbreviated as SF-MDA (Systemic Functional Multimodal Discourse Analysis). The chapter highlights the features of SFL theory which inform SF-MDA before presenting an exemplar analysis of a multimodal text, an internet webpage, using the analytical tools of SF-MDA. The last chapter in this part, Chapter 18, by Gerard O’Grady, outlines the relationship between SFL theory and Critical Discourse Analysis (CDA), particularly that developed by Norman Fairclough (1989, 2015). After tracing the development of CDA, O’Grady describes Fairclough’s methodology, pointing out its compatibility with SFL theoretical perspectives and analytical concepts. Towards the end of the chapter, O’Grady presents a critique of the criticisms of SFL-inspired CDA, and in the last part of the chapter he outlines some of the particularly productive areas of CDA research which make use of SFL theoretical tools.

I.5

Part III – SFL in Application

This third part of the volume presents several fields of research in which the theory of SFL has been applied, with the first three chapters (Chapters 19, 20, and 21) on different areas related to language development and learning. Geoff Williams’ chapter describes Halliday’s groundbreaking work on child language development (Chapter 19), and is

Downloaded from https://www.cambridge.org/core. Stockholm University Library, on 30 Oct 2019 at 11:18:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.002

6

WENDY L. BOWCHER, LISE FONTAINE, AND DAVID SCHÖNTHAL

organized around two main thrusts of the SFL approach: how research into language development informs the SFL theory of language, and how learning language goes hand in hand with learning culture and in developing one’s position and identity within society. Williams relates how this latter perspective, in particular, is demonstrated through research into mother–child interaction conducted by Ruqaiya Hasan and analyzed using her semantic networks (e.g. Hasan 1996). Chapter 20, by Heidi Byrnes, moves on to the application of SFL in the field of second language development. She first locates SFL within the general domain of second language acquisition research, and then discusses the application of several key features of SFL theory and L2 teaching and learning, noting, in particular, the value brought to the L2 teaching/learning environment of the SFL approach to the description of language and language in use, as well as its description of a lexicogrammar rather than the typical and ‘unsustainable’ separation of lexicon and grammar. She also raises the significance of SFL research into and description of grammatical metaphor in second language learning and teaching. In Chapter 21, Peter Mickan outlines how SFL theory has been applied within the field of general education. Mickan’s chapter covers work focusing on early childhood and primary school education, secondary school, tertiary education, and finally teacher-training and educational research. Underlying all these chapters is the principle that learning language and learning through language is a process of ‘learning how to mean’. In Chapter 22, John Bateman, Daniel McDonald, Tuomo Hiippala, Daniel Couto-Vale, and Eugeniu Costetchi note the long history of connection between SFL and computational linguistics, mentioning Halliday’s involvement in ‘some of the earliest attempts to achieve automatic translation systems in the 1950s’ and his key role in some of the ‘most well-known language-oriented systems to emerge in computational linguistics and Artificial Intelligence in the 1970s and 1980s’. The chapter discusses recent SFLrelated research, noting some of the challenges that a meaning-based theory of language poses for computational models, but also the distinct and far-reaching possibilities that SFL can offer the field. The next three chapters (Chapters 23, 24, and 25) are connected in terms of their focus on SFL in relation to science and medical research. Chapter 23, by Elissa Asp and Jessica de Villiers, concerns clinical linguistics. After briefly describing the field of clinical linguistics, Asp and de Villiers present an overview of research that falls within the SFL theoretical approach, including work on schizophrenia, neurodevelopmental and neurocognitive disorders, Alzheimer’s disease, and aphasia. They also discuss the significance of SFL-informed research and some future directions. The focus in Chapter 24, by Michael Halliday and David G. Butt, is ‘science’ and scientific language. They describe the part that language has played in the development of science and how scientific language, as a register, has come to construe knowledge and experience in a specific and

Downloaded from https://www.cambridge.org/core. Stockholm University Library, on 30 Oct 2019 at 11:18:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.002

Introduction

7

‘uncommonsense’ way, which they argue is in line with the social purposes of science. Another argument that the authors make is that science language, or ‘verbal science’, operates in a similar way to verbal art (see Miller, this volume) in that there is a ‘symbolic articulation’ – that is, conventional language choices realize metaphorical constructions of meanings to create novel understandings. Alison Rotha Moore, in Chapter 25, comprehensively reviews SFLinformed research into language and medicine. She describes the kind of health problems and medical contexts in which research has been conducted, such as HIV, emergency services, surgery contexts, and health curricula. She discusses the analytical tools used and then outlines the achievements SFL researchers have made in this field. The chapter also suggests some of the directions this kind of research can take and the possible knock-on improvements that could emerge in healthcare and in the healthcare system. The next chapter (Chapter 26), by Donna R. Miller, entitled Language and Literature, presents some of the most innovative work on the analysis of ‘verbal art’ available. Miller presents a historical recount of the field of stylistics and the place of Halliday’s and Hasan’s work in relation to this, noting the possible reasons why Halliday’s work has been acknowledged outside the circle of SFL scholars, whereas, surprisingly, Hasan’s has received little recognition. Several key influential figures emerge in her chapter, such as Jakobson and Mukařovský. The chapter describes Hasan’s systemic socio-semiotic stylistics (SSS) model and its value and insights for understanding the ‘art’ in verbal art, demonstrating that literature is not like other varieties of language, but is a special variety. In Chapter 27, Michele Zappavigna describes current work in the application of SFL tools to analyzing social media platforms such as Twitter, Facebook, Instagram, Weibo, etc. Specifically, she focuses on the way SFL tools can unlock the construction of ‘identity’ and ‘affiliation’ in such platforms, pointing out that this kind of investigation derives from Firth’s (1950) work on the language of persons and personalities and his concepts of ‘communion of feeling’ and ‘the user in uses’ (see also Martin 2009). The chapter also discusses issues involved in collecting social media data. Chapter 28, by Erich Steiner, focuses on translation studies and the usefulness of SFL ideas and concepts for theorizing and modelling the process of translation. His chapter compares translation with multilingual text production and interpreting, and discusses the relationship between translation, text variation, and paraphrase. It also includes a discussion on the SFL perspectives on equivalence, the translation of registers or text types, and the role of the translator in the process of translation. Steiner describes some of the SFL tools for text analysis that are relevant to translation and possible future directions of SFL-informed translation studies. The last chapter in the volume is Abhishek Kumar Kashyap’s chapter on language typology (Chapter 29). Kashyap briefly traces the development of

Downloaded from https://www.cambridge.org/core. Stockholm University Library, on 30 Oct 2019 at 11:18:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.002

8

WENDY L. BOWCHER, LISE FONTAINE, AND DAVID SCHÖNTHAL

the field of language typology, the kinds of questions that typologists ask, and some of the different theoretical approaches taken. The chapter covers the major contributions to language typology from an SFL perspective and applications of language typology to other fields such as translation, intercultural communication, and language teaching and learning.

References Fairclough, N. 1989. Language and Power. Harlow: Longman. Fairclough, N. 2015. Language and Power. 3rd ed. Abingdon: Routledge. Firth, J. R. 1950. Personality and Language in Society. The Sociological Review 42(1): 37–52. Halliday, M. A. K. 1985. Part A. In M. A. K. Halliday and R. Hasan, Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Geelong, Vic.: Deakin University Press. 1–49. Halliday, M. A. K. 2013. Putting Linguistic Theory to Work. In J. J. Webster, ed., The Collected Works of M. A. K. Halliday, Volume 11: Halliday in the 21st Century. London: Bloomsbury. 127–42. Halliday, M. A. K. and R. Hasan. 1976. Cohesion in English. London: Longman. Hasan, R. 1985. Part B. In M. A. K. Halliday and R. Hasan, Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Geelong, Vic.: Deakin University Press. 52–118. Hasan, R. 1996. Semantic Networks: A Tool for the Analysis of Meaning. In C. Cloran, D. Butt, and G. Williams, eds., Ways of Saying, Ways of Meaning: Selected Papers of Ruqaiya Hasan. London: Cassell. 104–30. Hasan, R. 2009. Rationality in Everyday Talk: From Process to System. In J. J. Webster, ed., The Collected Works of Ruqaiya Hasan, Volume 2: Semantic Variation: Meaning in Society and in Sociolinguistics. Sheffield: Equinox. 309–52. Khoo, K. M. 2016. ‘Threads of Continuity’ and Interaction: Coherence, Texture and Cohesive Harmony. In W. L. Bowcher and J. Y. Liang, eds., Society in Language, Language in Society: Essays in Honour of Ruqaiya Hasan. Basingstoke: Palgrave Macmillan. 300–30. Mann, W. C., C. M. I. M. Matthiessen, and S. A. Thompson. 1992. Rhetorical Structure Theory and Text Analysis. In W. C. Mann and S. A. Thompson, eds., Discourse Description: Diverse Linguistic Analyses of a Fund-Raising Text. Amsterdam: John Benjamins. 39–78. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, J. R. 2009. Realisation, Instantiation and Individuation: Some Thoughts on Identity in Youth Justice Conferencing. DELTA: Documentação de Estudos em Lingüística Teórica e Aplicada 25(SPE): 549–83.

Downloaded from https://www.cambridge.org/core. Stockholm University Library, on 30 Oct 2019 at 11:18:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.002

1 Firth and the Origins of Systemic Functional Linguistics Process, Pragma, and Polysystem David G. Butt 1.1

J. R. Firth (1890–1960): First Impressions and Paradoxes of Plain Statement ‘[M]eaning’ is a property of the mutually relevant people, things, events in the situation. Some of the events are the noises made by the speakers. (Firth 1964:111)

The ideas and principles of Professor John Rupert Firth are an essential source of what is important and distinctive about the development, after Firth’s death, of Systemic Functional Linguistics (SFL). Firth was the first Professor of General Linguistics in England (1944) at the School of Oriental and African Studies at London University (SOAS). He also headed the Department of Phonetics in London. Though first trained in history, between the outset of WWI and 1928, Firth had worked as a teacher of English and as a professor in British Imperial India (at Lahore, now Pakistan) as well as in Afghanistan and East Africa. He later returned to these communities to conduct further descriptions of languages. His students and colleagues are notable for the extent and depth to which they developed the study of languages of these regions, as well as languages of East and South East Asia. Firth emphasized the importance of de-Anglicization, and of looking back at one’s own language from the perspective of another culture. This is a form of ‘de-familiarization’ quite remarkable in a person who seemed a conservative Yorkshireman. Yet, as emphasized by Roman Jakobson after their two meetings, Firth shared with the pioneer of British linguistics, Henry Sweet (1845–1912), an ‘unusual courage to see the world’ with his own eyes ‘irrespective of the environmental usage, habit and predilection’ of conventional thinking (Jakobson 1966:242). For others, especially to phonemicists and morphologists in America, this ‘unusual courage’ appeared to be eccentricity, and a lack of

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

12

DAVID G. BUTT

commitment to being methodical or complete in his explanations. Thomas (2011:179–80) offers a recent synopsis of such reactions; but note the related misgivings expressed by one of Firth’s colleagues at SOAS (see Palmer 1968:4). The thesis of this chapter is that those emphasizing the eccentricity and ‘obscurity’ of the linguistic approach urged by J. R. Firth have misconstrued the rational basis of Firth’s motivations, and produced a ‘conventional wisdom’ that continues to obstruct our understanding of the potential of linguistics amongst the human sciences. Firth, in essence, sought a language-based linguistics. This is to say Firth would not follow other theorists who deferred to psychological, philosophical, or logical units as if such units offered some extrinsic grounding and escape from relativity – namely, the relativity inherent in one group of human speakers examining patterns of speech in another community. Furthermore, by contrast with Bloomfield, and then with Chomsky, Firth’s emphasis was on the linguist making statements of meaning (certainly not making statements against the possibility of such statements!). To these issues – the relativity of the language we use for describing languages and of the primacy of meaning – Firth was, in my view, punctiliously consistent. Both these issues are of increasing importance in human sciences: for instance, on the one hand, through a growing scepticism concerning universals in language and the commensurability of frameworks by which languages are described; and, on the other, through the growing emphasis on interpersonal meaning, that is, meaning over and beyond reference and truth conditions (see Section 1.3). Firth (1968:97) put his views succinctly: My own approach in general linguistics and especially in the study of meaning in purely linguistic terms dates back to about 1930 when the linguistic movement in philosophy was also arousing interest. My main concern is to make statements of meaning in purely linguistic terms, that is to say, such statements are made in terms of structures and systems at a number of levels of analysis: for example, in phonology, grammar, stylistics, situation, attested and established texts. I do not attempt statements about a speaker’s or writer’s thoughts and intentions, ideas and concepts – these are for other disciplines.

We might note at this point that there is little difficulty in citing what Firth recommended for linguistics, and also in establishing what he believed a linguist should abjure. Firth’s view is at the polar extreme from Chomsky’s claim that linguistics is a ‘branch of cognitive psychology’ (Chomsky 1972:1). Similarly, the idea that linguists might concede the domain of semantics to a philosopher of ‘intentions’, like J. R. Searle (1969), or that we pass pragmatics over to the maxims of a logician like H. P. Grice (1989), would have seemed a total abrogation of the roles and tool power of linguistics. For a start, such an approach assumes a diminution of linguistics, namely, that each of the strata of language could be contracted out and

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

13

managed separately and autonomously, with linguists left to ‘mop up’ the syntactic and lexicographical curiosities beside a few eccentric phoneticians. To explain my thesis concerning Firth and the role of linguistics in the human sciences, I will emphasize the Firthian notions of meaning and context of situation, with some inclusion of phonetics, phonology, and the issues surrounding ‘prosody’ – the latter supplying a metaphor for the pattern and process on all the levels of the linguistic order in the Firthian approach: viz. ‘semantic prosody’. I recognize that my views may be at odds with certain of the expressed views of very great scholars, some of whom had direct contact over years with Firth. Perhaps, being ‘in medias res’ can create its own difficulties in the sorting out of any history of ideas. On the other hand, I may simply be wrong on various matters due to my distance from the era, and due to my own limitations in linguistics. But one can only attend to the relevant texts, and strive to reconcile sources – working for a linguistic, scientific, and sociohistorical coherence. My reading is that such coherence has not been achieved, despite there being certain advantages for such reconstrual of Firth’s work in the current intellectual milieu. The argument here, through meaning and context, has not been an obvious pathway for positive evaluations of Firth: meaning and context of situation have even been a source of some embarrassment for those evaluating Firth’s ideas, for some of those inside the orbit of Firth’s influence, as well as for outside commentary (see Langendoen 1968; Lyons 1968). A recent overview by Kachru (2015:72–93) is a valuable addition to evaluations of Firth in that Kachru’s career spanned decades of Firth’s influence, including an exploration of the ways in which context might be invoked in linguistic description. A principled approach to context appears to be a primary step for Firth in the general problem of ‘turning language back on itself’, not merely an addendum or codicil. Firth emphasizes social process (the living of life) and the engaged, human body anticipating a structured future, the ‘conservation of the pattern of life’ (Firth 1957:143). The expression ‘Turning language back on itself’ was Firth’s characteristically concrete way of drawing attention to the ‘timeless and ineffable’ character of linguistic categories (Firth 1968:39), and of avoiding the veneer of technicality in the logician’s term ‘metalanguage’ (Halliday 2002 [1985]). Metalanguage too easily takes on a false status as if representing prelinguistic, or even presemiotic, categories. Firth’s formulation reminds us that meanings can only be rendered by another version of themselves: hence the ineffability. For example, to suggest that [+Animate] or [+Male] take us deeper into a semantic or logical atomism would be simply to overlook the ideologically loaded and relativistic value these terms have already in English (and certainly between speakers of different Englishes). The category terms that represent the complex of relations of a language should not be mental,

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

14

DAVID G. BUTT

acoustic, or even social primes that are extrinsic to languages. Linguistic categories are tools that bring out a pattern or regularity in relation to an immediate analytic purpose. They are therefore ‘ad hoc’ and not to be considered as existents in themselves. In fact, the positive and negative in Firth’s views are like points one can plot across a theoretical phase space (Cohen and Stewart 1995:198–210). Were one to miss the ‘co-ordinates’ of theory announced or implied in one paper, their repetition in an adjacent paper is awaiting any reader willing to track the ‘thematic motifs’ of this approach to the study of meaning. On the one hand, Firth can be alerting us to the idealization of the system in the work of Saussure and of Hjelmslev (the latter’s semiotic ‘calculus’; see Firth 1957:140); yet, on the other hand, Firth can be seen applying the notions of valeur and paradigm without compromise, in a strictly relational characterization of all linguistic units (see Palmer 1968:7). There is no contradiction in this; though, at first, paradox there may seem to be. On the Saussurean side, Firth rejects the superordinate structuralism implied by Saussure – the Russian critics have it right, we are told: Saussure presents the system as if ‘in rebus’ (Firth 1957:181, emphasis in original). In relation to Hjelmslev, Firth understands the motivation for a ‘semiotic calculus’ to serve semiotic endeavours in the way mathematics has served the physical sciences; but he rejects the practical necessity of what must be a step into greater abstraction or Platonism. Fresh formulations and succinct exempla (as in the systematic interpretation of Sitwell’s expression Emily coloured primulas; see Firth 1968:15–18) reward the careful, reflective reader with the step by step demonstration of Firth’s approach, and of the reasons for his rejection of psychological units in semantics. The economy and clarity of the demonstration affords an interesting comparison with Shklovsky’s attempt to explain the impact of Mayakovsky’s lexis in Russian poetry (see Shklovsky 1972:127–33). Firth’s ‘A Synopsis of Linguistic Theory 1930–1955’ offers a compressed summation, usefully examined in the context of the various papers of his junior colleagues (Firth 1962). In the Synopsis, Firth assists us in establishing the central tenets of his writings by quoting his own works with a careful ‘weighing up of every sentence’ (an assurance cited by his editor; see Palmer 1968:4). His editor was, in fact, Frank Palmer, a leading proponent of the ‘London School’. But Palmer (1968:4, emphasis in original), in a burst of frustration, characterizes the Synopsis as ‘even less coherent and consistent than de Saussure’s Cours de linguistique générale’ (one reason offered by Palmer is that the sub-divisions do not seem to align with any unified subject which each might purport to explain). Evaluations, even by those who worked close to Firth, fall across a spectrum of such bursts of frustration and bafflement: why not a fully worked out theory? Why no pages of the much anticipated book? How could one claim there is meaning made at all levels? How can linguistics operate without recourse to distinctions like thought in relation to

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

15

expression? Those at greater distance, in time and space, raise other issues – for example, the claim that it is inconsistent of Firth to state the anomalies of phonological practice using a phonemic alphabet. As an initial response to such reactions and interpretations of Firth’s ideas, it is advisable to consider two points: first of all, the problem of anachronism in seeing the past according to the fashion of speaking in succeeding decades; and secondly, to ask how a genuinely different way of seeing problems can achieve a rational evaluation in any science with a strong prevailing set of assumptions. Certain ways of seeing scientific and linguistic issues in particular have made it difficult for many linguists to construe what Firth was delivering with the brevity of ‘gnomic utterances’. Strevens emphasized, on the other hand, in his editorial comments (Firth 1964), that Firth wrote his papers for expert audiences of linguistic specialists, to whom the assumption of expertise was naturally extended. The problems arise with the beliefs a reader brings. If reading or hearing Firth in the late 1940s, the prevailing views were positivist and Bloomfieldean. ‘Language’ (Bloomfield 1933) espoused a behaviourist psychology and eschewed any ‘science’ in semantics – any science of meaning. Perhaps more problematic was the ideal of what could be a science at all. The discovery procedures of the era meant treating levels like phonology and grammar as if they needed to be managed autonomously: quite the inverse of Firth’s ‘meaning is made at all levels’. Bloomfield’s statement may not do justice even to Bloomfield’s own linguistic methods (e.g. his work on Menomini, including categories like ‘verbs of being’ (Bloomfield 1962:274); and ‘verbs of undergoing’ (Bloomfield 1962:298)). Firth’s reactions were patently clear, but bewildering to the adherents of Bloomfield: the phoneme is dead, and Nida’s work on morphology is ‘“nonsense” . . . added to “nonsense”’ (Firth 1957:170). A decade and more later, with Chomsky’s work, syntax is presented as formal and autonomous; meaning is deferred and passed over to philosophy; the individual is the domain of study; the assumption of a genetically based universal grammar is used against any evidence of the ‘typical actual’ of language behaviour; a language is the collection of sentences generated by the formal rules of the language; and intuitions of grammaticality become the least assailable form of linguistic testimony. As the adequacy of Chomsky’s assumptions in linguistics was variously contested, semantics returned but via formalisms from logic (entailment; presupposition), on the one hand, and from ‘speech act’ theory in philosophy, on the other. The study of context or pragmatics was also re-invented, but through a kind of ‘Alice in Wonderland’ idiom of maxims of conversational logic. These supposed reinvigorations of semantics and cultural analysis were two doses of North Atlantic ‘conventional wisdom’ which were, from their dominant spokespersons at least (i.e. Searle and Grice, respectively), without ethnographic evidence or any historical basis or cross-cultural complexity. Grice himself, in his idiom of gentle whimsy,

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

16

DAVID G. BUTT

is purported to have claimed that his maxims were just special versions of what a ‘decent chap ought to do’ (cited by Thomas 2011:219). Had not Malinowski and Firth already died, their reactions to these developments might well have ended them. But my point is that by the time Firth and Malinowski are relinquished to the status of ‘proto-pragmatists’ (Levinson 1983:xii), they have been occluded each decade for quite different reasons – firstly, due to assumptions that most have relinquished in linguistics; and secondly, because we now draw our semantic inspiration from speech act theory, Gricean maxims, the psychological load of relevance, or other sources extrinsic to the recording of actual linguistic exchanges. If we reflect on the decades of the dominant American linguistic theories after WWII, the idea of paradox can be invoked most strongly against each of the prevailing assumptions of each decade. Consider, for instance: the rejection by both Bloomfield and Chomsky of the study of meaning; the promotion of intuitive competence over observable behaviour; the dramatic genetic speculations about how recent language might have evolved; the idea that the basis of language might not have been for communication between people; and the final reduction of UG to recursion (and Merge) only (Chomsky and McGilvray 2012:16–20, 245). There were, then, at least on the face of things, both paradoxical statements and what might be called serious revisions. Yet, all these statements have been influential principles for a period in modern linguistics. One might be reminded then that dogma and rhetorical forcefulness characterize much of what achieves a high degree of visibility in science (see Brooks 2011: Chapter 6). Firth’s approach, with its emphasis on actual languages and variation, deserves fresh evaluation in the light of Halliday’s development of Firth’s polysystemic approach to linguistic description. Much as with the work of Sapir, important work can be brought back to a more rational evaluation.

1.2

Restricted Language

The efficacy and consistency of Firth’s linguistics can be seen clearly in his emphasis on ‘restricted languages’: the variation in relation to social purposes and varying social contexts. By setting the scale of analysis close to the ‘typical actual’ of social events, Firth’s approach helps resolve (or ‘dissolve’) many of the practical and theoretical conundrums of ‘doing linguistics’. These include, for example: (a)

whether linguists are accounting for a collective phenomenon or the ability of the speaking individual; (b) whether linguistics studies a unified relational system in an idealized social or psychological space, or a local, specific group of people with their own semantics and characteristic sociality; and

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

(c)

17

whether linguistics should attend to the dominant motif of a unity in which everything holds together (after Meillet’s dictum) or whether linguists should emphasize the evidence of change and variation (e.g. Halliday 1974; 1978; Harris 1981; 1990; Le Page and Tabouret-Keller 1985; Mufwene 2001).

Firth’s approach assists us in reconciling these dichotomies in what Halliday likes to describe as ‘eating our cake, and having it too’. By addressing register variation, the researcher takes up the participants as a reciprocating and dynamic system: ‘There can be no reciprocal comprehension if there is no situation’ (Firth 1962:30). Language is not about individual minds but the adhesion of significant actions. Being close to the anthropological stream of actions, there has to be an account of the social milieu, and of the way, for example, speakers construe the appropriateness of one channel over another. This is the germ of what Halliday’s context theory refers to as ‘mode’. In the first place, this is the issue of channel – how the text is carried and how the text is organized in relation to the demands of the channel. For instance, the choice of giving or demanding information may be a semantic generalization with direct relevance to the grammatical stratum (viz. the selection of mood in Western grammatical tradition). But this generalization needs to be related to whether one is acting face to face or in a more mediated mode of meaning. Similarly relevant to social meaning is whether the giving or demanding is processed and experienced in ‘real time’ or whether ‘delayed’ (i.e. offering the respondent time to weigh up a reply). Particular instances of giving or demanding information can only be accorded a linguistic value when contextualized, when embedded in the experiential matrix of specific social exchange.

1.3

‘Mutual Expectancies’: Pattern, Process, and Pragma

Firth’s principles appear to be all directed to bringing out the way meaningful practices in a culture ‘hang together’ or cohere through an ensemble of organized behaviours. From the access we have to the ‘aggregate of experience’ (Firth 1962:1), we can follow the social relations and their role in events down to the posture of body and voice in the reciprocation of a social situation. It is the expectancies that hold the social show together. These mutual expectancies can be regarded in probabilistic terms: when we get ‘A’, to what degree does that predict (or prehend) the presence of ‘B’, and of ‘C’, or of ‘B: C’, etc.? Clearly, this suggests we will be investigating aspects of text beyond the domains of normative grammar and lexicology, and beyond phonemic ideas of sound. Such an open sense of co-occurrence encompasses more too than the ideas that Firth passed down to us in collocation (the co-presence of word-like units) and colligation (the clustering of grammatical categories). Mutual expectancy opens up for

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

18

DAVID G. BUTT

consideration any regularity that carries a value in the text; and this breadth of conception leads Firth to step away from assuming the efficacy of conventional ‘units’, preferring to use the term ‘piece’. With this neutral ‘place-holder’ for sequence and order, Firth can propose statements that are not limited by the procrustean assumptions of classical, Western ‘syntagma’. One might ‘net in’ co-occurrences that are significant in characterizing the text, or the persona of an author, or the typical personae of a group with their restricted language, their trade language, their signs of solidarity or ‘insider’ talk, their total sociality, their semantic variety. Yet, it still needs to be asked, how then do we proceed under the guidance of these principles? To produce ‘statements of meaning’, the language needs the separation of its various polyphonic strands, of its ensemble of congruent levels. For Firth, the serial contextualization of a language piece is like the diffraction of white light through another medium: one sees that the totality of the white beam is actually a spectrum of waves all contributing at different scales, but with each essential to the patterning across all scales. Hence, ‘meaning is made at all levels’. And that cross-level relation must be handled according to ‘the differences that make a difference’ (Bateson 1982). The alliterative terms – patterns, pragma, and polysystem – supply three of the motifs that guide Firth’s approach to structure and function. Structure is that syntagmatic order of ‘mutual expectancy’ within a given social event – not the simple sequence of the structuralist’s fixed units. Function is the profile of relations that pertain to paradigmatic aspects of that order – essentially, where a ‘piece’ fits in all up and down the relations on one level and then, where crucial to the issue under investigation, across those relations from other levels that determine the value of the ‘piece’. For example, the expression He kept popping in and out all afternoon challenges segmental descriptions at the ranks in the grammar of verb/verbal group; but statements of meaning relating to this wording would also need to include that it only fits into specific social situations, between certain participants, face to face, and somewhere contextually between personal, confidential exchange and gossip. It is part of a network of relations that narrow down the potential of wording to what might be thought of as its place in a ‘restricted language’. Put quite simply, a polysystemic profile suggests that the piece could not appear just anywhere. It carries with it a penumbra of collocational and colligational patterns that reflect the habits of a social membership and of a personality: an instance of the ‘typical actual’ for a personality in a social moment. The linguistic importance and cross-disciplinary resonance of this observation can be brought out by again reflecting on ‘mutual expectancy’, and the more unusual, technical word from Whitehead’s ‘process’ interpretation of nature: namely, prehension (Whitehead 1979:379). Considered from a phonetician’s point of view, mutual expectancy encompasses much that linguistics up to today has overlooked about patterning in language.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

19

The spectrum of phonetic expectancies allows for ‘a closer approach to the social texture of language’ (Laver 1980:5). Laver, in quoting Firth and discussing variation, emphasizes all ‘recurrent, patterned, phonetic activity’ (Laver 1980:5) that characterizes a speech community, not just the distinctive contrasts which purportedly define its relation to a natural language. Laver’s approach, grounded in the phonetic leadership and teaching of David Abercrombie, illustrates the systemicization of voice quality and what Firth referred to as a total vocal ‘posture’ – a ‘vox’ – which carries the ‘habitus’, the patterns of living expressed by the posture: ‘Surely it is part of the meaning of being American to sound like one’ (Firth 1957:192). This connection to forms of life is explicit in Firth’s numerous footnotes to Wittgenstein’s later work (Glock 1996:329). Then there are the mutual expectancies between the complex of variables that produce the semantic clusters of different registers. Halliday (following Jean Ure and others) uses register as the key term for a language variety according to the immediate use. Firth used ‘restricted language’. By this he emphasized that nothing modifies the meaning of a wording more than a change of context. Language working to a particular purpose in a conventional social event takes on its valeur from the habitual appearance in that restricted environment. He even suggested that dictionaries could be organized around such restricted languages. This would mean that lexicographers would not need to think of every definition of a word as isolates, but that the range of a word or piece could be displayed by ‘serial contextualization’ (a term that could be applied to all levels of analysis, thereby doing the theoretical work of Saussure’s valeur). While register and restricted language are ways of describing the variation in meaning-making practices, in semantics, such variation is realized in choices at different levels of language. Such choices include probabilities of co-occurrence: e.g. hypotaxis vs. parataxis; transitive ‘relations by place and order, by particles and by case’ (Firth 1962:18); cohesion devices, regular or bunched in their dispersion; and the various ‘filters’ and ‘lenses’ of the ‘time camera’ of tense – that is of ‘grammatical time’ based on what linguistic systems actually ‘do’: Confining ourselves to English as the language of description, let us face the facts and admit that such words as time, past, present, future and all the rest of the ‘temporal’ nomenclature, have been employed with gross carelessness to describe notions supposed to characterize the verb . . . Each language has its own means of handling ‘experiential’ time, has its own ‘time-camera’ so to speak, with its own special view-finders, perspectives, filters, and lenses. It is childish to draw excessively over-simplified linear diagrams to deal with such linguistic structures and systems. The point is they are not timesystems but linguistic systems. (Firth 1962:19)

So the same complex layering of continua in phonetics was to be seen at other levels mutatis mutandis. Firth asks, for instance, where are the

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

20

DAVID G. BUTT

morphemic exponents of the grammatical roles of the Latin word pedibus (feet, ablative: on foot)? Namely, how should the word be divided to display the meanings it contributes? A single word isolate such as Latin pedibus might have to be considered in a sentence structure in which the categories in colligation would include gender, number and person, and the noun-substantive itself. But where are all these if the grammarian looks at the word pedibus itself? The exponents of the categories are ‘cumulative’ in the word and also discontinuous in the sentence. . . . The mutually expectant relations of the grammatical categories in colligation, however, cannot be regarded as necessarily having phonological ‘shape’. (Firth 1962:14)

The extension of meaningful relations through mutual expectancy also motivates the concept of semantic prosody, a concept that addresses many difficulties in text analysis and linguistic corpora, particularly in relation to any account of interpersonal meanings. In this concept we also see the lexical and grammatical discussed in terms of collocation and colligation. In computational linguistics, collocation becomes powerful because it can be easily given mathematical specification. Firth’s insight: you can ‘know a word by the company it keeps’ is well illustrated by corpus work that can be directed to diachronic and/or synchronic goals. The grammatical counterpart – colligation – is more elusive: I take it to net in the proximities of grammatical categories over and beyond the order that they require by agreement. This may be a new domain of research addressing the implicit patterns of choice across stylistics and discourse analysis. We might claim, as a result of restricted language and register studies, that we do not speak a grand system which we all hold mentally in a common collective: rather, we work through a patchwork of differently habituated social exchanges, all variously local or strange, with variable boundaries on to other exchanges. We are marked by our memberships and habits in relation to these exchanges. Hence, our personalities are on display in the ‘roles’ that we inhabit and by the verbal lines that we know and contribute. Firth reminds us, then, not only of our vocal posture but also of the need to know our rhetorical roles. In and out of various group memberships, we each constitute, metaphorically, our own ‘figure’ of speech. Thus, the collective and the personal are reconciled through semantic enquiry (Firth 1957:177). As we move from phonetic perspectives to contexts, semantics, and grammatical expressions, we have the tool power to consider problems that appear extrinsic to linguistics, but which, in fact, have a direct bearing on the forms that language takes around us, and within us. Class and ideology take on an extended significance, a significance brought out into plain sight by the statistical treatment of messages in the semantic variation work of Hasan from the 1980s (see Hasan 2009). The evidence is drawn from principal component analysis of consistency of choice across up to fifty semantic parameters. These parameters, in their combinations and

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

21

different configurations, represent strategies within interaction that clearly aligned with social class differences in the Australian urban context. In fact, the semantic parameters that distinguished class are illustrations of what Firth called a semantic prosody: a motivated consistency across diverse resources in the construction of meaning. Here the use of ‘mutual expectancy’ can be further appreciated as a radical theoretical motif. It forces us to consider not only ‘the company that a word keeps’, nor only the overt conventions of grammatical agreement and government (the theoretical compass points for navigating in synthetic languages like Latin and Greek), but all the ensemble effects that enact the social positioning of speakers – the fabric of discourse over and beyond the grammatical ‘rules’. Hasan’s central concern was how speech realized the social roles of class and gender. Her data were indicative of Bernstein’s social theory, with social roles being part of a lived experience coded by implicit expectations of how the world ‘hangs together’, and of how one plays a role within the conditions offered by class membership. The semiotic foundations of class, of codes, and of control were in mothers’ talk, particularly in relation to a continuum from individuation to the assumption of shared ‘local’ values. This is to say, following Firth, that there was a deep semantic prosody in the ways that mothers projected the child’s place in the world back to the child. One prosodic consistency was of a world of individuated access to evidence and the ‘appearance’ of negotiated authority; the other was of a strong acceptance of a settled order in which knowledge and authority came unquestioned with natural relationships: viz. in deference to age and intrinsically with the status of parents. This continuum was based on a configuration of each mother’s pattern of choices – that is, the degree to which their messages encoded the sources and significance of differing points of view (including the point of view of the child involved). As the explicit framing of ‘point of view’ diminished in mothers’ talk, the assumption of a world with a fixed order became more insistent in the thousands of messages that flowed around and through the child. It is hard to think of a more Firthian demonstration of the diversity of our modes of meanings: the ‘collective consciousness’ of Saussure’s social fact is not a unity in which all the parts hold together (after Meillet’s dictum). Rather, language becomes a means for engineering and maintaining covert difference around positional authority in ‘the living of life’ (Hasan, personal communication). By being continuous between human external and internal experience, we can see in the mothers’ talk a potent ideological medium. On this last point Firth (1968:199) quotes Whitehead (1938): The human body is that region of the world which is the primary field of human experience but it is continuous with the rest of the world. We are in the world and the world is in us.

The concepts of ‘mutual expectancies’ and Hasan’s semantic variation are congruent with the co-presence (or absence) of forms suggested by Whorf

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

22

DAVID G. BUTT

(1956:83) in the notion: ‘configurational rapport’. Essentially, this configuration means that certain elements of language co-pattern because they are part of a wider, insistent ‘fashion of speaking’ in the culture (Whorf 1956:158). It is by such ways of meaning that a culture manages experience. Whorf emphasized covert categories – those distinctions that are not overtly coded in the grammar, but which display a reactance – a consequence – in discourse texture. One needs experience in the culture to interpret these categories: simple examples might be what constitutes a female name in varieties of English, or what is a feminine name in forms of Chinese. Whorf also drew attention to cryptotypes – those aspects of the language that are restricted or influential for no obvious motivation. Consider such issues for English: for example, certain verbs that cannot be used transitively; the lack of a passivized form of the verb be in English, even though other verbs can substitute for be with full passive potential (see the ‘represents / is represented by’ test in identifying clauses in Halliday and Matthiessen 2014: Chapter 5). These phenomena may be thought of as analogous to the ‘hidden variables’ theory in physics, as discussed by Bohm (1952a;b): essentially an ‘implicate order’ behind the explicate order of phenomena accounted for by normative science (in this case, normative grammar).

1.4

So What Would Be a Firthian Approach in Situation, and for Sound?

1.4.1 Context of Situation In bringing attention to work developed before 1960 – work that merits attention and further development – I will draw on the contextual study by Mitchell (1957). One needs to work with ‘[p]rocesses and patterns of life in the environment’ that can be ‘generalised in contexts of situation . . . Order and structures are seen in these, and after examining distribution in collocations, “pieces”, words and morphemes, [these] may be arranged in ordered series, resulting in systems and sets of systems’ (Firth 1968:24). Consequently, one is not seeking a mere enumeration of environmental details (what Hasan distinguished as the ‘material situational setting’). There have to be some criteria of relevance, what in the environment leaps out to have a consequential role in the direction that meaning takes. Mitchell himself takes Firth’s use of ‘technical’ (as in a ‘technical language’) and turns it into a further, useful distinction: technical terms in Mitchell refer to those aspects of patterning which give rise to the particular character of a situation, while non-technical terms are relevant but generalizable across a number of contexts. Firth (1968:200) rounds off the issue of abstraction with: The abstraction here called context of situation does not deal with mere ‘sense’ or with thoughts. It is not a description of the environment. It is a set of categories in ordered relations abstracted from the life of man in the flux of events, from personality in society.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

23

Mitchell uses three ‘attested’ texts of sales in markets and shops in Cyrenaica. From these, he exemplifies the most significant contextual distinctions made by Firth. In particular, the final discussion of negotiations over a fast horse – a prestige purchase, involving long-standing traditions and community expertise – illustrates the practical necessity of beginning at context rather than treating context as a final form of pragmatic ‘tweaking’. Mitchell also raises a number of problems, issues to which I would also attest from my own collaborations in surgical care and psychiatric contexts. These problems include the following: the absence of talk on the activity; the presence of oblique talk; and the difference between talk on the job and talk explaining the job; and (I would add) talk that runs a bulletin and guide which, as a commentary, brings all participants to a shared understanding of how the process is progressing. Mitchell certainly emphasizes ‘mutual expectancy’ and even cites prehension (Mitchell 1957:39, 49, 54, 55), noting the cumulative effects of unfolding connections between relevant words and actions. In SFL, following Halliday, the process of choices unfolding has been referred to as logogenesis. The term encompasses the changing values in the text as new choices accord with what has gone before, and as these choices direct the changing expectancies as to what is to come in the light of the most recent choices in the potential. Consider how generic elements or stages are related to both order and succession (Mitchell 1957:43, 47). The link between ‘habitual’ patterns, collocation, and expectation is expressed by Mitchell (1957:55). An interesting distinction is also made plain between personalities and persons (Mitchell 1957:36–59): the difference is between the role you are playing and the physical being involved. Firth emphatically rejected the idea of individuals in language; but he invoked ‘personality’ (as did Sapir). These distinctions lead to Mitchell’s four-column tabulation of Text, Translation, Personality, and Stage. At other points there are signs of some emergence of the more recent style of Pragmatics: ‘essential conditions’ (Mitchell 1957:36) suggests J. L. Austin’s influence (although Austin’s 1955 lectures were not in print until 1962); and mutual expectancy appears to encompass overt connections like those that were later referred to by ‘adjacency pairing’ (Mitchell 1957:59). Mitchell (1957:34) is working out the ‘complex pattern of activity’ pertaining to a central pattern (the way ‘a line needs to be distinguished as to its role in either a rectangle, or as the radius of a circle’). In relation to the sale of the horse in the market, the exchange takes semantic directions that the outsider is unlikely to predict or construe. The quoting of a quatrain of traditional poetry is not unique to the Bedouin stages and styles of negotiation – many negotiations across cultures invoke gnomic sayings or apothegms. But the complex relationship between the horse’s speed and the rider’s good fortune, on the one hand, and the whorls on the horse’s flanks, on the other, are opaque for anyone not enculturated to Bedouin horsemanship. These issues are detailed (Mitchell 1957:70–1) and are strikingly parallel to Malinowski’s example of canoe racing, relying as they do

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

24

DAVID G. BUTT

on traditional formulas, the specific locale (mountain and water for Malinowski’s example), and the activity (competitive paddling). Some contact I have had with East Timorese cultures (see accounts of Tetum from Ainaro district, especially in de Corte-Real (1998); but also see Hull et al. (1998)) have produced significant parallels. For example, what we might categorize as texts of greeting or official welcoming have a complex multivalence: yes, they are performed as a ceremonial greeting to officials from another community; they typically involve a senior male ‘keeper of the traditional knowledge’ – a ‘lianai’. But they also urge the power of the local community in an intimidating manner. In short, they threaten or warn against any false action in relation to the local community. Footage of an Indonesian government representative ‘greeted’ in this way (before Timor-Leste achieved independence from Indonesian authority) has a very unsettling effect. The strikingly elaborate war canoes of the Trobriand Island communities trading in their precious shells or ‘kula’, as discussed by Malinowski and others, appear to have been similarly ambivalent: a situational meaning of respect of ritual and of ‘Don’t mess with us’. In relation to East Timor, de Corte-Real and I attended a meeting with an Indonesian military commander in Dili who directed de Corte-Real most civilly to stick to the literary aspects of the performance genre, not to the political and nationalist implications. The multifunctional meaning of action and wording was clear to the officer. What Mitchell achieves is a ‘context first’ outline of linguistically motivated categories. Mitchell does not here systemicize the sound or other (intervening) levels of analysis, although various issues of grammar and numerous observations on rendering the speech phonetically are noted throughout. But, unlike the strong accounts we have of prosodic analysis in published anthologies,1 illustrations of context categories have not been pursued to a similar degree. This is an indication of some bafflement amongst those working with the neo-Firthian traditions. Interest in the systemicization of context has produced significant work, especially that led by Hasan’s consistent attention to the problems of turning ‘context of situation’ into a concept with greater tool power. Now there are a number of nodes of contextual modelling, each distinctive: in the UK (e.g. Cardiff ); in Hong Kong; and in Germany; as well as in Australia. Firth is clear on the point of systemic representation, including in relation to context. Language needs to be dispersed, like white light, across its spectrum of different ‘wave lengths’, and addressed as a polysystemic phenomenon. Such an approach appears to yield the most practical descriptions since there is something universal about language: its organization as a system of systems:

1

See articles by Allen (1953); Halliday (1959; 1963); Henderson (1987); Mitchell (1975); Palmer (1970); Robins (1957).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

25

I have suggested that language is systemic. . . . We may assume that any social person speaking in his own personality will behave systematically, since experienced language is universally systemic. Therefore, we may study his speech and ask the question, ‘What is systemic?’ (Firth 1957:187)

1.4.2 Sound While Firth and Halliday are both cited for systemicization in phonology and the semantics of intonation, the consistency of their work (along with the work of Abercrombie at Edinburgh) is notable when viewed against the current trends in phonological and phonetic debate. In an attempt to set the record right, Palmer emphasizes that Firth’s distinctive approach can be tracked in publications from 1934 to 1937 (see Palmer 1970:x–xi). The polysystemic principle was argued in Firth’s (1935:51) discussion of Marathi, where he points out that eight /n/ sounds needed to be recognized as linguistically and functionally distinct, and therefore not the same unit even if seemingly identical phonetically. This was part of the broader contextual principle: namely, that one had to recognize not only the phonetic contexts, but also ‘lexical and grammatical functions’ (Firth cited in Palmer 1970:xi). The value of an item was dependent on the systemic character of ‘recurrent contexts’. So too, there were the issues of y and w prosodies; the urging of the importance of the syllable and of extended phenomena in phonological analysis; and the notion of ‘articulation types’. Palmer recommends the completeness of Henderson’s (1949) analysis of Siamese in that she shows a full hierarchy of prosodies: in sentence; sentence parts; polysyllables and sentence pieces; in syllables and syllable parts; and in consonant and vowel units. Firth’s arguments against rigid phonemic methods concerned the unreality of segmentation and of discovery procedures. These arguments were bewildering to linguists of his day, despite the vigorous critique of the phoneme concept by Twaddell in the USA (see Anderson 1985) and the questions raised by other linguists in the USA (Palmer (1970:x) cites Harris and Hockett in this regard). Today, views homogeneous with those of Firth are part of the ‘natural’ background in the study of phonetics. Since Firth emphasized that the /d/ that was word initial was not the same as /d/ word final since the two acoustic elements operated in distinct systems (i.e. with differing ‘valeurs’), then we might regard this as a rigorous application of relational thinking (suggesting some affinity with the extreme relationalism of Hjelmslev, a linguistic alter ego with whom Firth enjoyed frequent exchanges and debates). We can see that Firth’s declarations against Saussure’s work – namely, that it creates a system ‘in rebus’ – need to be seen alongside what may be the strictest application of Saussure’s relational universe (see the comment on valeur by Palmer 1968:xx). The problems urged by Firth in relation to phonetics and phonology are usefully explained and criticized in an evaluation of Linguistic Thought in

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

26

DAVID G. BUTT

England 1914–1945 (see Love 1988:151–62). After setting out Firth’s approach, Love argues for the efficacy of (and necessity of ) our use of abstractions and ‘scriptic conventions’, contrary to Firth’s attitudes. Without such conventions, Love claims, Firth is unable to separate the ‘preordained structure’ from the ‘unique-ness’ of an utterance (Love 1988:163–4). More significant is the claim that Firth’s principles would open ‘the floodgates’ to comparisons right across a language (Love 1988:162) without any counter principle to limit analysis. I disagree with Love’s ultimate judgements: first of all, Firth’s focus on ‘restricted languages’ suggests that the ‘floodgates’ to the whole language are an artificial or hypothetical, rather than a practical, concern. Second, one needs to keep in mind that the machinery of analysis is ad hoc for Firth; yet, all discussion has to start somewhere. Consequently, yes, you start your discourse where you meet the discourse of others – hence, grammar in Europe will set out from terms like transitivity, mood, and tense, although the end of discussion may be in changing the valeur of these concepts, or with the development of other categories altogether. In phonology, this may mean starting out from the ‘scriptic conventions used for stating and citing utterances’ (Love 1988:164). I would suggest that the problem for Firth is then how to improve upon the specific account, not how to ensure an absolute final account, or to supply exhaustive paradigms, as Love appears to assume. Like parenting (and probably all sciences), it is a case of achieving a ‘good enough’ or improved result. Love’s final remarks do not give sufficient weight to the pragmatist implications of Firthian ‘ad hoc-ery’ and to the role given to the operational status of linguistic techniques. The role of the term ineffability also seems to me undervalued for its significance (see by contrast the emphasis in Halliday 2002 [1985]). It is surprising to see Firth being criticized by Love for underestimating the difficulties involved in ‘turning language back on itself’ when so much of Firth’s writing is directed to bringing out the implications of just those difficulties. Firth’s insights attempt a ‘renewal of connection’ with the facts and their contiguities in an actual language – much as Whitehead originally conceived of intellectual process as having to take flight up to the thin air of abstraction and then to return to a grounding in the realia of the experiential matrix that is our primary resource for demarcating sense from non-sense.

1.5

Posture and the Communicating Body

Firth consistently refers to the body going out to the world – of the ears being active in exploring sounds; of the brain as a guide to acting and moving in ‘situations’; of the human as engaged in the pursuit of a ‘joy’ at structure. Firth sets out in linguistics with the explicit assumption of this

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

27

active engagement with the world, quite the opposite of the passive, tabula rasa notion often foisted upon empiricists in human sciences. Firth’s citing of sources, however, needs to be regarded for its own cultural context and the ways in which Firth’s approach is oriented to current thinking. Firth cites Charles Sherrington’s work for its emphasis on movements and proprioception as ways into understanding emotions – the ‘felt ME’ – and as the way of building relations with other humans. Firth responds positively to Sherrington’s view that the brain is a ‘manager of muscle’, a view in accord with the fact that current analysis suggests that the cerebellum, although only 10 per cent of the brain’s volume, encompasses more than fifty per cent of its neurons. More centrally, Firth argues for the continuities between human inner and outer worlds, with the brain creating a form of ‘mutual grip’ between the world and us. This side of Firth’s thinking is congruent with the position of neuropsychologists like Trevarthen (1998; see also Panksepp and Trevarthen 2009). Trevarthen, having worked with Sperry, Bruner, and Halliday, has argued for a related going out to the world in neonates, a bridge-building through intersubjectivity to person-ness. Trevarthen (in Stensæth and Trondalen 2012) summarizes his own thinking thus: In 1974, emphasising the rhythmic properties of expressions, I said, ‘when a newborn is alert and coordinated, its still very rudimentary movements have, nevertheless, the pace as well as the form of activities such as looking, listening, and reaching to touch, from the start. This can be perceived and reacted to unconsciously by an older person. As the person approaches the infant, acting gently and carefully as people tend to do instinctively to a baby, then all the emanations from this approach have rhythmical properties that are comparable with those inside the movement-generating mechanisms of the infant’s brain. From this correspondence I believe the infant builds a bridge to persons.’

Firth’s concepts of ‘posture’ and ‘person’ suggest a general account of a communicating body; and certainly this is an area in which the emphatic statements of Firth go under-interpreted. Contemporary discourses on ‘embodiment’ are a reprise on ideas emerging from 1906 to the 1950s. From 1960 to 1990, a narrower metaphor for brains became dominant in the developments of artificial intelligence: the image of a disembodied computer engaged in the perhaps 10 to the power of 120 moves of a game of chess. The contemporary emphasis on the body is hardly an innovation, but a return to an earlier engagement with ideas recently occluded by that ‘starkly narrow’ agenda in American linguistics during the period after the 1950s (Levinson 1983:xii). Narrow claims concerning the task of linguistics – claims concerning universal grammar, mental organs, poverty of the stimulus, autonomous syntax, genetically based competence, etc. – exerted a hegemonic control over what could be declared ‘science’ in linguistics (see e.g. Ellis 1993).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

28

DAVID G. BUTT

1.6

Prehension in Science and the Arts: The Zeitgeist

Prehension is a term central to Whitehead’s theory of natural, cosmic cohesion: the interconnectedness between all ‘occasions of experience’. Aspects of this approach attracted numerous important scientists (e.g. C. H. Waddington in biology; Joseph Needham in bio-chemistry, also a specialist on science and civilization in China; Ludwig von Bertalanffy in evolutionary systems thinking; Charles Hartshorne in philosophy). For most scientists such a notion could have been judged as not more than a promissory note on the emergence of influence across vast spaces, between different scales of events, and in a continuum between outer and inner experiences. For the claims across so many phenomena, Whitehead’s work has even been scoffed at for obscurantism and ‘metaphysics’ (a term of abuse between many who claim to be in natural sciences). Curiously, we all now live with experimental confirmation of electron ‘entanglement’, that is, between sub-atomic particles at ‘out of contact’ distances (Aspect et al. 1982; DeWitt 2004:288–305; Whitaker 2006). Consequently, the idea of ‘negative prehension’ – Whitehead’s idea that each region of the universe bears the influence of the configuration of the rest of the universe – should be revisited as prescient, not absurd. It is interesting to note that a leading theorist of prosodic theory (Allen 1970 [1951]) actually used ‘action at a distance’ as an argument against cases of assimilation and dissimilation in phonemic theory. This illustrates, at least, that linguists did observe the epistemological aspects of physical theory in reflecting on linguistics as a science. The metaphor of prehension may seem a curious choice for Firth – a philosopher’s term cited by a linguist who is vehemently against ‘arm chair’ philosophizing in linguistics. Yet here too, there is significant consistency. Whitehead’s project was nothing less than reversing the misleading consequences of atomisms and ‘fallacies of misplaced concreteness’. He was Chancellor at London University in the 1930s. Did Whitehead and Firth converse on this matter? – I do not know. But Firth took the title of Whitehead’s 1938 book, ‘Modes of Thought’, into his own writing: as ‘Modes of Meaning’. Firth’s oft-used expression for applying the contextual framework was to extract ‘pattern’ from ‘the general mush of goings on’; again, both key idioms from Whitehead’s writings between the World Wars. Within the linguistics of text, the idea of prehension is well motivated and practical. Firth’s own search for differences which make a difference semantically only needed computers to track ‘latent patterns’: an approach taken up indefatigably by John Sinclair, a supervisee and colleague of Halliday. Sinclair’s work (a real legacy in its own methods and style) was much guided by Firth’s concept for seeking new patterns in lexis and new pieces in grammar, and by ‘semantic prosody’ (Sinclair 1966; see also reference to ‘latent patterning’ in Sinclair and Coulthard 1975). Every choice in a text sets up a new probability about what is impending, and at the same time

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

29

modifies the value of what has preceded, namely, the choices that have brought the logogenetic unfolding to this point. The expansion of computer-based corpus studies has not only activated Firth’s concepts of collocation and colligation, but has also given a practical, probabilistic character to register studies and for ‘restricted languages’ (Halliday and James 1993; Teich 2003; Bartsch et al. 2005; Matthiessen 2015b).

1.7

Conclusion

Unlike most other influential theories of language and linguistics with active proponents since WWII, Systemic Functional Linguistics (SFL) is continuous with linguistic history: its proponents do not claim to have made a dramatic or ‘revolutionary’ break, but rather, a rational integration of linguistic experiences including an emphasis on languages beyond European studies. Furthermore, unlike the formal theories of our era, and quite contrary to the way it is referred to from the ‘outside’, SFL involves no neologisms or terms without deep roots in linguistic or pedagogical or rhetorical traditions. Halliday’s studies of Chinese have always provided a lens through which meanings in European and world languages are brought into a depth of field not characteristic of theories that set out from the Western classical tradition. Consider, for example, the assumption around Subject deletion or ‘Pro-drop’ in generative linguistics. Those working in SFL need to be clear about the potential, and the limits, of their tradition without the distraction of boundary disputes or apocryphal sketches of linguistic history. For limits there are. It might be argued, for instance, that despite the efforts of Halliday and Hasan (1985) with their work on context networks, SFL practitioners as a cohort have not delivered on the explication of the context to meaning relation. These longcited notions may not have been brought up to what Matthiessen (2015a:178) refers to as ‘an industrial-strength representational system’ – the concepts have not been doing sufficient work in the discourse descriptions of SFL. They have not added sufficient value where it might have made an important contribution, for example, to narrowing down the referential domain of computational linguistics. Nevertheless, the notion of ‘situation’, along with the idea of language variation according to use (i.e. register), have already had a significant influence in enlivening linguistics and language education in England, Australia, and elsewhere. Then there is the work on semantic variation (Hasan and Cloran 1990), on Rhetorical Structure Theory (see especially Mann and Matthiessen 1983), on computational modelling developed in a number of centres (in Germany, at Macquarie in Australia, and by two generations of the Cardiff group of SFL researchers), as well as the general linguistic approach to the mind set out and illustrated in Construing Experience through Meaning (Halliday and Matthiessen 1999). All are examples of

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

30

DAVID G. BUTT

the robustness of the SFL model. Furthermore, the ideas around context have insulated SFL from the autonomous formalisms that Duranti states had taken over from the anthropological linguistics in America (see Duranti 2009:3). Firth was first of all a naturalist of languages – the languages of Eastern and African cultures. Firth was focused on instances of language rather than on the essence of what language ‘is’. Like Henry Sweet before him, Firth’s efforts were directed to shifting the axis of linguistic discussion away from the focus on a Eurocentred, reconstructed classical tradition towards the ‘typical actual’ of current speech. His interests were data driven and problem oriented: from apprehending the polyphony of human articulation to improving the system of ‘shorthand’ for rapid recording of speech. In his writings he tended to reprise his speeches before learned societies, a spoken context in which much could be taken for granted. On the other hand, he wrote two clear and prescient books on language matters for a wider public that had, between the Wars, a demonstrated keenness to absorb knowledge from experts. Between linguists, it might be said that he insisted on consistency of theory, part of which meant being vitriolic at any reifying of theoretical abstractions. Firth was politically conservative, yet found common ground with the Russian critique of Genevan structuralism as well as with Halliday’s Marxist perspective on linguistics. Most remarkable, perhaps, is that he demanded of his students and colleagues that they defamiliarize their worlds through the semantics of a non-European language, adopting the lens of linguistic evidence and only looking back at English with a renewed, extended grammatical imagination.

References Allen, W. S. 1953. Relationship in Comparative Linguistics. Transactions of the Philological Society 52(1): 52–108. Allen, W. S. 1970 [1951]. Some Prosodic Aspects of Retroflexion and Aspiration in Sanskrit. In F. R. Palmer, ed., Prosodic Analysis. London: Oxford University Press. 82–90. Anderson, S. R. 1985. Phonology in the Twentieth Century: Theories of Rules and Theories of Representations. Chicago: University of Chicago Press. Aspect, A., J. Dalibard, and G. Roger. 1982. Experimental Test of Bell’s Inequalities Using Time Varying Analyzers. Physical Review Letters 49(25): 1804–7. Bartsch, S., R. Eckart, M. Holtz, and E. Teich. 2005. Corpus-based Register Profiling of Texts from Mechanical Engineering. Proceedings of the Corpus Linguistics Conference CL 2005 1(1). Available online at: www .birmingham.ac.uk/research/activity/corpus/publications/conference-arch ives/2005-birmingham.aspx. (Last accessed 25/07/2017.)

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

31

Bateson, G. 1982. Difference, Double Description and the Interactive Designation of Self. In F. A. Hanson, ed., Studies in Symbolism and Cultural Communication. Kansas: University of Kansas Publications in Anthropology. 3–8. Bloomfield, L. 1933. Language. New York: Holt, Reinhart and Winston. Bloomfield, L. 1962. The Menomini Language. New Haven: Yale University Press. Bohm, D. 1952a A Suggested Interpretation of the Quantum Theory in Terms of ‘Hidden’ Variables, I. Physical Review Letters 82(2): 166–79. Bohm, D. 1952b. A Suggested Interpretation of the Quantum Theory in Terms of ‘Hidden’ Variables, II. Physical Review Letters 82(2): 180–93. Brooks, M. 2011. The Secret Anarchy of Science. London: Profile Books. Chomsky, N. 1972. Language and Mind. New York: Harcourt Brace Jovanovich. Chomsky, N. and J. McGilvray. 2012. The Science of Language: Interviews with James McGilvray. Cambridge: Cambridge University Press. Cohen, J. and I. Stewart. 1995. The Collapse of Chaos: Discovering Simplicity in a Complex World. London: Penguin. de Corte-Real, B. 1998. Mambae and Its Verbal Art Genres: A Cultural Reflection of Suru-Ainaro, East Timor. PhD Thesis, Macquarie University. DeWitt, R. 2004. Worldviews: An Introduction to the History and Philosophy of Science. Oxford: Blackwell. Duranti, A. 2009. Linguistic Anthropology: A Reader. Chichester: WileyBlackwell. Ellis, J. 1993. Language, Thought and Logic. Evanston: Northwestern University Press. Firth, J. R. 1935. The Technique of Semantics. Transactions of the Philological Society 34(1): 36–72. Firth, J. R. 1957. Papers in Linguistics 1934–1951. London: Oxford University Press. Firth, J. R. 1962. A Synopsis of Linguistic Theory 1930–55. In J. R. Firth et al., eds., Studies in Linguistic Analysis: Special Volume of the Philological Society. Oxford: Blackwell. 1–32. Firth, J. R. 1964. The Tongues of Men and Speech. Edited by P. D. Strevens. London: Oxford University Press. Firth, J. R. 1968. Selected Papers of J. R. Firth 1952–59. Edited by F. R. Palmer. London: Longmans. Glock, H. J. 1996. A Wittgenstein Dictionary. Oxford: Blackwell. Grice, H. P. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University Press. Halliday, M. A. K. 1959. The Language of the Chinese ‘Secret History of the Mongols’. Oxford: Blackwell. Halliday, M. A. K. 1963. Intonation in English Grammar. Transactions of the Philological Society. 62(1): 143–69. Halliday, M. A. K. 1974. Language and Social Man. London: Longman.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

32

DAVID G. BUTT

Halliday, M. A. K. 1978. Language as Social Semiotic: The Social Interpretation of Language and Meaning. London: Edward Arnold. Halliday, M. A. K. 1984. Language as Code and Language as Behaviour: A Systemic-Functional Interpretation of the Nature and Ontogenesis of Dialogue. In R. Fawcett, M. A. K. Halliday, S. M. Lamb, and A. Makkai, eds., The Semiotics of Language and Culture: Language as Social Semiotic, Vol. 1. London: Pinter. 3–35. Halliday, M. A. K. 2002 [1985]. On the Ineffability of Grammatical Categories. In J. J. Webster, ed., The Collected Works of M.A.K. Halliday Volume 1: On Grammar. London: Continuum. 291–322. Halliday, M. A. K. and Z. L. James. 1993. A Quantitative Study of Polarity and Primary Tense in the English Finite Clause. In J. Sinclair, M. Hoey, and G. Fox, eds., Techniques of Description: Spoken and Written Discourse. London: Routledge. 32–66. Halliday, M. A. K. and R. Hasan. 1985. Language, Context and Text: A Social Semiotic Perspective. Geelong: Deakin University Press. Halliday, M. A. K. and C. M. I. M. Matthiessen. 1999. Construing Experience through Meaning: A Language-based Approach to Cognition. London: Cassell. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. London: Routledge. Harris, R. 1981. The Language Myth. New York: St. Martin’s Press. Harris, R. 1990. On Redefining Linguistics. In H. G. Davis and T. J. Taylor, eds., Redefining Linguistics. New York: Routledge. 18–52. Hasan, R. 2009. The Collected Works of Ruqaiya Hasan, Volume 2: Semantic Variation: Meaning in Society and in Sociolinguistics. Sheffield: Equinox. Hasan, R. and C. Cloran. 1990. A Sociolinguistic Interpretation of Everyday Talk between Mothers and Children. In M. A. K. Halliday et al., eds., Learning, Keeping and Using Language: Selected Papers from the 8th World Congress of Applied Linguistics, Sydney, 16–21 August 1987. Amsterdam: John Benjamins. Henderson, E. J. A. 1949. Prosodies in Siamese: A Study in Synthesis. Asia Major New Series 1: 189–215. Henderson, E. J. A. 1987. J. R. Firth in Retrospect: A View from the Eighties. In R. Steele and T. Threadgold, eds., Language Topics: Essays in Honour of Michael Halliday. Amsterdam: John Benjamins. Hull, G. 1998. The Basic Lexical Affinities of Timor’s Austronesian Languages: A Preliminary Investigation. Studies in Languages and Cultures of East Timor 1: 97–202. Jakobson, R. 1966. Henry Sweet’s Path toward Phonetics. In C. E. Bazell et al., eds., In Memory of J. R. Firth. London: Longmans. 242–54. Kachru, B. B. 2015. Socially Realistic Linguistics: The Firthian Tradition. In J. J. Webster, ed., The Bloomsbury Companion to M. A. K. Halliday. London: Bloomsbury. 72–93. Langendoen, D. T. 1968. The London School of Linguistics. Cambridge, MA: MIT Press.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

Firth and the Origins of SFL

33

Laver, J. 1980. The Phonetic Description of Voice Quality. Cambridge: Cambridge University Press. Le Page, R. B. and A. Tabouret-Keller. 1985. Acts of Identity: Creole-based Approaches to Language and Ethnicity. Cambridge: Cambridge University Press. Levinson, S. 1983. Pragmatics. London: Cambridge University Press. Love, N. 1988. The Linguistic Thought of J. R. Firth. In R. Harris, ed., Linguistic Thought in England 1914–1945. London: Duckworth. 148–64. Lyons, J. 1968. Theoretical Linguistics. Cambridge: Cambridge University Press. Mann, W. C. and C. M. I. M. Matthiessen. 1983. Nigel: A Systemic Grammar for Text Generation. Marina del Rey: Information Sciences Institute, University of Southern California. Matthiessen, C. M. I. M. 2015a. Halliday on Language. In J. J. Webster, ed., The Bloomsbury Companion to M. A. K. Halliday. London: Bloomsbury. 137–202. Matthiessen, C. M. I. M. 2015b. Register in the Round: Registerial Cartography. Functional Linguistics 2(9): 1–48. Mitchell, T. F. 1957. The Language of Buying and Selling in Cyrenaica: A Situational Statement. Hesperis 44: 31–71. Mitchell, T. F. 1975. Principles of Neo-Firthian Linguistics. London: Longman. Mufwene, S. S. 2001. The Ecology of Language Evolution. Cambridge: Cambridge University Press. Palmer, F. R. 1968. Introduction. In F. R. Palmer, ed., Selected Papers of J. R. Firth 1952–59. London: Longmans. 1–11. Palmer, F. R. 1970. Prosodic Analysis. London: Oxford University Press. Panksepp, J. and C. Trevarthen. 2009. The Neuroscience of Emotion in Music. In S. Malloch and C. Trevarthen, eds., Communicative Musicality: Exploring the Basis of Human Companionship. Oxford: Oxford University Press. 105–46. Robins, R. H. 1957. Malinowski, Firth, and the ‘Context of Situation’. In E. Ardener, ed., Social Anthropology and Language. London: Tavistock. 33–46. Searle, J. R. 1969. Speech Acts. Cambridge: Cambridge University Press. Shklovsky, V. 1972. Mayakovsky and his Circle. New York: Dodd, Mead and Co. Sinclair, J. 1966. Beginning the Study of Lexis. In C. Bazell, J. Catford, M. A. K. Halliday, and R. Robins, eds., In Memory of J. R. Firth. London: Longman. 148–62. Sinclair, J. and R. M. Coulthard. 1975. Towards an Analysis of Discourse: The English Used by Teachers and Pupils. Oxford: Oxford University Press. Stensæth, K. and G. Trondalen. 2012. Dialogue on Intersubjectivity: An Interview with Stein Braten and Colwyn Trevarthen. Voices: A World Forum for Music Therapy 12(3). Available online at: https://voices.no/index .php/voices/article/view/682/568#31. (Last accessed 29/05/2017.)

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

34

DAVID G. BUTT

Teich, E. 2003. Cross-linguistic Variation in System and Text. Berlin: Mouton de Gruyter. Thomas, M. A. 2011. Fifty Key Thinkers on Language and Linguistics. New York: Routledge. Trevarthen, C. 1998. The Concept and Foundations of Infant Intersubjectivity. In S. Bråten, ed., Intersubjective Communication and Emotion in Early Ontogeny. Cambridge: Cambridge University Press. 15–46. Whitehead, A. N. 1938. Modes of Thought. New York: The MacMillan Company. Whitehead, A. N. 1979. Process and Reality: An Essay in Cosmology. New York: Macmillan. Whitaker, A. 2006. Einstein, Bohr, and the Quantum Dilemma: From Quantum Theory to Quantum Information. Cambridge: Cambridge University Press. Whorf, B. L. 1956. Science and Linguistics. In J. B. Carroll, ed., Language, Thought and Reality. Cambridge, MA: MIT Press. 207–19.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:12:20, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.003

2 Key Terms in the SFL Model Jonathan J. Webster

2.1

Systemic Functional Linguistics

A Systemic Functional approach looks at how language functions to make meaning in context of situation.1 Meaning is represented paradigmatically in terms of systems of choice related to what is being talked about (i.e. ideational); how those interacting are relating to one another through what they say (i.e. interpersonal); and how ideational and interpersonal meanings are turned into discourse (i.e. textual). Halliday and Matthiessen (2014:33, emphasis in original) maintain that ‘functionality is intrinsic to language: that is to say, the entire architecture of language is arranged along functional lines. Language is as it is because of the functions in which it has evolved in the human species.’ A functional approach to the study of language offers insight into how language is learnt and how language eventually evolves into the adult language system with these three major components of meaning or metafunctions: ideational, interpersonal, and textual.2

2.2

Meaning in SFL Theory

Language is the instantiation of an indefinitely large meaning potential through acts of meaning which simultaneously construe experience and enact social relationships. Acts of meaning are also acts of identity, occurring in contexts of situation. By means of my ‘act of meaning’, I participate

1

The selections for further reading included in the footnotes of this chapter are adapted from Bloomsbury’s The Essential Halliday (Halliday 2009) edited by J. J. Webster, which includes selected extracts and additional readings from the ten volumes of Halliday’s Collected Works for twenty key concepts in Systemic Functional Linguistics.

2

For further reading on ‘functions and use of language’, see Halliday 2003a:298–322; Halliday 2003b:47, 51–6, 68–74, 81–7, 270–80; Halliday 2007a:41–2, 50–3, 56–7; Halliday 2007b:88–92, 120–2.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

36

J O N AT H A N J . W E B S T E R

in an act of interpersonal exchange, communicating my sense of my own identity, my worldview, my interpretation of experience. An act of is an instance of meaning formed out of an infinite meaning potential for reflecting on the world and interacting with others in it.3 Language is a semiotic system.4 It is the means by which meanings are created and exchanged. ‘Since language is the leading edge of meaning’, writes Halliday (2013:211), ‘the leading edge of semiotics is linguistics.’ Halliday (2013:194) describes a language as being more than a semiotic system – it is also ‘a system that makes meanings’: a semogenic system. As he explains, [t]he usual way we talk about language is by saying that language ‘expresses’ meaning, as if the meanings were already there – already existing, in some formation or other, and waiting for language to transpose them into sound, or into some kind of visual symbols. But meaning is brought about by language; and the energy by which this is achieved, the source of its semogenic power, is grammar. (Halliday 2013:194–5)

In his book Reinventing the Sacred, Kauffman (2008:193–4) writes that ‘while the human mind, central to our human embodied agency, is sometimes algorithmic and sometimes computes, it does some things we do not yet understand; it makes meanings’. Halliday (personal communication) describes language as ‘a basic human resource with potentially immense power, which is hidden, partly because people are genuinely not aware of how much they are, in fact, depending on it’. We depend on language to construe the world around us and describe our feelings within, and exchange this meaning with others. Meaning, in SFL theory, is not limited to referential meaning, i.e. word meanings. We use language not only to construe experience but also to enact social relationships, and create the discourse. A text is an instance of meaning, a construct of meaning that is formed out of a continuous process of choice among the innumerable interrelated sets of semantic options organized into three main functional components or metafunctions: ideational, interpersonal, and textual.5 Operating in parallel, the three metafunctions comprise the total meaning potential of a language.

3

For further reading on ‘act(s) of meaning’, see Halliday 2002a:201, 206, 354–6; Halliday 2002b:50, 52; Halliday 2003a:171, 174, 355–74, 375–89; Halliday 2003b:11–12, 14–15, 18–20, 113–43, 212–18, 239, 245–6, 249–50, 327–52; Halliday 2005a:198–202; Halliday 2013:253, 264.

4

For further reading on ‘semiotics’, see Halliday 2002a:196–218, 384–418; Halliday 2002b:23–84, 150–2; Halliday 2003a:2–7, 93, 113–15, 116–24, 131–7, 147–51, 171, 192–8, 199–212, 213–31, 275–7, 355–74, 375–89, 390–432; Halliday 2003b:6–27, 90–112, 140–3, 157–95, 212–26, 250–66, 281–307, 327–52; Halliday 2004:43–4, 53–5, 102–34, 216–25, 198–202; Halliday 2007a:81–96; Halliday 2007b:179–86, 193–6, 259–63.

5

For further reading on ‘metafunction’, see Halliday 2002a:21–36, 390–2; Halliday 2003a:15–18, 248–50, 277–8; Halliday 2003b:209–25, 332–3, 335–6, 338–41, 343–4, 346, 348–9; Halliday 2005a:200–2, 215–22; Halliday 2007b:183–4.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

37

These sets of semantic options constitute a semantic system or system of meaning potential which is distinguishable from other semiotic systems by the fact that it is based on grammar. The semantic system is one of three overlapping levels, or strata, which together comprise the whole linguistic system. Between the semantic system above and the phonological and morphological realization below is the lexicogrammar.6 What enables the meaning-making potential in language is the lexicogrammar. While the three metafunctions are arguably universal to every language – every language user needs to be able to use language to construe experience, enact social relationships, and create discourse – the means by which meaningful choices are lexicogrammtically realized is languagespecific. The complex realization of options from the three metafunctions becomes evident in the clause as a lexicogrammatical unit.

2.3

Grammar in SFL Theory

SFL is first and foremost a theory about how language works at the level of grammar.7 ‘[T]hinking about meaning means thinking grammatically’, writes Halliday (2013:207). A grammar is a theory of experience of everyday life. It is that abstract stratum of coding between meaning and expression; it is a resource for making meaning. Just as linguistics is language about language – or ‘language turned back on itself’ (Firth 1957:181) – grammatics is a theory of grammar; it is a theory for explaining how the grammar constructs a theory of experience. Grammatics is theorizing about a theory; it is a theory of a second order, a part of a more general theory of meaning.8 In categorizing the grammar, the grammarian comes at the task from three perspectives, each of which corresponds to a different stratum. First, there is the higher stratum of semantics. Here, the grammarian’s perspective is from above. Second is the stratum of lexicogrammar, where the perspective is from around. The third perspective is from below and looks at the morphological and phonological realization of meaning. In a functional grammar, priority is given to the perspective from above, as form follows function, and the meaning of an expression will decide its phonological and morphological realization.9

6

For further reading on ‘semantic systems’, see Halliday 2002a:196–218, 310–11; Halliday 2002b:23–8, 45–52; Halliday 2003a:323–54; Halliday 2003b:90–8, 109–12, 115–25, 281–94; Halliday 2007a:345–6; Halliday 2007b:131–3, 143–4, 153, 158, 164–6, 183–4, 186–95, 256–7.

7

For further reading on ‘theory and description’, see Halliday 2002a:37–42, 58–61, 72–7, 86, 98–9, 106–17, 158–72, 396, 403–6, 414–15; Halliday 2003a:7–15, 37–47, 199–212, 327–30; Halliday 2004:53–8; Halliday 2005a:227–38; Halliday 2005b:156–63; Halliday 2006:294–322; Halliday 2007a:136–9, 149.

8

For further reading on ‘grammatics’, see Halliday 2002a:296–8, 365–6, 369–73, 384–6, 416–17; Halliday

9

For further reading on ‘trinocular vision’, see Halliday 2002a:398, 402, 408–9; Halliday 2003a:202–5, 254–5, 266;

2003a:264–5, 274–6, 286, 362, 373, 385; Halliday 2005a:213–38. Halliday 2005a:231.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

38

J O N AT H A N J . W E B S T E R

To the ‘architecture’ of grammar, Halliday and Matthiessen (2014:55) attribute the dimensions that define the overall semiotic space of lexicogrammar, the relationships that inhere in these dimensions – and its relationship to other sub-systems of language – to semantics and to phonology (or graphology). Thus, according to systemic functional theory, lexicogrammar is diversified into a metafunctional spectrum, extended in delicacy from grammar to lexis, and ordered into a series of ranked units.

The fundamental categories for the theory of grammar are unit, structure, class, and system. Rank is the scale on which the units are ranged. Rankshift occurs when a given unit is transferred to a lower rank. A structure is made up of ordered elements. Sequence is one formal exponent of the more abstract notion of order. Delicacy has to do with the depth of detail, ranging along a cline from least delicate at one end, i.e. primary, to those small infinities at the opposite ‘where distinctions are so fine that they cease to be distinctions at all’ (Halliday 2002a:48), i.e. secondary. Structural types include configurational, prosodic, and periodic. Experiential meaning can be accounted for in terms of the configuration of process, participant, and circumstance. Interpersonal meaning tends to be prosodic, involving intonation. Textual meaning is more periodic, with the flow of discourse understood less in terms of discrete constituents than wave-like movements.10

2.4

Realizing Ideational Meaning

Based on the findings from his study of the English language, Halliday observed how each functional component or sub-component produces its own distinct dimension of structure. For example, experiential meaning, i.e. the ‘construing experience’ function (a sub-component within the Ideational Metafunction), is realized as the structural configuration of process as an integrated phenomenon involving participant(s) and circumstance(s). In Systemic Functional Grammar (SFG), the structural configuration of process, participant, and circumstance is referred to as a clause’s transitivity structure. We talk about our experience of the world in terms of processes, plus the participants and circumstances that enter into them. Processes are typically realized as verbs which may describe an action, or a feeling, or a state of being, or a way of behaving, either happening in the world around us or within our own consciousness. Processes are often accompanied by

10

For further reading on ‘structure and rank’, see Halliday 2002a:40–9, 75–81, 95–105, 106–17, 118–26, 196–218; Halliday 2002b:24–5, 27, 79–80; Halliday 2005a:29–36; Halliday 2005b:xii–xxix, 154–63, 249–51.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

39

mention of who or what is participating, and possibly also of the circumstances – where/when/why/how. The same experience may be described differently in different languages. Even within the same language, the same experience may be expressed very differently, suggesting some difference in how the same event is construed. For example, if instead of saying This reminds us that . . . the speaker says This is a reminder that . . ., the speaker has reconstrued the mental process to remind – into a participant, a reminder, with all the qualities of noun-ness. There is no one way to construe what is going on. The same happening can be reconstrued differently depending on the speaker’s perspective on the event being described. As Halliday (2003a:16, emphases in original) puts it, ‘the grammar is not merely annotating experience; it is construing experience – theorizing it, in the form that we call “understanding”.’ How one construes the world and what is happening will be evident in how you talk about the world. While processes may be grouped into different types on semantic grounds – processes of ‘doing’, ‘being’, ‘sensing’, ‘saying’, etc. – there is no clear reason for stopping at this point and not identifying further distinctions until we have distinguished between individual words. Instead Halliday distinguishes types of processes on both semantic and lexicogrammatical grounds. Processes also may be distinguished on the basis of who/what qualifies to be a participant in that process. For example, because sensing verbs are verbs of consciousness, the one sensing, the Senser, must be either human or human-like, in other words, have a sense of consciousness. Unlike a material process, the second participant in a mental or verbal process may be a that-clause. A relational process may also take a that-clause, but only if the noun as Subject is equivalent in meaning to what is expressed in the that-clause, e.g. The concern is that rain might ruin the picnic. Corresponding to the different process types – material, mental, relational, verbal, etc. – are different role designations. While it is reasonable to talk about participating entities such as Actor and Goal in a material process (or Agent and Affected if analyzed according to an ergative interpretation), it would not be sensible to use the same role labels for participants in some other kind of process, such as a relational process, in which one is describing the attributes of some entity. Besides experiential meaning, another sub-component within the Ideational metafunction deals with inter-clausal logico-semantic relations, such as whether clauses of equal grammatical standing are combined paratactically, or whether one main clause is hypotactically combined with a dependent clause. Besides taxis relations, logico-semantic relations also include whether the clauses are related through elaboration (e.g. exemplification, clarification), extension (e.g. addition, alternation), enhancement (e.g. embellishment, qualification), or projection (e.g. indirect/ direct speech).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

40

J O N AT H A N J . W E B S T E R

2.5

Representing Interpersonal Meaning

Language functions to facilitate exchange, both of information and of goods and services. The interpersonal metafunction deals with our use of language to interact with our partners in the exchange. In any exchange, there is give and take. When we ask a question – Is it raining? – we anticipate an answer. If my communication is successfully understood by my listener(s), then they will know what I expect of them, and, depending on their ability and willingness, respond appropriately to what I have requested. If we want someone to do something for us – Please turn down the volume on the television – then such a request would typically employ an imperative. A declarative typically expresses a statement, typically giving information; an interrogative expresses a question, typically asking for information; an imperative expresses a directive; and an exclamative expresses an exclamation. Such choices are realized structurally in English by what Halliday calls the Mood element, consisting of the Subject and Finite. The sequence of Subject before Finite realizes a declarative, e.g. The man fell; Finite before Subject realizes a yes/no interrogative, e.g. Did the man fall? In fact, the relationship between mood and illocutionary act is not one-to-one but instead many-to-many. While a declarative typically expresses a statement intended to give information, it may also function to make a request of the listener. For example, the statement It sure is cold here could in fact be a request to someone to turn off the air-conditioner. To understand the speaker’s intended meaning, the situation context and the linguistic co-text are essential. Modality is another aspect of interpersonal meaning related to the expression of the speaker’s attitude about what they are saying. In English, modality is expressed by the use of modal adjuncts (possibly by certain adverbs like surely, possibly, or by various thematic structures such as it is possible that, or there is a possibility that), or through a small set of verbs known as modal auxiliaries, e.g. can, may, might. Intonation also plays a significant role in expressing ‘the particular tone of assertion, query, hesitation, doubt, reservation, forcefulness, wonderment, or whatever it is, with which the speaker tags the proposition’ (Halliday 2002a:205). Intonation may be analyzed as a complex of three phonological systems, or as Halliday puts it ‘systemic variables’: tonality, tonicity, and tone, which are interdependent with rhythm.11

11

For further reading on ‘intonation’, see Halliday 2002a:55, 78, 90–1, 192–3, 205–7, 262–4, 269–70; Halliday 2002b:27–9, 32–6, 204–5, 232–3, 255; Halliday 2003b:50–1, 106–7, 162, 177, 184–9, 233, 317–19; Halliday 2004:69–71; Halliday 2005a:77–8; Halliday 2005b:57–70, 106–7, 139–40, 155–6, 161, 192–5, 213–15, 218, 237–86, 287–92; Halliday 2007a:71–3, 101, 158–9.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

2.6

41

Representing Textual Meaning

The Textual Metafunction is concerned with textual meaning, or what gives texture to a text. Texture is what makes a text into a coherent piece of language, as opposed to simply being an unorganized string of sentences. The more cohesive and coherent the text is, the greater is its texture. On the one hand, cohesion deals with how successive sentences are integrated to form a text, i.e. inter-sentence texture. Cohesion in a text increases as the elements within a text become more mutually dependent on one another for their interpretation. Cohesive ties cross clause and sentence boundaries. Coherence, or intra-sentence texture, on the other hand, has to do with fit to context. The organization of the message to fit the context comprises two aspects: one aspect, which Halliday refers to as the ‘hearer angle’, relates to the organization of the message so that it ties up with the preceding text, with that which the hearer has already heard about, i.e. the Given; the second aspect, the ‘speaker angle’, relates to how the message is organized around what the speaker wants to say, or what Halliday calls ‘Theme’. Each clause is a proposition which contributes new information to the text as a whole. Unlike Given information which is recoverable from the preceding text or the immediate context of situation, New information is something not previously mentioned; it is not recoverable. Together, Given plus New information constitute information structure. Intonation and rhythm, especially the pitch contour of speech, figure prominently in the information system, foregrounding the new information (New) from that which is otherwise recoverable from the discourse and its context of situation (Given). New information typically culminates in an element – i.e. the Focus – which is recognizable from the tonic prominence which it receives. This connection between tone group and information unit notwithstanding, information structure is as relevant to the analysis of written texts as it is to spoken. In both spoken and written discourse, the Given refers to something recoverable or locatable in the text or the context of situation. In terms of thematic structure in English, the Theme occurs in clause-initial position. Because it is a matter of choice about what comes first in the clause, it is therefore meaningful. Where the choice is typical, we say it is ‘unmarked’. In a declarative clause, the grammatical subject is the unmarked theme. In a yes/no question, the unmarked theme is the request for the answer ‘yes’ or ‘no’, i.e. the finite element of the verb: e.g. Did you eat? In a WH-question, the unmarked theme is the request for information, i.e. the interrogative ‘WHelement’: what, who, how, why, etc. In an imperative, the unmarked theme is either the addressee(s), as in (You) leave quietly! (‘you’ being understood even if not mentioned), or addressee(s) and speaker, as in Let’s leave quietly!

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

42

J O N AT H A N J . W E B S T E R

Besides the grammatical Subject, other grammatical units which may function as experiential/topical Theme include those that contribute to the content of the utterance, from complements to circumstantial adjuncts, and even the predicate in an imperative clause. The Theme is underlined in each of the following examples (1) to (5): (1)

Bill left the party early.

(2)

I don’t know.

(3)

This afternoon we will go see a movie.

(4)

The truth they could not accept.

(5)

The one who baked the cake is my uncle.

The grammatical Subject as Theme of a declarative clause is typical and unmarked, e.g. examples (1), (2), and (5), even when it contains an embedded clause as postmodifier of the nominal-as-Subject. An adjunct expressing circumstance is less typical and more marked, e.g. example (3). The complement as Theme, e.g. example (4), is more marked because a complement is least likely to be made thematic. The Theme of a clause is first and foremost about the content of the message, but may also express other meanings as well. For example, in the sentence, But Mary, I do love your cooking. The initial conjunction But is textual Theme, the vocative Mary is interpersonal Theme, and the grammatical Subject I is topical or ideational Theme. In English, this sequence of thematic entities – textual, interpersonal, ideational – appears to be the norm. Thematic progression concerns the successive choices of what is rendered thematic. Daneš (1974) identified three typical patterns of thematic progression: linear, continuous, and derived. Linear thematic progression occurs when the new information found in the Rheme of one clause occurs as Theme in the next clause. The flow of information is New!Given. In the following example, from a journal article appearing in Science (Bradshaw and Holzapfel 2006), notice how the Theme of refers to the new information in and ; and the Theme of refers back to the new information from .

Over the past 40 years, species have been extending their ranges toward the poles and populations have been migrating, developing, or reproducing earlier in the spring than previously (1–4). These range expansions and changes in the timing of seasonal events have generally been attributed to ‘phenotypic plasticity’ – that is, the ability of individuals [ to modify their behavior, morphology, or physiology in response to altered environmental conditions (5, 6).] Phenotypic plasticity is not the whole story. However, recent studies show that over the recent decades, climate change has led to heritable,

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

43

genetic changes in populations of animals as diverse as birds, squirrels, and mosquitoes (see the first figure).

In continuous thematic progression, the same theme is repeated over subsequent clauses. This is illustrated in the following paragraph from a speech given by Malala Yousafzai addressing the Youth Assembly at the United Nations on her sixteenth birthday, 12 July 2013. Note the repeated thematic references to those who shot her: Dear Friends, on the 9th of October 2012, the Taliban shot me on the left side of my forehead. They shot my friends too. They thought that the bullets would silence us. But they failed. And then, out of that silence, came thousands of voices. The terrorists thought that they would change our aims and stop our ambitions but nothing changed in my life except this: Weakness, fear and hopelessness died. Strength, power and courage was born. I am the same Malala. My ambitions are the same. My hopes are the same. My dreams are the same. (Yousafzai 2013)

Derived themes are themes of successive clauses which may be regarded as related to some superordinate topic or hypertheme. For example, in a narrative account, we may notice repeated thematic references to significant chronological points in the story, e.g. Early in the morning, By late afternoon, or In the evening.

2.7

Formalizing Paradigmatic Relations

In the late 1950s Halliday worked in the pioneering Cambridge Language Research Unit: In this context it became necessary to represent grammatical features in explicit, computable terms. I wanted to formalize paradigmatic relations, those of the system; but I did not know how to do it – and I totally failed to persuade anyone else of this! (Halliday 2005a:138)

Subsequently, however, system networks were developed as a way to represent meaning potential not simply syntagmatically as an inventory of sequentially ordered items, but instead primarily as paradigmatically organized sets of options. A system network is comprised of only AND and OR relations between options. No distinction is made between linguistic information and non-linguistic information; all information is stored and processed the same way, as interconnected options in a vast network. Information processing consists of the transmission of activation along pathways defined by the network, and changes in probabilities attached to options.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

44

J O N AT H A N J . W E B S T E R

Halliday (2003a:327) defines semantic networks as ‘a hypothesis about patterns of meaning’. Semantic networks • • •

specify the range of alternatives at the semantic stratum; relate the hypothesized categories to some general social theory or ‘theory of behaviour’; relate those same categories to the categories of linguistic form. (see Halliday 2003a:334)

The great problem with the system network as a form of representation, as Halliday himself points out, is that it is a very static form of representation. In an interview with Kress, Hasan, and Martin, Halliday commented: It freezes the whole thing, and then you have to introduce the dynamic in the form of paths through the system. Your problem then is to show how the actual process of making paths through the system changes the system. (Halliday in Martin 2013:110)

Each act of meaning, ‘each instance, no matter how minutely, perturbs these probabilities and so changes the system’ (Matthiessen 2015:212). Halliday uses the term ‘instantiation effect’ to describe how our acts of meaning feed back into the system. With advances that have been made in corpus building, quantitative profiling, or the assignment of probabilities to choices within the system network, has become a realistic goal for linguistic description.12 A related concept is the notion of ‘markedness’. In a skew where one option is more frequent, this option is quantitatively unmarked, and the less frequent option is marked. Where the options are equally probable, neither is the ‘unmarked term’. Besides being quantitatively unmarked, an option may be simpler and therefore identified as being formally unmarked.13 In Saussurean terms, acts of meaning are ‘parole’, and the system of meaning potential which these acts instantiate is ‘langue’. However, unlike the Saussurean distinction between langue and parole, Halliday sees acts of identity (parole) and meaning potential (langue) not as two distinct classes of phenomena, but instead as only a difference in the stance taken by the observer. Langue is parole seen from a distance, parole is langue up close and in its context. What some perceive to be a dichotomy between langue and parole, Halliday instead views as a cline of instantiation ranging between system and instance. As Halliday (2005a:248) explains,

12

For further reading on ‘quantifying language’, see Halliday 2002a:70–2, 92–4, 166, 168–9, 352–68; Halliday 2003a:23–6, 122, 253, 404–13, 425–6, 430; Halliday 2005a:8–9, 13–19, 42–62, 63–75, 76–92, 93–129, 130–56, 157–90, 235–8; Halliday 2006:209–48; Halliday 2007a:310–16.

13

For further reading on ‘markedness’, see Halliday 2002a:305, 320–1, 326, 376–7; Halliday 2002b:28–38, 199–200, 205; Halliday 2003b:342–3; Halliday 2005a:22–3, 81, 88, 91, 96–7, 101–2, 131–2; Halliday 2005b: 5–54, 55–109, 110–53, 154–63, 193–5, 203, 220–31, 249–61, 264–86, 288–9; Halliday 2006:5–174, 209–48, 330–2.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

45

Instantiation is a cline, modelling the shift in the standpoint of the observer: what we call the ‘system’ is language seen from a distance, as semiotic potential, while what we call ‘text’ is language seen from close up, as instances derived from that potential. In other words, there is only one phenomenon here, not two; langue and parole are simply different observational positions.

To illustrate his point, Halliday gives the analogy of climate and weather. The weather is what we experience on a daily basis. Over time, we generalize this daily occurrence of the weather into our sense of what the climate is. So the climate becomes a theory about the weather. Similarly, like the notion of climate, the system of meaning – meaning potential – is a theory of the text – the instantiation of meaning potential (see Halliday 2008:79–80). Just as the weather and climate are not two distinct and separate phenomena, so too parole (instances of language) and langue (system of language) are one and the same phenomenon.

2.8

Language in Flux

Language is always in flux. It is dynamic and changing. SFL views the fuzziness and indeterminacy which one observes in actual usage to be both necessary and positive, enabling language to achieve its richness. Halliday describes indeterminacy in language as occurring (i) where distinctions in meaning are more continuous than discrete; (ii) where meanings become fused to the extent that one cannot select between them; and (iii) where some domain of experience may be construed in contradictory or competing ways. This inherent fuzziness and indeterminacy cannot be ignored, but instead deserves to be accounted for in the grammar.14 The direction of this change is towards growth. The meaning space is constantly expanding. As society evolves, so too does language. New ways of doing demand new ways of meaning, i.e. new registers.15 Phylogenetic evolutionary change in human language is one of three histories identified by Halliday, the other two being the ontogenetic development of each individual’s language, and the logogenetic unfolding of a particular text. 14

For further reading on ‘indeterminacy in language’, see Halliday 2002a:399–402, 409–10; Halliday 2002b:33, 51, 139–40, 145–6; Halliday 2003a:54–5, 254–5, 266–7; Halliday 2005a:204–7, 211, 226–30; Halliday 2007b:193, 200.

15

Halliday defines a register as ‘a syndrome, or cluster of associated variants; and again only a small fraction of the theoretically possible combinations will actually be found to occur’ (2002b(1990):168) rather than the obligatory incidence of particular features. Dialects are identified by their users. Codes are patterns or speech habits of speakers of the same language. For further reading on ‘varieties and variation in language: dialect, register, code’, see Halliday 2002b:17, 168–70, 231–4; Halliday 2003a:255–6, 268, 360, 362–3, 382–3, 416–17; Halliday 2005a:225–6, 248, 263–4; Halliday 2005b:214–16; Halliday 2007a:29–31, 240–3, 296–300; Halliday 2007b:5–40, 85–8, 93–7, 103–7, 115–16, 129–30, 138–9, 140–2, 147, 174–5, 181–3, 196–7, 205–9, 235, 242–3, 252–5, 259–61.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

46

J O N AT H A N J . W E B S T E R

Phylogenetic evolutionary change is well illustrated by the language of science, in which, as Halliday explains, there has been ‘a steady drift [in the grammar of scientific discourses] towards things’ (Halliday 2004:129, emphasis in original). Things have been foregrounded at the expense of qualities, processes, and relations. While this nominalizing grammar ‘has given us enormous powers over our physical and biological environment’ (Halliday 2004:47), so much so that ‘the scientist can make the world stand still, or turn it into one consisting only of things, or even create new, virtual realities’ (Halliday 2004:viii), still it has done so at the risk of alienating learners and turning science into ‘the prerogative of an elite’ (Halliday 2004:179). Over the course of history, as the need arose for new ways to theorize the human experience, humankind has relied on the power of language ‘to reconstrue commonsense reality into one that imposed regularities on experience and brought the environment more within our power to control’ (Halliday 2004:xvii). Halliday describes this metaphor-making potential, i.e. grammatical metaphor, as ‘a concomitant of a higher-order, stratified semiotic – once the brain splits content into semantics and grammar, it can match them up in more than one way’ (Halliday 2004:123). It is as though there has been a ‘partial freeing of the lower-level systems from the control of the semantics so that they become domains of choice in their own right’ (Halliday 2002b:131). Halliday (2013:78, emphases in original) further explains this metaphormaking potential as follows: Metaphor, whether in its grammatical or its lexical sense, is a cross-coupling between the semantics and the lexicogrammar. In lexical metaphor, which is metaphor in its traditional sense, this is the replacement of one lexical item (word or phrase) by another in the realization of a given meaning . . . In grammatical metaphor, one grammatical category is replaced by another – a word class, a structure unit, and often both; for example, in place of ‘she didn’t know the rules, so she died’ we have ‘her ignorance of the rules led to her death’. Both grammatical and lexical metaphors are characterized by semantic junction (this is the basis of the distinction between metaphor and simile).

Historically speaking, this metaphor-making potential has been achieved over ‘three successive waves of theoretical energy’ (Halliday 2004:46) – generalization, abstractness, metaphor – each ‘tak[ing] us one step further away from ordinary everyday experience’ (Halliday 2004:47), but at the same time each step may be thought of as having ‘enlarged the meaning potential by adding a new dimension to the total model’ (Halliday 2004:46). Whether it is the ‘art’ in verbal art, or the ‘science’ in verbal science, what is being crafted through the metaphor-making potential available in language are hypotheses or models about the world experienced around us

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

47

and in us. Such is the semogenic power of grammar, that it enables us to define ‘the basic experience of being human’. Hasan refers to the hypotheses articulated in verbal art as themes about some aspect of social life achieved through second-order semiosis – a kind of double articulation – ‘so that one order of meaning acts as metaphor for a second order of meaning’ (Hasan 1985:100). Halliday adopts Mukařovský’s term ‘de-automatization’ to describe this ability ‘to interpret the grammar in terms that go beyond its direct realizational function’ (Halliday 2002b:139), allowing ‘the wholesale recasting of the relationship between the grammar and the semantics’ (Halliday 2004:19). Halliday’s notion of grammatical metaphor resembles what Kauffman (2008) refers to as ‘Darwinian preadaptations’. Darwinian preadaptations in biological evolution occur when pre-existing phenotypic features are detoured to new uses, ‘explor[ing] . . . the “adjacent possible”, and thereby expand[ing] the range of actuality in unforeseen and unforeseeable ways’ (Shaviro 2008). Just as science does not have all the answers when it comes to predicting the partially lawless evolution produced by Darwinian preadaptations, so too the ceaseless creativity in language, while enabled by grammar, can never be fully predicted by grammar – especially if that grammar is algorithmic, computational, or connectionist in character. Semantics and grammar are coupled together through what Halliday (2004:94) calls ‘a relation of congruence’, in which things are realized as nouns, processes as verbs, qualities as adjectives, relators as conjunctions, circumstances as adverbs or prepositional phrases. However, we can de-couple these congruent relations, and re-couple them in whatever way best suits our purpose. Examples of grammatical metaphor include length, which is ‘a junction of (the quality) “long” and the category meaning of a noun, which is “entity” or “thing”, [and motion, which is] a junction of the (the process) “move” and the category meaning, again, of a noun’ (Halliday 2004:xvi–xvii).16 The following three-step recipe illustrates how to turn a congruent statement into its metaphorical alternative: Congruent: If the item is exposed for long, it will deteriorate rapidly. (i) Nominalize the Process (exposed ) exposure; deteriorate ) deterioration) (ii) Make the Medium of that Process a ‘possessive’ modifier (the item / it ) of the item) (iii) Express the relation between the two events (If . . . then ) will result in) Metaphor: Prolonged exposure will result in rapid deterioration of the item.

16

For further reading on ‘grammatical metaphor’, see Halliday 2002a:219–60, 346–8, 358–60, 397; Halliday 2002b:23–84, 160, 164, 219–23, 226; Halliday 2003a:130–4, 139–76, 248–70, 282, 284–5, 384, 388, 415, 419–23; Halliday 2003b:339–40, 347–9, 367–9; Halliday 2004:7–23, 32–43, 49–101, 102–34, 143, 147–52, 156–7, 162, 171–9, 190–7, 214–16, 220–5; Halliday 2005a:42–62, 63–75, 196–212, 213–38; Halliday 2006:325–33, 339–44; Halliday 2007a:63–80, 105–10, 117, 123, 126–8, 301–3, 354–67, 379, 381; Halliday 2007b:239, 243–4, 278.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

48

J O N AT H A N J . W E B S T E R

However, the choice between congruent and metaphorical is not binary, but complementary. Rather, as Halliday (2004:84–5) illustrates, there may exist a series of agnate forms at intermediate points along a cline of metaphoricity: (i) (ii) (iii)

(iv)

(v) (vi) (vii)

Osmolarity increases, so putrescine is rapidly excreted. (clause nexus: paratactic) Because osmolarity increases, putrescine is rapidly excreted. (clause nexus: hypotactic) That osmolarity increases has the effect that putrescine is rapidly excreted. (clause: two rankshifted clauses, finite) Osmolarity increasing leads to putrescine being rapidly excreted. (clause: two rankshifted clauses, non-finite) Increasing of osmolarity causes rapid excreting of putrescine. (clause: two nominal groups, verb as Head) Increase of osmolarity causes rapid excretion of putrescine. (clause: two nominal groups, mass noun as Head) Increases of osmolarity cause rapid excretions of putrescine. (nominal groups, count noun as Head)

Besides this complementarity between congruent and metaphorical, there are other complementarities as well, such as that between speech and writing, or the meta-complementarity between lexis and grammar. These complementarities make it possible for our common-sense grammars of daily life to accommodate multiple and possibly even contradictory perspectives on the same set of phenomena. In SFL theory, beyond the language-internal strata of semantics, lexicogrammar, and phonology, there exists the language-external contextual level. The context of culture represents the systemic potential for choice depending on how the context of situation is defined with respect to field, tenor, and mode. What is going on, i.e. field, plays a role in determining choices related to transitivity structure, or process, participant, and circumstance. Who is involved, i.e. tenor, deals with interpersonal choices from the systems of mood and modality. How the exchange takes place will influence choices involving cohesion, as well as thematic and information structures.17

17

For further reading on ‘context of culture and context of situation’, see Halliday 2002a:29, 35, 201, 211, 217, 221, 225–31, 243, 246, 263, 283–5, 311, 357, 359, 405; Halliday 2002b:38, 44, 51–64, 150–2, 229–34, 243–4, 251, 254; Halliday 2003a:154–6, 185, 195–7, 210, 273, 279, 298–9, 358, 362, 382, 420; Halliday 2003b:81, 87, 95, 101, 111, 121, 134, 204, 207, 286–95, 302–4; Halliday 2005a:207–8, 217, 225, 238, 249, 256, 260, 266; Halliday 2005b:199, 306–37; Halliday 2006:10, 13, 16, 20, 64, 355–7; Halliday 2007a:85–7, 94, 96, 271–90, 298, 300, 307, 311, 349, 354–67, 368–82; Halliday 2007b:59, 62, 77, 82, 90–7, 105, 110–20, 127, 130, 133–7, 140, 142, 172, 180–2, 184, 187, 192–9, 203, 209, 235, 242, 258–9, 262.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

2.9

49

A Marxist Orientation

Listing what he identifies as the seven distinctive features of SFL,18 Halliday includes ‘Marxism’. Marxism, he writes, has ‘always [been] part of my own thinking: values of language and language varieties, language as political process and political tool; linguistics as (critical component of ) a science of meaning ’ (Halliday, personal communication). Subsequently, in an interview,19 Halliday suggests that in order to study language in Marxist terms, one should begin by questioning the views of language that were then current in a non-Marxist society. So, for example, when he returned to England from China in the 1950s, language studies and linguistics privileged or prioritized standard languages over dialects; written language over spoken; classical languages over modern ones; formal language over colloquial; dominant languages over emergent ones; literary language over everyday language; majority languages over minority languages; ideational meaning over interpersonal. While acknowledging work being done in dialectology, and on minority languages, Halliday (personal communication) nonetheless argues, ‘The view of language which was predominant was essentially derived from all those highly-valued forms of language rather than the others.’ Having noted this skewed emphasis on highly valued forms as compared with ‘socially-induced low visibility forms of language’, Halliday, along with other members of the Communist Party Linguistics Group20 sought to unpack how these priorities had influenced the representation of grammar in mainstream linguistics. They focused their attention on the languages of ex-colonial societies which were struggling to gain status as national languages. Their aim was to give value to these forms of language. As Halliday explains in the same

18

In a letter, Professor M. A. K. Halliday identified the following seven distinctive features of SFL: ‘(1) Quantitative studies and probability: This has been a constant thread since “Linguistics and machine translation” (it was even raised in my PhD thesis on the Secret History), with references also to Shannon Weaver’s theory of information (now at last recognized for its importance, e.g. in recent work by Terrence Deacon). And as far as I know nobody has followed up the work I did with collaboration from Zoe James on the probabilistic nature of grammatical systems. (2) Metafunction: again a recurrent theme, with the point that metafunction determines the way that languages have evolved, and the insistence that the interpersonal and textual metafunctions are equally fundamental, along with the ideational, to the functioning of language as a semiotic system, the form taken by grammatical structures maximizing the possibilities for different meanings to combine freely with one another. (3) Historical contexts of linguistics: Wang Li’s and Firth’s departments were unusual (even unique?) in building the history of linguistics into their teaching of the subject; Wang Li the Chinese tradition, Firth the European (and also the Indian, though I never learnt so much about that); more recently, linguistics in the context of the history of ideas, relation of the human to the natural sciences (linguistics itself being also physical, biological and social). (4) Marxism: early with Jeff Ellis, Dennis Berg, Jean Ure and others, mostly existing only in typescript; not made explicit in the McCarthy era (I had lost too many openings because of it in those years!) but always part of my own thinking: values of language and language varieties, language as political process and political tool; linguistics as (critical component of ) a science of meaning . (5) Unity of lexis and grammar. (6) Continuity of protolanguage. (7) Thinking in terms of patterns of analogies, complementarities, compatibility with other views and theories (where others only see contractions)’ (Halliday, personal communication).

19

This interview was conducted by Annabelle Lukin, David Butt, and myself with Professors Halliday and Hasan in their home in 2012.

20

Jeffrey Ellis, Jean Ure, Dennis Berg, Trevor Hill, and Peter Wexler.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

50

J O N AT H A N J . W E B S T E R

interview, ‘In my own background at that time, my own thinking at that time, I would say “Yes, I was definitely trying to adopt what I would consider a Marxist approach, and indeed, I still would”.’ Clarifying what he means by Marxist thinking, Halliday emphasizes that he is ‘not at all saying Marxist means as it were there even in embryo in the words of Marx himself’. Rather, ‘if Marxist thinking means anything, it simply means clear and honest and objective thinking about, first of all, whatever it is that you are focusing your attention on’. One cannot avoid the influence of one’s personal ideology on one’s work in science. To argue otherwise and say that ideology should not affect our science is itself an ideological position. While acknowledging that Marxism is part of the background to his own thinking about language and society, Halliday is quick to add that this does not commit anyone who does Systemic Functional Linguistics to working in left-wing politics. Nevertheless, there is obvious compatibility between an SFL perspective on language and the Marxist view that consciousness and language are evolutionary and ‘are intertwined because of the social basis of the origins of both’ (Holborow 1999:17). In the same interview with Lukin, Butt, and Webster, Hasan described the view both she and Halliday have taken of language as being ‘intensely social’: Human beings are incapable of living, surviving alone. In this lies their humanity. Everything in them is created through being part of a society. Either a reaction to it, or a following of it, whichever form you take.

As he states in his interview with Kress, Hasan, and Martin in 1986 (Martin 2013:118), Halliday saw Firth’s approach to the study of language as being ‘perfectly compatible’ with a Marxist linguistics: It seemed to me that, in fact, the ways in which Firth was looking at language, putting it in its social context, were in no way in conflict with what seemed to me to be a political approach. So that it seemed to me that in taking what I did from Firth, I was not separating the linguistic from the political. It seemed to me rather that most of his thinking was such that I could see it perfectly compatible with, indeed a rather necessary step towards, what I understand as a Marxist linguistics.

In a recent contribution to the Bloomsbury Companion to M. A. K. Halliday, Halliday (2015:99) adds the following about how a Marxist orientation impacted his own thinking: It seems to me that this overall conceptualization of language, which had been developing slowly in my own thinking across several decades, is essentially – though not aggressively – marxist in its orientation. If I never proclaimed this out loud, this was because it would be too much open to misunderstanding: there are too many different ideas about what ‘marxist’ means, and most people nowadays wouldn’t think it was worth discussing. One attribute of my ideas that is at least compatible with a marxist ideology

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

51

is the way they have always developed in conversation with other people, often as a by-product of my activities as a teacher; I have tried to acknowledge those who have been part of this enterprise, though I am conscious of having done so only very inadequately. It is impossible to track the provenance of scientific ideas, and with our modernist ways of thinking we attach too much importance to the individual anyway. It was my privilege to encounter so many congenial and thoughtful colleagues.

Halliday (2015:98) also credits a Marxist orientation with having motivated the ‘appliable’ emphasis of Systemic Functional Theory: I hoped that what I was trying to achieve as a linguist might make some contribution to improving the human condition, however minuscule and oblique. This is what I meant by calling the theory ‘appliable’. The term is less specific than ‘applicable’, which denotes applicable to some specific task, and therefore less immediate, and more indirect; its relevance is less obvious, but more long term. But other than this feature of being appliable, what other aspect of the theory might be considered as marxist?

2.10

Appliable Linguistics

Practising appliable linguistics has indeed been a driving force behind the development of Systemic Functional Linguistic Theory. Describing what he means by ‘an appliable linguistics’, Halliday (2013:128) writes: An appliable linguistics, as I understand it, is a theory which tackles problems and tries to answer questions – but questions that are asked, and problems that are raised, not by professional linguists so much as by other people who are in some way concerned with language, whether professionally or otherwise. There are large numbers of such people: educators, translators, legal and medical specialists, computer scientists, students of literature and drama, . . .; and it is their ‘take’ on language that is being addressed, at least to the point of clarifying what sorts of questions can usefully expect to be asked, and whether or not there is any hope of coming up with an answer.

Early in Halliday’s academic career, he taught at Edinburgh. The fact that many of Halliday’s students were likely to become teachers in the Scottish school system prompted Halliday’s interest in discovering what linguistics had to offer to these teachers as well as what could be learned from those teaching at secondary and primary levels of education. When Halliday moved back to London in 1963 to serve as Director of the Communication Research Centre at University College, his experience working on several projects related to educational materials development (Breakthrough to Literacy, Language in Use, and, indirectly, Language and Communication) and his involvement in the Linguistics Properties of Scientific English project contributed to his groundbreaking description of English grammar, entitled ‘Notes

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

52

J O N AT H A N J . W E B S T E R

on transitivity and theme in English’, which was published over three successive issues of the Journal of Linguistics in 1967–1968 (see Halliday 1967a, 1967b, 1968). Halliday saw this development of a grammar for educational purposes as addressing the need for social accountability. Teachers would often ask Halliday about the language experience of children before they came to school. Their questions raised Halliday’s own interest in how children develop language. Halliday saw language development as involving learning language or leaning how to mean, learning which takes place through language, and learning about language. The better informed the language teacher is about language, the more successful their pupils’ learning is likely to be. Halliday’s interest in studying the language of science, for example, grew out of his concern ‘to find the source of the difficulties faced by learners of science’ (Halliday 2004:xx). One can only help the learner, insisted Halliday, to the extent that one understands how the discourse works. Halliday’s search led him back into the history of scientific discourses, out of which he discovered ‘new strategies evolving: new ways of organizing the grammar as a resource for making meaning’ (Halliday 2004:xv).21 More recently Systemic Functional Linguistics Theory has proved its usefulness in attempts at using natural language as the means for achieving intelligent computing. A Systemic Functional form of semantic representation has been employed in projects related to data fusion, fuzzy reasoning, and the ability to construe the context of situation by inference from the text.22

References Bradshaw, W. E. and C. M. Holzapfel. 2006. Evolutionary Response to Rapid Climate Change. Science 312(5779): 1477–8. Daneš, F. 1974. Functional Sentence Perspective and the Organization of the Text. In F. Daneš, ed., Papers on Functional Sentence Perspective. Prague: Academia. 106–28. Firth, J. R. 1957. Papers in Linguistics 1934–1951. London: Oxford University Press. Halliday, M. A. K. 1967a. Notes on Transitivity and Theme in English: Part 1. Journal of Linguistics 3(1): 37–81. Halliday, M. A. K. 1967b. Notes on Transitivity and Theme in English: Part 2. Journal of Linguistics 3(2): 199–244.

21

For further reading on ‘language teaching, learning and development’, see Halliday 2002a:323–4, 349–51; Halliday 2003a:228–30, 273–4, 378–9, 384, 397–404, 429–30; Halliday 2003b; Halliday 2005b:297–305; Halliday 2007a; Halliday 2007b:63–4, 75–81, 118, 128–9, 175–6, 193–5, 212–13, 223–30.

22

For further reading on ‘linguistic computing’, see Halliday 2003b.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

Key Terms in the SFL Model

53

Halliday, M. A. K. 1968. Notes on Transitivity and Theme in English: Part 3. Journal of Linguistics 4(2): 179–215. Halliday, M. A. K. 2002a. Collected Works of M. A. K. Halliday, Vol. 1: On Grammar. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2002b. Collected Works of M. A. K. Halliday, Vol. 2: Linguistic Studies of Text and Discourse. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2003a. Collected Works of M. A. K. Halliday, Vol. 3: On Language and Linguistics. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2003b. Collected Works of M. A. K. Halliday, Vol. 4: The Language of Early Childhood. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2004. Collected Works of M. A. K. Halliday, Vol. 5: The Language of Science. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2005a. Collected Works of M. A. K. Halliday, Vol. 6: Computational and Quantitative Studies. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2005b. Collected Works of M. A. K. Halliday, Vol. 7: Studies in English Language. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2006. Collected Works of M. A. K. Halliday, Vol. 8: Studies in Chinese Language. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2007a. Collected Works of M. A. K. Halliday, Vol. 9: Language and Education. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2007b. Collected Works of M. A. K. Halliday, Vol. 10: Language and Society. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2008. Complementarities in Language. Edited by J. J. Webster. Beijing: Commercial Press. Halliday, M. A. K. 2009. The Essential Halliday. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2013. Collected Works of M. A. K. Halliday, Vol. 11: Halliday in the 21st Century. Edited by J. J. Webster. London: Bloomsbury. Halliday, M. A. K. 2015. The Influence of Marxism. In J. J. Webster, ed., The Bloomsbury Companion to M. A. K. Halliday. London: Bloomsbury. 94–100. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Routledge. Hasan, R. 1985. Linguistics, Language and Verbal Art. Oxford: Oxford University Press. Holborow, M. 1999. The Politics of English. London: SAGE. Kauffman, S. 2008. Reinventing the Sacred: A New View of Science, Reason, and Religion. New York: Basic Books. Martin, J. R. 2013. Interviews with M. A. K. Halliday: Language Turned Back on Himself. London: Bloomsbury. Matthiessen, C. M. I. M. 2015. Halliday’s Conception of Language as a Probabilistic System. In J. J. Webster, ed., The Bloomsbury Companion to M. A. K. Halliday. London: Bloomsbury. 203–41.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

54

J O N AT H A N J . W E B S T E R

Shaviro, S. 2008. Reinventing the Sacred (Stuart Kauffman). Available online at: www.shaviro.com/Blog/?p=636. (Last accessed 16/05/17.) Yousafzai, M. 2013. The Full Text: Malala Yousafzai Delivers Defiant Riposte to Taliban Militants with Speech to the UN General Assembly. Independent, 12 July 2013. Available online at: www.independent.co.uk/news/world/ asia/the-full-text-malala-yousafzai-delivers-defiant-riposte-to-taliban-mili tants-with-speech-to-the-un-8706606.html. (Last accessed 05/06/17.)

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:07:52, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.004

3 Semantics Miriam Taverniers

3.1

Introduction

Semantics can generally be described as ‘the study of linguistic meaning’, and this goes for Systemic Functional Linguistics (SFL) too. Beyond this safe but rather uninformative definition, there are many different types of interpretations of what (a) semantics can be, how big it is, what place it has in a more general model of language, and how it is related to other dimensions of the model. And this, too, goes for SFL. Whereas in general, linguistic models can be distinguished in terms of how they view ‘linguistic meaning’, within SFL different conceptions of ‘semantics’ exist side by side, and different views of semantics are seen as complementarities that make the theory rich, flexible, and adaptable (to different purposes). In other words, it is not possible to give an overview of what ‘semantics’ means (no pun intended!) in SFL. What is possible, and much more interesting, is to explore how semantics can potentially be viewed in SFL, how those different views of semantics can be understood, and how semantics has been modelled in SFL. The aim of this chapter is thus twofold: to survey different possible conceptions of semantics in SFL by elucidating how these conceptions are related to one another (i.e. by fleshing out what each conception highlights with respect to a certain architectural dimension of the theory), and to explore how specific semantic analyses and models which have been proposed in SFL can be understood against the background of these possible conceptions. Section 3.2 explores different views of semantics in SFL. These conceptions are put into a broader theoretical perspective in Sections 3.3 and 3.4. Section 3.5 then looks at some recent specific semantic models in SFL and places them against the background built up in Sections 3.2 to 3.4.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

56

MIRIAM TAVERNIERS

3.2

The Basics of Semantics in SFL: What, Where, How, and Why?

In order to explain what ‘semantics’ can mean in SFL, we will consider the following questions: (1) What can ‘linguistic meaning’ be in a systemic functional perspective? and Where is ‘semantics’ located in the overall architecture of the model? (2) How can semantics be modelled? (3) Why is semantics to be recognized as something in its own right? Why is it theorized and designed as it is?

3.2.1

The Basis of Valeur, Stratification, and Meaning-Making: WHAT and WHERE is Semantics? The basic equation of ‘semantics’ with ‘meaning’ can be extended in two directions. ‘Meaning’ does not occur as such: (i) it is expressed in a certain way, i.e. linguistic meaning is encoded or ‘realized’ in linguistic ‘forms’ which are further expressed in sound or in writing, and (ii) it is always ‘meaning-in-acontext’, i.e. something (and this is not restricted to language) is meaningful in a specific ‘context’. The conception of meaning between form and context is the view that is highlighted in the familiar image of stratification in SFL, with semantics as a stratum between context and lexicogrammar. The two dimensions of ‘meaning’ pointed out above are also present in the tool of the system network. The concept of ‘meaning’ as ‘that which is realized in form’ underlies the relationship between a valeur in a network, i.e. a systemic option (or rather, a path of systemic choices) and its realization in a structure (specified in a realization statement, see e.g. Schulz and Fontaine, this volume). In the tool of the system network, the other, contextual side of ‘meaning’ is present too, since systemic options that are grouped in one network are those that are available in a particular setting, i.e. in the context of a particular rank or unit, for example, the options of process type which are available at clause level. In this view ‘meaning’ is the value – or valeur – of an option in relation to other options that are available in a context, a value that is realized in a structure, which thus is a token of that value. Hence two basic conceptions of ‘meaning’ in SFL can be distinguished by looking at where the relation of ‘realization’ or encoding occurs in the model, and this is between strata, on the one hand, and between systemic options and realization statements, on the other. In Figure 3.1 these two conceptions of meaning/semantics are given in [1] and [2].1 1

Note that the duality of the initial two conceptions of meaning that are pointed out here is widespread in linguistics: the former, as the type of meaning that is interfacing with context, has also been called ‘contextual meaning’, or ‘extra-linguistic meaning’ (and more specific sub-types of this are ‘reference’, ‘ontological meaning’, ‘speech act meaning’, etc.), whereas the latter has been called ‘intra-linguistic’ or ‘formal meaning’ (with ‘sense’ as a sub-type), or has been defined in relation to grammar as the ‘semantics of grammar’.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005



ics nt

meaning

semantics

context



lexicogrammar

form

wordings

ph gra

[2]

paradigmatic valeur

value syntagmatic structure

token

realization statement

realization statement

systemic option

Figure 3.1 Four types of conceptions of ‘meaning’ in SFL

[1]

meanings

ram

doings

co n

context

sem a



lex -g

xt te

phon/

[3]

content plane

expression plane

[4]

language as a whole as ‘meaning-making’ as a ‘meaning potential’

58

MIRIAM TAVERNIERS

A third conception of meaning (Figure 3.1:[3]) appears when a differentiation between an expression side and a content side of language is prioritized: in that case ‘meaning’ is the content that is expressed in sounds/ writings. This idea underlies the basic structuralist split between ‘content’ and ‘expression’ in language, i.e. the two sides of a linguistic sign in Saussure’s model, which Hjelmslev (1963) called content and expression ‘planes’. In addition to those three views of ‘meaning’, there is a broader view (Figure 3.1:[4]) which is highlighted in typical characterizations of language from an SFL perspective: language as a whole is defined as a ‘meaning potential’, and learning a language is learning ‘to mean’ (see the title of Halliday 1975). Halliday’s adaptation of the familiar verb to mean to refer to this overall meaning of language highlights this fourth, more general conception of meaning: it is not just the semantic stratum that ‘means’, nor the content plane, but language as a whole makes it possible for us to ‘mean’ in the various contexts of our human lives. Hence we arrive at four basic conceptions of meaning/semantics: [1] meaning as valeur; [2] meaning as one stratum in relation to other strata in a stratified view of language; [3] meaning as the content side of language; and [4] meaning as what characterizes language in general as a meaning-making resource. These conceptions have been highlighted, focused on, and combined in different stages in SFL, but they also lie at the basis of a primary distinction between the Cardiff model of grammar and Halliday’s model of grammar. In the Cardiff model the distinction between system and structure is seen as primary, and falling together with the dividing line between semantics and lexicogrammar. Semantics is seen as the stratum of the system networks, whose options are then realized one stratum below, in lexicogrammatical structures. In other words, in this model semantics is paradigmatic and lexicogrammar is syntagmatic (see Schulz and Fontaine, this volume). In the traditional SFL model, which is focused on in this chapter, the ideas of meaning as valeur, meaning as an intermediate stratum, meaning as the content plane of language, and meaning as what characterizes language as a whole are not incompatible at all, but are intrinsically intertwined, and are all useful and necessary, as we will see below. In this view the different conceptions of meaning pointed out above are all valid side by side. At the same time, it has been explored what a semantics as a stratum in its own right can be and how this can then be modelled, a topic we turn to in Section 3.2.2.

3.2.2

Three Interrelated Conceptions of HOW Semantics Can Be Modelled In this subsection we focus on the conception of meaning as a stratum in its own right, i.e. that conception of meaning for which the nominal term ‘semantics’ is used in SFL. Afterwards (Section 3.3) we will return to how this semantics is related to other dimensions that are called ‘meaning’ or ‘semantic’ in the broader perspective sketched.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

59

Asking the question of what a (separate) semantic stratum can be, more specifically, correlates with asking what this stratum can contain, or how this stratum can be modelled in relation to other – especially surrounding – strata. In the visual metaphor of stratification, with the strata represented as cotangential circles (since Martin and Matthiessen 1991; see Figure 3.1), the higher levels of the language system are the more abstract ones, and the lower ones the more concrete realizations. In keeping with this orientation, semantics in general can be conceived of as ‘higher order valeur’: a ‘valeur’ which can be recognized above the lexicogrammar, as a more abstract meaning. Three complementary ways of modelling this ‘higher order valeur’ can be distinguished. I will initially characterize them here, and we will then return to the different types of modelling in more detail when looking at specific semantic analyses (Section 3.3) after we have addressed the ‘why’ question (Section 3.2.3). (a)

2

Semantics as topological meaning: In one view, semantics is the stratum at which areas of the lexicogrammar are regrouped into semantic domains. This is the case when distinct lexicogrammatical phenomena (in different networks and/or at different ranks or units) realize a similar motif at a higher, semantic level. In a topological model (see Martin and Matthiessen 1991) meanings are not organized systemically (or ‘typologically’), in networks of distinctive valeurs, but in terms of their likeness along one or more dimensions: phenomena that are similar in one or more respects are conceived of as areas that are closer to one another and as belonging to the same domain in a larger multidimensional space.2 The method of (re)grouping lexicogrammatically distinct phenomena into more abstract domains or components has often been used in SFL in addition to modelling meanings in networks.3 Two topological semantic concepts that are familiar in SFL are the metafunctions and the motif of logico-semantic relations. The metafunctions are first and foremost ‘semantic components’ of language (e.g. Halliday 1977), and they are groupings of phenomena which are dispersed over different networks at the lexicogrammatical stratum, for instance, networks for different ranks (with, e.g., ideational meanings realized in clauses, in clause complexes, and in groups). Logico-semantic relations, i.e. expansions and projections

In this respect the topological modelling of (higher-order) meanings is similar to the method of semantic maps which is used in functional typology (e.g. Haspelmath 2003).

3

Note that the ‘regrouping’ of distinct lexicogrammatical phenomena into topological ‘domains’ does not necessarily imply that those domains are conceived as pertaining to a different (i.e. usually ‘higher’) stratum. Martin and Matthiessen (1991) talked about regroupings within a stratum, with a focus on the lexicogrammar. This has been interpreted by Halliday (1996:15) as suggesting a view of lexicogrammar as typologically organized and semantics as topologically organized strata. This is also the view highlighted here, although it should be borne in mind that the topological/typological distinction does not correlate with semantics/lexicogrammar per se (indeed, it perturbates throughout the system as a fractal motif – see Section 3.4.2 below on fractality and 3.4.3 on the related concept of an extravagant theory of language).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

60

MIRIAM TAVERNIERS

and their subtypes, can also be realized by different lexicogrammatical means, in different units (see Halliday’s (1985) synoptic table of ‘expansion’, and Halliday and Matthiessen’s (1999) more extended treatment of the various manifestations of ‘expansion’ and ‘projection’ across lexicogrammatical means). (b) Semantics as discourse-structural meaning: In a second type of view, semantics is the stratum at which patternings at the level of discourse are modelled as patterns of unfolding text which are larger than the structures recognized at the level of lexicogrammar, where the maximal unit, in terms of size, is the clause complex. Thus in this sense, the SFL concept of cohesion is an intrinsically semantic concept (see Halliday and Hasan 1976:4), and by extension, since it is cohesive relations that create texture and hence make a text, a text is by definition a semantic unit. Note that ‘discourse-structural’ meaning is here intended as a type of structure which is different from lexicogrammatical structure, the latter being structure in the traditional, narrower sense. In this vein, too, cohesion is often characterized as a non-structural phenomenon. (c) Semantics as higher-level systemic meaning: In a third conception of semantics, the semantic stratum itself is organized in terms of system networks that are superimposed upon lower lexicogrammatical networks. In this view, options in semantic networks are realized by options in lexicogrammatical networks. This conception underlies the model of speech functions in SFL (see Halliday 1984). speech function is a semantic system with options such as ‘command’ (asking for goods and services) or ‘question’ (asking for or about information). Those semantic options can be realized by (various) lexicogrammatical options from the system of mood: e.g. a ‘command’ can be realized as an imperative (Open the window), or as an interrogative (Could you open the window for me?). Figure 3.2 summarizes those three design alternatives of a semantic stratum in the overall architecture of SFL. These alternatives are not mutually exclusive but complementary. In much recent work on semantics in SFL, the organization of the semantics in terms of system networks has been an important objective, but in many cases this systemic modelling went hand in hand with and/or was inspired by insights from topological and discourse-structural views of semantics, as we will see below (Section 3.4).

3.2.3 A Closer Look at Stratification: The WHY of Semantics The answer to the first why question – Why is it necessary to recognize a semantic stratum? – is that a semantic stratum is needed in order to account for ‘variability’ between expression and content functioning in different contexts. Recognizing a semantic stratum is necessary when there is no one-to-one relationship between expression and content, i.e. when a coupling between a form/structure, on the one hand, and a meaning, on the

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

61

context

semantics

[1]

[2]

[3]

lexicogrammar

Figure 3.2 Three design alternatives in modelling a semantic stratum: topological [1], systemic/typological [2], and discourse-structural [3]

other hand, is not enough. For some systemicists or from one perspective, this is always the case, for others it is the case for specific contexts (as we will see below). In order to come to an understanding of how the recognition of a stratum of semantics is motivated in relation to this concept of variability, it is necessary to first clarify what exactly this variability is, and how different research purposes can put this variability into different types of perspectives.

3.2.3.1 Variability and the Internal Stratification of the Content Plane Theoretically, if there is no one-to-one relation between content and expression, there are two types of situations which are possible, and these two can also occur in combination. This is visualized in Figure 3.3. At the left (Situation [1]) is the type of situation that is familiar from the systemic model of the lexicogrammar, with systems in which each valeur (or endpoint in a systemic path) is tied to a specific expression form, i.e. there is a one-to-one relation between meaning and form. In the centre are two complementary possibilities of variation: one-to-many, where one meaning is dispersed over different expression forms (Situation [2]), or many-to-one, where one form can be the expression of different types of meanings (Situation [3]). At the right is the combination of Situations [2] and [3].4 4

Lamb (1962) (whose stratificational theory has been a source of inspiration in SFL) refers to those different situations of variability between strata as ‘composite realization’ (=[2] here), ‘portmanteau realization’ (=[3]), and ‘interlocking diversification’ (=[4]).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

62

MIRIAM TAVERNIERS

meanings [1]

[2]

[3]

[4]

f forms Figure 3.3 One-to-many [2], many-to-one [3], and many-to-many [4] conceptions of variability between meanings and forms in relation to the default view of one-to-one meaning-to-form couplings within lexicogrammar [1]

In research contexts where a semantic stratum comes into play (to be explored in Section 3.5), the analysis and description of the meaning of a linguistic sign as a content that is tied to an expression in a (lexicogrammatical) system (Situation [1] in Figure 3.3) is not sufficient or adequate. In those cases, it is useful to ‘un-couple’ content and expression, and to recognize an additional, higher level of content which is not covered in the content in the lexicogrammatical system.5 In more general terms, one specification of a content linked to its expression is not enough: the content plane of language is split into two strata, i.e. lexicogrammar and semantics. This is referred to by Halliday as an ‘internal stratification of the content plane’ (Halliday 1976, 1998a; also see Taverniers 2011 for a further exploration). The additional interface in the content plane, i.e. between the strata of semantics and lexicogrammar, is set up in order to account for variability between contents and expressions.

3.2.3.2

Semantics as an Interface and Trinocular Perspectives on Variability The description of the internal stratification of the content plane above focuses on the relation between lexicogrammar and semantics as two content strata of language. However, semantics is not only inserted above lexicogrammar, but also ‘below context’: it is a stratum that is wedged between lexicogrammar with its systemic model of linguistic forms and what lies outside language, i.e. the non-linguistic context. In this sense, the two content strata are interface strata, and semantics is an interface stratum through which language interacts with extra-linguistic context, i.e. with eco-social environment (see Halliday 2013). We saw above that the stratification of the content plane into semantics and lexicogrammar makes it possible to account for variability between contents and expressions. This variability is an inherent feature of language functioning in different contexts: it is in the interaction with different contexts, and in the actualization of language in specific instances, that variation arises. 5

In Hjelmslev’s (1963) terms, a ‘connotative semiotic’ is then recognized (i.e. a semiotic that has a higher-order content plane) (see Taverniers 2008).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

63

More specifically, stratification makes it possible to create ‘meanings’ that are adapted to specific contexts and that are beyond what is (fixed or ‘codified’) in the (lexicogrammatical) system of a language. Stratification makes it possible to use ‘forms’ in ways that go beyond their valeurs in lexicogrammatical systems, for instance, to mean several things at the same time, i.e. to be creative in a myriad of ways with the finite means that are available in the formal units of the language. Thus the concept of stratification is how SFL theorizes the way in which language interacts with non-linguistic context. Language appears in its material expression (i.e. in sounds and writing), but as a ‘semiotic’ system it does not interact with the eco-social environment ‘directly’ through sounds and writings: there are intermediate layers of content between linguistic expressions and the contexts in which language functions. First, there has to be a level of ‘meaning’ (there is a separation between ‘content’ and ‘expression’, which is the principle of double articulation); and second, in order to account for variability, there have to be two content strata (this is the separation between ‘semantics’ and ‘lexicogrammar’ within the content plane) – and this is a hypothesis about how language functions in ‘context’. In keeping with the visual metaphor of strata as organized with context at the top and the expression in sound or writing at the bottom, any type of phenomenon can be viewed from three perspectives. Focusing on semantics, this stratum can thus be seen ‘from above’ (from context), ‘from below’ (from lexicogrammar), and ‘from roundabout’ (from its own position in the stratified model). The possibility of those complementary perspectives is referred to as ‘trinocular’ vision or perspective (Halliday 1977, 1996), or ‘trinocularity’ (Halliday 2009:79–80). Different types of questions arise about what a semantics can be, depending on the type of stratal perspective that is taken on semantics as an interface. This type of perspective, together with the type of variability between lexicogrammar and semantics that is focused on, determines the more specific reasons for setting up a semantics within the architecture of SFL, and also how this semantics can be conceived of.

3.2.3.3

A Specific Motivation for Setting Up a Semantics: Grammatical Metaphor The need for a separate semantic stratum came into sight in various research perspectives in SFL in which (a type of ) variability between content and expression came to be highlighted. One of those, which has been highly influential in the theory as a whole, is the area of grammatical metaphor.6

6

Other areas in relatively early SFL studies which point to a semantic stratum in addition to a lexicogrammatical one include the following: stylistics and socio-semantic variation (esp. with the notion of a de-automatization of a grammar à la Mukařovský, e.g. Halliday 1982); language development (ontogenesis) (with the view that the adult language system contains more strata – an extra content stratum? – than the proto-language of the child); and mood (with variability between mood choices and socio-semantic roles, on the one hand, and between mood types and lexicogrammatical realizations, on the other hand).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

64

MIRIAM TAVERNIERS

The concept of grammatical metaphor instrinsically has to do with the lack of a one-to-one relationship between meaning and form, and it is already present, in a pre-theoretical way, in earlier concepts such as markedness and foregrounding, as well as congruent (typical) and incongruent (atypical) links between semantics and lexicogrammar. Grammatical metaphor becomes the more general label for what had already been partly recognized in the interpersonal component, on the one hand (where one speech function can be realized in different ways), and similar types of variation in the ideational component, on the other hand. Halliday (1984) calls the typical, default links between speech functions and their realizations in lexicogrammar (through the system of mood) ‘congruent’, and the less typical ‘alternative’ realizations ‘metaphorical’. In this sense a command can be realized congruently by an imperative (Please, open the window for me), or metaphorically by a modalized interrogative (Could you open the window?). In Halliday’s (1984) first presentation of ‘grammatical metaphor’, such incongruent realizations of speech functions are called ‘metaphors of mood’. Later, in Halliday (1985), within the interpersonal component, a second type of metaphor is recognized, viz. ‘metaphors of modality’, in which a modal meaning is realized, not by means of a modal element in the clause structure (i.e. a modal verb or a modal adjunct: It will (probably) be a good season) – which is regarded as the congruent realization of modal meanings, but rather by other means which are beyond the clause structure and which give a more ‘explicit’ wording of the modality intended (as in I think it will be a good season; I suppose it will be good season; It is obvious that it will be a good season; Everyone says it will be a good season). Figure 3.4 presents a visual presentation of interpersonal metaphors in terms of the variability between content and expression as sketched above. In presenting the concept of grammatical metaphor, Halliday (1985) explicitly combines two complementary views of variability between forms and meanings. He describes the traditional concept of what is ‘metaphorical’ as highlighting a ‘many-to-one’ type of view of the relation between meaning ‘command’

‘probability’

projecting relational clause It is obvious that …

modalized interrogative Could you … imperative*

modalized declarative You should …

*congruent realization

modal elements will, probably*

projecting mental clause I think, I suppose …

*congruent realization

Figure 3.4 Interpersonal grammatical metaphors of mood (left) and modality (right) theorized within a view of variation of one-to-many

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

‘part of an animal’*

‘part of a piece of furniture’

‘event’

leg expressed as clause* *literal meaning

65

expressed as nominal

*congruent realization

Figure 3.5 The traditional view of metaphor (left), and an alternative view which is equally inspiring in the study of grammatical metaphor (right)

and expression: there is one form, and this form has multiple meanings, among which are at least one literal and at least one metaphorical interpretation. This is the conception that prevails in thinking about metaphor in general, and especially in traditional views of lexical metaphor (e.g. legs can be ‘parts of animals’, i.e. have a literal meaning, or ‘parts of furniture’, as in the lexical metaphor table legs). However, Halliday argues, the complementary view, i.e. of ‘one-to-many’, is equally important in analyzing grammatical metaphor (see Figure 3.5, which shows the relation between those two views). Thus in introducing ‘ideational grammatical metaphor’, Halliday (1985) gives an example of a clause which realizes a process with its participants (i.e. a processual or ‘event’ meaning), and then shows that this processual meaning could be realized by a nominal group, which functions as a participant in another process configuration. The first example offered by Halliday concerns the congruent wording Mary saw something wonderful, compared to the incongruent A wonderful sight met Mary’s eyes or Mary came upon a wonderful sight, where the process of ‘Mary seeing something wonderful’ is construed as a nominal group, a wonderful sight. In explaining grammatical metaphor as ‘one-to-many’ in addition to ‘many-to-one’, Halliday both connects grammatical metaphor with the traditional concept of metaphoricity (‘many-to-one’) and brings the new concept of grammatical metaphor more in line with what had earlier been recognized in relation to typicality in the interpersonal metafunction (where the ‘one-to-many’ view was the initial source of inspiration: one speech function such as ‘command’ can have several realizations such as ‘imperative’, ‘interrogative’, etc.). In terms of the stratal type of perspective, the approach to grammatical metaphor that is initially taken is that from below: the starting point is a form, and the question is what this form ‘means’, and then this ‘meaning’ is analyzed as having other

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

66

MIRIAM TAVERNIERS

(alternative, metaphorical) ways of being realized. Section 3.5.1 explains how flipping the traditional conception of metaphor, starting from above and asking how one meaning can be realized, has led to a more comprehensive investigation of grammatical metaphor.

3.3

Wider Perspective I, Flexibility: Enter Metaredundancy and Probability

The overview of initial conceptions of ‘semantics’ in SFL has shown that there is no single explanation of what (a) semantics is, where it is to be found, how it can be modelled, if it is regarded as a separate stratum, and why it should be recognized as such. Even in those cases where the focus was on a (separate) semantic stratum, the description was couched in terms of ‘certain contexts’, or ‘certain research purposes’ for which such a semantics comes into view. Indeed, even a cursory reading of the systemic functional literature will reveal that the familiar systems of the clause, viz. process type, mood and modality, and theme and information, are sometimes called ‘lexicogrammatical systems’, and sometimes ‘semantic systems’. This (seeming) ‘indeterminacy’ in the description has to do with the very conception of the interface between semantics and lexicogrammar as by definition inherently flexible. In order to understand this flexibility, it is necessary to put the notion of stratification in a wider perspective, and to take into account another relation that has hitherto not been mentioned, viz. the relation between the system and actual instantiations, or what is called ‘instantiation’ in SFL. This wider perspective is that of viewing language as a dynamic open system, and bringing in the concepts of ‘metaredundancy’ and ‘probability’.

3.3.1 Metastability and Metaredundancy Since the mid-1980s, and under the influence of work by Lemke (1984), Halliday has come to view language as a ‘dynamic open system’, or a ‘metastable system’. Such a system persists in interacting with different contexts by being open, i.e. being maximally adaptable, and thus by constantly changing, while at the same time avoiding change (i.e. it remains stable at a meta-level). We saw above that the concept of stratification is how SFL interprets the way in which language interacts with context or its eco-social environment. The model of stratification also provides the ideal basis to talk about how language can be metastable, but the concept of dynamic openness puts the idea of stratification into a wider perspective. As a semiotic system, language is able to continue to function because of the combination of two features: the availability of ‘critical contrasts at every level’, and ‘complex arrangements of articulation of these contrasts’

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

67

(Lemke 1992), i.e. complex arrangements of the relations between those options in the different strata. Thus, language achieves metastability through the particular type of relationship between its coding layers or strata. Those layers consist of things going together in a predictable way (to limit change and thus ensure stability), while at the same time they contain some ‘gaps and contradictions’ which provide room for innovations, which originate as adaptations to changes (i.e. new requirements) in the context (Lemke 1995:175). Focusing on the relation between semantics and lexicogrammar, the ‘things going together in a predictable way’ are stable meaning–form relations (such as one-to-one relations), while the ‘gaps’ can be meanings that do not have one default realization (meanings that are not (yet) codified) and the ‘contradictions’ meanings that (come to) have multiple different realizations, or the other way around. Thus, the notion of complex arrangements between options in the different strata corresponds to the presence of various types of variability that we distinguished above. This type of relation between layers, based on such complex arrangements of variability, is what characterizes ‘semiotic’ dynamic open systems, of which language is the paramount example, and is called ‘metaredundancy’. In a pair of layers with only one-to-one meaning–expression couplings, the layers are ‘redundant’ vis-à-vis each other (there is no need to differentiate them, since the one is always completely predicted by the other). When gaps and contradictions are present, the relationship is said to be ‘metaredundant’: in the system as a whole, there is redundancy to a certain/minimal extent, and there is room for new connections between layers, i.e. there is room for creativity or adaptations to new contexts. Innovations arise on top of and by virtue of the existing system, and it is also in this sense that the extra layer needed to model them is metaredundant with the existing system. Within the perspective of language as a dynamic open system, the concept of metaredundancy provides a new interpretation of the relationship between strata, which is traditionally called ‘realization’ in SFL (see Halliday 1991; Lemke 1995). In this perspective, the verb realize, too, is reinterpreted as ‘redound’. What metaredundancy adds to realization, besides the notion that it is an explicitly dynamic concept, is the fact that it is a relationship of accumulative nesting, i.e. it is not sequential but, rather, bidirectional. Thus, lexicogrammar redounds with phonology/ graphology; semantics redounds with the relation between lexicogrammar and phonology/graphology; and context redounds with the relation between semantics which redounds with the relation between lexicogrammar and phonology/graphology. This is the view that is most relevant when focusing on the content side of language, i.e. ‘meaning’. The visual image of the strata with the cotangential circles (which appeared around the time that the theory of dynamic open systems became influential in SFL) ranging from smaller (for the lower systems) to larger (for the higher systems) represents exactly this view. However, while realization imposes a direction

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

MIRIAM TAVERNIERS

xt te on

se

xt te on

s tic an m

s tic an m

s tic an m

s tic an m

ar

m am gr

lex

ico

mar am

gr

lex

ico

se

se

se

c

c

68

m am gr

lex

ph

phon

phon

ph

ra /g

ico

ar

lex

ico

ar

m am gr

ra /g

Figure 3.6 Metaredundancy relationships between strata, starting from the top focusing on the content side of language (left), and starting from the bottom focusing on the expression side of language (right)

on the relation (i.e. it forces us to choose one thing that realizes the other, as with the terms ‘signifier’ and ‘signified’), metaredundancy is neutral in this respect. Accordingly, the opposite view with metaredundancy cycles starting from the lowest stratum becomes a complementary frame of interpretation. Here, phonology redounds with the metaredundancy between lexicogrammar and semantics, and if context is also taken into account, phonology redounds with the metaredundancy between lexicogrammar and the metaredundancy between semantics and context. This view, as Halliday (1992) argues, is the most relevant in research focusing on phonology. Figure 3.6 is an attempt to capture accumulative metaredundancy cycles in a three-dimensional way. The conception of language as a dynamic open system puts the motivation for distinguishing between lexicogrammar and semantics into a wider perspective. It is because of the requirement of adaptability of the system to different contexts that the distinction between lexicogrammar and semantics is needed, because it is exactly through the flexible relation between lexicogrammar and semantics that language can be open, and that gaps and contradictions, as room for innovation, can be allowed. In other words, it is through variability between meanings and forms that language can continue to function. It is through the specific type of relationship of semantics, not just to lexicogrammar, but to the rest of the lower layers, that language has semogenic power, that it is a ‘meaning potential’.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

69

In addition to the new perspective on the distinction between semantics and lexicogrammar, the conception of language as a dynamic open system also provides a wider frame of interpretation for the observation (see Section 3.2.1) that different conceptions of ‘meaning’ are possible and relevant at the same time in SFL. The four different conceptions of ‘meaning’ distinguished above have a place in the model of language as a dynamic open system: •

• •

•

‘Meaning’ as the stratum of semantics itself: this can now be viewed in various ways, at least two of which are relevant here. (i) Within the content plane of language, it is the stratum that redounds with lexicogrammar, and focusing on content, the interface with lexicogrammar is the place where variability between meanings and forms is made possible. (ii) In language (content + expression), semantics is the highest stratum, which metaredounds with the rest of the language system. This view of semantics is proposed in SFL in connection with Hjelmslev’s notion of a connotative semiotic (e.g. Halliday 1991). Here, semantics is seen as a connotative layer which interacts with an existing semiotic (in Hjelmslev’s terminology, the latter is a denotative system and functions as the expression of the connotative content). This is the idea of a new stratum coming into existence on top of and by virtue of the existing system. ‘Meaning’ as the content plane of language: this is what metaredounds with the expression side, with phonology. ‘Meaning’ as language as a whole: in this view language is interpreted as a meaning potential (a potential that allows us to function in different contexts): this is language metaredounding with context. ‘Meaning’ as the valeurs in system networks: this view is the least clear in the visualizations with layers. However, if the concept of language as a dynamic open system is carried through, this, too, is to be seen as a level that redounds with the structures. For the valeurs are only an objectification of language by the researcher. The relationship between the valeurs and the structures is a relationship with much stability (one-toone relations), but here too, there is room for innovation in order for language to persist as a metastable system.

These different conceptions of ‘meaning’ can be highlighted in the framework of language as a dynamic open system. The validity of all those views within SFL creates what can be seen as ‘indeterminacy’, as in, for example, alternative characterizations of the familiar networks of mood, process type, and theme as ‘semantic’ in some works, and as ‘lexicogrammatical’ in others. What counts is that both of these ‘content’ strata are meaning (as a verb, i.e. meaning-making). This is explicitly captured in the idea of metaredundancy: semantics metaredounds with the meaning that is already in the relationship between lexicogrammar and phonology. It is in this sense, too, that grammatical metaphor is theorized as ‘stratal tension’ (e.g. Martin 1992) (i.e. tension between a literal and a metaphorical meaning), because both the lexicogrammar and the semantics convey meaning.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

70

MIRIAM TAVERNIERS

3.3.2 Time, Instantiation, and Probabilities In discussing meaning and semantics in the framework of language as a dynamic open system, so far in Section 3.3 the focus has been on the status of strata and the relations between them, prioritizing the relation of realization and its reinterpretation as metaredundancy. We have seen how the concept of metaredundancy places the views of ‘meaning’ distinguished above in a different theoretical perspective: both the types of ‘meaning’ that are distinguished and the idea of variability between meanings and forms are elucidated through metaredundancy. As a next step in exploiting this theoretical frame of interpretation, a further dimension needs to be taken into account, viz. time and the concomitant notion that is called ‘instantiation’ in SFL. We will see that the same framework also provides a deeper understanding of the kinds of ‘semantics’ that first appeared at the horizon of SFL’s ventures into semantic theory (see Section 3.2.3.3). One dimension that plays a crucial role in the picture of language as a dynamic open system is that of ‘time’, and it will be noted that the description above is inevitably couched in terms of temporal meanings (witness expressions such as ‘persist’, ‘constant’, ‘change’, ‘new’, ‘always’, ‘at the same time’). Metastability is an inherently dynamic concept: a system that is metastable is stable by ‘constantly changing’. Hence metastability is in fact ‘dynamic stability’, although this term may seem a contradiction. This apparent contradiction is due to the role of time, or rather, the way in which, as researchers, we observe and objectify language, which only occurs to us in its usage in unfolding time. What appears as a difference between two states of language that are objectified for the sake of research (i.e. two time slices across which language does not appear as ‘stable’) may appear as actualizations of the same meaning potential if observed from a further perspective, from a greater time depth. Vice versa, what is objectified as a stable system (in research, and also for the purposes of learning a language as a technique) is just that: an objectification (abstraction) of a system that is constantly in flux in a myriad of actualizations in different situations. This is the paradox of dynamic stability. As a meaning potential (which is virtually stable and thus learnable), language predicts what forms and meanings can be actualized in a certain context, and at the same time it is responsive to new requirements in those contexts, because it contains gaps and contradictions. Hence language is constantly open to allow and incorporate what appear as ‘innovations’ within its system, but what are in fact nothing more than actualizations of the system in interaction with different environments. In relation to the dimension of ‘time’, those innovations – which may or may not become part of the stable system over time – can be conceived of in different ways, depending on whether or not they are incorporated in the system and depending on the research perspective that is taken. The research perspective refers to the time depth that is taken, i.e. what time frame of language is made the object of study: language as a codified system or language in

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

71

vivo in one particular instantiation (a text) or a particular setting. Innovations can be conceived of as transformations of the system, as changes, as deviations, as metaphors, as ‘complexifications’ of the system (adding a further choice in the existing system, i.e. a more fine-grained distinction). It will be noted that the different types of motivations for a separate semantics that we have considered exemplify some of those different conceptions of ‘innovations’. In relation to the notion of ‘variability’ (between meanings and forms), innovations can also be conceived of in different ways, which are complex to describe because the dimension of time (and the research focus in terms of time depth) also plays an intrinsic role. ‘Change’ can be based on diversification (from one-to-one to one-to-many or many-to-one), or it can be a change in the existing proportionalities between options, e.g. options may become more or less ‘at risk’ in certain environments, while other options die out and new options appear. In essence, all change has to do with changing probabilities in a system (see Halliday 1991). Any innovation in language can be purely ‘accidental’, it can become part of a situation-specific system (e.g. a register), or it can become part of the overall system. In this sense, each instance potentially has an influence on the system. Halliday (e.g. 1985) often uses a comparison with the weather. Each weather situation is an instance of a climate, but also influences that climate: when there are more wet winter days, the winter climate will become wetter too. A change that becomes part of a system needs to be incorporated into existing networks. If the change is based on diversification, this leads to a complexification of the system wherein more finegrained distinctions will have to be set up. When language changes (i.e. evolves), it is the couplings between layers that change, or more precisely the relations of variability between strata: relations are opened up and new relations appear, or, in other words, strata are uncoupled and recoupled again. Grammatical metaphor, which has been a key reason in SFL for distinguishing between the strata of semantics and lexicogrammar, is conceived of in exactly those terms: it is seen as ‘a cross-coupling (decoupling, and recoupling in a different alignment) between the semantics and the lexicogrammar)’ (Halliday 2008:16). Note also how ‘metaphor’ in general is seen as the creation of new form-meaning couplings which come into being on top of and by virtue of existing couplings.

3.4

Wider Perspective II, a Design Rationale Based on Multiperspectivism and Fractality: Enter Extravagance

Throughout the step-by-step characterization of ‘meaning’ and ‘semantics’ in SFL in this chapter, there has been a prevailing emphasis on ‘different views’, ‘different possible conceptions’, and ‘perspectives’. This feature has

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

72

MIRIAM TAVERNIERS

been explicitly thematized in SFL, at a meta-level, i.e. at the level of how we think about how we can model language, or the level of our research practice as linguists. Because it plays such an important role in understanding how semantics is conceived of in SFL, it is useful to consider how this idea of different views is given shape at the meta-level. There are two interacting aspects to this feature, viz. ‘multiperspectivism’ (Section 3.4.1) and ‘fractality’ (Section 3.4.2). The combination of these leads to a specific type of theory that is called an ‘extravagant’ theory (Section 3.4.3).

3.4.1 Multiperspectivism An inherent feature of SFL is that it is ‘multiperspectival’. Its theory of language is based on a number of dimensions along which differentiations can be made (see Taverniers 2002 for a further exploration of ‘differentiating dimensions’), the most important of which are the distinction between metafunctions, between different strata (stratification, as well as the relationship of realization or metaredundancy), between ranks, between system and instance (instantiation), and between syntagmatic and paradigmatic modelling (see Halliday and Webster 2009). Within each of those dimensions, a linguistic phenomenon can be looked upon in different ways, and different perspectives are possible, depending on the viewpoint and the focal depth: •

•

Perspectives can differ in terms of directions, when views from different vantage points are possible. The stratal types of perspectives mentioned above are a case in point: any phenomenon viewed in terms of stratification can be looked upon from below, from above, or from roundabout. Hence trinocularity is just one instance of this feature of multiperspectivism. Perspectives can also differ in terms of focus. The investigator can single out one focus point in a set of complementary views (e.g. look at a phenomenon from one specific metafunction). Another example of differences in focus is that where focal length differs in terms of time depth, which is captured in the notion of ‘instantiation’ in SFL.

Any linguistic phenomenon can thus be looked at from different perspectives, within one dimension, and also in terms of different dimensions, and this leads to a theory that is explicitly based on complementarities. The combination of complementarities leads to a richer theory about a complex phenomenon, and it is through ‘shunting’ perspectives (see Halliday 1961:254) that the analyst can come to a better understanding and a more detailed description of a phenomenon. Grammatical metaphor can again be mentioned as a case in point. We saw above how Halliday called for a shift in the perspective that is taken on the variability between lexicogrammar and semantics that is inherent in metaphor: a shift from the traditional view of ‘one expression has

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

73

context

semantics ‘sequence of events’

*

clause complex

clause

‘event’

‘entity’

*

*

(nominal) group functioning in a clause

word functioning in a group

*congruent realization

lexicogrammar y/graphology phonolog Figure 3.7 Shunting perspectives in theorizing grammatical metaphor as a tension between lexicogrammar and semantics

multiple meanings’ (a metaphorical and a literal one) to a view in which one meaning can be expressed in various ways. By shunting the perspective in this way (see Figure 3.7 and compare Figure 3.5 above), and by looking at metaphor from the semantics, SFL has been able to develop a more comprehensive theory of ideational metaphor. This development went in various steps, each further step adding more conceptual depth to the theory of grammatical metaphor – but all steps hinge on the flipped perspective compared to the more traditional view of ‘metaphoricity’. First different types of realizations were conceived of in terms of degrees of congruence (i.e. one meaning, such as an event meaning, can be expressed in various ways, from more to less metaphorical). Concomitant with this, the view developed that an ideational metaphor can be ‘unpacked’ in different steps, from relatively incongruent to more congruent ways of wording the same meaning. Then the different realizations were put in a type of implicational hierarchy, from most ‘relational’ or

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

74

MIRIAM TAVERNIERS

context

semantics ideational meaning

interpersonal meaning

textual meaning

combined syntagm

lexicogrammar y/graphology phonolog Figure 3.8 Shunting perspectives in theorizing the metafunctions as semantic components of the system of language

‘processual’ meanings to condensed meanings with the noun as the end point (see Halliday 1998a:211). This in turn laid bare different possible points in a general ‘drift towards thinginess’ that characterizes those registers which often make use of packaging meanings in a condensed form, such as scientific discourse. Another example of taking different perspectives is related to the concept of the metafunctions. The metafunctions can be viewed as semantic zones, which are realized in different lexicogrammatical areas. Alternatively, each syntagm, i.e. each structure at the level of lexicogrammar, has all three metafunctional meanings. This is a familiar SFL view of the multi-tiered syntagm as a concerted meaning, i.e. a fusion of layers of metafunctional meanings. Figure 3.8 shows how these two views on the metafunctions are complementary, and it is clear that both are needed in order to come to a deeper, richer understanding of the notion of the metafunctions as

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

75

‘semantic components of the system’ (while at the same time (i.e. metaredundancy at play!) depending critically on the bundling of systems at the lexicogrammatical stratum).

3.4.2 Fractality By combining complementarities in order to come to a richer theory of the complex phenomenon that language is, sometimes recurrent patterns are discovered, and this feature is referred to in SFL as ‘fractality’ or ‘fractal resonance’ (Martin 1995). Thus, fractality is an additional design feature which is exploited at the meta-level, the level of the linguistic practice itself. Put simply, a pattern that appears in one view, in one focus, at one level – i.e. a type of modelling that is useful in accounting for a phenomenon within that view – may also be recognized elsewhere, in another view that appears by shunting the perspective. The newly recognized pattern is then said to ‘resonate’ with the first one. Focusing on the meta-level, resonance means that a model, a type of distinction, which works well for one area, one view, or one dimension, is used as an inspiration to explore a different area, view, or dimension of language. The resonance may be between different ranks or units. This is the case with the semantic motifs of expansion and projection (see Section 3.2.2). A meaning such as causality (as a subtype of expansion: enhancement) can be realized between clauses within a clause complex (This happened because that happened.), within clauses through relational processes (This caused that.), and within groups (The cause of that was . . .) (see Halliday and Matthiessen 1999:222–6 on this and related types of fractal resonance). Such resonating lexicogrammatical phenomena can be grouped together into a topological domain, or a ‘(transcategorical) semantic domain’ (see the notion of semantic domain in Halliday and Matthiessen 2004:593–4). The resonance may also be inter-stratal. This kind of parallelism is explored when the model of lexicogrammar, which is relatively familiar, is used as an inspiration to tackle the stratum of semantics. This type of inter-stratal fractality, which ties in well with the notion of metaredundancy between levels of the system, will become clearer in Section 3.5, which focuses on recent semantic models in SFL.

3.4.3 Extravagance We have seen how different conceptions of semantics are not just possible alternative views (see Sections 3.2.1 and 3.2.2), but exist side by side and moreover, all appear, naturally, in a view of language as a dynamic open system based on metaredundancy relations between its layers (see Section 3.3.1). What has now been added, in Section 3.4, is the idea that this varied view of semantics is ‘fostered’ in SFL. A theory which is based on complementarities and fractal patterns resonating across its components, is

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

76

MIRIAM TAVERNIERS

called an ‘extravagant’ theory by Halliday (e.g. 1998b), and it is argued that linguistic theory as a meta-semiotic can only be extravagant, because language itself is extravagant (see Halliday 2008). Language itself is extravagant by having order while at the same time constantly allowing disorder in its adaptation to different contexts. The linguistic theory that attempts to understand language in all its facets therefore cannot but be extravagant as well. Thus, the meta-semiotic is itself a dynamic open system, with some things that have become conventionalized, but with gaps (evidently, and luckily!) and, importantly, contradictions (alternatives, complementarities). And note that drawing the parallel between language and linguistics (see Halliday 1998b) is pointing to a fractal resonance between the two.

3.5

Recent Semantic Endeavours in SFL

3.5.1 Mapping Semantic Models in SFL The aim of this section is to give a brief but systematic overview of semantic models that have developed in SFL, most of them since the 1990s. The individual models will not be presented in detail since the focus will be on elucidating the nature of those models against the explanation of ‘meaning’ and ‘semantics’ that is sketched above, and in doing so, on highlighting the specific contribution and hence ‘place’ of each semantic model in relation to the theory as a whole. More specifically, this section will show how different semantic models in SFL each flesh out a specific dimension of a semantic stratum, i.e. a facet of semantics which comes into view by taking a specific perspective. The perspectives that are taken on semantics can be explained as originating from choices that are made in terms of the complementarities/options that have been disentangled in the preceding sections, viz. •

• •

the way in which semantics is modelled in addition to the aim of networking (i.e. topologically or discourse-structurally, or both) (see Section 3.2.2); the role and the type of variability between semantics and lexicogrammar (see Section 3.2.3.1); the type of stratal perspective that is taken (see Section 3.2.3.2).

We will draw upon the various types of complementarities distinguished above in exploring how recent semantic models are designed in SFL. In doing so, we will come to a theoretically grounded understanding of recent conceptions of semantics in SFL, of what a semantics can be, and of how it can be modelled. In doing so, further on, we will once more re-visit the crucial concept of stratification, and further flesh it out on the basis of the specific models of semantics that will be under focus. In turning to recent approaches to semantics in SFL, there is one tendency that becomes clear: semantics has become ever more important in

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

77

the theory as a whole (see Butler and Taverniers 2008). This is in vein with a key aim in SFL since its inception, viz. to make a theory that prioritizes meaning, to set up a system that is as semantic as possible (a ‘semanticky’ grammar, see Halliday and Matthiessen 2014:31), or a grammar that is ‘pushed’ as far as possible towards the semantics (see Halliday 1985:xix), but the ‘semanticization’ of the model lies more specifically in the fact that the modelling of the ‘semantic stratum’ has become an important research aim across different areas and metafunctions. In the later, more fully-fledged semantic endeavours in SFL, there is continuity and also a major new research focus. There is continuity from the initial motivations for separating semantics and lexicogrammar in that central themes have remained important to the present day: grammatical metaphor, and the semantics of speech function (as seen in Hasan’s model of message semantics, see Section 3.5.3). Socio-semantic variation and a focus on text also remain important. At the same time, it is possible to recognize an oscillation in a specific approach to semantics, which is inherently in accordance with the overall architecture of SFL, viz. the modelling of (separate) semantic ‘networks’ which then interact with the lower networks in the lexicogrammar. This had already started in earlier work on the interpersonal metafunction focusing on the relation between speech function and mood. In more recent models this approach is generalized to other interpersonal domains (e.g. stance and sourcing of stances), to the ideational metafunction and to the analysis of discourse. Hence, what we see in the design of the stratum of semantics is fractality, with the method of system networks being fractally extended from the lexicogrammar to the semantics (see Matthiessen 2009:14–15). Beyond this strong ‘systemic’ motif, the other two types of modelling are also important sources of inspiration, viz. ‘topological’ and ‘discoursestructural’ designs. As indicated above, those three types of designs are complementary, and although the ultimate aim may be a systemic model (as in recent semantic models in SFL), the topological and discoursestructural views of semantics may be used as a primary source of inspiration to disentangle ‘meanings’ which can then be ‘networked’ in a more full-blown, systemic semantic model. It turns out that these different types of designs, topological and discourse-structural, can be seen as very primary lines of thought for conceptualizing ‘semantics’ (in addition to a third one that will be introduced below). In other words, they are not only alternative types of design, but they are also two complementary pre-systemic approaches to ‘semantics’, and each of them is based in a specific vision of what a semantics can be. In all approaches to semantics, the question is, What ‘higher’ meaning can be recognized, above the lexicogrammatical networks, that forms an interface between lexicogrammar and context? In a ‘topological’ approach, this question is explored in terms of different levels of ‘abstraction’. The ‘higher meaning’ is of a different level of coding in two ways: it is not exactly

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

78

MIRIAM TAVERNIERS

what is already encoded in the lexicogrammar (otherwise no further stratum of semantics is necessary), rather it goes beyond the lexicogrammatical meaning; and it is closer to non-linguistic context, with which it interacts (see semantics as an interface stratum). For example, in the interpersonal metafunction, ‘command’ is a higher-level, more abstract meaning compared to a lexicogrammatical category such as ‘imperative’. The link with the topological design is this: as soon as a higher-order meaning such as ‘command’ is recognized, the further question can be asked how this can be linked back to the lexicogrammar. It then becomes clear that there are different facets to realizing ‘commands’, i.e. ‘command’ at the stratum of semantics is not just a semantic rendering of ‘imperative’ at the stratum of lexicogrammar. More specifically, commands can also be realized by other lexicogrammatical means than the imperative (such as different types of modal verbs and/or adjuncts, see Figure 3.4). And hence in drawing this connection between topological conception and differentiating semantics on the basis of level of abstraction, we come full circle with the idea of variability between two content strata (or between meanings and forms). In a ‘discourse-structural’ approach, the question of what higher meaning can be recognized is explored in terms of different levels of ‘patterning’. The ‘higher meaning’ is of a different level of patterning in two ways: it does not just ‘consist of’ the structures and building blocks that are available in the lexicogrammar, rather it goes beyond this lexicogrammatical patterning; and it interacts with context. In this perspective, the text or discourse becomes crucial at the level of semantics, since it is through texts that humans interact with context (not through clauses). In modelling semantics as how we interact with context through texts, one important cluster of questions therefore will be how discourse is organized, how it is built, and what the basic discourse units are. In relation to the method of the system network, this question translates more specifically into a quest for the unit that is the entry condition for semantic systems. Seen from below, i.e. in terms of how semantics is beyond lexicogrammar, the crucial question becomes what type of patterning takes place above and beyond the highest lexicogrammatical unit, i.e. beyond the clause or clause complex. Importantly, again, the discourse units do not ‘consist of’ lexicogrammatical units such as clause and clause complex, but rather, they are realized by different lexicogrammatical means (see Halliday and Matthiessen 2004:587). In discourse, these lexicogrammatical means occur sequentially, in unfolding text. Hence what will be important in the discourse-structural conception of semantics is how a discourse pattern emerges from a sequence of lexicogrammatical patterns unfolding in text. And this is how the link with variability can be made here: a discourse pattern does not have a one-to-one relation with one lexicogrammatical unit or pattern, but rather is realized through a sequence of (recurring) lexicogrammatical forms in unfolding text. The abstraction-based and pattern-based approaches to semantics which are taken in recent semantic models are summarized and presented

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

co

grammar

antics

s em

grammar

co

s em

79

lex i

antics

Semantics

lex i

abstraction-approach to semantics

pattern-approach to semantics

design

topological

discourse-structural

higher-order meaning

higher-order abstraction: closer to context

larger-order patterning: structures bigger than lexicogrammatical structures

semantics as interface

what we can mean in between what we can do & what we can say

larger structures not composed of lexgram building blocks, but realized by lexgram resources

Figure 3.9 Abstraction-based and pattern-based approaches to semantics, highlighting a topological and discourse-structural design, respectively

visually in Figure 3.9. This figure shows how the approaches can be understood as interpretations of the variability relation between a higher meaning and a dispersed realization in the lexicogrammar. As indicated above, the abstraction-based and pattern-based approaches have been complementary sources of inspiration in modelling a semantic stratum, and in setting up networks for this stratum. This has been the case across the metafunctions, but there are differences in the roles and relations between these two approaches for the different metafunctions. We will now turn to specific semantic models in SFL. We will first look at each metafunction, paying attention to what exactly the abstraction and patterning approaches have resulted in for those metafunctions, i.e. in setting up semantic models that flesh out the specific nature of each metafunction. After that we will turn to a third approach to theorizing a semantics, complementary to the abstraction-based and pattern-based conceptions, viz. a register/probability-approach, which has been applied more globally across the metafunctions.

3.5.2 Semantics from a Textual Perspective At the level of lexicogrammar the textual metafunction comprises the systems of theme and information, which are realized in the clause, and

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

80

MIRIAM TAVERNIERS

cohesion, which is a label for a number of sub-systems (such as reference, conjunction) which are realized non-structurally. The textual metafunction can be characterized by the following distinctive features. It (i) (ii)

(iii)

is concerned with the creation of texture; is regarded as a second-order metafunction, because it allows meanings from the other two metafunctions, interpersonal and ideational, to be brought together in syntagms and in larger, coherent wholes; does not only comprise structural resources (theme and information) like the other metafunctions do, but also non-structural ones.

Now, these defining features of the textual metafunction will help us understand how this metafunction is conceived of at the semantic level. The textual metafunction is intrinsically concerned with texture, and so the ‘patterning approach’ is a natural route for approaching textual semantics. In fact, the system of cohesion, with its ‘non-structural’ realization across clauses (e.g. lexical cohesion realized by lexemes spread across the text) is already a recognition of a higher-order pattern (beyond the ‘structural’ resources), and this conception ties in with the idea of the textual metafunction as a second-order metafunction. In his model of discourse semantics, Martin (1992) reinterprets three sub-systems of cohesion at the semantic level (which for him is the level of discourse, of higher-order patterning). In this framework, cohesion is not seen as part of a second-order (textual) metafunction, but as occurring at a higher stratum, that of discourse semantics. In discourse semantics, cohesive resources are seen as belonging to different metafunctions: there are ideational, interpersonal, and textual discourse-semantic systems. Each of those higher-order semantic systems is a pattern-based reinterpretation of a component of the earlier system of cohesion, which was already regarded as ‘different’ from other lexicogrammatical systems, because cohesive resources are non-structural. In addition to thus recognizing the ‘pattern’dimension of cohesion, the overall reconception of the notion of ‘cohesion’ as a higher stratum (rather than a ‘second-order’ metafunction) can be seen as a remodelling of cohesion in terms of degree of ‘abstraction’. In this sense, the pattern-based and abstraction-based conceptions of semantics are both integral aspects of the model of discourse semantics.

3.5.3 Semantics from an Interpersonal Perspective With regard to the interpersonal metafunction, a system of speech function is set up above the lexicogrammatical ones of mood and modality, in order to deal with how speech functions are realized by dispersed lexicogrammatical resources, and above (see Section 3.2.2) this was given as a good example of the abstraction-based approach to semantics. An elaborate model of speech functions which is context-specific (the context being interactions between mothers and their children) is Hasan’s (1996) message

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

81

semantics. However, a general speech functional model has been part of the theory for a long time (see an early overview in Halliday 1984) and has been incorporated into textbooks and presentations of the theory.7 In analyzing the mapping between speech functions and their realizations in lexicogrammar, some meaning–form couplings can be seen as the ‘default’ ones and this leads to the concept of interpersonal grammatical metaphors of mood (see Section 3.2.3.3). However, in another view, such ‘typicality’ is not relevant or is at least problematic (in relation to mood as well as modality) (see Hasan 2010:287). Speech functions also feature in a pattern-approach to semantics, which highlights relations between them in sequences, such as pairs consisting of an initiating and a responding move (as in the traditional concept of adjacency pairs) and in larger stretches of discourse, especially dialogue. This interpersonal dimension which focuses on the scaffolding of interaction is usually referred to as negotiation or exchange, as in the framework of discourse semantics (Martin 1992; Eggins and Slade 2005; Martin and Rose 2003). In addition to speech function and exchange structure, a further interpersonal area for which a conception of a semantic stratum is essential is the expression of evaluation, which is modelled in the sub-theory of appraisal. Appraisal deals with how attitudes and values are conveyed, how those values are sourced, and how interactants are aligned in relation to those values (White 2015). One source of inspiration (see White 1999) for setting up appraisal theory was the concept of interpersonal grammatical metaphors of modality, which revealed that one interpersonal meaning such as a modal value of ‘probability’ (see examples in Section 3.2.3.3) can be realized in various ways in the lexicogrammar (see Figure 3.4). In appraisal theory, this conception is extended to a range of other types of interpersonal values in addition to modality, for instance, affective meanings and value judgements, which are realized in dispersed ways, through interpersonal resources such as forms of modality, through experiential resources – explicitly or through connotation, and/or through logical resources (e.g. a conjunction but signalling a concessive meaning of counter-expectation). Hence appraisal theory is a good example of an abstraction-based approach to semantics which (re-)organizes various types of lexicogrammatical means at a higher level in topological areas. Appraisal meanings can also be looked at from a pattern-approach to semantics, which then focuses on how attitudes are negotiated in a text and how different alignments are set up across the text (e.g. Martin and Rose 2003; Martin and White 2005). A specific patterning, i.e. sequential, aspect of appraisal that is useful to mention is that of ‘semantic prosody’. The combination of various appraisal resources in a text can form a pattern which sets a tone or an attitudinal mood which is spread across a stretch of discourse, and 7

For overviews in textbooks, see, for instance, Thompson 1996: Chapter 4; Martin et al. 1997: Chapter 3; Halliday and Matthiessen 2014: Section 4.1; Matthiessen 1993: Section 5.1.2.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

82

MIRIAM TAVERNIERS

which unfolds with varying degrees of strength, just like musical prosody (see Martin and White 2005:59). For instance, certain words can trigger positive or negative connotations in other words that occur in their neighbourhood (see Louw 1993). Note, too, that the featuring of ‘connotation’, both as a dimension of appraisal (the evocation, rather than explicit construal, of evaluative meanings) and in the concept of semantic prosodies, as an intrinsic aspect of interpersonal semantics, resonates with a general conception of the stratum of semantics as a ‘connotative’ layer added to the ‘denotative’ system of language (i.e. semantics interpreted in terms of Hjelmslev’s model of a connotative semiotic, see Section 3.3.1 above).

3.5.4 Semantics from an Ideational Perspective At the level of lexicogrammar, the ideational metafunction deals with the building blocks of representational content through which experience is construed, especially process types at the level of the clause, and with relations between experiences, especially clause combining dealt with in the systems of taxis and logico-semantic relations. Within the ideational component, the former is called the experiential metafunction, the latter the logical metafunction. At the level of semantics, the ideational metafunction with both subcomponents is reinterpreted in Halliday and Matthiessen’s (1999) model of the ideation base of language, which contains basic building blocks of experience (‘experiential’), and combinations of them (‘logical’). In this sense, the model of the ideation base is in effect a semantic model of the rank scale, the primary options ranging from ‘elements’ (which can be processes, participants, or circumstances) through ‘figures’ (by default realized by clauses in the lexicogrammar) to ‘sequences’ (by default realized by clause complexes). In linking the semantic system to the lexicogrammar, the notion of typicality plays a crucial role, as indicated in the expression ‘by default’ in the previous sentence. Hence the ideation base incorporates a model of ideational grammatical metaphor, which crucially hinges on the notion of a rank scale (nominal groups are more condensed than clausal realizations, and a clause is more condensed than a clause complex). Thus, the ideation base is an ontological or phenomenological semantics, i.e. a semantics of what types of entities, qualities, and relations between entities there ‘are’; or what types of ‘phenomena’ humans conceptualize (significantly, the entry point for Halliday and Matthiessen’s semantic network of the ideation base is called ‘phenomenon’). It is clear that this is an abstraction-based approach to semantics, the semantics here being an interface between language and cognition or conceptualization, and again what is modelled in the semantics are topological zones (types of ‘phenomena’) which can be realized in dispersed ways in lexicogrammatical resources. One specific component of their model, viz. the relations of expansion and projection, which are seen as ‘transcategorical semantic

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

83

domains’, has been referred to above (Section 3.2.2) as an example of a topological approach to semantics, and the dispersion of realizations across the lexicogrammar as an example of fractality (see Section 3.4.2). Expansion was already regarded as dispersed through the lexicogrammar in IFG1 (see especially Halliday 1985:306–7). The components of expansion and projection (grouped as ‘sequence’ in the ideation base) lend themselves to a pattern-approach to semantics, precisely because it includes a concept of sequencing (projecting and projected element; expanded and expanding element), and thus are also useful in a discourse perspective. There are at least two discourse-patterning dimensions that are relevant: the unfolding of sequential occurrences of projections and expansions through a text; and the fractal ‘replication’ of such relations as also holding between larger units of text, such as between clause complexes and between paragraphs (in the same sense as hyper- and macro-Themes). This sequencing through relations of expansion and projection at text level is incorporated as conjunction in Martin’s discourse semantics (see Martin 1992; Martin and Rose 2003). It is a focus of earlier studies in the framework of Rhetorical Structure Theory (see Matthiessen and Thompson 1988; Mann et al. 1992; Matthiessen 2002).

3.5.5 Semantics and Register This section focuses on the relation between semantics and context. Here the question is what semantics realizes from a higher stratum, and how the relation between semantics and context can be modelled. Semantics, in this view, is the stratum of registers as ‘semantic strategies’ (Matthiessen 2009:219), strategies being selections of options that are available in a particular (institutional) setting and that, together, form a ‘procedure’ or ‘technique’ for functioning in that setting. The higher-level meaning that is modelled at the semantic stratum in this case is thus a technique as a combination of strategies, i.e. a grouping of meanings which can be made in that context, and which are realized through different lexicogrammatical resources that are ‘at risk’ in a particular setting. Resources that are ‘at risk’ in a context are those that are more likely to be selected because they redound with certain aspects of the context. Because registers are defined through the combination of options that are ‘at risk’ in a setting, they are ways of setting the probabilities in the lexicogrammar stratum (see Halliday and Matthiessen 2014:29), or, put differently, it is the specific combination of those options at risk which are grouped at the semantic stratum as sets of strategies, forming a ‘technique’ that works in a particular setting.8 8

It will be noted that in this way, semantics as register bears a fundamental similarity to that component of interpersonal semantics which models ‘speech functional’ meanings through which social semiotic ‘roles’ are enacted. This similarity is not surprising, since a register is a procedure for ‘functioning’ in a specific context, i.e. taking a specific ‘role’ in an institutional setting. The similarity has been noted before in SFL, but it has not been studied systematically (but rather has been seen as an ‘inconsistency’ or an unresolved issue). See Butler (2003), who also

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

co

grammar

MIRIAM TAVERNIERS

antics sem

84

lex i

actualization-approach to semantics design

probabilistic

higher-order meaning

higher-order grouping of strategies that ‘work’ in a context

semantics as interface

registers as gatekeepers: pre-selection of lexgram resources ‘at risk’ in a context

Figure 3.10 An actualization approach of semantics, highlighting registers as gatekeepers between context and lexicogrammar

This view of semantics, which I will refer to as an ‘actualization approach’, can be seen as a third conception of semantics, complementary to the pattern and abstraction approaches. It is visualized in Figure 3.10. This approach is shown as a probabilistic interpretation, where at the semantics there is a conglomerate of strategies, which together form a procedure for functioning in certain contexts, and which activate certain, but not other, lexicogrammatical options which are thus ‘at risk’ in this context (shown by the check marks). Options that are not activated are indicated in a lighter shade of grey and are thus backgrounded. Here semantics can be seen as a gateway between context and lexicogrammar, and each register as a gatekeeper. More specifically, semantics is a gateway between the language as potential, as a general code, and a specific context, and a register is what is relevant, what is ‘activated’ in a context. As complementary to abstraction and patterning-approaches to semantics, this approach can be called an actualization view, because it focuses on how language as a potential is actualized in specific techniques that ‘work’ in specific settings. Note that register is also conceived of as a connotative semiotic (see Butler 2003:383–90), a conception which again

refers to Gregory’s (1967) theory of register in which both dimensions pointed out are seen as two different aspects of the interpersonal component, viz. one which relates to speech functions through which the relation between the interactants are enacted (‘personal tenor’), and another which deals more broadly with the purpose of the text (‘functional tenor’). In later work, the ‘purpose of the text’ is not tied to the interpersonal metafunction, but is realized across the different metafunctions.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

85

resonates with the view of semantics as a connotative semiotic mentioned previously (see Sections 3.3.1 and 3.4.3). A final point to consider, in relation to semantics and register, is the question of how exactly this gatekeeping function of registers between lexicogrammar and context can be modelled. In SFL the relation of language with context is interpreted in terms of the metafunctions. What exactly, in the context, activates particular choices in a text is conceived of as belonging to three metafunctional dimensions: (1) the content of what is being talked about or the type of activity that is taking place (field – ideational); (2) the relationship between the interactants (tenor – interpersonal); and (3) the role that language itself is playing as a medium (mode – textual). In this way the complementarity between three metafunctions resonates fractally through the different strata (see Hasan 2009:174), also including context. This functional interpretation of language in relation to context is referred to as the context-metafunctions hook-up in SFL, and remains controversial to the present day (e.g. Hasan 2009, 2014).9

3.5.6 Mapping Semantic Models in SFL: Overview and Conclusion The various specific conceptions of semantics that have come into view in relatively recent semantic models in SFL are visually summarized in Figure 3.11. Three complementary approaches to semantics were distinguished which each bear a particular relation to the design tool of the system network (and the conception of language as a meaning potential, a network of options). The abstraction and patterning approaches were both called ‘pre-systemic’: they lead to topological areas of higher-order meanings or larger-sized patterning in the semantics, which have dispersed realizations in the lexicogrammar. Making internal distinctions within those zones and patterns forms a first step into setting up networks at the level of the semantics, i.e. above the lexicogrammar, and this is how the different semantic models pointed out above developed. For each of the metafunctions, the abstraction and patterning approaches led to specific systems, which are summarized below the stratal images in Figure 3.11. The actualization approach to semantics has a different relation to the systemic dimension of language. Semantics is here conceived of as a gateway between the overall code of language and specific contexts. A register as a grouping of semantic strategies thus sets the probabilities in the system – preselecting options across the different networks, 9

One problem, pointed out by Hasan (2009), is that when linguists determine what it is, in the context, that activates particular choices in a text, they are already reasoning from the text (or with an imaginary text in mind), i.e. they always reason from language. Hence context thus perceived is what Hasan calls ‘relevant context’, and this is different from the more general non-linguistic ‘eco-social context’ which features in other interpretations of semantics. It will be noted that this conception of ‘relevant context’ is itself a consequence and an inherent feature of language as a semiotic dynamic open system with relations of metaredundancy between its strata, hence also between semantics and context-as-seenfrom-semantics.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

textual

se m

co

actualization-approach

ideational ideation base

speech function (message semantics)

identification

Rhetorical Structure Theory

focus on

enactment of roles and stances & how they are negotiated

creation of texture

building blocks through which we construe experience & the relation between them

type of semantics

interactive semantics/ socio-semantics

(discourse semantics: textual and across metafunctions)

ontological/ phenomenological/ conceptual semantics

l ua text

al tion idea

sequencing of projection & expansion relations between larger units of text

so rper inte

na

l

ideational metaphor conjunction

g rammar

(hyper-Theme, macro-Theme)

te n o r

e

co

periodicity

od

a n t ic s

exchange structure + negotiation of attitudes

m

field

interpersonal metaphor negotiation

reference chains tracking participants [earlier work on cohesion reinterpreted across the metafunctions]

t contex

appraisal

pattern

g rammar

a n t ic s

t contex

co

pattern-approach

se m

abstraction

g rammar

se m

co

interpersonal

specific metafunctional semantic models

a n t ic s

t contex

g rammar

a n t ic s se m

abstraction-approach

lex i

t contex

lex i

lex i

approach to semantics

MIRIAM TAVERNIERS

lex i

86

Figure 3.11 Overview of semantic models in SFL in terms of differentiating dimensions in the overall architecture (stratification, instantiation, metafunctional complementarity); different design principles (systemic, topological, and discourse-structural); and the role of variability between meanings and forms

and it links upward, to ideational, interpersonal, and textual aspects of the context that are relevant and that redound with what is activated from language. At the bottom right, the figure contains an attempt to visualize the role of register as a semantic gatekeeper. It is useful, as promised, to return once more to the crucial concepts of stratification and metaredundancy, and briefly re-visit them in relation to the types of ‘semantics’ that were explored in this section. It has become clear above that the abstraction, patterning, and actualization approaches to semantics are complementary. What I would like to highlight here is that this complementarity also has a theoretical significance in relation to how we understand stratification/metaredundancy. Metaredundancy relations (see Figure 3.6 above) can be conceived of as ‘contextualization relations’ (see Thibault 2004, who makes the same connection): each layer in a semiotic system puts another (set of ) layer(s) into context. Stratification, too, and in a more concrete sense, is a model for theorizing how language (and its various coding layers) is ‘contextualized’. What the analysis of ‘meaning’ and ‘semantics’ in this chapter shows is that the contextualization which semantics provides as a stratum that is of a higher-order nature compared to lexicogrammar can mean at least three things: it can be

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

87

higher-order abstraction (‘meanings’ that are closer to eco-social context, analyzed in topological areas); it can be higher-order patterning (‘meanings’ that are larger in size, analyzed in sequential patterns); or it can be actualization (‘meanings’ that are relevant, analyzed in registers). We have now disentangled the dimensions of abstraction, patterning, and actualization, for the sake of analysis, and for the mere sake of attempting to understand ‘language’, and, more specifically, in this chapter, what ‘semantics’ can be. In language itself, all those contextualizations occur holistically and simultaneously in each instance of meaning-making. As a final point, let us make the circle full and return to language as a dynamic open system which is metastable, i.e. which continues to function by changing. Each of the dimensions of semantics, viz. abstraction, patterning, and actualization, vastly enhance the openness of language and thus the overall meaning potential: they are ways of organization (contextualizations!) that are not predicted by the lexicogrammar. At the same time, they have a constraining function. This is clearest in the case of the actualization approach, which highlights the role of registers as gatekeepers, but it is also true for the abstraction and patterning dimensions, by definition, because semantics interfaces between lexicogrammar and context. There is bottom-up constraining of semantics towards context, because only those ‘meanings’ are available that can be realized in lexicogrammar with its internal organization (such as the ‘amalgamation’ of different metafunctions in each syntagm), and there is top-down restraining of semantics towards lexicogrammar, because only those ‘meanings’ appear (topologically or as patterns) in semantics which the eco-social environment allows (e.g. the types of speech functions are restrained by the types of social roles in which we interact; the types of ‘phenomena’ that are distinguished are restrained by the entities we find in our environment – this is why this dimension of semantics is called ‘ontological’). In Figure 3.11, many-to-one lines of variability between context and semantics have not just been added to the register approach, but also to the other two approaches. And thus the function of semantics as enhancing and constraining at the same time finally brings us back to the concept of variability, with which we started our exploration of the ‘why’ of semantics.

3.6

Summary

This chapter focused on ‘semantics’ and the related concept of ‘meaning’ in SFL. It set out to explain what semantics is, how it is organized, where it is found, and why it is regarded as something in its own right which is worth looking at. Those questions were explored through a two-pronged approach which determined the two-fold aim of this chapter: (1) on the one hand, to distinguish different conceptions of ‘semantics’ and to disentangle the various theoretical distinctions and perspectives which play a role in thinking

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

88

MIRIAM TAVERNIERS

about semantics in SFL, and, (2) on the other hand, to look at various specific proposals for recognizing a semantics in SFL and to place them against the theoretical background of possible conceptions of semantics. We started off in Section 3.2 by looking at where ‘meaning’ is situated in the model of the system network and stratification in SFL, and by distinguishing three related ways in which semantics is designed in SFL. The ‘why’ question led us to the notion of variability relations between semantics and lexicogrammar, and we saw how grammatical metaphor was an important initial motivation for recognizing a (separate) stratum of semantics in SFL. In Section 3.3, conceptions of meaning and semantics in SFL were placed against a wider theoretical background, in two steps. First the concept of variability was reconsidered against the background of viewing language as a semiotic dynamic open system which is characterized by metastability and by metaredundancy relations between its coding levels. In a second step, the apparent indeterminate or multi-faceted view of meaning and semantics in SFL was explained in relation to a design rationale that is based on multiperspectivism and fractality. In Section 3.4 we turned to specific semantic models in SFL. In connecting how exactly semantics is fleshed out in those models to the different conceptions of semantics which had been disentangled in the previous sections, three basic approaches to semantics were distinguished, viz. abstraction, patterning, and actualization. Those were linked to the ways in which a semantics can be designed (topological, systemic/typological, and discourse-structural), to the role of variability, and to stratification and metaredundancy.

References Butler, C. S. 2003. Structure and Function: A Guide to Three Major Structuralfunctional Theories, Volume 2: From Clause to Discourse and Beyond. Amsterdam: John Benjamins. Butler, C. S. and M. Taverniers. 2008. Layering in Structural-functional Grammars. Linguistics 46(4): 689–756. Eggins, S. and D. Slade. 2005. Analysing Casual Conversation. 2nd ed. Sheffield: Equinox. Gregory, M. 1967. Aspects of Varieties Differentiation. Journal of Linguistics 3: 177–274. Halliday, M. A. K. 1961. Categories of the Theory of Grammar. Word 17: 241–92. Halliday, M. A. K. 1975. Learning How to Mean: Explorations in the Development of Language. London: Arnold. Halliday, M. A. K. 1976. Functions and Universals in Language. In G. Kress, ed., Halliday: System and Function in Language. London: Oxford University Press. 26–31.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

89

Halliday, M. A. K. 1977. Text as Semantic Choice in Social Contexts. In T. van Dijk and J. S. Petöfi, eds., Grammars and Descriptions. Berlin: Mouton de Gruyter. 176–225. Halliday, M. A. K. 1982. The De-automatization of Grammar: From Priestley’s ‘An Inspector Calls’. In J. M. Anderson, ed., Language Form and Linguistic Variation: Papers Dedicated to Angus McIntosh. Amsterdam: John Benjamins. 129–59. Halliday, M. A. K. 1984. Language as Code and Language as Behaviour: A Systemic-functional Interpretation of the Nature and Ontogenesis of Dialogue. In R. P. Fawcett, M. A. K. Halliday, S. Lamb, and A. Makkai, eds., The Semiotics of Culture and Language, Volume 1: Language as Social Semiotic. London: Pinter. 3–35. Halliday, M. A. K. 1985. Introduction to Functional Grammar. London: Arnold. Halliday, M. A. K. 1991. Towards Probabilistic Interpretations. In E. Ventola, ed., Functional and Systemic Linguistics: Approaches and Uses. Berlin: Mouton de Gruyter. 39–61. Halliday, M. A. K. 1992. How Do You Mean? In M. Davies and L. Ravelli, eds., Advances in Systemic Linguistics: Recent Theory and Practice. London: Pinter. 20–35. Halliday, M. A. K. 1996. On Grammar and Grammatics. In R. Hasan, C. Cloran, and D. G. Butt, eds., Functional Descriptions: Theory in Practice. Amsterdam: John Benjamins. 1–38. Halliday, M. A. K. 1998a. Things and Relations: Regrammaticising Experience as Technical Knowledge. In J. R. Martin and R. Veel, eds., Reading Science: Critical and Functional Perspectives on Discourses of Science. London: Routledge. 185–235. Halliday, M. A. K. 1998b. Linguistics as Metaphor. In A.-M. Simon-Vandenbergen, K. Davidse, and D. Noël, eds., Reconnecting Language: Morphology and Syntax in Functional Perspectives. Amsterdam: John Benjamins. 3–27. Halliday, M. A. K. 2008. Opening Address: Working with Meaning: Towards an Appliable Linguistics. In J. J. Webster, ed., Meaning in Context: Implementing Intelligent Applications of Language Studies. London: Continuum. 7–23. Halliday, M. A. K. 2009. Methods – Techniques – Problems. In M. A. K. Halliday, and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 59–86. Halliday, M. A. K. 2013. Meaning as Choice. In L. Fontaine, T. Bartlett, and G. O’Grady, eds., Systemic Functional Linguistics: Exploring Choice. Cambridge: Cambridge University Press. 15–36. Halliday, M. A. K. and R. Hasan. 1976. Cohesion in English. London: Longman. Halliday, M. A. K. and C. M. I. M. Matthiessen. 1999. Construing Experience through Meaning: A Language-based Approach to Cognition. London: Cassell. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2004. Introduction to Functional Grammar. 3rd ed. London: Edward Arnold. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Routledge.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

90

MIRIAM TAVERNIERS

Halliday, M. A. K. and J. J. Webster. 2009. Keywords. In M. A. K. Halliday and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 229–53. Hasan, R. 1996. Semantic Networks: A Tool for the Analysis of Meaning. In C. Cloran, D. G. Butt, and G. Williams, eds., Ways of Saying, Ways of Meaning: Selected Papers of Ruqaiya Hasan. London: Cassell. 104–31. Hasan, R. 2009. The Place of Context in a Systemic Functional Model. In M. A. K. Halliday and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 166–89. Hasan, R. 2010. The Meaning of ‘Not’ is Not in ‘Not’. In A. Mahboob and N. Knight, eds., Appliable Linguistics. London: Continuum. 267–306. Hasan, R. 2014. Towards a Paradigmatic Description of Context: Systems, Metafunctions, and Semantics. Functional Linguistics 1(9): 1–54. Haspelmath, M. 2003. The Geometry of Grammatical Meaning: Semantic Maps and Cross-linguistic Comparison. In M. Tomasello, ed., The New Psychology of Language: Cognitive and Functional Approaches to Language Structure. Mahwah: Erlbaum. 213–42. Hjelmslev, L. 1963. Prolegomena to a Theory of Language. Translated by F. J. Whitfield. Madison: University of Wisconsin Press. Lamb, S. M. 1962. Outline of Stratificational Grammar. Berkeley: University of California. Lemke, J. L. 1984. Semiotics and Education. Toronto: Victoria University. 23–62. Lemke, J. L. 1992. New Challenges for Systemic-functional Linguistics: Dialect Diversity and Language Change. Network 18: 61–8. Lemke, J. L. 1995. Textual Politics: Discourse and Social Dynamics. London: Taylor & Francis. Louw, W. 1993. Irony in the Text or Sincerity in the Writer? The Diagnostic Potential of Semantic Prosodies. In M. Baker, G. Francis, and E. Tognini-Bonelli, eds., Text and Technology: In Honour of John Sinclair. Amsterdam: John Benjamins. 157–76. Mann, W. C., C. M. I. M. Matthiessen, and S. A. Thompson. 1992. Rhetorical Structure Theory and Text Analysis. In: W. C. Mann and S. A. Thompson, eds., Discourse Descriptions: Diverse Analyses of a Fund-raising Text. Amsterdam: John Benjamins. 39–78. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, J. R. 1995. Text and Clause: Fractal Resonance. Text 15(1): 5–42. Martin, J. R. and C. M. I. M. Matthiessen. 1991. Systemic Typology and Topology. In F. Christie, ed., Literacy in Social Processes. Darwin: Northern Territory University, Centre for Studies in Language and Education. 345–83. Martin, J. R., C. M. I. M. Matthiessen, and C. Painter. 1997. Working with Functional Grammar. London: Arnold. Martin, J. R. and D. Rose. 2003. Working with Discourse. London: Continuum.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

Semantics

91

Martin, J. R. and P. White. 2005. The Language of Evaluation. New York: Palgrave Macmillan. Matthiessen, C. M. I. M. 1993. Lexicogrammatical Cartography: English Systems. Tokyo: International Sciences Publishers. Matthiessen, C. M. I. M. 2002. Combining Clauses into Clause Complexes: A Multi-faceted View. In J. Bybee and M. Noonan, eds., Complex Sentences in Grammar and Discourse: Essays in Honor of Sandra A. Thompson. Amsterdam: John Benjamins. 235–320. Matthiessen, C. M. I. M. 2009. Meaning in the Making: Meaning Potential Emerging from Acts of Meaning. Language Learning 59: 206–29. Matthiessen, C. M. I. M. and S. A. Thompson. 1988. The Structure of Discourse and ‘Subordination’. In J. Haiman and S. A. Thompson, eds., Clause Combining in Grammar and Discourse. Amsterdam: John Benjamins. 275–329. Taverniers, M. 2002. Systemic-functional Linguistics and the Notion of Grammatical Metaphor: A Theoretical Study and the Proposal for a Semiotic-functional Integrative Model. PhD Thesis, Ghent University. Taverniers, M. 2008. Hjelmslev’s Semiotic Model of Language: An Exegesis. Semiotica 171: 367–94. Taverniers, M. 2011. The Syntax–semantics Interface in Systemic Functional Grammar: Halliday’s Interpretation of the Hjelmslevian Model of Stratification. Journal of Pragmatics 43(4): 1100–26. Thibault, P. J. 2004. Agency and Consciousness in Discourse: Self–Other Dynamics as a Complex System. London: Continuum. Thompson, G. 1996. Introducing Functional Grammar. London: Arnold. White, P. R. R. 1999. Beyond Interpersonal Metaphors of Mood: Modelling the Discourse Semantics of Evaluation and Subjectivity. Paper presented at the 11th Euro-International Systemic Functional Workshop, July 1999, Ghent University. White, P. R. R. 2015. Appraisal Theory. In K. Tracy, C. Ilie, and T. Sandel, eds., The International Encyclopedia of Language and Social Interaction. Hoboken: Wiley. 1–7.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 15:31:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.005

4 The Clause An Overview of the Lexicogrammar Margaret Berry

This chapter was to have been written by Geoff Thompson, but sadly he died before he could write it. I would like what I have to say to be regarded as my tribute to Geoff. Certainly I shall be drawing deeply on his work, particularly on the third edition of his Introducing Functional Grammar (Thompson 2014).

4.1

Introduction

The chapter is intended to cover material which in most introductions to SFL takes at least three chapters – long chapters. Inevitably, therefore, I shall have to cut corners and oversimplify. I will give references to places where more detailed discussions can be found. I shall assume knowledge of the key concepts of SFL outlined in the preceding chapters (see in particular Webster, this volume): the metafunctions, the notion of choice, systems, system networks, rank and strata, and particularly the stratum of lexicogrammar. I shall be concerned with choices from the three main metafunctions – experiential, interpersonal, textual – and with the part these play in the structure of the clause. The chapter will be from the perspective of ‘mainstream’ SFL. Relevant discussion from the perspective of another version of SFL – Cardiff Grammar – can be found in Schulz and Fontaine, this volume. Following Halliday and Matthiessen (2014), the language of exemplification will be English, with only occasional footnotes referring to other languages. SFL prides itself on being an ‘appliable’ linguistics. Applications are to be discussed in later chapters of this volume. However, since for me the most interesting application of the kind of analysis I am going to be discussing is the application to discovering how children learn to write in different I am grateful to Chris Butler, Jeff Wilkinson, and the editors of this volume for comments on the first draft of this chapter. Of course I alone am responsible for any errors or misrepresentations.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

93

registers, I shall have this application always in mind. Most of my examples will be taken from the work of an eleven-year-old girl. I will call her Charlotte. Charlotte is in her first year at secondary school and is enjoying the freedom she is being given to experiment with different styles of writing for different purposes. She has provided me with a horror story, an account of a day in her life, a history essay, work from her health and nutrition course, and work from her science course. There will be no room to discuss her differing registers fully, but I will make clear which of my examples are taken from which of her pieces of work. I am very grateful to ‘Charlotte’ for allowing me to use her work in this way, and also to her mother for acting as go-between.

4.2

The Experiential Metafunction

The first metafunction to be discussed is the experiential metafunction. The experiential metafunction is the function of language to represent our experience of the world, to say what we want to say about the happenings and states of affairs of the world and our responses to them. From the perspective of the clause, SFL assumes that these happenings and states and responses are represented as ‘processes’, with ‘participants’ in those processes and ‘circumstances’ attendant on the processes. Examples (1a) to (1c) are taken from Charlotte’s account of a day in her life. (1a)

I participant

pressed process

the snooze button participant

(1b)

I participant

fed process

my hamster participant

(1c)

I participant

’ll leave process

the house participant

twice more circumstance

at eight circumstance

The experiential metafunction is responsible for the basic constituents of a clause. (1b) is assumed to have three basic constituents, one process and two participants; (1a) and (1c) each four, one process, two participants, and one circumstance. We shall see later that the interpersonal and textual metafunctions affect the structure of the clause in other ways. SFL recognizes different types of process, each with its own types of participant: material, mental, relational, verbal, behavioural, and existential processes. The first three of these are usually regarded as the three main types, the other three as minor (e.g. Halliday and Matthiessen 2014:215). Halliday and Matthiessen report on a study of what they call ‘a registerially mixed sample of texts’ in which material processes were the most frequent, relational processes the second most frequent, and

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

94

MARGARET BERRY

material process behavioural process clause

mental process verbal process relational process existential process

Figure 4.1 Types of process represented as a system network

mental processes third. There will be room in this chapter only to discuss the three main types.1 The types of process are presented as options in a system relevant to the clause (see Figure 4.1). The options in the system are given the names of the types of process. However, it is important to note that what is really being chosen is a kind of package deal, a configuration of a process together with relevant participants. Hopefully this will become clear as I discuss the main types of process in detail.

4.2.1 Material Processes Material processes are those which involve physical actions, such as pressed, fed, and leave in examples (1a) to (1c) above. In (2a) and (2b), from Charlotte’s history essay, the processes rushed and turned are again material processes, as are grasp and emits in (2c) and (2d) from Charlotte’s horror story. (2a)

The English rushed boldly forward

(2b)

The Normans turned their horses

(2c)

Cold clammy hands grasp my own

(2d)

A pearly aura emits from their body

A material process will have a participant representing the doer of the action. SFL calls this participant the ‘Actor’. In (2a) to (2d), the Actors are respectively The English, The Normans, Cold clammy hands, and A pearly aura. It is possible however that, although a process has an Actor, the Actor will not necessarily appear in the clause. In (3), a continuation of (1b), the Actor of trekked is not mentioned in its own clause, but we know that the Actor is I as this is recoverable from the previous clause.2

1

Halliday and Matthiessen (2014:215) comment that ‘[t]he minor process types appear to vary more across languages

2

|| indicates a clause boundary. I am using standard SFL notation, as set out in Halliday and Matthiessen (2014:ix–xi).

than the major ones’.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

(3)

95

|| I fed my hamster || and trekked downstairs ||

Often a material process will have a second participant representing the person or thing that the action is done to. SFL calls this the ‘Goal’. In the first half of (3) my hamster is the Goal, in (2b) their horses, in (2c) my own. Some material processes on the other hand have only an Actor with no Goal. (2a), (2d), and the second clause of (3) include circumstances, but have no second participant; there is no person or thing that the action is done to. We are here of course recognizing the traditional grammar distinction between ‘transitive’ and ‘intransitive’, the difference between having an object and not having an object, but SFL is focusing more on the semantics than traditional grammar. SFL in fact uses the term ‘transitivity’ more broadly than traditional grammar, to refer to the whole area of processes, their participants, and their circumstances. When analyzing texts, SFL uses ‘probes’ (e.g. questions) to distinguish the different types of process and participant. The probes for material process are: (a) Can the clause be rephrased in the form What X did was (to) . . .? (b) If the process is taking place at the present moment, is the most natural form in which to refer to it that of the continuous present (i.e. the be + ing form)? The processes in the two clauses of (3) pass both these tests. What I did was to feed my hamster and What I did was to trek downstairs are both perfectly acceptable. And if we switch the processes into the present, the most natural way to refer to them would be I am feeding my hamster and I am trekking downstairs. Other forms would have different implications. For instance, I feed my hamster and I trek downstairs would seem to refer to habitual processes rather than single processes taking place at the present moment.3 On the other hand, the process of ‘knowing’ in (4), from Charlotte’s health and nutrition essay, fails both tests. (4)

I knew mango juice had a thick consistency

What I did was to know mango juice had a thick consistency sounds very strange; ‘knowing’ is not really a form of ‘doing’. And if we switch the process into the present, I am knowing mango juice has a thick consistency sounds equally strange. Much more likely would be I know mango juice has a thick consistency. The two processes in (3) then are material processes. The ‘knowing’ process in (4) is not. (Example (4) will be discussed later in Section 4.2.2 on mental processes.)

3

The point about the continuous present being the most natural way of referring to a single present process is perhaps truer of some registers than of others. If a register has something of the character of a commentary, the simple present form can be used to refer to a present process. (2c) and (2d) are from a passage in Charlotte’s horror story that reads like a commentary on what is going on. In this context, Cold clammy hands grasp my own and Cold clammy hands are grasping my own would seem to be equally possible ways of referring to the present process, as are A pearly aura emits from their body and A pearly aura is emitting from their body.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

96

MARGARET BERRY

The first of the probes for material process can also be used as a probe for the participant ‘Actor’. If the clause can be rephrased in the form What X did was (to) . . ., where a constituent of the clause naturally replaces X, then that constituent is representing the Actor. Applying this to (2a) and (2b), repeated here for ease of reference, (2a)

The English rushed boldly forward

(2b)

The Normans turned their horses

we get What the English did was to rush boldly forward and What the Normans did was to turn their horses. The English and The Normans are the Actors of their respective processes. The probe for ‘Goal’ is an extension of the probe for material process and Actor. If the clause can be rephrased in the form What X did to Y was (to) . . ., where a constituent of the clause naturally replaces Y, then that constituent is representing the Goal of the process. (2b) can be rephrased as What the Normans did to their horses was turn them. Their horses is the Goal of the process of ‘turning’. There is no equivalent to Y in (2a). (As we have already seen, in traditional grammar terms, the process in (2b) is transitive, the process in (2a) is intransitive.) Applying this to examples (1a) to (1c), again repeated here for ease of reference, (1a)

I pressed the snooze button . . .

(1b)

I fed my hamster

(1c)

I’ll leave the house . . .

we get What I did to the snooze button was press it and What I did to my hamster was feed it. The snooze button and my hamster are the Goals of their respective processes. However, there is a problem with (1c): i.e. What I’ll do to the house is leave it. Here, I is not really doing anything to the house. The snooze button is presumably in a different position as a result of the pressing. And a fed hamster is presumably different from an unfed hamster. But there is no change in the house as a result of the leaving. SFL regards the house as not a Goal but an instance of another kind of participant which it calls ‘Scope’. The processes in (1a) and (1b) have Goals and so are transitive. The process in (1c) is regarded as intransitive, but having Scope. A Scope resembles a Goal in that typically it occurs immediately after the process and in that it does not include a preposition. But it represents the domain to which the process relates, rather than something the process is done to. Another example of a Scope (a made-up example this time) would be the hill in (5a). (5a)

Jack climbed the hill

(5b)

Jack climbed rapidly

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

97

Figure 4.2 More delicate choices within material processes

Both (5a) and (5b) are regarded as intransitive. (5a) is intransitive with Scope. (5b) is intransitive without Scope – it just has a circumstance.4 I have discussed the probes for material process, Actor, and Goal in detail, in order to show how probes work. Unfortunately there will not be room in this chapter to discuss in the same way the probes for the other categories I am about to cover. For more on probes, see Fontaine (2013:85–91); Bartlett (2014:48–82). Moving on to discuss other distinctions within material process, I will quote from Thompson (2014:96), who says ‘there are many different suggestions for ways in which [material processes] can be subcategorized at more delicate levels’. In fact Thompson himself (2014:95–7, 111–14), Berry (1975:151–2, 154–61), and Bartlett (2014:48–58) offer different selections from among the possible distinctions. The most comprehensive account is that of Halliday and Matthiessen (2014:224–44, see especially their Figure 5–10, on page 229), but even that does not cover all the distinctions that have been suggested. The distinctions I have chosen to present in this chapter are those which in the past I have found most helpful in distinguishing different registers. Figure 4.2 shows the distinctions I have discussed here so far and the distinctions I am going on to discuss. I have already discussed the transitive (having both Actor and Goal) and intransitive (having Actor but no Goal) system and the further + scope and scope system. The transformative/creative system, the next system down in Figure 4.2, distinguishes between processes that bring Goals into existence (creative) and those which do something to existing Goals (transformative). Thompson’s (2014:96) examples of this are as in (6a) and (6b). (6a)

I’ve just made the Christmas puddings

(6b)

My Mum never eats Christmas pudding

(6a) is creative as the process of making actually brings the puddings into existence. (6b) is transformative as the pudding exists before the eating of it (or in this case the not eating of it!). Examples (7a) and (7b) are both from Charlotte’s health and nutrition essay.

4

For discussion of different types of Scope, see Halliday and Matthiessen (2014:239–42).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

98

MARGARET BERRY

(7a)

Mango juice would add thickness to my smoothie

(7b)

I used raspberries

(7a) is presumably creative as the process of adding creates the thickness. But (7b) is transformative as the raspberries existed before Charlotte used them in her smoothie.5 A set of distinctions that I have found particularly useful in distinguishing different registers is that between concrete and abstract, with the further distinction between animate and inanimate (again see Figure 4.2.) All the examples of material processes I have given so far would count as concrete material processes. But Halliday and Matthiessen (2014:243) note that material processes may represent abstract doings and happenings. They give examples from a passage of financial news, which include (8a) and (8b). (8a)

AT&T’s stock slid Tuesday [sic!]

(8b)

The disappointing forecast, which came as AT&T posted firstquarter results . . ., dampened the enthusiasm . . .

It may seem a contradiction in terms to say that material processes can be abstract, but processes such as those in (8a) and (8b) do pass the tests for material processes discussed above: i.e. What AT&T’s stock did was to slide, and What AT&T did was to post first-quarter results. And if we switch the processes into the present, the most natural form for them would be AT&T’s stock is sliding, and AT&T is posting first-quarter results. Of course one may want to regard such examples as metaphorical, but analyzing them as abstract versions of material processes shows the kind of metaphor involved. Examples of abstract material processes from Charlotte’s work, both from her history essay, are (9a) and (9b). (9a)

My evidence comes from the Primary Source C

(9b)

This point links to my next point6

Concrete material processes may be either animate or inanimate, this distinction showing mainly in their Actors. As I explained in the introduction to Section 2, although the names of the options make it appear that we

5

As well as discussing the creating or transforming of Goals in transitive clauses, as I just have, Halliday and Matthiessen (2014:231) discuss the creation of Actors in intransitive clauses. This is why I have followed them in presenting the transformative/creative system as simultaneous with the transitive/intransitive system, instead of presenting the transformative/creative system as dependent on the selection of transitive.

6

Halliday and Matthiessen may want to regard example (9b) as a relational clause – see what they say, for example, (2014:265) about These plates went from the head to the tail. However, for me (9b) passes the tests for material processes. We can say What this point does is link to my next point and This point is linking to my next point, though these forms are perhaps unlikely in the register of Charlotte’s history essay.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

99

are simply classifying processes, in fact we are classifying package deals consisting of configurations of process plus participant/s. Examples (1a) to (1c) above, from Charlotte’s account of a day in her life, have the animate Actor I. Examples (2c) and (2d) above, and (10a) to (10d) below, all from Charlotte’s horror story, have inanimate Actors.7 (10a) Mist is swirling about me (10b) A ghostly light penetrates the gloom (10c) The wind will begin to howl like a wolf (10d) The candles will flicker The first paragraph of Charlotte’s horror story includes five material processes. These all have inanimate Actors. And material process clauses with inanimate Actors continue to feature in later paragraphs.

4.2.2 Mental Processes The second type of process to be discussed here is that of mental processes. Mental processes are not really processes of doing, but rather processes of sensing. As we saw above, the sensing process of ‘knowing’, as in example (4), fails both tests for material processes. (4)

I knew mango juice had a thick consistency

We cannot really say either What I did was to know mango juice had a thick consistency or I am knowing mango juice has a thick consistency. ‘Knowing’ is not a form of ‘doing’. And, if we switch the process into the present, the most likely form for it would be the simple present – I know mango juice has a thick consistency – not the continuous present which is most usual for material processes. The tests for material processes can thus be negatively used for mental processes. Material processes pass the tests, mental processes fail them. (For more on probes for mental processes, see Bartlett 2014:64–5; Fontaine 2013:87.)8 SFL recognizes four different types of mental process, as shown in Figure 4.3.9

7

I am here disagreeing with Halliday and Matthiessen (2014:250), who say that in material clauses ‘the distinction between conscious and non-conscious beings simply plays no part’. It is true that, in the class of material processes viewed as a whole, Actors may be either conscious (animate) or non-conscious (inanimate), but this is not true of all individual members of the class. Some material processes will normally have an animate Actor (e.g. trek in example (3)), while others will normally have an inanimate Actor (e.g. flicker in example (10d)). There seems to be a cline of material processes from this point of view. Examples with untypical animacy are usually viewed as metaphorical. For discussion, see Berry (1975:151–2, 155).

8

Also see below. A mental process can take a clause as phenomenon. A material process cannot take a clause as goal.

9

For a more detailed system network showing more delicate choices within mental processes, see Halliday and Matthiessen (2014:258).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

100

MARGARET BERRY

Figure 4.3 A more delicate choice within mental processes

Perceptive processes include such processes as ‘seeing’, ‘hearing’, ‘feeling’. Examples (11a) and (11b) are from Charlotte’s horror story, (11c) from her account of the day in her life. (11a) I can hear her rasping breath (11b) I heard the cry of my best friend (11c) You can see the parents spectating on the sidelines Emotive processes include such processes as ‘loving’, ‘hating’, ‘enjoying’, and ‘fearing’. (12a) and (12b) are from Charlotte’s day in her life, (12c) from her health and nutrition work. (12a) You hate sprinting (12b) Since then I have loved sprinting (12c) I love the smell and taste of lime Cognitive processes include such processes as ‘thinking’, ‘believing’, ‘understanding’, and ‘conjecturing’, as well as the process of ‘knowing’ from example (4). (13a) and (13b) are from Charlotte’s health and nutrition work, (13c) from her history essay. (13a) I thought the mango juice would counteract the tart flavour of the raspberries (13b) I thought it would make an attractive red (13c) I have concluded that the Norman tactics are the most important reason for the Norman victory10 Desiderative processes include such processes as ‘wanting’, ‘wishing’, and ‘hoping’. (14a) is the only desiderative process I have found in Charlotte’s work, from the day in her life. (14b) and (14c) are from Thompson (2014:100).

10

It could be said that conclude passes the tests for material process. However, although there has not been room to discuss it fully here, a further difference between mental and material processes is that mental processes can take a clause as complement, while material processes cannot (see footnote 9). Conclude can take a clause, as in example (13c).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

101

(14a) I hope my dream will come true (14b) I don’t want any trouble (14c) You may crave a cigarette Mental processes will have a participant whose function is the sensing. SFL calls this the ‘Senser’. Typically the Senser will be animate, indeed human. All the examples I have so far given of mental processes have human Sensers – the I’s and you’s. When a mental process occurs with an inanimate Senser, this is seen as an example of personification. Example (15) is from Thompson (2014:98). (15)

We used to have a car that didn’t like cold weather

The thing that is sensed is called the ‘Phenomenon’. This may be a simple Phenomenon, realized by a nominal group, such as any trouble in (14b), a cigarette in (14c), cold weather in (15). Or it may be more complex, involving another clause which itself contains another process. (11c), repeated here, has the Phenomenon the parents spectating on the sidelines. (11c) You can see the parents spectating on the sidelines The Phenomenon of see includes another process; what is seen is the parents spectating on the sidelines. Halliday and Matthiessen (2014:256) call this type of complex phenomenon a ‘macrophenomenon’. In a macrophenomenon the second process will be realized by a non-finite verb, usually the ‘-ing’ form. Another type of complex phenomenon, called by Halliday and Matthiessen a ‘metaphenomenon’, is found in (13a) to (13c) and (14a), again repeated here. (13a) I thought the mango juice would counteract the tart flavour of the raspberries (13b) I thought it would make an attractive red (13c) I have concluded that the Norman tactics are the most important reason for the Norman victory (14a) I hope my dream will come true In a metaphenomenon, the second process will be realized by a finite verb, and the metaphenomenon itself can be introduced by that. In (13c) the metaphenomenon actually is introduced by that. A that could be inserted at the beginnings of the metaphenomena in (13a), (13b), and (14a). Halliday and Matthiessen (2014:256) distinguish two subtypes of metaphenomena: ‘idea’ type and ‘fact’ type. An ‘idea’ type introduces a new idea. A ‘fact’ type represents something that has already happened. All Charlotte’s metaphenomena – in (13a) to (13c) and (14a) – are of the ‘idea’ type. Halliday and Matthiessen’s (2014:256) examples of the ‘fact’ type are as in (16a) and (16b).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

102

MARGARET BERRY

(16a) He saw that they had left (16b) He regretted (the fact) that they had left All four main types of mental process – perceptive, emotive, cognitive, and desiderative – can take simple Phenomena, but they vary in the types of complex Phenomena with which they are associated. Macrophenomena occur with perceptive and emotive processes, but not with cognitive or desiderative processes. All four types can take metaphenomena, but perceptive and emotive processes typically take ‘fact’ type, while cognitive and desiderative processes typically take ‘idea’ type. For more on the possibilities for the various types of mental process, see Halliday and Matthiessen (2014:256); Bartlett (2014:64–5).

4.2.3 Relational Processes The third type of process to be discussed is that of relational processes. (17a) and (17b), from Charlotte’s horror story, and (17c) and (17d), from the day in her life, are all examples of clauses with relational processes. (17a) Everything is hazy and unclear (17b) I am a ghost (17c) The snooze button was a saviour (17d) We are big breakfast people In each clause a relationship is set up between an entity (Everything, I, The snooze button, We), and a quality (hazy and unclear, a ghost, a saviour, big breakfast people). Thompson (2014:101) comments: Strictly speaking, neither of the basic experiential terms, ‘process’ and ‘participant’, is completely appropriate for this category. There is no process in the normal sense of ‘something happening’; and although there are always two concepts – one on each side of the relationship – there is only one participant in the real world.

A description such as hazy and unclear is ‘hardly a prototypical participant’ (Thompson 2014:101). And in (17b) a ghost is not something separate from I, but simply a description of I. But Thompson (2014:101) continues: ‘However, no grammatical term will cover equally well all the phenomena to which we need to apply it, so we will continue to talk about process and participants.’ SFL recognizes two main types of relational process: ‘attributive relational processes’ and ‘identifying relational processes’. (17a) to (17d) above are all examples of attributive relational processes, in which a description is attributed to an entity or group of entities.Attributive relational processes are each assumed to have two participants: a ‘Carrier’, the entity or entities described, and an ‘Attribute’, the description related. In (17a) to (17d) the Carriers are

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

103

Everything, I, The snooze button, and We. The Attributes are hazy and unclear, a ghost, a saviour, and big breakfast people. The Attribute may be adjectival, as in (17a), or a nominal group, usually indefinite, as in (17b) to (17d). Examples of identifying relational processes are (18a), from Charlotte’s day in her life, (18b) and (18c), from her horror story, and (18d) and (18e), from her history essay. (18a) Breakfast is the most important meal of the day (18b) The only reminder of my original features are my eyes (18c) Dying was the easy part (18d) My argument is that with power comes clumsy judgement (18e) My third reason for why William won the Battle of Hastings is the Norman tactics In an identifying relational process clause, one entity is identified in relation to another. Breakfast, for instance, is identified as the most important meal of the day. Usually something specific is identified by reference to a more generalizable category. Breakfast is something specific, which is identified by reference to the generalizable category of the most important meal of the day. SFL calls the more specific participant the ‘Token’, while the more generalizable category is called the ‘Value’. It is not always easy to decide which is Token and which is Value. I would say that in (18a) to (18c) the Tokens are Breakfast, my eyes, and Dying, with the most important meal of the day, The only reminder of my original features, and the easy part as Values. (Perhaps a clue to the fact that my eyes is Token is that Charlotte has made the verb agree with my eyes rather than with what precedes it. My computer does not approve of this and has underlined are in green!) Thompson (2014:105) includes example (19). (19)

The explanation is that it is forbidden by the second law of thermodynamics

He labels the that-clause as Token and The explanation as Value. By analogy with this, Charlotte’s (18d) would have the that-clause as Token and My argument as Value. Does that mean that in (18e) the Norman tactics would be Token and My third reason . . . would be Value? Identifying relational processes are similar to attributive relational processes in that, although there appear to be two participants, there is only one real-world entity. Breakfast and the most important meal of the day both refer to the same real-world entity, though the relation between them provides a new perspective. However, identifying relational processes differ from attributive relational processes in that their participants are realized by definite nominal groups or the equivalent, instead of the indefinite nominal groups of attributive relational processes. Also, the identifying relational process clauses are reversible in a way that attributive relational

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

104

MARGARET BERRY

Figure 4.4 A choice within relational processes

process clauses are not. It would be perfectly possible to say The most important meal of the day is breakfast or My eyes are the only reminder of my original features, instead of the versions in (18a) and (18b), and still sound naturally idiomatic. While it is just about possible to reverse some attributive relational processes – e.g. Big breakfast people are we – the result would sound marked and highly rhetorical. There is a great deal more that could, and should, be said about relational processes, but space does not permit. I will close this subsection simply by saying that I think Thompson (2014:101–5, 122–7) is particularly helpful on relational processes. For more detailed discussion, see Halliday and Matthiessen (2014:259–300). Figure 4.4 shows the one choice within relational processes that I have had room to discuss. There are many more. See Halliday and Matthiessen (2014:264).

4.2.4 Experiential Matters Not Covered in This Chapter There are a number of experiential matters that I have not been able even to touch on in this chapter. I have already indicated that I have not been able to discuss the three minor types of process and their participants – behavioural, verbal, and existential processes. See Halliday and Matthiessen (2014:300–10). I have also not been able to discuss circumstances. For a helpful table of possible types of circumstance, see Fontaine (2013:80); Halliday and Matthiessen (2014:313–14). Most of my examples have been of clauses with active voice; I have not been able to discuss their passive voice counterparts. Thompson does discuss passive counterparts to active clauses – see index to Thompson (2014) for detailed references. My examples have mainly been prototypical examples – for discussion of more problematic examples, see e.g. O’Donnell et al. (2009); Gwilliams and Fontaine (2015). I have not said anything at all about an alternative way of looking at all this – the ergative model as opposed to the transitive model. See Thompson (2014:139–42); Halliday and Matthiessen (2014:332–55).

4.3

The Interpersonal Metafunction

SFL assumes that, simultaneously with making experiential choices of the kind discussed in Section 4.2, we also make interpersonal choices – choices

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

105

Figure 4.5 A simplified MOOD network for English

which have to do with the way in which we interact with other people – and that these interpersonal choices also have an effect on the structure of the clause. Halliday (2002:203) says of the experiential options that they will tend to be realized particulately – that is, a structure that represents experiential meanings ‘will tend to have this form: it will be a configuration, or constellation, of discrete elements’. We have already seen that the experiential choices lead to the basic constituents of a clause, their number, and their nature. Halliday (2002:205) says of interpersonal meanings that they will tend to be realized prosodically – that is, the interpersonal meaning is ‘strung throughout the clause as a continuous motif or colouring’. This section will attempt to show what Halliday means by this.

4.3.1 MOOD SFL recognizes two main kinds of clause rank interpersonal choice, which are usually discussed under the headings of MOOD and MODALITY. MOOD choices will be discussed in this subsection, MODALITY choices in the following subsection. MOOD choices have to do with the forms we use when indicating the kind of interaction in which we are engaged; e.g. whether we are making statements or asking questions or giving commands. The main choices are shown in Figure 4.5.11 Only independent clauses have access to these choices, not subordinate clauses. Concepts such as ‘declarative’, ‘interrogative’, and ‘imperative’ are of course familiar in most approaches to the study of language. SFL’s particular take on them is to present them as options from which choices can be made, and to relate them both to the semantics in the stratum above and to the lexicogrammatical structure of the clause. Relating them to the semantics is not a simple matter. While it is true that in the unmarked cases questions are realized by interrogatives,

11

For a more detailed version of this network, see Thompson (2014:60). For an even more detailed version, see Halliday and Matthiessen (2014:162).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

106

MARGARET BERRY

statements by declaratives, and commands by imperatives, this is by no means always so. Commands, for instance, can in certain circumstances be realized by interrogatives or declaratives. Full discussion of this is beyond the scope of the present chapter (but see Berry 2016). For a full discussion of the relation between MOOD and SPEECH FUNCTION, see Martin (1992: Chapter 2). Also, of course questions etc. can be realized intonationally (see Debashish, this volume). I am in this chapter concerned with the effect on the lexicogrammar. As far as the structure of the clause is concerned, realizations of the MOOD options mostly affect the ‘Predicator’, the constituent which mainly carries the transitivity process, and the ‘Subject’, a constituent which carries one of the transitivity participants. (Which participant will depend, for instance, on whether the clause is active or passive.) The choice between ‘major’ and ‘minor’ is realized by the presence or absence of a Predicator. In example (20), from Charlotte’s day in her life, there are two major clauses, followed by a minor clause. (20)

I pressed the snooze button twice more, but eventually clambered out of the warmth of my bed. A typical school morning.

The first clause has the Predicator pressed and the second clause the Predicator clambered. But in the third clause, a minor clause, there is no explicit process. Charlotte uses minor clauses in her horror story and in the day in her life, but in her other writings minor clauses appear only as headings of sections and subsections. The choice between ‘indicative’ and ‘imperative’ is realized by the presence or absence of a Subject. The first clause in (20) has the Subject I and is indicative. The second clause would also be regarded as indicative. Although the Subject does not actually appear in that clause, it is recoverable from the previous clause. On the other hand, the main clause in (21a), again from Charlotte’s day in her life, is imperative. There is no Subject of Imagine. Similarly (21b), from Thompson (2014:58), is imperative. There is no Subject of Answer.12 (21a) Imagine you are standing on a race track (21b) Answer no more than three of the following questions For the realization of the choice between ‘declarative’ and ‘interrogative’, we have to consider a possible split in the Predicator between the Finite part of it and the main Lexical part. Examples (22a), (22b), and (22c) are all from Charlotte’s horror story. 12

It is possible for imperatives to be accompanied by things that look like Subjects. Thompson (2014:59) gives the example You listen to me, young man. However, Thompson argues that these are not ‘normal’ Subjects. The question of what is meant by ‘Subject’ in SFL is a complex one. For varying views, see Halliday and Matthiessen (2014:147–50); Fawcett (1999); Fontaine (2013:109–15).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

107

(22a) My eyes sparkle with tears (22b) I shall now tell you my story (22c) Will my mother cry at the sight of me All three clauses are major clauses; they all have Predicators. And all three are indicative; they all have Subjects. But in (22b) and (22c) the Predicators are split between a Finite part and a main Lexical part: shall . . . tell and will . . . cry. In (22a) the Predicator is not split – sparkle is itself a finite verb – the Finite part and the main Lexical part have been conflated. To return to the realization of the choice between ‘declarative’ and ‘interrogative’, in a declarative clause the Subject precedes the Finite part of the Predicator, but in an interrogative clause the Subject follows the Finite part. (22a) and (22b) are both declarative while (22c) is interrogative. This is in fact my working definition of the Subject, that it is the constituent in indicative clauses that inverts with the Finite verb to show whether the clause is declarative or interrogative.13 Where there is already an auxiliary verb to carry the finiteness, it is simply a matter of inversion. We can turn the declarative (22b) into an interrogative simply by inverting the I and the shall – Shall I now tell you my story. We can turn the interrogative (22c) into a declarative simply by inverting the Will and my mother – My mother will cry at the sight of me. Where there is not already an auxiliary to turn a declarative into an interrogative, it is necessary to import one. To turn (22a) into an interrogative, we would need to import do – Do my eyes sparkle with tears. (For more on Subject and Finite, see Fontaine 2013:110–20.) What I have just said is truer of yes/no interrogatives than of wh-interrogatives. While it is the case that wh-interrogatives usually involve inversion of Subject and Finite verb, this is not always so. What really realizes a wh-interrogative is of course the presence of a wh-word – who, what, where, when, why, how. When the wh-word is itself the Subject, there is no inversion. As well as the main MOOD options already discussed, there are other possibilities. For instance, tag phrases may be used. A clause with declarative form may be given an interrogative tag – e.g. My eyes sparkle with tears, don’t they or I’ll now tell you my story, shall I. An imperative may be ‘softened’ by a tag – Imagine, will you, that you are standing on a race track or Imagine you are standing on a race track, will you. As usual there is much more that could be said. But hopefully I have said enough to show what Halliday means by saying that the realizations of

13

I am here taking a more syntactic, less semantic view of the Subject than is probably usual in SFL these days. In my view, we still need a syntactic perspective if we are to account for the ways in which the MOOD options are realized. But see the references in footnote 12.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

108

MARGARET BERRY

interpersonal meanings are ‘strung throughout the clause’. Experiential options give rise to discrete constituents, those representing processes, participants, and circumstances. The realizations of the interpersonal MOOD options may involve the presence or absence of an experiential constituent, the splitting of the constituent that carries the process, the ordering of the Finite and the Subject, and/or the addition of a tag, and the latter may be in the middle of the clause or at the end, as in the examples I have just given.

4.3.2 MODALITY ‘Modality’ is the name which SFL gives to the whole area of clause rank speaker/writer assessment of what is being said. If a clause is communicating information, the assessment may be of the probability that the information is true. In the made-up examples (23a) to (23c) different assessments are made of the degree of probability. (The assumption is that all of them are spoken on hearing someone arriving.) (23a) That may be John now (23b) That will be John now (23c) That must be John now. Or if a clause is negotiating an action, the assessment may be of the desirability or practicability of the action. (24a) to (24c), again made-up, offer different assessments of the desirability of going to the meeting. (24a) You may go to the meeting (if you wish) (24b) You ought to go to the meeting (it’s an important one) (24c) You must go to the meeting (it’s crucially important) Assessments of the truth of information are termed by SFL ‘modalization’. Assessments of the desirability/practicability of an action are termed by SFL ‘modulation’. (In other approaches to linguistics, modalization is often termed ‘epistemic modality’, while modulation is often termed ‘deontic modality’.) Each of these main categories can be subdivided. Assessments of information may be about the probability of the truth of information, as in (23a) to (23c). Or they may be about the usuality of the truth of the information, as in (25). (25)

John is usually home by now

Assessments of actions may be about the desirability/permissibility/obligatoriness of the actions, as in (24a) to (24c). Or they may be about the practicability of the action, in terms of the willingness/ability of the Actor, as in (26a) and (24b).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

109

Figure 4.6 A simplified MODALITY network for English

(26a) I will go to the meeting (if you want me to) (26b) I can go to the meeting (I’ve nothing else scheduled for that time) Charlotte’s modalities are mostly of the willingness/ability subtype of the modulation type of modality, as in (27a) to (27d), all from her horror story. (27a) Ghosts can’t rest unless their death was fair (27b) I will not do this (27c) I could feel the blood trickling down my chest (27d) I cannot stop myself However, she does include a few assessments of probability and one assessment of usuality. (28a) and (28b) are from the horror story, (28c) to (28e) from the day in her life. (28a) They might try to console themselves (28b) Maybe their hearts will start to hammer in their chests (28c) You may be wondering why I say this (28d) You may share my opinion (28e) Usually breakfast is toast and peanut butter There is a further modality choice which cuts across the ones discussed so far (see Figure 4.6).14 Cutting across the choice between modalization and modulation is a choice among low, median, and high. This has to do with the strength of the assessment. In examples (23a) to (23c), repeated below, (23a) represents

14

For a more detailed version of this network, see Thompson (2014:77). For an even more detailed version, see Halliday and Matthiessen (2014:185).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

110

MARGARET BERRY

an assessment of low probability, (23b) an assessment of median probability, and (23c) an assessment of high probability. (23a) That may be John now (23b) That will be John now (23c) That must be John now In examples (24a) to (24c), again repeated below, (24a) represents an assessment of low obligation, (24b) an assessment of median obligation, and (24c) an assessment of high obligation. (24a) You may go to the meeting (if you wish) (24b) You ought to go to the meeting (it’s an important one) (24c) You must go to the meeting (it’s crucially important) The modality options may be realized in a number of ways. In most of the examples I have given so far, the realization is by modal auxiliary verbs, such as may, will, and must. Alternatively the realization may be by modal Adjuncts. Or by a combination of modal auxiliary and modal Adjunct. Alternative realizations of low probability modalization are shown in (23a), repeated below, and (29a) to (29d). (23a) That may be John now (29a) That’s possibly John now (29b) Perhaps that’s John now (29c) That may possibly be John now (29d) That may be John now perhaps Where there is a modal Adjunct, such as possibly or perhaps, this may be at the beginning of the clause, as in (29b), or in the middle as in (29a) and (29c), or at the end as in (29d). Hopefully this can be regarded as another illustration of what Halliday means by saying that the realizations of interpersonal meanings are ‘strung throughout the clause’.15 Bartlett (2014:113) provides a helpful table summarizing modality options and showing their realizations.

4.4

The Textual Metafunction

The third metafunction to be discussed is the textual metafunction. This is the function of language to weave together the experiential and

15

Halliday and Matthiessen (2014:183–4) say ‘when we move around the languages of the world, we find a great deal of variation in the grammaticalization of modality and other types of interpersonal judgement’. They go on to discuss some of the differences.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

111

interpersonal meanings and present them as text. Bartlett (2014:131) says: ‘If experiential and interpersonal elements are the threads of a text, we can say that the textual metafunction refers to the way in which these elements are woven together to make a patterned cloth.’ This section will consider the part that the structure of the clause plays in the weaving. Halliday (2002:208) says that textual meanings will tend to be realized ‘culminatively’: ‘What the textual component does is to express the particular semantic status of elements in the discourse by assigning them to the boundaries; this gives special significance to “coming first” and “coming last”.’ As far as the clause is concerned, SFL regards two pairs of concepts as being relevant to ‘coming first’ and ‘coming last’: Theme and Rheme, and Given and New. Theme is relevant to ‘coming first’ in the clause, and New is relevant to ‘coming last’, though the relevance of the latter is indirect, as I shall explain. When considering the weaving together of experiential and interpersonal meanings into the text, the questions to be asked are: Which bits of meaning are regarded as Theme and so come first? Which bits of meaning are regarded as New and so come last? This section is mainly going to be about Theme and Rheme, but first a brief word about Given and New.16

4.4.1 Given and New Mainstream SFL assumes that Given and New are realized phonologically through intonation (see Debashish, this volume, for a more detailed discussion). However, it is necessary to say a little here in order to explain the connection with the structure of the clause. SFL recognizes an ‘information unit’ which is realized by a ‘tone group’ – that is by a particular pattern of pitch movement. In the unmarked case a tone group will be co-extensive with a clause. Again in the unmarked case, a tone group will consist of a stretch of relatively level pitch followed by the main pitch movement. The likelihood is that the relatively level stretch will be presenting information as Given – that is, as information that is already known – while the main pitch movement will be presenting information as New, or particularly newsworthy. The most usual place for the main pitch movement is on the last experiential constituent of the clause. Because of this, the last experiential constituent of the clause gains an association with newness or newsworthiness, and even in written language this is where one tends to place what represents the main point of what one is saying. In a recent article I wrote example (30a). The publisher’s copy editor changed it to (30b). I changed it back to (30a). The methods of development

16

Textual meaning is sometimes regarded as ‘second-order’ meaning, since it is about the arrangement and relative prominence of bits of experiential meaning and interpersonal meaning. For discussion, see Matthiessen (1992).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

112

MARGARET BERRY

were the main point, and the new point in that paragraph, of what I was saying, and I wanted them to have end focus. (30a) This means that, like Child C, Writer 2 in both his passages has clear methods of development (30b) This means that, like Child C, Writer 2 has clear methods of development in both his passages It is probably for similar reasons that Charlotte preferred (18b), repeated below, to (31a), even though this meant placing the Token after the Value (see Section 4.2.3 above). Having introduced her eyes in New position in (18b), she is then able to continue with (31b), the eyes now being represented as Given, with the tears now represented as New. (18b) The only reminder of my original features are my eyes (31a) My eyes are the only reminder of my original features (31b) They sparkle with tears ‘Coming last’, then, is significant in the clause, as a result of the association, in the unmarked case, between the last experiential constituent of the clause and the main pitch movement of the tone group, this in turn being associated with New information. This indirect association enables the weaving of experiential meaning into the text to take account of the relative givenness/newness of the experiential meanings on offer.17

4.4.2 Theme and Rheme While New is relevant to coming last in the clause, Theme is relevant to coming first. Theme is in fact the name that SFL gives to the beginning of a clause. The rest of the clause is the Rheme. One might expect that what comes first in the clause will always be Given. But things are not as simple as that. In example (31b), They, which comes first, is indeed Given, in the sense that its referent has been mentioned in the previous clause. But in (32a) and (32b), again from Charlotte’s horror story, the most given items – I and it – do not come first. Given and Theme are in principle distinct, though they may co-occur. (31b) They sparkle with tears (32a) Tonight I shall visit their house (32b) Like a glimmer of hope, it represents the tiny spark of humanity left in the souls that wander the graveyard

17

For different definitions of ‘Given’ and ‘New’, see Berry (forthcoming).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

113

Halliday and Matthiessen (2014:89) say that the Theme is ‘that which locates and orients the clause within its context’. The orienting may take a number of different forms. For instance, the Theme, the first experiential constituent in the clause, may be indicating that the text is staying with a topic entity already mentioned, as in (31b). Or it may indicate a moving on in time, as in (32a). Or it may simply be setting the scene, as in (32b).18 We have already seen that first position in the clause plays an important part in English in realizing some of the interpersonal options. For a yes/no interrogative, the finite verb will usually be in first position; for a whinterrogative, it will usually be the wh-word; and for an imperative, the lexical verb may be in first place. In a declarative clause, the most usual thing to happen will be that the Subject will be at the beginning. When this happens, SFL calls it ‘Unmarked Theme’. The first three clauses of the second paragraph of Charlotte’s horror story all have Unmarked Themes. (33a) I am a ghost (33b) I wear a long white dress (33c) My hair flows down my back in silvery strands We then get the clauses about her eyes, quoted above, and then in the rest of the paragraph the main clauses all have I Subjects, these being in first experiential place and so being Unmarked Theme. This is the paragraph in which Charlotte establishes the persona she has adopted for the story. The repeated I’s as Unmarked Theme keep this persona firmly in view.19 Subject in first position then is the unmarked order for a declarative clause. But other elements of the clause may precede the Subject for special effects, in which case they are regarded as ‘Marked Themes’. Perhaps the most common kind of Marked Theme is where an Adjunct, a constituent representing a transitivity Circumstance, precedes the Subject. Examples (34a) to (34d) are all from the day in Charlotte’s life. (34a) Blearily I exposed one eye to the sunlight streaming through my blinds (34b) Before, I was driven to school

18

Opinions differ in SFL as to how much of a clause should be regarded as the beginning. Is the Theme just the very first experiential constituent, or should it include any other experiential constituent near the beginning that has an orienting function? I have said that They in (31b) has an orienting function in that it shows the text is staying with a topic entity already mentioned. But I in (32a) and it in (32b) also have this function. Should they not also be included in the Themes of their respective clauses? If one takes the notion of orienting function seriously, there are grounds for saying that the Theme extends right up to the main lexical verb. Indeed in a language such as Spanish the Theme may even include the lexical verb as orienting functions carried by early parts of clauses in English are carried by verbal inflections in Spanish. For discussion, see Berry (1996).

19

Fries (1981) links Theme with what he calls the ‘method of development’ of a text. The pattern of the experiential constituents selected to be Theme shows how the subject matter of the text is being organized and developed.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

114

MARGARET BERRY

(34c) Suddenly it’s lunchtime (34d) On such days I’ll spend the hours in peaceful tranquillity Sometimes a whole subordinate clause may precede the Subject. Charlotte makes a good deal of use of this option. Here are just a few examples, (35a) from the horror story, (35b) from the day in the life, (35c) from the health and nutrition work, and (35d) from the history essay. (35a) As she drifts closer, I can hear her rasping breath (35b) After breakfast has long passed, I’ll leave the house at eight (35c) And as my smoothie was summery, I thought it would add a refreshing aftertaste (35d) Although this is a small detail, it shows that William cared about his army Rather less common are instances of a Complement preceding a Subject. (The term ‘Complement’ is being used here to refer to an element which represents a transitivity participant such as ‘Goal’ or ‘Attribute’, which in English would more usually occur after the main lexical verb. In other words it is the Object/Complement of traditional grammar.) I did not find any instances of Complement Marked Theme in Charlotte’s writing. But in (36), Lucy, a sixteen-year-old friend of Charlotte’s, uses both Adjunct Marked Theme and Complement Marked Theme. (36)

In groups we worked out our things we needed to test to be able to use as evidence to prove our hypothesis right or wrong. Later on after having the results then we would work out the cross sectional area. The width we measured 3 times and the depth 11 times to get it accurate. The channel velocity we tested three times and used two separate experiments which were with the hydro prop and recording the time it took for the dog biscuit to travel across 10 metres of the river at each site. We visited 4 different sites at each point of the river.

The paragraph begins with an Adjunct Marked Theme In groups. Lucy is clearly very aware of the group nature of the work in which she is engaged, and this is picked up in the we Subjects. She also uses Adjunct Marked Theme to indicate a shift in time reference – Later on after having the results then. Complement Marked Themes are found in The width we measured 3 times and The channel velocity we tested three times. And if we regard the depth 11 times as an elliptical clause – the depth [we measured] 11 times – we would have another Complement Marked Theme here. In the first sentence of the paragraph, Lucy mentions our things we needed to test. Then later in the paragraph, she uses Complement Marked Themes to highlight the things tested.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

115

Earlier in this subsection, I drew attention to the part played by Theme in realizing interpersonal options such as yes/no interrogative, wh-interrogative, and imperative. So far in relation to declarative clauses, I have been discussing the arrangement of experiential meanings. But declarative clauses too can have interpersonal material in their Themes. Modal Adjuncts, of the kind discussed in Section 3.2, can occur at the beginning of a clause. (28b) and (28e), repeated below, provide examples of this. (28b) Maybe their hearts will start to hammer in their chests (28e) Usually breakfast is toast and peanut butter Halliday and Matthiessen do not regard Modal Adjuncts as Marked Themes. That term is reserved for experiential material that precedes the Subject. Instead they write of ‘multiple Theme’ (e.g. Halliday and Matthiessen 2014:107). In (28b) Maybe their hearts would all be regarded as Theme, with an interpersonal component Maybe and an experiential component their hearts. Similarly in (28e) Usually breakfast would all be the Theme, with an interpersonal component Usually and an experiential component breakfast. In each case the Theme would be regarded as Unmarked, since the Subject is the first experiential component, but as multiple Theme, since it contains interpersonal material as well as experiential. Something else that can occur at the beginning of a clause, without Halliday and Matthiessen regarding it as Marked Theme, is a word or phrase which is specifically functioning to show how the text fits together. Conjunctions would count here, as would conjunctive Adjuncts, such as for instance, in addition, and therefore. In (37), from Charlotte’s horror story, Besides would be regarded as a conjunctive Adjunct. (37)

Besides, I could never kill my parents

Where a multiple Theme includes material from all three metafunctions, the likely order is specifically textual Theme followed by interpersonal Theme followed by experiential Theme. Charlotte has a rather nice multiple Theme in the day in her life. (38)

I don’t know about you but, controversially in the evening when I’m going to bed, I don’t feel tired

But would be textual Theme, controversially would be interpersonal Theme, and in the evening when I’m going to bed would be experiential Theme. The experiential Theme would be Marked Theme as it precedes the Subject I of the main clause. Figure 4.7 summarizes the declarative clause options I have been discussing in this subsection. As usual there is a great deal more that could be said. Thompson (2014:147–81) is particularly helpful on Theme.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

116

MARGARET BERRY

Figure 4.7 Thematic options for declarative clauses

For a more detailed network for Theme, see Thompson (2014:170). For a much more detailed network, see Halliday and Matthiessen (2014:106).

4.5

Concluding Remarks

This chapter has discussed: • • •

choices from the experiential metafunction, showing how these lead to the number and nature of the basic constituents of a clause; choices from the interpersonal metafunction, showing how the realizations of these are threaded in among the basic constituents; choices from the textual metafunction, showing how these relate to what comes first and what comes last in the clause.

Many interesting questions have had to be left undiscussed, for reasons of space. However, I have given references to where more detailed discussions can be found.

References Bartlett, T. 2014. Analysing Power in Language: A Practical Guide. London: Routledge. Berry, M. 1975. An Introduction to Systemic Linguistics, Volume 1: Structures and Systems. London: Batsford. Berry, M. 1996. What is Theme? A(nother) Personal View. In M. Berry, C. S. Butler, R. P. Fawcett, and G. Huang, eds., Meaning and Form: Systemic Functional Interpretations. Meaning and Choice in Language: Studies for Michael Halliday. Norwood: Ablex Publishing Corporation. 1–64.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

The Clause

117

Berry, M. 2016. Interpersonal Meanings, with Particular Reference to Getting People to Do Things. Paper for seminar, Vigo, February 2016. Berry, M. forthcoming. ‘Actually Given’ Versus ‘Presented as Given’ and ‘Actually New’ Versus ‘Presented as New’: What Happens when the ‘Presented as’ Gets out of Step with the ‘Actually’? Paper presented at the 2nd Round Table on Communicative Dynamism, Namur, Belgium, September 2016. Fawcett, R. 1999. On the Subject of the Subject in English. Functions of Language 6(2): 243–73. Fontaine, L. 2013. Analysing English Grammar: A Systemic Functional Introduction. Cambridge: Cambridge University Press. Fries, P. H. 1981. On the Status of Theme in English: Arguments from Discourse. Forum Linguisticum 6: 1–38. Gwilliams, L. and L. Fontaine. 2015. Indeterminacy in Process Type Classification. Functional Linguistics 2(8): 1–19. Halliday, M. A. K. 2002. Modes of Meaning and Modes of Expression: Types of Grammatical Structure and Their Determination by Different Semantic Functions. In J. J. Webster, ed., On Grammar: Collected Works of M. A. K. Halliday. Vol. 1. London: Continuum. 196–218. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Routledge. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Matthiessen, C. M. I. M. 1992. Interpreting the Textual Metafunction. In M. Davies and L. Ravelli, eds., Advances in Systemic Linguistics: Recent Theory and Practice. London: Pinter. 37–81. O’Donnell, M., M. Zappavigna, and C. Whitelaw. 2009. A Survey of Process Type Classification over Difficult Cases. In C. Jones and E. Ventola, eds., From Language to Multimodality: New Developments in the Study of Ideational Meaning. London: Continuum. 47–64. Thompson, G. 2014. Introducing Functional Grammar. 3rd ed. London: Routledge.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 23 Aug 2019 at 07:38:48, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.006

5 The Rooms of the House Grammar at Group Rank Lise Fontaine and David Schönthal

5.1

Introduction One morning I shot an elephant in my pyjamas How he got into my pyjamas, I’ll never know! (Groucho Marx1)

Jokes like Groucho Marx’s famous ‘elephant in my pyjamas’ illustrate how we process units at an intermediate level between words and clauses (see Fontaine 2013). This is also evidenced by other ambiguous structures such as I saw the man from next door, where we get a different meaning depending on the way in which the units relate to each other, i.e. the man from next door is the person that I saw or I was next door when I saw the man. The distinction centres on whether the man from next door functions as one unit or two. The main point here is that there is an internal structure to the clause; it cannot be seen as a string of words. As Halliday and Matthiessen (2014:362) explain, ‘Describing a sentence as a construction of words is rather like describing a house as a construction of bricks, without recognizing the walls and the rooms as intermediate structural units.’ Understanding these intermediary units, i.e. the rooms of the house, is essential to our understanding of meaning. Within the framework of Systemic Functional Linguistics (SFL), these units are viewed from the perspective of the function they serve in a higher unit, which is, in this case, the clause. It is generally accepted that between the level or rank of clause and word there are two different types of unit, group and phrase, which are roughly equivalent. Halliday maintains that ‘a phrase is different from a group in that, whereas a group is an expansion of a word, a phrase is a contraction of a clause’ (Halliday and Matthiessen 2014:362–3). The differences between these two will be explained in detail throughout Section 5.2 and 5.3. However, we can think of the difference in

We would like to thank Margaret Berry for her very useful comments on drafts of this chapter. 1

From the film Animal Crackers (1930).

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

119

the following terms: groups form around a particular type of word, for example, a noun forms a group called the nominal group, such as the beautiful scenery, where the other words in the group somehow modify the noun, in this case scenery. A phrase, in contrast, is less like a relationship among words and rather combines two components through complementation, as in the prepositional phrase by the lake, where the lake is seen as a complement to the preposition by. In this chapter we explore these two types of intermediary unit in SFL. We take as given that there are units larger than the word and smaller than the clause. There is generally considerable agreement in terms of the descriptions of these intermediary units (see McDonald 2017 for an excellent overview of the historical development of groups). However, there is also room for debate. The aim of this chapter is to critically examine the theoretical reasons for including two fundamentally different types of grammatical unit between clause and word. The main question we ask is whether these reasons hold in all cases, i.e. is it theoretically justified to maintain two types. The way we will approach this is as follows. In the next section, we provide an overview of the existing description of the units below the clause as currently represented in the theory. Following this, Section 5.3 will present and evaluate three main criteria for classifying a unit as either a group or a phrase. These distinctions, drawn primarily from Matthiessen (1995) and Halliday and Matthiessen (2014), include (i) the concept of (primary) class and the relation between the functional potential of the unit and the ‘head word’; (ii) univariate versus multivariate structures; and (iii) the role of rankshifted units. In considering these perspectives on the units at the intermediate rank between words and clauses, we conclude that there is no significant theoretical or practical value in maintaining two different types of unit at this level. We argue that it is important to ask questions such as those we propose here in order to evaluate the strength of the position of the theory and its usefulness in an appliable theory of language.

5.2

An Overview of Grammatical Units below the Clause

In this section, we provide an overview of the grammatical units which function between the rank of clause and word. In Halliday’s important 1961 paper ‘Categories of the Theory of Grammar’, he describes the relationship between grammatical units and the concept of rank as follows: The category set up to account for the stretches that carry grammatical patterns is the ‘unit’. The units of grammar form a hierarchy that is a taxonomy. . . . The relation among the units, then, is that, going from top (largest) to bottom (smallest), each ‘consists of’ one, or of more than one, of the unit next below (next smaller). The scale on which the units are in fact ranged in the theory needs a name, and may be called ‘rank’. (Halliday 1961:251)

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

120

LISE FONTAINE AND DAVID SCHÖNTHAL

Table 5.1 Example of the rank scale Rank Scale

Example

Clause unit Group unit Word unit

The brown foxes were jumping over the laziest dog [The brown foxes] [were jumping] [over the laziest dog] [The / brown / foxes] [were / jumping] [over / the / laziest / dog]

The rank scale is described in Halliday and Matthiessen (2014:5) as ‘a hierarchy of units, related by constituency’. The principle is then that any unit at a given rank consists of units of a lower rank, e.g. a clause unit consists of one or more group units and a group unit consists of one or more word units.2 The terms, e.g. clause and word, reflect the type of unit at each rank and each one then has different sub-classes, i.e. different types of clause, group, and word. This is illustrated in Table 5.1 above. In this example, we have included only the rank immediately ‘above’ the group unit, i.e. clause, and immediately below, i.e. word, in order to simplify the illustration. It is also possible for a unit to be embedded, or rankshifted, for example, when a group unit contains a clause or another group unit. In addition to the ranking of these units, every unit can form a complex which means that it is possible for any unit to be multiplied recursively into a unit complex such as a clause complex, group complex, or word complex. While complexing is not central to the rank of unit between clause and word, there is an important distinction to be made between a group, such as the nominal group, and a word complex. This is a point that we will return to in the discussion of the nominal group in Section 5.2.1 and in our evaluation of the distinction between group and phrase in Section 5.3. However, very briefly, it is worth explaining that while groups are often discussed as groups of words, the group is a unit that consists of words (units of a lower rank) and may include rankshifted units, as will be explained below. It is a structure that has a relatively fixed order of constituents. A word complex, on the other hand, is a linear arrangement of two or more words (units of the same rank) involving some kind of dependency, where the order is not fixed, i.e. a different order results in a different meaning, as shown in the expressions in italics in examples (1) and (2).3

2

(1)

the county plans to spread the money among a mixture of government securities and bank investments (EnTenTen13, SketchEngine)

(2)

I once worked in an investment bank (EnTenTen13, SketchEngine)

As introduced later in this chapter, we use the term ‘group unit’ to refer to the units at group rank, i.e. both groups and phrases alike. Further, for a full account of how rank relates to the key SFL concepts of stratum, delicacy, and realization, see Berry (2017).

3

For details of SketchEngine, see Kilgarriff et al. (2014) or online: www.sketchengine.co.uk.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

121

The distinction is described by Halliday and Matthiessen (2014:437) as follows: ‘[t]reating the group simply as a “word complex” does not account for all these various aspects of its meaning’. They illustrate this point by comparing ‘railway ticket office staff, which could be explained as a (univariate) word complex, [with] that of these two old railway engines, which could not’ (Halliday and Matthiessen 2014:437, emphases in original). In addition, there is also a phonological rank scale (e.g. Halliday 1961; Halliday and Matthiessen 2014; also see Debashish and Bowcher, this volume). According to Halliday and Matthiessen (2014:15), the tone group ‘does a great deal of work in the construal of meaning: it organizes continuous speech as a sequence of units of information’. In studying spoken language, an understanding of the tone group is essential. See Halliday and Matthiessen (2014) for a fuller account of the rank scale and Fawcett (2010) for a critical review. In Halliday’s first account of the rank scale, he explains the hierarchical ranking of grammatical units as follows: For English, for the two units between sentence and word the terms ‘clause’ and ‘phrase’ are generally used. It is at the rank of the phrase that there is most confusion – because there are here the greatest difficulties – in the description of English; one reason is that in English this unit carries a fundamental ‘class’ division, so fundamental that it is useful to have two names for this unit in order to be able to talk about it: I propose to call it the ‘group’, but to make a class distinction within it between ‘group’ and ‘phrase’. (Halliday 1961:252–3)

Thus, Halliday proposes two types of unit, group and phrase, at the same rank on the scale, effectively having two different types of unit with no difference in rank. We will use the term ‘group unit’, as in Table 5.1, as an umbrella term for units at this rank. This avoids having to say ‘groups and phrases’ which is not only lengthy, but there is only one phrase and it seems reasonable to have a single term for each rank along the scale. In this paper, we have opted for the term ‘group’ rather than ‘phrase’ simply because it is by far the more common term in Systemic Functional Linguistics and as we will argue below can be applied to all intermediary units. The difference between groups and phrases is described by Halliday and Matthiessen (2014:262–3) as follows: A phrase is different from a group in that, whereas a group is an expansion of a word, a phrase is a contraction of a clause. Starting from opposite ends, the two achieve roughly the same status on the rank scale, as units that lie somewhere between the rank of a clause and that of a word.

This suggests that a group is in some sense more ‘word-like’ whereas a phrase is more ‘clause-like’. This is a perspective that will be challenged in Section 5.3. It is important to note here that work by Fawcett (1980, 2010) and Tucker (1998, 2017) presents perhaps the most detailed account of

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

122

LISE FONTAINE AND DAVID SCHÖNTHAL

Figure 5.1 Word classes in SFL (adapted from Halliday and Matthiessen (2014:75))

these intermediary units, although this is from a slightly different SFL theoretical approach, the Cardiff Grammar (see Schulz and Fontaine, this volume). There is another difference between group and phrase that must be covered briefly before moving on to the description of each class of unit at the group unit rank. This concerns the metafunctional nature of group units. In SFL, the approach to lexicogrammatical description at the group unit rank parallels the description of the clause in terms of the three main metafunctions. There is an assumption that all group units express, at least to some extent, experiential, interpersonal, and textual meaning. Halliday and Matthiessen (2014:361) explain that ‘[a]lthough we can still recognize the same three components, they are not represented in the form of separate whole structures, but rather as partial contributions to a single structural line’.4 When it comes to logical meanings, however, we find that phrases, unlike groups, are said to not express logical meanings. This metafunction explains one of the key differences between groups and phrases. While groups have both experiential and logical structure (Halliday and Matthiessen 2014:361–2), phrases do not express logical structure (Halliday and Matthiessen 2014:425). This is a distinction we will return to in Section 5.3 so that we can critically examine some of the implicit assumptions that underlie this claim. One final distinction must be discussed in relation to the description of units in SFL. We mentioned above that within SFL, units are defined in terms of the function they serve to express in the unit above, rather than by structural similarities that groups might serve. This is what motivates Halliday’s classification of units, and it explains why, for example, adverbials are classed differently than adjectives, but also why nominal groups and prepositional phrases are seen as different classes in English, whereas in other languages they might not be. The three primary classes of word are shown in Figure 5.1, where each primary class corresponds to one of the ‘three main classes of group: nominal group, verbal group and adverbial group’ (Halliday and Matthiessen 2014:362, emphases in original). These classes are defined by the functions served in the clause. For example, there is a tendency for nominal groups to function as Subject (or Complement) and/or Actor (or Goal), whereas adverbial groups serve

4

A detailed account of the metafunctions at clause rank is given in Berry (this volume).

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

123

to express Adjunct or Circumstance roles. In the early development of SFL, Halliday (1961:261, emphases in original) explains class and unit as follows: A class is always defined with reference to the structure of the unit next above, and structure with reference to classes of the unit next below. A class is not a grouping of members of a given unit which are alike in their own structure. In other words, by reference to the rank scale, classes are derived ‘from above’ (or ‘downwards’) and not ‘from below’ (or ‘upwards’).

As stated above, the principal distinction between units at this rank concerns two main features: (i) the secondary type of class which forms the basis for the unit, e.g. adverb vs. conjunction, and (ii) the nature of the relationship between the elements of the group (e.g. the nominal group consists of nominal elements). There are five groups generally recognized within SFL (but see Fawcett 2010 and Schulz and Fontaine, this volume, for some differences). Each has as its head element, a member of the relevant class. For example, as suggested in Figure 5.1, the nominal group can have as its head element a noun (Thing) or another nominal, e.g. an adjective (Epithet), as will be explained below. Groups are formed by expansion of the head through modification. The five groups are listed below in (3) to (7), along with illustrative examples where the base (head) element is underscored. The example given in (8) is a prepositional phrase. The phrase in SFL is not seen as having a word basis but rather as a type of reduced clause. This is due to the relational function of the preposition and the complementation function of the nominal group. It is difficult to claim that the complement nominal group, in this case the box, modifies the preposition in any way. All examples in (3) to (8) have been attested in EnTenTen13 via SketchEngine. (3)

a beautiful poem, nominal group (Ngp)

(4)

was eating, verbal group (Vgp)

(5)

very quickly, adverbial group (Advgp)

(6)

right under, prepositional group (Pgp)

(7)

just as, conjunction group (Cgp)

(8)

in the box, prepositional phrase (PP)

In what follows, each group and phrase will be described in turn. Following this, in Section 5.3, we will examine the criteria for the distinction made here between the unit of group and phrase.

5.2.1 Nominal Group The discussion of groups has so far presented the units in terms of logical meanings, i.e. expansion through modification. However, this seems to have been based on the idea of the group as word complex rather than on

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

124

LISE FONTAINE AND DAVID SCHÖNTHAL

Table 5.2 The nominal group (EnTenTen13, SketchEngine)5 Deictic

Numerative

Epithet

Classifier

Thing

Qualifier

Those

two

popular

lunch

stops

with an amazing view

the idea of constituency. As Matthiessen (1995:662) points out, when viewed experientially, what is construed is not ‘an expansion of a Thing (which is the logical perspective) but as a configuration of roles representing different aspects of a participant – a multivariate structure’. The nominal configuration includes the following experiential elements: Deictic, Numerative, Epithet, Classifier, Thing, and Qualifier. An example is given in Table 5.2. Each element of the Ngp contributes its own function. The Thing element functions as the ‘semantic core’ (Halliday and Matthiessen 2014:383). It classifies the entity being referred to (see ‘cultural classification’ in Fawcett 1980; Tucker 1998). In most cases, it also serves as Head element in the logical structure, but there are cases where the two strands of meaning diverge, as will be explained below. The Deictic has a determiner function in the sense of indicating phoric relations (see Martin 1992) and indicates whether the entity being referred to is specific or not and indeed whether the addressee should be able to identify the referent. In this sense, it has the potential to carry an implicature of definiteness or uniqueness in so far as the Deictic element can specify the referent in some way (i.e. which thing it is). The Numerative, as its name suggests, serves to specify the quantity or amount of the thing being referred to. Epithets and Classifiers are modifiers that either describe or depict the referent. Epithets are typically adjectives that function as qualities of the referent, e.g. a lovely visit, whereas Classifiers are typically nouns that sub-classify the referent along with the Thing element, e.g. an office chair. This distinction is not strict, and it is possible for there to be some overlap in this area. See Halliday and Matthiessen (2014:376–8) for a detailed discussion. The Qualifier element is defined generally as the element that occurs after the Thing element (Halliday and Matthiessen 2014:381). By its nature, the Qualifier is always expressed by an embedded (rankshifted) element, but this is most frequently a PP. The relationship between the Thing element and the Qualifier is different from that of the Classifier. With the Classifier, it is reasonable to describe the relationship as an ‘is-a’ relationship, e.g. ‘an office chair’ is a type of chair as compared to ‘a chair for the office’. However, with the Qualifier, the relationship is relational. Halliday and Matthiessen (2014:382) describe it as having a characterizing function

5

The original expression included only a as a Deictic: in order to include a Numerative, two, the Deictic was changed to a plural one, those, and stop was made plural as well. Finding a naturally occurring example with all elements included was very difficult.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

125

where ‘the characterization here is in terms of some process within which the Thing is, directly or indirectly, a participant. It may be a major process, i.e. a relative clause; or a minor process – a prepositional phrase’. There is a strong relation between Thing and the Head of the Ngp in English. In SFL, we expect that the Head of the Ngp will be conflated with the Thing element, as in Table 5.3, as this is the most usual instance (Halliday and Matthiessen 2014:390). This combination makes sense somehow in that the semantic core of the referring expression is what will occur in the role of Head. The main distinction between Thing and Head for Halliday and Matthiessen (2014:392) is that Thing is an element of experiential structure and Head is an element of logical structure. The element that expresses the Head determines the type of nominal expression, which could conflate with the Thing element, e.g. common noun, proper noun, or personal pronoun, but there could be other elements as Head, for example, Deictic or Numerative. In other words, the Head and Thing are not necessarily conflated in the Ngp. For example, consider the analysis of a cup of tea in Table 5.4. In this case, the Head is a Deictic or Numerative and the expression is called a ‘measure nominal’ (Halliday and Matthiessen 2014:375). Halliday’s Ngp structure also includes the potential for characterization through the Qualifier in experiential terms. The analysis of the children in blue hats in Table 5.5 shows the Qualifier element being expressed by a PP, which will be discussed below. One difficulty with the presentation in Halliday and Matthiessen (2014) is that much of the fine detail concerning the experiential structure of the Ngp is missing where it concerns the Qualifier. Halliday and Matthiessen (2014:390) state that the distinction between Premodifier and Postmodifier is not functional, but rather depends ‘on the rank of the modifying term’, i.e. whether it is rankshifted (embedded) or not. The distinction for Halliday Table 5.3 Experiential and logical analysis of the white cup the

white

cup

Deictic

Epithet

Thing

Modifier

Head

Table 5.4 Analysis of a ‘measure nominal’ (adapted from Halliday and Matthiessen 2014:392) nominal group

a

experiential structure logical structure

cup

of

tea

Numerative Premodifier

Head

Thing Postmodifier

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

126

LISE FONTAINE AND DAVID SCHÖNTHAL

is in the information structure, with the Postmodifier having ‘the greater potential as news’ (Halliday and Matthiessen 2014:390). See Fontaine (2017a) for a detailed discussion of the problems with the logical analysis and also Fawcett (2010) for an alternative without these problems (see also Davidse 2004; Ghesquière 2014).

5.2.2 Verbal Group The structure of the Vgp in English is complex for a variety of reasons: the way in which the Finite element works in English, the way verbs combine to express complex (secondary) tenses, and the way ‘words’ combine to form complex verbal lexemes (e.g. ‘to make up’). All Vgps have at least an Event element, which corresponds to a lexical verb. Finite Vgps also have a Finite element which serves to give the clause a point of reference; it provides a bounded limit to the clause. The Finite may either be conflated with the first auxiliary verb in the Vgp (see Table 5.6) or, in the absence of any auxiliaries, directly with the main verb. The question of headedness within the Vgp could be seen as depending on the perspective taken: for the interpersonal metafunction, it would be the Finite element and for the experiential metafunction, it would be the Event element, which is expressed by the main lexical verb (see Table 5.6). Arguably, the Finite element also carries textual meaning which is similar to that of the Deictic Determiner in the Ngp (see Halliday and Matthiessen 2014:397). In addition to these two elements, the Vgp could include one or more Auxiliary

Table 5.5 Analysis of the children in blue hats (adapted from Halliday and Matthiessen 2014:383) Nominal group the Logical structure Experiential structure

children

Premodifier

Head

Deictic

Thing

in

blue

hats

Postmodifier Qualifier Prepositional Phrase Process

Range Nominal group

Logical structure

Premodifier

Head

Epithet

Thing

Experiential structure

Table 5.6 The verbal group (EnTenTen13, SketchEngine) might

have

been

being

paid

Finite/AUXMOD

AUXPERF

AUXPROG

AUXPASS

Event

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

127

elements and/or a Polarity element. Although the Auxiliary element, AUX, is optional, there may be more than one type expressed: e.g. modal, perfective, progressive, and/or passive auxiliaries, as illustrated in Table 5.6 (see also Table 5.13 for an additional example). There has been some debate in the literature as to whether or not the Vgp should be seen as a group unit. Notably, Fawcett (1980, 2000a, 2000b, 2010) has argued for treating all verbal items as elements of the clause, i.e. at clause rank rather than below clause rank. Fawcett (2000a, 2000b) outlines four main arguments which relate to the treatment of phrasal verbs, the role of the Finite (or Operator) in mood structure, the problem of discontinuous items such as adverbial insertion, and theoretical simplicity in the system descriptions. There is no space here to enter into this debate, but see Morley (2000) for counter-arguments and Quiroz (2017) for a useful discussion of the theoretical considerations of the issues. For a discussion of the Vgp in other languages, see McDonald (2017); CaffarelCayron (2017).

5.2.3 Adverbial Group Halliday does not discuss the experiential structure of the Advgp in detail, but only offers a very brief discussion of the logical structure of this unit (see Halliday and Matthiessen 2014:419). For a very detailed account, see Tucker (1998), although the theoretical underpinnings differ from Halliday’s. As a group unit based on the lexical class of adverb, the Advgp is described in logical terms by the Head element in relation to the modifying elements as shown in Table 5.7. Experientially, however, this is somewhat problematic given the potential for this group to have a rankshifted postmodifier, as shown below (see Halliday and Matthiessen 2014:422) ‘[T]he embedded clause serves to represent a standard of comparison’ (Halliday and Matthiessen 2014:492, referring to Fries 1977). The Head element of the Advgp can be expressed by ‘an adverb denoting a circumstance as Head – for example, a circumstance of time (e.g. yesterday, today, tomorrow) or of quality (e.g. well, badly, fast, quickly, slowly). Advgps serving as modal Adjunct have an adverb denoting an assessment as Head – for example, an assessment of time (e.g. still, yet, already) or of intensity (e.g. really, just, only, actually)’ (Halliday and Matthiessen 2014:419).

Table 5.7 Description of the adverbial group (adapted from Halliday and Matthiessen 2014:422) much Logical Experiential

more

Modifier Extent

Temperer

quickly

than I could

Head

Postmodifier

Quality

Standard

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

128

LISE FONTAINE AND DAVID SCHÖNTHAL

See Tucker (1998; 2017) for an alternative analysis, including a convincing alternative account for adjective groups (for which see below).

5.2.4 Prepositional Group The Head element of the Pgp is a preposition which can be modified. This is in contrast to the prepositional phrase, which will be discussed below. Examples from Halliday and Matthiessen (2014:423) include: right behind, not without, all along, way off as in right behind the door, not without some misgivings, all along the beach, way off the mark. In these cases, right, not, all and way modify their respective preposition, and there is a kind of logical relationship of modification such that the meaning of the preposition is altered by the modifier. Some instances of prepositions appear complex in form, especially in written language, where they are represented orthographically by more than one word. For example, in front of and for the sake of (from Halliday and Matthiessen 2014:423) are considered to be complex prepositions, which are effectively single lexical items similar to compound words, i.e. they are not analyzed compositionally. The Pgp is only discussed in terms of expansion of the preposition, i.e. in terms of the logical structure of Head and modifier. Halliday and Matthiessen (2014) do not provide an account of the experiential structure. It is possible that there is no real motivation for a Pgp since it might be best accounted for as a word complex (see Matthiessen 1995:626). Also see Fawcett (2010) for an approach that has unified the Pgp and phrase into one unit.

5.2.5 Conjunction Group The conjunction group is described by Halliday and Matthiessen (2014:423) as consisting of a single conjunction such as and, or, or but, but also including more complex6 ones such as as soon as, by the time, or in case. As with other types of word classes, they ‘can form word groups by modification, for example, even if, just as, not until, if only’ (Halliday and Matthiessen 2014:423, emphases in original). Furthermore, conjunctions serve to express three main functions: there are ‘binders’ (i.e. subordinating conjunctions) such as which, who, or where (Halliday and Matthiessen 2014:482); ‘linkers’ (i.e. coordinating conjunctions) such as and, or, or but (Halliday and Matthiessen 2014:454); and ‘continuatives’, such as well, also commonly known as discourse markers (Halliday and Matthiessen 2014:107). However, although Halliday and Matthiessen (2014) consider the conjunction group as an element below the clause, it works quite differently 6

Although this depends on the lexical representation assumed. If these are single lexemes, i.e. multi-word expressions constituting effectively a single ‘word’, then their morphological composition may be more complex in terms of the formation of the item, but nevertheless only a single item in the lexicon.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

129

from the other groups. We do find conjunctions within groups, as in the Ngp the cows and horses, but there is no potential in these cases for modification and therefore it is not possible for a conjunction group to appear below the clause. The conjunction group only occurs between clauses. Thus, unlike other groups, it seems restricted to the clause rank, i.e. it cannot be embedded or rankshifted. For these reasons, we will consider it as a unit between clauses rather than below the clause, i.e. a unit at clause rank only. Finally, as with the Pgp, the conjunction group may well be better accounted for as a word complex rather than a group.

5.2.6 Prepositional Phrase As mentioned above, next to the groups discussed so far, there is also one phrase, which differs from groups in that it is considered to be a contraction of a clause rather than an expansion of a word (see Halliday and Matthiessen 2014:362–3). This phrase is the prepositional phrase, such as on the burning deck (from Halliday and Matthiessen 2014:424). It is analyzed as a contraction of a clause because the preposition arguably is no different from a verbal non-finite predicator, which Halliday and Matthiessen (2014:425) illustrate with the examples near/adjoining (the house), without/not wearing (a hat), or about/concerning (the trial). An analysis of a PP as a minor Process is given in Table 5.8. A PP, according to Halliday and Matthiessen (2014:311–12), is an odd sort of hybrid construction. It has a nominal group inside it, as a constituent, so it looks bigger than a group; and yet it is still not quite a clause. In English, this nominal group inside a prepositional phrase is no different from a nominal group functioning directly as a participant in a clause, and in principle every nominal group can occur in either context.

A further feature that, according to Halliday and Matthiessen (2014:425), distinguishes the PP from groups, is that ‘they have no logical structure as Head and Modifier, and cannot be reduced to a single element’. Thus, PPs always have to consist of a prepositional element (the minor Process) and a completive (Range), whereas Ngps such as apples, for example, can consist of the Thing element only. We will critically engage with these distinguishing features between phrases and groups in Section 5.3 below.

Table 5.8 The prepositional phrase Click

on

Process

the image you’ve chosen Location

Process

Range

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

130

LISE FONTAINE AND DAVID SCHÖNTHAL

5.2.7 Adjective Group The final group unit to be mentioned here briefly is the adjective group. The status of adjectives and the adjective group is an area of debate within SFL literature (see Tucker 1998, 2017; Fontaine 2013, 2017a). As mentioned above, the adjective class is a subtype of nominal (see Figure 5.1). In an example such as You’re very lucky in Table 5.9 (taken from Halliday and Matthiessen 2014:391), the Complement, very lucky, is thus considered as a Ngp where the Head element is expressed by an Epithet. Therefore, in IFG, there is no need for the adjective group as another group unit, but see Tucker (1998).

5.2.8

A Worked Example of the Clause in Terms of Units below the Clause To end this section, we will consider the group unit analysis of the clauses in Table 5.10 and Table 5.11. Also see Berry (this volume) for a discussion of the three metafunctions at clause rank. Table 5.9 Epithet headed nominal group (adapted from Halliday and Matthiessen 2014:391) You

‘re

very

lucky

Carrier

Process: relational

Attribute

Nominal Group

Verbal Group

Nominal Group Premodifier

Head

Epithet

Table 5.10 Sample clause analysis (EnTenTen13, SketchEngine) the

sharp

gorgonzola

Carrier Nominal Group Deictic

Epithet

Thing

was Process: relational Verbal Group Finite/ Event

an

unusual

choice

for

a

filling

Attribute Nominal Group Deictic

Epithet

Thing

Qualifier Prepositional Phrase Process

Range Nominal Group Deictic

Thing

Table 5.11 Clause analysis with Advgp (EnTenTen13, SketchEngine) The

backups Goal

can

be

Nominal Group Deictic

performed

Process: material

Thing

Verbal Group Finite/AUXMOD

AUXPASS

very

quickly

Circumstance: Manner Adverbial Group Event

Temperer

Quality

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

131

While these two analyses illustrate the group unit structure of two basic examples, other more complex examples would give rise to further complex issues in the description of these units. Unfortunately, it is beyond the scope of this chapter to deal with these here, but see Fontaine (2013) for a more detailed discussion of the structure of group units.

5.3

Criteria for Distinguishing between Group and Phrase

Having outlined the current description of the various intermediary units, this section will take a much closer look at the distinction between groups and phrases. There are three main distinctions between these two units.7 These are discussed in various sources, but the presentation in this section will draw primarily on Matthiessen (1995, especially 627) and Halliday and Matthiessen (2014). We will first consider the relationship between class membership and how groups are formed. Then, we will examine the distinction between groups and phrases in terms of multivariate vs. univariate structures. In the final part of this section, we consider how each type of group unit relates to the rank scale.

5.3.1 Notion of Class Membership The principle of class of unit works on the assumption that the sub-classes are similar in kind in terms of function in the rank above, e.g. while nominals, verbals, and adverbials as group units are on the same rank, they are not the same class because they serve different functions in the class above (participant, process, and circumstance respectively). One of the ways in which groups are said to work differently from phrases is that ‘[groups] are groups of words of the same primary class (nominals, verbals, or adverbials), so their functional potential is related to the Head word’ (Matthiessen 1995:627). The description of word class was given above in Figure 5.1. In this section, we consider whether this description holds for each of the group units presented above. While there is an argument to be made for identifying a three-part distinction in the clause, i.e. participant, process, and circumstance (interpreted very broadly), it is more difficult to relate this to the structure of a unit. Nevertheless, there is a clear ‘preference’, if we can call it so, for Ngps to be the unit expressing participants in the clause and for processes to be expressed by Vgps. As concerns circumstances, these are most frequently expressed by Advgps or PPs, which suggests that PPs are more functionally appropriate as members of the adverbial class. In addition, PPs can express a participant role in certain clauses, i.e. as Attribute in relational attributive clauses, or Recipient in verbal clauses. 7

These criteria do not address the criterion related to the function the unit serves in the unit above, but see Halliday and Matthiessen (2014:362–4). We thank Margaret Berry for this point.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

132

LISE FONTAINE AND DAVID SCHÖNTHAL

If we recall the word classification discussed above in Figure 5.1, the Pgp (e.g. right behind, not without, all along) should, following the principle of class membership, be made up of a group of words of the same primary class, but if prepositions are in the verbal primary class and adverbs are in the adverbial primary class, then the Pgp cannot be said to form a group. Indeed, there is very little evidence for either a prepositional group or a conjunction group, since they could be treated as word complexes at the word rank. There is a potential problem with these ‘compositional hierarchies’ (Halliday and Matthiessen 2014:22). As McDonald (2017:252) explains, ‘Because the conceptualisation of classes of word is dependent on their function in classes of group, which are in turn dependent on their function in the clause, only those word classes that commonly function as the Heads of groups are “assigned” to a discrete group class.’ This is precisely why the adjective class is a sub-class of nominal (see above). It is not entirely unreasonable to class the adjective as a type of nominal, since its function is always in relation to some noun, and its function is far more directly related to expressing participant meaning in the clause than process meaning or circumstance meaning. The apparent similarities between adjectives and adverbs are considered as structural (i.e. morphological) rather than functional, although see Tucker (1998). Recall that unit is not defined by similarity of structure: see Berry (1975); Fawcett (2010) for a discussion of the differences between class and type descriptions. It is difficult to consider prepositions as a sub-class of verb other than in the sense that they profile a relation (see Langacker 2016). The reason for this is that in terms of class, its function in the unit above, i.e. in the PP, and thereby in the clause, is not to express a process but rather to express a circumstance. In fact the preposition rarely has a function in the clause (see example (9) below and also Fontaine 2017b). In terms of class, the prepositional units could belong to the sub-class of adverbials rather than verbals. Berry (1975:76–7) maintains a useful distinction between class (defined by function) and type (defined by structure), which allows, for example, clarity when discussing the class of a unit (e.g. Advgp) vs. the type of formal item (e.g. p, preposition, or c, completive). See also Fawcett (1980, 2010), who also maintains such a distinction in terms of functional elements and structural units and items (also see Schulz and Fontaine, this volume). For the primary class of nominals, we do find within the Ngp word classes identified as nominal (e.g. determiner, numeral, adjective, and noun as shown in Figure 5.1), but also adverbial as in example (9), where the Ngp is in bold.8 (9)

8

The air outside feels weird and troubled.

Items such as ‘outside’ that have no complement are not treated as prepositions in SFL but rather as adverbials. This is somewhat problematic, but see Fontaine (2017b) for a discussion of such prepositional items in SFL.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

133

Here, since the qualifier outside is expressed by an adverbial class of word, in such cases the Ngp is not formed of words of the same class as the primary nominal class. Even if we argued that outside belonged to the verbal class, as a preposition, the argument is unchanged. While it is not clear that the Ngp or the Pgp is solely composed of elements belonging to the same class, it does appear that the Vgp adheres to this class principle, since the Vgp is effectively a group of verbal words (e.g. might have been working). Even considering the case of phrasal verbs, the group includes only verbal units since Halliday and Matthiessen (2014:413) define phrasal verbs as ‘lexical verbs that consist of more than just the verb word itself’. The treatment of multi-word expressions which may include nominal units (e.g. kick the bucket) is significant since how they are classed makes a difference to the theory as a whole, requiring an account of lexical representation (see Fontaine (2017b) for a discussion of this point), however this is beyond the scope of the current chapter. The concept of class seems to be based on how the secondary class defines what can occur as Head element in the logical structure of the group. There are three main problems with this: (i) The relationship of secondary class to head does not hold in all cases. In the verbal class, prepositions cannot function as Head in a Vgp. (ii) There is an implication that all classes can be determined in terms of logical meanings (i.e. having a Head element), which excludes the PP. (iii) This view of class suggests that each class has constituency; by the rank scale, however, constituency is best accounted for by experiential structure rather than logical structure, i.e. the clause (at rank) is also a class, and while it has an experiential structure, it has no logical structure. To summarize this section, while the class principle for groups applies to Vgps, it is not strictly the case for Ngps. Furthermore, we find that it could be argued that the class principle suggests that prepositions and adverbials belong in the same primary class due to their functional potential in the clause.

5.3.2 Univariate vs. Multivariate Structures The second distinguishing feature between groups and phrases is the type of ‘variate’ structure they have, i.e. whether they are univariate or multivariate. The two types of structure describe the type of relation between units (e.g. clauses, groups, words). Units on the rank scale are defined in terms of constituency in the sense of being ‘composed of’ units from the rank below, including the possibility of a rankshifted unit from the same or from a higher rank. A univariate structure is defined as one where the individual elements all have the same functional relationship (Halliday and Matthiessen 2014:390). A multivariate structure, on the other hand, consists of units with distinct functional relations (Matthiessen et al. 2010:148). Roughly, groups are fundamentally seen as univariate structures, while phrases are seen as multivariate structures which have no univariate

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

134

LISE FONTAINE AND DAVID SCHÖNTHAL

structure, although, as we shall see, this split is often blurred, which brings into question the distinction between group and phrase units. We can compare univariate and multivariate structures following Matthiessen et al. (2010:145, emphasis in original), who describe univariate structures as structures which are ‘generated as an iteration of the same functional relationship’. Indeed, Matthiessen (1995:627) states that ‘[groups] have a univariate structure with a Head, which is the only obligatory element, and optional Modifiers’. This is a principle that is generally accepted throughout the SFL literature but has never really been examined critically, except perhaps in Fawcett (2010). Two examples of univariate structures are given with the nominal word complex in (10) and the Vgp in (11). (10)

investment trust cash management account

(11)

will have been eaten

In example (10), we find a series of five nouns, and in example (11) a series of four verbs. This type of structure is therefore seen as serial and recursive, involving an interdependency which can be accounted for through the logical metafunction in terms of logical relations or modification (see Martin 1996). However, the example in (10) is not a complete Ngp: it is missing something. Even if we consider this as a word complex, as the Head element of the Ngp, it is not the only obligatory element. It must be grounded as an instance (see Langacker 2016). Actual instances of Ngps with similar word complexes are given in (12) and (13). In both cases, it should be clear that the rankshifting of the person’s and with a 3 per cent annual rate of return prevents a description of these expressions as word complexes. (12)

the person’s income management account (EnTenTen13, SketchEngine)

(13)

a cash management account with a 3 per cent annual rate of return (EnTenTen13, SketchEngine)

As mentioned above, multivariate structures are said to differ from univariate ones in that the structure is ‘a configuration of elements each having a distinct function with respect to the whole’ (Matthiessen et al. 2010:145, emphasis in original). This configuration relates to constituency as discussed above in the sense that there is a kind of generic structure involved which is configurational and non-recursive. This type of structure is accounted for by the experiential metafunction. In this sense, we can describe the Ngp as a configuration of the following elements: Deictic + Numerative + Epithet + Classifier + Thing. Halliday and Matthiessen (2014:437, emphasis in original) describe multivariate structure as follows: Groups have developed their own multivariate constituent structures with functional configurations . . . Here the elements are (i) distinct in function, (ii) realized by distinct classes, and (iii) more or less fixed in sequence. A configuration of such a kind has to be represented as a multivariate

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

135

structure. Treating the group simply as a ‘word complex’ does not account for all these various aspects of its meaning. It is for this reason that we recognize the group as a distinct rank in the grammar.

These three points can be seen as properties of multivariate structures, i.e. their elements have their own function, their elements belong to different classes, and their overall structure is relatively fixed. If we consider the Vgp, it is less clear that it has any similar kind of multivariate structure. Using these two types of structure to distinguish groups from phrases is centred on one key claim and that is that groups have both structures while phrases have only one. This is explained by McDonald (2017:251) as follows: The group incorporates two principles of structure: univariate, in which a single kind of functional relationship is seen as multiplied recursively, for example modification involving a Head and one or more Modifiers; and multivariate, in which a number of different functional roles can be recognised.

It is often argued that the PP is the only unit to have only a multivariate structure, i.e. no univariate structure. However, this is also true of the clause; it is a multivariate structure. In this section, we will consider the criteria of variate structure in terms of how it applies to the group units described above. Halliday and Matthiessen (2014:390) maintain that the Ngp is a ‘multivariate structure: a configuration of elements each having a distinct function with respect to the whole’, which is illustrated in Table 5.12. When compared to the Vgp, as shown in Table 5.13, it seems that the experiential configuration includes elements with very little distinction in terms of class, and that the arrangement appears more similar to the iterative structure of example (10). Quiroz (2017:304–5) points out that ‘a univariate structural interpretation does not map onto the multivariate organisation of the verbal group in any self-evident way’. Fontaine (2017a) has argued against a univariate analysis of the Ngp, although according to different criteria. If the Ngp is more multivariate in nature, the Vgp seems more univariate. Halliday and Matthiessen (2014:398) explain that ‘the verbal group is also structured logically, but in a way that is quite different from, Table 5.12 Nominal group (adapted from Halliday and Matthiessen 2014:364) those

two

splendid

old

electric

trains

Deictic

Numerative

Epithet1

Epithet2

Classifier

Thing

Table 5.13 Verbal group (adapted from Halliday and Matthiessen 2014:397) couldn’t

have

been

going to

be

being

eaten

Finite

Auxiliary1

Auxiliary2

Auxiliary3

Auxiliary4

Auxiliary5

Event

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

136

LISE FONTAINE AND DAVID SCHÖNTHAL

and has no parallel in, the nominal group. The logical structure of the verbal group realizes the system of tense.’ The nature of this unit is not clear and open to debate. Furthermore, with the exception of the Vgp, all group units have the potential to include elements that are rankshifted. These rankshifted elements are said to have no logical relationship with the Head element of the unit, e.g. the Qualifier without sugar in the Ngp a cake without sugar does not have a functional relationship to the Head cake. It is this feature which casts some doubt on the univariate structure of groups. Halliday and Matthiessen (2014:425) take the position that ‘prepositional phrases are phrases, not groups; they have no logical structure as Head and Modifier, and cannot be reduced to a single element’. We can conclude from this that phrases do not have univariate structure (although there is only one such unit at group rank). Given that most groups also have this as potential (except, as already stated, the Vgp), we would have to consider whether we need a nominal group for nominal units without rankshifted elements and a phrase for one that do, i.e. nominal group and nominal phrase, as is the case for prepositions (i.e. prepositional group and prepositional phrase). A similar case could be made for adverbial units at group rank, for example, comparative adverbs such as more quickly than he can, which is more phrase-like than group-like. However, it is much easier to take the position that any unit with the potential for rankshifted elements has a more configurational nature to it and therefore has no univariate structure. Doing so, however, has important theoretical consequences, since it would suggest that, by definition, the Vgp is a group, and that all other units are in fact phrases.

5.3.3 Notion of Rank Scale With groups, ‘rankshifted units can serve as Postmodifiers in Ngps and Advgps (but not in verbal ones) but they are not an obligatory part of the structure as they are in prepositional phrases’ (Matthiessen 1995:627). We have already seen that the Ngp has the potential to include rankshifted units, i.e. to express Qualifiers as well as some Deictics and Epithets; however, the Vgp does not have this potential in English. In the description of the PP in Section 5.2, it was clear that rankshifted units are a key feature of this unit.9 In this section, we will consider the extent to which the optional/obligatory status of rankshifted units is useful to distinguish between groups and phrases. Before discussing how each unit ‘behaves’ in terms of the rank scale, we will briefly outline Halliday’s five principles of the rank scale (taken from Halliday and Matthiessen 2014:9–10, emphases in original): 9

The role of rankshifting is debatable here but without a clear status of lexical representation within SFL theory, along with a robust debate about the rank scale, e.g. ‘be willing/keen/eager to do; be afraid/scared to do’ (Halliday and Matthiessen 2014:427 note 17), but see Tucker (1998). Further discussion of this is beyond the scope of this chapter.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

137

There is a scale of rank in the grammar of every language. That of English (which is typical of many) can be represented as: clause phrase/group word morpheme (b) Each consists of one or more units of the rank next below. For example, Come! is a clause consisting of one group consisting of one word consisting of one morpheme. (c) Units of every rank may form complexes: not only clause complexes but also phrase complexes, group complexes, word complexes and even morpheme complexes may be generated by the same grammatical resources. (d) There is the potential for rank shift, whereby a unit of one rank may be down-ranked (downgraded) to function in the structure of a unit of its own rank or of a rank below. Most commonly, though not uniquely, a clause may be down-ranked to function in the structure of a group. (e) Under certain circumstances it is possible for one unit to be enclosed within another; not as a constituent of it, but simply in such a way as to split the other one into two discrete parts. (a)

In theory, all group units adhere to the first principle, which situates the primary classes of unit in terms of constituency. Principles (b) and (d) will be discussed below, whereas (c) and (e) will not, since the potential to form a unit complex (e.g. through coordination) and the potential to interrupt a unit have no significant bearing on the nature of the class of unit. As we will show, however, the requirement for a group or phrase to consist of the rank below (principle (b)) and the potential for rankshifting (principle (d)) are relevant to the classification of intermediary units. In terms of constituency, principle (b) states that a given unit must contain at least one unit of the rank below it. This principle holds for all group units, and it is difficult to ignore the role that word class (as per Figure 5.1) plays in determining the nature of the unit at this rank, even though originally class is meant to be related to the function served in the unit above. However, as we have seen for the Ngp, there are some difficulties as concerns Ngps that have an adjective as Head. If, for example, we were to eliminate the logical structure at the group unit rank (Fontaine 2017a) and to uncouple adjectives from the nominal word class, then we would find that the Ngp adheres in a more parallel way to the other primary classes. In other words, the current anomaly in the verbal primary class, which includes both verbs and prepositions, connected respectively to a Vgp and a PP, would be addressed in the nominal primary class by including a noun (nominal) group, an adjective group, a determiner group, and a numeral group. This is perhaps unnecessarily complex, and it is not suggested here as a position that should be adopted. However, this raises

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

138

LISE FONTAINE AND DAVID SCHÖNTHAL

important questions. The point being made here is that this is an area that requires attention and development in the theory. Principle (d) states that rankshifted units can serve as postmodifiers in Ngps and Advgps (although not in verbal ones), but that they are not an obligatory part of the structure as they are in PPs. However, as we have seen above, in the case of Advgps with comparative adverbs, the rankshifted postmodifier is obligatory, as in the PP. If we accept an adjective group as Tucker (1998, 2017) does, this also applies to comparative adjectives (e.g. kinder than most people). This distinguishes Vgps from all other units at group rank, since they cannot have a postmodifier and therefore the potential is not there. Another distinguishing feature relates to the fact that Vgps cannot be rankshifted. As Matthiessen (1995:715) points out, ‘Unlike nominal groups, adverbial groups and prepositional phrases, verbal groups serve a single set of functions in the clause and they cannot be rankshifted.’ Hence, these two principles of the rank scale lead us to ask whether there is any need to maintain a distinction between phrase and group at group rank, and whether the Vgp belongs as a unit at this rank. If we adhere to the rank scale principles, we find that many of the assumptions about groups and phrases are challenged, and the picture that emerges is quite different from what might have been expected. Having completed the analysis of the three principle differences between groups and phrases, we will now consider their function in the unit above and the unit below (or same rank) in order to get a full sense of groups and phrases as a unit of rank.

5.3.4 Evaluation of the Group Units Having now considered the three main criteria, we reach the following description. It appears that none of these units maintains all three criteria. There is room for debate in the details. The only unit that stands out as clearly different is the Vgp. While each criterion discussed above is helpful for better understanding the nature of each unit, as a set, the picture is less clear. A summary of the outcome of the three criteria as applied to each unit is given in Table 5.14 and Table 5.15. Here we have added a fourth criterion which questions whether Table 5.14 Experiential structure at and below the clause Class of group unit

Function in experiential structure of clause

Nominal

Participant, Circumstance Process Circumstance Circumstance, Participant

Verbal Adverbial Prepositional

Function in experiential structure below the clause Qualifier in the Ngp, certain types of determiner (Deictic, Numerative, etc.), Range in PP None (no potential for this) Intensifier in Advgp (see Temperer in CG) Qualifier in the Ngp, Standard in Advgp (and Adjgp) as Scope or Finisher in CG

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

139

Table 5.15 Comparison of units following three criteria

Class of unit

Are all word elements of the unit of the same class?

Does the unit have both univariate and multivariate structures?

Are rankshifted units optional?

Can the unit be reduced to a single element? No, if you assume phoricity must be indicated (e.g. *boy is nice vs. boys are nice, the boy is nice), i.e. it must be a grounded instance of a type (see Langacker 2016). Can only be reduced to one element in non-finite clauses, otherwise both the Finite and Event elements are necessary. Yes

Ngp

No, not all are nominal

No, only multivariate

Yes

Vgp

Yes, all are verbal

Yes, both, but this is highly debatable

No rankshifting potential

Advgp

Yes, all are adverbial

No, for comparative adverbials, rankshifting is obligatory

PP

Yes, all are ‘verbal’

Both for simple adverbials, but only multivariate for comparatives No, only multivariate

Adjgp (if we accept that there is a need for this unit)

No, if adjectives are classed as nominals, but yes, if classed as adverbials

No, only multivariate

This depends on assumptions about prepositions, but generally rankshifting is obligatory No, for comparative adjectives, rankshifting is obligatory

No, cannot be reduced to a single element. In SFL intransitive prepositions are considered adverbial.

Yes

the unit can be reduced to a single element, i.e. a single word. As the tables show, it is very difficult to clearly distinguish between phrases and groups. Many of the features that are meant to account for PPs also apply to groups, with the exception of the Vgp. In fact, when all units and criteria are considered, the Vgp is the only unit to stand out from the others so distinctly.

5.4

Concluding Remarks

This chapter set out to consider the nature of units at the rank between clause and word (i.e. the group rank) and to examine the theoretical reasons for distinguishing two different units at this rank. What this has revealed is that depending on the criteria used, comparing units can reveal a variety of outcomes. There are very clearly different types of unit at the group rank, but whether the nature of the units is substantially different remains a

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

140

LISE FONTAINE AND DAVID SCHÖNTHAL

question. We wanted to find out whether it is theoretically justified to maintain two types, i.e. group and phrase. Based on our investigation, one clear result shows through and that is that the Vgp is significantly different from the other group units. It cannot be rankshifted nor does it have the potential to include rankshifted elements as part of its unit. Its organizing principle seems to prefer univariate relations, and it is the only group to have this strong preference. Discussions in the SFL literature related to structure can be somewhat challenging at times. The reliance on univariate and multivariate concepts for defining units is not justified. It might be more productive to think of relations rather than structures when it comes to discussions related to univariate-ness and multivariate-ness. This would allow us to account for the principle organizing system of units. The clause, for example, like the Ngp, is a configuration of elements, but the Vgp seems far less like this and far more serial in nature. In a sense, it makes little difference what these units are called, i.e. whether group or phrase. Each has its own internal structure and functional elements. Given that the term ‘group’ is so prevalent in SFL theories, and given that there does not seem to be any justifiable reason for differentiating between groups and phrases, we suggest to use the term ‘group’ for all the different types of group units at the rank between clause and word, including the prepositional phrase.

References Berry, M. 1975. An Introduction to Systemic Linguistics, Vol. 1: Structures and Systems. London: Batsford. Berry, M. 2017. Stratum, Delicacy, Realisation and Rank. In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 42–55. Caffarel-Cayron, A. 2017. The Verbal Group in French. In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 319–38. Davidse, K. 2004. The Interaction of Identification and Quantification in English Determiners. In M. Achard and S. Kemmer, eds., Language, Culture and Mind. Stanford: CSLI Publications. 507–33. Fawcett, R. 1980. Cognitive Linguistics and Social Interaction: Towards an Integrated Model of a Systemic Functional Grammar and the Other Components of an Interacting Mind. Heidelberg: Julius Groos and Exeter University. Fawcett, R. 2000a. In Place of Halliday’s ‘Verbal Group’, Part 1: Evidence from the Problems of Halliday’s Representations and the Relative Simplicity of the Proposed Alternative. Word 51(2): 157–203. Fawcett, R. 2000b. In Place of Halliday’s ‘Verbal Group’, Part 2: Evidence from Generation, Semantics and Interruptability. Word 51(3): 327–75.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

The Rooms of the House

141

Fawcett, R. 2010. A Theory of Syntax for Systemic Functional Linguistics. Amsterdam: John Benjamins. Fontaine, L. 2013. Analysing English Grammar: A Systemic-functional Introduction. Cambridge: Cambridge University Press. Fontaine, L. 2017a. The English Nominal Group: The Centrality of the Thing Element. In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 267–83. Fontaine, L. 2017b. On Prepositions and Particles: A Case for Lexical Representation in Systemic Functional Linguistics. Word 63(2): 115–35. Fries, P. H. 1977. English Predications of Comparison. In R. DiPietro and E. Blansitt, eds., The Third LACUS Forum 1976. Columbia: Hornbeam Press. 545–56. Ghesquière, L. 2014. The Directionality of (Inter)subjectification in the English Noun Phrase: Pathways of Change. Berlin: Mouton de Gruyter. Halliday, M. A. K. 1961. Categories of the Theory of Grammar. Word 17(2): 241–92. Halliday, M. A. K. and C. M. I. M Matthiessen 2014. Introduction to Functional Grammar. 4th ed. London: Edward Arnold. Kilgarriff, A. et al. 2014. The Sketch Engine: Ten Years On. Lexicography 1: 1–30. Langacker, R. 2016. Nominal Structure in Cognitive Grammar: The Lublin Lectures. Edited by A. Głaz, H. Kowalewski, and P. Łozowski. Lublin: Marie Curie-Skłodowska University Press. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, J. R. 1996. Types of Structure: Deconstructing Notions of Constituency in Clause and Text. In E. H. Hovy and D. R. Scott, eds., Computational and Conversational Discourse: Burning Issues – An Interdisciplinary Account. Heidelberg: Springer. 39–66. Matthiessen, C. M. I. M. 1995. Lexicogrammatical Cartography: English Systems. Tokyo: International Language Sciences Publishers. Matthiessen, C. M. I. M., K. Teruya, and M. Lam. 2010. Key Terms in Systemic Functional Linguistics. London: Continuum. McDonald, E. 2017. Form and Function in Groups. In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 251–66. Morley, D. 2000. Syntax in Functional Grammar: An Introduction to Lexicogrammar in Systemic Linguistics. London: Continuum. Quiroz, B. 2017. The Verbal Group. In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 301–18. Tucker, G. 1998. The Lexicogrammar of Adjectives: A Systemic Functional Approach to Lexis. London: Cassell Academic. Tucker, G. 2017. The Adjectival Group. In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 284–300.

Downloaded from https://www.cambridge.org/core. The Librarian-Seeley Historical Library, on 20 Aug 2019 at 08:33:15, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.007

6 Context and Register Wendy L. Bowcher

6.1

Introduction

This chapter presents an overview of the concepts of context and register; their history in Systemic Functional Linguistic (SFL) theory, the relation between the two concepts, and the most influential perspectives and models developed in this area by SFL researchers.

6.2

Pre-SFL to SFL Theory: A Brief History of the Concepts of Context and Register

6.2.1 Context In the early stages of the development of SFL theory, ‘context’ referred to the semantic level of language; there was phonology, grammar and lexis, and context (see Halliday 1961; Ellis 1966; Gregory 1967). At that time, the extra-linguistic environment was known as ‘situation’ (Halliday 1961; Halliday et al. 1964). Gregory makes the distinction between situation and context in this way: By context is understood the correlation of formally described linguistic features, grouping of such features within texts and abstracted from them, with those situational features themselves constantly recurrent and relevant to the understanding of language events. Situation is an aspect of the description of language events, not a level of language or linguistics. Context is seen as a level of language, as its concern is with certain patterns and pattern correlations which are part of . . . linguistic behavior. (Gregory 1967:178)

I am grateful to Edward McDonald for comments on an earlier draft of this chapter. I alone am responsible for any shortcomings.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

143

During the 1960s, as the ideas of what would become SFL theory began to take shape, context, in the sense of semantics, came to be called ‘meaning’ or simply ‘semantics’, and situation came to be referred to as ‘context of situation’. The changes, however, were not merely terminological, as we will see, but resonated with the changing shape of Halliday’s thinking on the architecture of language. The coinage of the term ‘context of situation’ is attributed to Malinowski, whose ethnographic work in the Trobriand Island communities led him to realize that the concept of the context of an utterance needed to encompass the linguistic, situational, and cultural context of a language (Malinowski 1923:306). Malinowski’s research complemented other work at the time, such as Ogden and Richards’ (1923) discussion of the relation between signs and meaning, and their suggestion that a description of ‘sign situations’ should come from observation of a corpus of instances rather than individual or ‘exceptional’ cases (Ogden and Richards 1923:19) – a scientific, not an ‘intuitive’ approach (Ogden and Richards 1923:20). Within this scholarly climate, Malinowski made important claims regarding the relation between meaning, language, and situation: ‘Language is essentially rooted in the reality of the culture, the tribal life and customs of a people, and . . . it cannot be explained without constant reference to these broader contexts of verbal utterance’ (Malinowski 1923:305). However, being an anthropologist and not a linguistic theoretician, he did not develop the concept of ‘context of situation’ within an explicit theory of language. Firth, a linguist who worked with Malinowski in the 1930s when Malinowski ‘was especially interested in discussing problems of languages’ (Firth 1950:43) stated clearly that, while the term context of situation was ‘first widely used in English by Malinowski’, it became a ‘key concept in the technique of the London group’1 for the study of language (Firth 1950:42). Firth differentiated Malinowski’s concept of context of situation from his own; Malinowski’s concept was essentially a material idea: ‘an ordered series of events’ (Firth 1950:43). For Firth, however, the concept of context of situation was ‘best used as a suitable schematic construct to apply to language events . . . a group of related categories at a different level from grammatical categories but rather of the same abstract nature’ (Firth 1950:43) through which the meanings of language in use could be interpreted. From the start, Firth could see the connection between this abstract concept and language use and how the concept might become part of the description and analysis of language. For example, he presents the following utterance: ‘Ahng gunna gi’ wun fer Ber’’ (I’m going to get one for Bert)

1

The London Group or London School refers to those linguistic scholars in Britain taught and influenced by J. R. Firth. These scholars were also often referred to as ‘neo-Firthians’.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

144

WENDY L. BOWCHER

and asks: What is the minimum number of participants? Three? Four? Where might it happen? In a pub? Where is Bert? Outside? Or playing darts? What are the relevant objects? What is the effect of the sentence? ‘Obvious!’ you say. So is the convenience of the schematic construct called ‘context of situation’. It makes sure of the sociological component. (Firth 1950:44)

Because Firth saw language as inextricably linked with the sociality of living, the study of language had to take into account ‘man’s active participation in the world’ (Firth 1957:2). However, Firth was careful to avoid including in his schema ‘the description of mental processes or meaning in the thoughts of the participants’ and ‘any consideration of intention, purport or purpose’ (Firth 1957:9). Such processes were not ‘ignored’ per se, rather, [a]s we know so little about mind and as our study is essentially social, I shall cease to respect the duality of mind and body, thought and word, and be satisfied with the whole man, thinking and acting as a whole, in association with his fellows. (Firth 1957:2)

With this view of language, context, and ‘the whole man’, the kind of language that was to be studied from Firth’s point of view was ‘actual language text’. Further, the primary sets of relations for enquiry were ‘the interior relations connected with the text itself’, which include the syntagmatic and paradigmatic relations within language elements; the situational relations, which include the ‘interior relations within the context of situation, the focal constituent for the linguist being the text’; the ‘analytic relations set up between parts of the text’; and ‘special constituents, items, objects, persons or events within the situation’ (Firth 1957:5). As for the interior relations of the context of situation, Firth (1957:9) proposed the following: (a)

The participants: persons, personalities, and relevant features of these. (i) The verbal action of the participants. (ii) The non-verbal action of the participants. (b) The relevant objects and non-verbal and non-personal events. (c) The effect of the verbal action. Context of situation was clearly connected, in a scientific sense, with what was observable in the way in which language was used. Text, moreover, was considered a constituent of the context of situation: ‘The placing of a text as a constituent in a context of situation contributes to the statement of meaning since situations are set up to recognize use’ (Firth 1957:11, emphasis in the original). This stance extended to the meaning of a word; a word was considered to be intimately tied to its use within a context of situation, thus setting up a correlation with a specific way of speaking and a specific set of

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

145

words within a given context – ‘collocations are actual words in habitual company’ (Firth 1957:14). Indeed, what emerges from Firth’s writings is that context of situation was invoked as a technique for making ‘statements of meaning’. That is, context of situation could be used as a means of understanding and accounting for the what, the how, and the why of a speech event (see Hasan 2014:3), and was an applicable concept in the study of language activity. This applicability was demonstrated by Halliday (1959/ 1974) who included an analysis of the context of situation based on Firth’s schema in his analysis of a vernacular text in early Mandarin Chinese. Ellis (1966:79) has suggested that ‘context of situation’ is ‘one of Firth’s decisive contributions to linguistic theory’, but he notes that because Firth ‘left this concept in many ways unelaborated’ there were problems which needed to be considered in order to develop linguistic ‘categories, as powerful and as general as possible, to relate the level of form with that of situation’ (Ellis 1966:79). There is much in Ellis’ work that points to the originative scholarly thinking on the concept of context of situation in the late 1950s and early 1960s (and of course, not just within the London School, although that is our focus here), and which also points forward to the concept as it has been eventually theorized in the SFL model of language. For example, he discusses the concepts of ‘potential’ and ‘instance’, citing McIntosh (1961) who introduced the terms ‘potential’ and ‘actual’, and the notion of ‘scale of delicacy’ in relation to linguistic analysis, notation, and semantic focus (see also Halliday 1961). The issue of ‘relevancy’, which has to do with minimizing intuition and maximizing the generality and intersubjectivity of contextual and registerial descriptions, was also an important theme of discussion (see Catford 1965; Ellis and Ure 1974; Gregory 1967; also raised by Ogden and Richards 1923). Generally, context of situation became a key explanatory feature of language in use and generalizable across languages, largely inspired by the observation that the language people use differs in relation to its conditions of use (see Ellis 1965).

6.2.2 Register The recognition that language varies according to how it is used played a significant role in forming the concept of ‘register’ – the term ‘register’ as used in linguistics being attributed to Reid (1956). It goes without saying that one does not speak the same with one’s mother as with one’s friends or husband, nor does a sports commentator speak the same way when commentating a game as when ordering a meal in a restaurant. Firth highlighted this in his advice to the Air Ministry in suggesting the type of Japanese that would be most appropriate to learn in order to take up an effective position against them during World War II: When I was consulted by the Air Ministry on the outbreak of war with Japan, I welcomed the opportunity of service for the Royal Air Force

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

146

WENDY L. BOWCHER

because I saw at once that the operating reconnaissance and fighter aircraft by the Japanese could be studied by applying the concept of the limited situational contexts of war, the operative language of which we needed to know urgently and quickly. We were not going to meet the Japanese socially, but only in such contexts of fighting as required some form of spoken Japanese. (Firth 1950:43–4)

Language varies according to the occasions on which it is being used, and speakers of a language typically have no difficulty recognizing this fact and managing their ‘repertoires’ of language use accordingly. The relationship between register and context of situation became an important focus of discussion and research in a number of published papers in the mid twentieth century (e.g. Ellis 1966; Ellis and Ure 1974; Halliday et al. 1964), and the value and applicability of the term soon became widespread, particularly in the field of language teaching (e.g. Halliday et al. 1964; White 1974; Carter 1978; Ure 1982). Halliday et al. (1964), for instance, referred to register as ‘varieties of language’ that ‘cover the total range of our language activity’, with the further claim that ‘it is only by reference to the various situations, and situation types, in which language is used that we can understand its functioning and its effectiveness’ (Halliday et al. 1964: 89). Ellis (1966) proposed a range of different technical terms covering what he saw as subcategories of register: ‘register’, a linguistic category, and ‘division of idiolect’, distinguished by formal features and correlating with types of situations; ‘register range’, or the total repertory of registers that a person may use; ‘register choice’, or the specific register a person chooses out of his or her repertory of registers; and ‘register-features’, or the features of the language spoken in a specific situation. Register is thus defined as linguistic variation that is ‘use-based’ in contrast to (or perhaps complementary to) variation that is ‘user-based’, such as dialectal variation. This is not to say that a register cannot point to the identity of the user within the situation. Consider, for example, that using the register of sports commentary typically points to the speaker as being a ‘sports commentator’ (although not always), but the major difference between these two types of variation is that register primarily points to the identity of the situation, whereas dialectal variation points to the identity of the speaker. Nevertheless, both register and dialect make contact within the situation; speaking one’s original dialect may be considered disadvantageous in certain situations, such as job interviews, and so some speakers may choose a more standard dialect (see Godley and Escher 2012) in those situations. Speakers are generally aware that they need to speak with relevance to what is taking place and with whom they are interacting, which involves speaking the register of that situation. Because registers are language varieties, they are linguistically defined. That is, ‘it is by their formal properties that registers are defined’ such that Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

147

if two samples of language activity from what, on non-linguistic grounds, could be considered different situation-types show no differences in grammar or lexis, they are assigned to one and the same register: for the purpose of the description of the language there is only one situation-type here, not two. (Halliday et al. 1964:89)

6.2.3 Register in Relation to Context The relationship of register (linguistic variation) to context of situation (situational variation) is of paramount importance in SFL theory. But in the developmental days of SFL theory, how to theoretically account for this relationship was much disputed, with debates often centering on ‘relevancy’, the most useful and descriptive parameters of context of situation, and the relation between register and the nature of the language system itself. As to the description of situational features, various schemas were suggested. Ellis (1966), for instance, suggests several features of the situation: role (participant roles); field (type of activity); formality (informal or formal); mode (spoken or written); and thesis, or the ‘event . . . to which the utterance refers’ (Ellis 1966:84). Halliday et al. (1964:90–2) suggest ‘Field’ or ‘what is going on: to the area of operation of the language activity’; ‘Mode’, or ‘the role played by the language activity in the situation’, such as whether the language is spoken or written; and ‘Style’ ‘which refers to the relations among the participants’. Gregory (1967) weighed in on the debate with a critique of the various attempts at categorizing the situational features and with a proposal of his own categories: Field of Discourse; Mode of Discourse; and Tenor of Discourse. While Gregory preferred the term ‘diatypic variety’ to ‘register’, his introduction of the term ‘tenor’ in place of ‘style’ (the term used in Halliday et al. 1964) has been standard in SFL ever since (see Gregory 1967:195). With regard to relevancy, we have already noted that in Firth’s description of context of situation he refers to ‘relevant features’ and ‘relevant objects’ etc. But just how to determine what is relevant in terms of situational variables was important to unlocking the systematic relationship between language and context so that this relationship could be analyzed more objectively. In response to the issue of ‘relevancy’, Gregory contends: Those situational elements which are potentially contextually relevant to given linguistic forms or groups and complexes of forms are discovered, or ‘invented’, by commutation, by changing, as Catford (1965:36) noted, situational features and observing what textual changes take place, by changing an item or items in the text and observing what situational change occurs. This entails careful and continuous contrasts amongst the records, substantial and situational, of related series of language events. (Gregory 1967:179)

These ‘careful and continuous contrasts’ are indeed important for a scientific approach to understanding the relationship between language and Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

148

WENDY L. BOWCHER

context, but Halliday took things a step further than this. Halliday’s understanding of language as social activity led to important theoretical tenets associated with the concepts of context of situation and register in SFL theory. One of these is the stratal nature of language and the inclusion of context as a stratum. This is a legacy of Firth’s description of language into three strata (Firth 1957; although see also Lamb (1966), whose stratificational grammar was influential in Halliday’s thinking): graphology/phonology, grammar, semantics, and his proposal for the extra-linguistic level of context of situation. Because of his view that language is social activity, Halliday has not considered context to be an isolated concept, but one that is fully integrated into the description of language: The linguistic system has evolved in social contexts . . . the system is a meaning potential, which is actualized in the form of text . . . [and] the situation [is] embodied or enshrined in the text not piecemeal, but in a way which reflects the systematic relation between the semantic structure and the social environment. (Halliday 1977:199)

Another theoretical tenet is the concept of ‘realization’. Realization is an ‘interstratal relationship’ (Halliday 1992:20) and is modelled as a ‘metaredundancy’ relation. That is, with reference to the strata of meaning, wording, and sound, meaning is not just realized by wording and then wording is realized by sound, like a chain of ‘this, then that’. Rather, ‘meaning is realized by the realization of wording in sound’ or from the reverse perspective, ‘sounding realizes the realization of meaning in wording’ (Halliday 1992:24; see also Taverniers, this volume). Hasan refers to this as a ‘realization/activation’ relationship, where, for instance, ‘contextual features activate . . . meaningwording’ and meaning-wordings realize contextual features (Hasan 2014:45). Of key importance in the model of SFL is Halliday’s hypothesis that text is ‘a semantic unit’ and a ‘continuous process of semantic choice’ realized in lexicogrammatical choices (Halliday 1977:193–5). The semantic system is organized under three main categories: ideational (experiential and logical), interpersonal, and textual meanings (see Taverniers, this volume). These meanings pattern in specific ways across texts and are the means through which the social situation is ‘embodied’ in the text (Halliday 1977). The semantic system itself is thus an interface between the social system (the various contexts of situation) and the grammatical system of the language. In situations of language in use, certain linguistic choices are favoured or ‘at risk’. Halliday has demonstrated that this relationship is patterned and non-random, and not absolute or categorical, and although his early work expressed this relation in a rather ‘deterministic’ way (e.g. Halliday 1977), it is clear that the idea of ‘at risk’ or ‘favoured’ options does not suggest a one-to-one relation between the features of context and the activated linguistic features: The patterns of determination that we find between the context of situation and the text are a general characteristic of the whole complex that is formed by a text and its environment. We shall not expect to be able to show that the options embodied in one or another particular sentence are determined by the field, Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

149

tenor and mode of the situation. The principle is that each of these elements in the semiotic structure of the situation activates the corresponding component in the semantic system, creating in the process a semantic configuration, a grouping of favoured and foregrounded options from the total meaning potential that is typically associated with the situation types in question. This semantic configuration is what we understand by the ‘register’: it defines the variety . . . that the particular text is an instance of. The concept of register is the necessary mediating concept that enables us to establish the continuity between a text and its sociosemiotic environment. (Halliday 1977:203, emphasis added)

In English, Halliday demonstrates that ideational meaning is systematically, although not entirely, realized through the grammatical system of transitivity, with logical meaning realized through systems for clause complexing, interpersonal meaning largely through the mood system, and textual meaning primarily through systems of theme and information as well as through cohesive devices. Thus, from context through the strata of language and vice versa, Halliday has shown an activation-construal relation. Meaning resides not at one level (e.g. semantics) but is a property of all levels of language and into context (see Halliday 1985; Firth 1957; Berry, this volume; Butt, this volume; Taverniers, this volume; Webster, this volume). Thus, ‘relevancy’, and accounting for what is relevant in terms of features of the context of situation in relation to register, as it turns out, requires a systematic, realization/activation model of language and the social system, and this is evident in Halliday’s finely tuned systemicfunctional architecture of language. Crucial in the process of realization is that of instantiation, an ‘intrastratal relation’ wherein observation and description of language are moved along a scale from potential (the system) to instance (the text). That is, it is the same phenomenon that is being examined, but it is being examined from a different ‘depth of vision’. Halliday explains this by making an analogy with the relationship between climate and weather (Halliday 1992). Further, it is through instantiation that the language system remains ‘metastable’: ‘persist[ing] only through constantly changing by interpenetration with [the] environment’ (Halliday 1992:26). It is this metastability that allows for a high degree of predictability in situations of language use. That is, There is no situation in which the meanings are not to a certain extent prescribed for us. There is always some feature of which we can say, ‘This is typically associated with this or that use of language’. Even the most informal spontaneous conversation has its strategies and styles of meaning. (Halliday 1985:40; cf. Cloran 1987)

Because registers are varieties of language according to how language is being used in a situation, the instantiation of a register is a configuration of linguistic meanings, or ‘a continuous process of semantic choice’ (Halliday 1977:195). The SFL concept of text as a ‘semantic’ unit is important in the overall understanding of the relationship between context and register. Halliday explains: Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

150

WENDY L. BOWCHER

A text has a generic structure, is internally cohesive, and constitutes the relevant environment for the selection in the ‘textual’ systems of the grammar. But its unity as a text is likely to be displayed in patterns of ideational and interpersonal meaning as well. A text is the product of its environment, and it functions in that environment. (Halliday 1977:195, emphasis added; also see Hasan 1985)

This observation is echoed throughout SFL theory. Text is not simply a language event but a ‘sociological event’ (Halliday 1978:139). It is ‘naturally occurring language use . . . having a social function, and possessing the attributes of texture and structure’ (Hasan 2014:4). Moreover, because a text represents a specific register, its unity is a characteristic of that register. That is, while a text may involve different ‘movements’, collectively these movements hang together in a way that is characteristic of a specific register. Thus, how context of situation and register are related has to do with those features of the language system that are likely to be ‘at risk’ in a given situation. Halliday has pursued a line of inquiry which aims to specify those aspects of the context of situation which affect or ‘rule’ (Halliday 1977:19) what speakers choose from the options that make up the semantic system, the metafunctions. His theory models ‘the systematic relationship between language and the environment’ (Halliday 1977:19) by proposing that the context of situation is a semiotic construct and ‘an instance, or instantiation, of meanings that make up the social system’ (Halliday 1977:19). These are the ‘conditions’ (to use an earlier term) under which linguistic choices are made. Variation in the conditions correlates with variation in what is said. It follows, then, that register, being variation in language use, is modelled as a ‘semantic configuration’ and located at a different stratum than context of situation. That is, register is located at the semantic level, not above it. Shifting in register means re-ordering the probabilities at the semantic level . . . whereas the categories of field, mode and tenor belong one level up. These are the features of the context of situation; and this is an interface. But the register itself I would see as being linguistic; it is a setting of probabilities in the semantics. (Thibault 1987:610, emphases in original)

With regard to text and its relation to context of situation and to register, Halliday makes the following claim: The text is a continuous process. There is a constantly shifting relation between a text and its environment, both paradigmatic and syntagmatic: the syntagmatic environment, the ‘context of situation’ (which includes the semantic context – and which for this reason we interpret as a semiotic construct), can be treated as a constant for the text as a whole, but is in fact constantly changing, each part serving in turn as environment for the next. And the ongoing text creating process continually modifies the system that engenders it, which is the paradigmatic environment of the text. (Halliday 1977:198)

Halliday’s concepts of register and context (including context of situation) and their place in SFL theory are shown in Figure 6.1. Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

SYSTEM

INSTANCE

sociocultural environment

REALIZATION: activation/construal

CONTEXT

LANGUAGE

language system

151

semantics

specific context of situation contexts of situation

context of situation type

registers

register type

text (semantic unit)

lexicogrammar phonology

Figure 6.1 Register and context in SFL theory (adapted from Halliday 1999:8)

In Figure 6.1 we can see two basic features of SFL theory: the hierarchy of strata (Context – Language) and the cline of instantiation (System – Instance). As already noted, the strata are related through the relation of realization/construal: context is realized in language and language construes context, and this is indicated by the vertical lines. The cline of instantiation (the horizontal axes) depicts varying perspectives on the same phenomenon: a specific language event represents a selection from the language system; a specific context of situation represents a selection from the sociocultural context. Between the system and instance ends of the cline are ‘types’ or categories of contexts of situation correlating with types of texts or registers. Instantiation and realization are fundamental principles explaining how language changes and is also maintained, that is, how it is an ‘open dynamic system’ in that each instantiation (whether that be of context or language) ‘resets’ the ‘overall probabilities’ of the respective systems (Halliday 1992:27).

6.3

Context and Register: Developments and Perspectives

Within SFL theory, there have been several developments and variations of Halliday’s conception of context and register. This section briefly describes the most influential of these.

6.3.1 Hasan: The Essentialness of Context Hasan has consistently demonstrated and validated the SFL centrality of context for an understanding of language. Whether in analyzing lexicogrammatical choices, semantic systems, contextual parameters, or cultural consequences of language, she has repeatedly argued for keeping context in view, no matter what the analytical goal, if one is to understand the workings of language: ‘Seeing language as a form of human social action does not preclude attempts to understand the logic of its form and vice versa’ Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

152

WENDY L. BOWCHER

(Hasan 2005:68, emphasis in original). She has bemoaned the idea that context is often used merely to illuminate the nature of parole, [but] not of langue . . . [or] . . . a good device for ‘mopping up’ some of the problems that inhere in an essentially extra-social view of language. . .where context becomes a mundane ‘reality’ to be taken for granted, while language becomes a mysterious mental organ, and grammar a body of knowledge encrypted in the human brain at birth. (Hasan 2001:3, italics in the original)

Rather, Hasan has argued that context is central to the ontogenesis, phylogenesis, and logogenesis of language (see Hasan 2001), and is part of ‘a productive principle, reflection on which [has] enabled SFL to offer a scientific description of “how language works”’ (Hasan 2005:55–6). In sum, Hasan’s work speaks to her view that ‘there can be no language without context’ (Hasan 2001:8). Overall, Hasan’s contribution has added detail, precision, and depth to Halliday’s conception of context and register, not least because she has questioned many of the assumptions and terminology associated with these two concepts (see Lukin 2016 for a lengthy discussion of Hasan’s contribution to the development and description of context). An example of her questioning includes an early discussion on the concept of relevancy. Although Halliday theorized the probabilistic correlation between features of language and features of context, it was in probing the idea of relevancy and its relation to context and register that Hasan has proposed that relevant context, or the context of situation, is embedded in a material situational setting (MSS): ‘Situation-type is an abstraction from the totality of material situational setting’ (Hasan 1973:275; also see Hasan 1981:110, Hasan 1995:219). Her later work has honed this idea: those features of the social situation which are ‘illuminated’ in the text (Hasan 1995:219, 2005:61) realize the relevant contextual parameters of field, tenor, and mode, the features of the context of situation. The MSS, on the other hand, is not part of the context of situation, but acts as ‘a dormant source for affecting the verbal goings on’ (Hasan 1981:110). Features of the MSS enter the description of the context of situation when they are directly referenced, no matter how minor that reference may be. Hasan has proposed that both the MSS and the context of situation could be encapsulated in the overarching terms Action, Reflection, and Contact (ARC): ‘I think of ARC as relevant to any form of joint social practice, whether this involves language or not. By contrast, field of discourse, tenor of discourse and mode of discourse are . . . specifically discourse related’ (Hasan 2001:7; also see Hasan 2014:11–13). The concept of Material Situational Setting has been linked with various other concepts in the theory, including that of institutionalization, which

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

153

relates to the degree of negotiation possible within a situation due to the convergence (or not) of different semiotic systems through which a situation typically unfolds. Hasan argues: Social processes which are institutionalized would logically be multiply coded semiotically . . . And what, from the point of view of verbal coding, may be seen as just a material situational setting, can from the point of view of the social process be seen as situation semiotically coded through a series of distinct codes. (Hasan 1981:116)

Bowcher (1999) builds on this idea by outlining a set of questions to probe the context of a text to assist in ‘measuring’ the degree of institutionalization in a given context. For instance, in a court of law, layout of the room, seating arrangements, functionally designated furniture (e.g. the witness box), placement of certain individuals, forms of dress, ritualized procedures and actions, formulaic expressions, etc., multiply code social roles and relations, activities, and the order of those activities. There is little room for individual negotiation of these features of the situation, thus indicating a highly institutionalized context. Hasan early on flagged the connection between MSS and contextdependent and context-independent text (Hasan 1973), an issue which she returns to in Hasan (1999), and which is also taken up in detail by Cloran (1994, 1999), who shows how a single text may move in and out of a stronger or lesser connection with the material environment in which it takes place. Cloran categorizes these different ‘moves’ in terms of rhetorical units. A ‘rhetorical unit’ (RU) analysis is different from rhetorical structure theory (RST) analysis (see Mann and Thompson 1988; Mann et al. 1992) and ‘takes as its point of departure Hasan’s message semantics’ (Cloran 1999:196–7; also see Low and Fung, this volume). It involves analyzing the way in which messages group together to form distinct configurations of semantic features, an RU being a grouping of semantic features in a text. The semantic structure of a text is thus shown to have a hierarchy: a text consists of one or more RUs, and an RU consists of one or more messages (Cloran 1999:197). At a more local level, Hasan has interrogated specific terminology. An example of this is her discussion of the term ‘activity’: In ordinary life, the word has many meanings and each seems clear in its ‘context’, but what exactly did it mean in the description of field of discourse? Here it seemed to have multiple values: it was not clear if the word referred to precisely the same phenomenon in its various appearances, such as ‘social activity’, ‘relevant activity’, ‘language activity’. There were also ‘descriptive references’ such as ‘what is going on’, and ‘the area of the operation of the language activity’. Sometimes ‘the whole activity’ was said to consist of two kinds of ‘activities’, a ‘language activity’ which ‘assisted’ ‘the whole event’, in which case it would seem that the ‘whole event’ was to

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

154

WENDY L. BOWCHER

consist of both ‘language activity’ and some other kind of ‘activity’ which was not linguistic. At other times, the ‘whole of the relevant activity’ could be accounted for ‘practically’ by ‘language activity’. (Hasan 2014:6)

Such querying has led to more rigorous accounts of, for instance, the concept of action in relation to field and mode (see Hasan 1999, 2014; Bowcher 2013, 2014), but there is still more work to be done in this regard as evidenced by Hasan’s comment: ‘I have long felt the need to explore the differences between “act”, “action” and “activity”: this would bring greater order in understanding field as the parameter concerned with “doing” of some kind’ (Hasan 2014:51 note k). Hasan’s conception of context and register as ‘configurations’ of features has been an important influence on how researchers now research and model the relationship between context and text. With reference to context of situation, Hasan prefers the term ‘configuration’ to ‘combination’ because the features of context do not simply ‘combine’; ‘rather, contextual configuration is like a chemical solution, where each factor affects the meanings of the others’, thus claiming an ‘interdependence between the three parameters’ (Hasan 1995:231). The concept of contextual configuration (CC) being ‘an account of the significant attributes of [a given] social activity’ (Hasan 1985:56) is central to understanding two key features of the semantic nature of text: its structure and texture. Texture refers to the relations among the meanings of a text and is determined through an analysis of a text’s ‘cohesive harmony’, which relates to the cohesive ties among elements of a text (see Taboada, this volume). Text structure is termed ‘generic structure potential’ (GSP),2 as it is used to describe the structure of not only a specific text type but also a range of other related text types. Each member of that range of text types will have some structural properties in common with other members: no individual text type will have the same structural shape as any other, and none will be entirely different. The entire range of such text types will constitute a single register family. (Hasan 2014:9–10, emphases in original)

A GSP has both obligatory and optional elements. The obligatory elements are defining in terms of the text’s register, while the optional elements indicate variation among texts belonging to a given register. Hasan has demonstrated how the CC can be used to ascertain obligatory and optional elements of a text as well as the sequence of these elements and their iteration (Hasan 1985:56). While both GSP and texture reside at the semantic level of language, GSP is said to act as a link between the texture of a text

2

GSP was originally called ‘generalised structure potential’ (see Hasan 1978, 2014:51 note i).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

155

and the relevant context of a text (Hasan 1985:99). With regard to the relation between texture and context, Hasan comments that ‘situation type, at a high degree of specificity, is relevant to texture; you could see it as the motivating force of texture. But by the same token, the facts of texture construe the very detailed aspects of the situation in which the text came to life’ (Hasan 1985:115). Hasan also observes that cohesive chains may ‘display a close relationship to the structural movement of the text’ (Hasan 1985:115). This has been demonstrated in work such as Butt et al. (2010), Cloran (1999) (in relation to Rhetorical Units and Material Situation), and Lukin (2010) (see Khoo 2016 for a survey of work on Cohesive Harmony). Hasan’s GSP was innovative, and perhaps a concept before its time, as it has not been fully understood within SFL. Some critics have suggested it is a static or ‘synoptic’ representation of text structure (see Hasan 1995), but the concept of GSP offers a means of accounting for structure in terms of both variation and stability across registers, or text types, and in relation to features of context of situation. Hasan’s argument has always been that ‘the elements of text structure cannot be defined by reference to the rank status or sequential ordering of the lexicogrammatical units which have the function of realizing these elements’ (Hasan 1978:229). Rather, the controls upon the structural make-up of a text are not linguistic in origin . . . instead, the control is contextual: the nearest non-linguistic analogue of a text is not a logico-mathematical formula, but a non-verbal social event. A text is a social event whose primary mode of unfolding is linguistic. (Hasan 1978:229)

Ultimately, a GSP statement ‘explicitly signals those features whose selection would be the realization of some systematic variation across the derived structures: the derived structures do not vary accidentally; they vary with predictable perturbations in the configuration of the underlying context’ (Hasan 2014:10). Various scholars have utilized Hasan’s concept of GSP including Cloran (2016), in relation to Rhetorical Units, and more recently Bortoluzzi (2010), Bowcher (2015), Bowcher and Liang (2016), and Cheong (2004), in relation to the structure of multimodal texts. Other ideas associated with the concept of contextual configuration include ‘interdependence’ and ‘permeability’. With reference to context, interdependence refers to the fact that what one does with whom and in what manner are connected. Interdependence is significant in the process of realization as it means that there is no one-to-one relation between contextual variables and features of the language, i.e. field is not singularly realized in the experiential metafunction, etc. (Hasan 1995). However, Hasan points out there are ‘default’ linguistic realizations of the contextual features, and that language, being an open-dynamic system, necessitates ‘the possibility of departures from the highly probable’ (Hasan 2014:8). Hasan has been critical of those SFL scholars unsatisfied with the probabilistic nature of the realization relation between the contextual variables and

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

156

WENDY L. BOWCHER

the semantic systems (see Hasan 1995). She points out that the ‘predicting’ power of the contextual parameters with the metafunctions is a reflection of the nature of the relationship between social variables and linguistic choices, a mode of explanation wholly acceptable in sociolinguistics, where ‘predictions about linguistic features correlating with situational ones such as age, gender, geographical and/or social provenance, have typically been stated in probabilistic terms’ (Hasan 2014:7; see also Hasan 1995, 1999). With regard to the concept of ‘permeability’, Hasan posits that this has to do with the ‘conditioned environments’ and the ‘regularity of a set of relations’ that show how meanings are ‘expanded’ in the sense that the meaning ‘space’ of certain linguistic categories is permeated by the meanings of another category, and complementary to this, certain linguistic categories ‘may abandon’ their meaning space and permeate that of other categories (Hasan 2016:passim). Permeability exists between categories at the same stratum of language, with ‘the diagnosis of permeability [being] assisted by inter-stratal relation’ (Hasan 2016:374). Realization is thus ‘needed to recognize the relation of permeability’ (Hasan 2016:374), but realization is of a different kind of relation. The concept of permeability is thus different from both realization and interdependency.3 Another important contribution made by Hasan to the study of context and register is her work on the paradigmatic representation of context. For this she uses system networks, the modus operandi for representing the paradigmatic nature of different features at different language strata (see Webster, this volume). Hasan has argued that ‘the design of the system network is well suited to contextual parameters’ (Hasan 2014:14) as much as it is to representing choices at other strata. This does not mean this form of representation is necessarily straightforward (see Bowcher 2014 for a discussion of some of the issues): networks can get highly complex, and the different levels of abstraction require a different ‘value’ assigned to the descriptive choices. Essentially, however, system networks are ‘a form of argument’ and ‘are a consistent means of checking what is the better motivated proposal in linguistic description [in that] [t]he network either accounts for the linguistic variation and its consequences, choice by choice, or it does not’ (Butt 2001:1825, emphasis in original). While there has been considerable progress in the development of system networks for each of the contextual variables of field, tenor, and mode, including work by Berry (2016), Bowcher (2007, 2013, 2014), Butt (2004), and Hasan (1999, 2009, 2014), there is still much work that needs to be done in specifying their features and applying them to the study of context and register. An example of a system network for contextual field is shown in Figure 6.2.

3

See Hasan (2016) for a discussion and demonstration of her concept of permeability, and Miller and Bayley (2016) for work on the concept of hybridity.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

157

praccal ACTION conceptual

natural (Sensible) irrealis (Intelligible) specialized FIELD

SPHERE OF ACTION quodian instuonal individualized

bounded spao-temporal locaon

connuing immediate

PERFORMANCE OF ACTION

longitudinal goal orientaon

overt unconscious constant variable

Figure 6.2 System network for field (from Bowcher 2014:203)

6.3.2 Matthiessen: Registerial Cartography The importance of register and context in the study of language is underscored in the following comment by Matthiessen: Languages are aggregates of registers, and they evolve through registers. Registers emerge as adaptations to new contextual pressures on languages . . . and they may fade away as contextual conditions change: the registerial make-ups of languages keep evolving, changing the character of languages in the course of evolution. (Matthiessen 2014a:7; also see Matthiessen and Teruya 2016:212)

Matthiessen and colleagues are developing a context-based typology of registers, described as ‘a large-scale registerial cartography for a wide range of languages’ (Matthiessen 2015:2). The research aims to

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

158

WENDY L. BOWCHER

examine, describe and theorize registers according to Halliday’s trinocular vision . . . supplementing the view ‘from above’ – from contexts, with the views ‘from below’ – from lexicogrammar and phonology (or graphology), and ‘from roundabout’ – from the level of semantics itself, the level at which the variation takes place in the first instance (in terms of the ‘meanings at risk’ in different contexts). (Matthiessen 2014a:8, emphasis in original)

This work on mapping registers derives directly from Jean Ure’s unpublished work on register classification, which focused primarily on field and mode (Matthiessen 2015:3–5). The starting point for classifying registers is context, since registers are categorized by reference to their contexts of situation (Matthiessen 2015). Thus, features from all parameters of context of situation are relevant to categorizing a register, although so far in this project, it is the contextual variable of field that has been given the most attention. Since field concerns the ‘nature of the activity’, Matthiessen uses the concept of activity, or social process, as his starting point. He distinguishes three broad types of processes which he defines in the following way: Semiotic processes (i.e. ‘meaning’ processes – semiotic processes constitutive of context, manifested through social processes) (b) Semiotic processes potentially leading to social processes (i.e. ‘meaning’ leading to ‘doing’) (c) Social processes (i.e. ‘doing’ processes – social processes constitutive of context, semiotic processes facilitating (i.e. ‘meaning’ facilitating ‘doing’)). (Matthiessen 2014b:170–1)

(a)

These three superordinate categories capture a range of contexts: from those which are entirely constituted by language through to those which are almost entirely materially construed but in which language plays a facilitative role. Within these three broad categories are eight primary fields of activity: ‘expounding’, ‘reporting’, ‘recreating’, ‘sharing’, and ‘exploring’ fall within category (a); ‘recommending’ and ‘enabling’ fall within (b); and ‘doing’ falls within (c). Each of these primary fields can be subdivided to produce more delicate descriptions of activity types. Figure 6.3 shows the eight primary fields of activity (in the inner circle) with further subcategorization into secondary fields (in the outer circle). The three broad categories, however, are not shown. The figure itself represents a ‘typology’ of fields of activity. In Figure 6.3, we can see that each primary field of activity can be further subdivided into more delicate choices such as in the case of ‘recommending’ into ‘promoting’ or ‘advising’. As more analyses are conducted, greater degrees of delicacy of these socio-semiotic processes will emerge. For instance, more delicate distinctions for the category of ‘expounding’ are given in Figure 6.4 (also see Matthiessen 2015:10).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

159

Figure 6.3 Matthiessen’s map of the different ‘fields of activity’ (from Matthiessen 2014a:11)

Figure 6.4 More delicate distinctions for the category of ‘expounding’

In registerial mapping, as in any account of register, contextual descriptions go hand in hand with analyses at the semantic and lexicogrammatical level. However, Matthiessen also makes use of rhetorical structure theory (RST) (Mann and Thompson 1988; Mann et al. 1992) in analyzing the logicosemantic relations between parts of a text. RST is a descriptive framework for analyzing text structure in terms of the patterns of relations that hold between parts of a text, the hierarchical arrangement of text elements, and the communicative role played by text structure. Matthiessen adapts RST for a more systems-focused view of text relations. He proposes a system termed logico-semantic relation. This system has three primary systems: nuclearity, logico-semantic type, and orientation. The system of nuclearity is derived directly from RST and concerns the

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

160

WENDY L. BOWCHER

distinction between ‘relations linking the text segments as equal in status . . . or as unequal’ (Matthiessen 2014a:15); logico-semantic type presents choices between projection and expansion (see Halliday and Matthiessen 2014; Berry, this volume); and orientation presents a choice between ‘linking two text segments as representations of experience (“external”) or as interaction moves (“internal”)’ (Matthiessen 2014a:15; also see Halliday and Hasan 1976). Analyses involve shunting across strata to build a picture of the linguistic and contextual features of different registers in order to demonstrate and describe how they differ or overlap with each other (Matthiessen 2015). There are various publications in which the registerial cartography project is described and exemplified. For instance, Matthiessen (2014b) presents a useful description and exemplar analysis utilizing the registerial mapping techniques in relation to developing an SFL ‘appliable discourse analysis’ (ADA), and Matthiessen and Teruya (2016) work through several different texts which display different types of registerial indeterminacy, such as ‘blends’, ‘ambiguity’, and ‘overlap’. Matthiessen (2015) presents analyses of several text types in order to illustrate the nature of the analyses involved in registerial mapping. He also presents a discussion of how registerial mapping can be applied in the fields of educational linguistics and in healthcare communication studies. In the latter area, Matthiessen argues that the mapping of healthcare discourses, particularly those concerned with ‘communication in emergency (or accident and emergency) departments in large hospitals’ (Matthiessen 2015:44) can be used to ‘identify the registers that a patient is likely to have to engage with both within and outside institutions of healthcare . . . [and] trace patient journeys through a hospital department’ (Matthiessen 2015:45). Matthiessen’s work on registerial cartography engages directly with research on genre typology and topology conducted by Martin and colleagues (see Section 6.3.3). Matthiessen acknowledges ‘many connections with the very rich and detailed work on “genre agnation”, typically “genres” of writing . . . within the “genre model” (Matthiessen 2014b:173).

6.3.3 Martin: Context: Genre and Register In modelling the relationship between context and language, Martin (1992) utilizes the Hjelmslevian concepts of ‘connotative’ and ‘denotative’ semiotic and ‘planes’ of analysis (Hjelmslev 1961). Context is modelled as a connotative semiotic, a semiotic system which is abstract and requires a different semiotic system through which it is expressed, this latter system being thus denotative. In Martin’s model, the connotative plane is stratified into genre (context of culture) and register (context of situation). Together, these make up the ‘content plane’. Language is then modelled as the ‘expression plane’, the denotative system. Thus, register in Martin’s model is not treated as a text type, as we have seen in Halliday’s, Hasan’s, or

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

161

Matthiessen’s conception of register. Rather, register is located at the stratum of context and is ‘constituted by the contextual variables field, tenor and mode . . . [and] is the name of the metafunctionally organized connotative semiotic between language and genre . . . a semiotic system in its own right’ (Martin 1992:502). The contextual plane of genre is proposed as a means of accounting for the goal or purpose of a social process in the sense that analysis at this level focuses ‘on making explicit just which combinations of field, tenor and mode variables a culture enables, and how these are mapped out as staged, goal-oriented social processes’ (Eggins and Martin 2012:175). The term ‘genre’ replaced Martin’s earlier term ‘functional tenor’, which had been borrowed from Gregory (1967; see Martin 1999). In Martin’s early work, functional tenor was treated as a variable different from the other contextual variables of field, tenor, and mode, and was considered useful in describing the social purpose of a text (Martin 1999:28). Martin argues that separating functional tenor (genre) from the other contextual variables had ‘the advantage of consolidating Halliday’s suggestion that field was naturally related to ideational meaning, tenor to interpersonal meaning and mode to textual meaning’ (Martin 1999:27; see also Martin 1992:505). An important difference between genre and register in Martin’s model is that whereas register is metafunctionally organized, i.e. is organized into three parameters of field, tenor, and mode, which correlate with the semantic systems of ideational, interpersonal, and textual meanings, genre is not metafunctionally organized. Martin argues that this means that ‘texts can be classified in ways which cut across metafunctional components in language’ (Martin 1992:505) in the sense that a text can be classified as a specific genre-type, with a specific schematic structure, and within the structural components of the text different values of field, tenor, and mode may be realized at different stages in the genre. Thus, while register is a ‘metafunctionally diversified’ ‘reading’ of context, genre provides a ‘metafunctionally transcendent’ ‘reading’ (Martin 2012a:276); it is a plane above register. Further, genre becomes a ‘pattern of register patterns’ in that it ‘shapes’ register ‘by conditioning the way in which field, mode and tenor are recurrently mapped onto one another in a given culture’ (Martin 2012d:64). Martin’s stratified connotative semiotic model of context with language as denotative semiotic is shown in Figure 6.5. Martin has also proposed the term ‘macro-genre’ to refer to texts which contain more than one genre. In (Martin 2012b) he uses a secondary school student’s written geography report entitled ‘Endangered Species’ to illustrate how several genres may be drawn on to produce one macro-genre. An interesting direction for research into macro-genres is the exploration of not only how different genres within a macro-genre relate to each other, but ‘into how many and what kinds of macro-genre’ different genres may occur (Martin 2012e:313). The relation among the analytical planes as shown in Figure 6.5 is one of realization, with the relation of realization shown by the double-headed

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

162

WENDY L. BOWCHER

connotative semiotic stratified context plane

expression form

tenor genre

field mode discourse semantics

phonology/ graphology

lexicogrammar

stratified content form

expression form

denotative semiotic Figure 6.5 Martin’s model of stratified context plane (connotative semiotic) with language as expression form (denotative semiotic) (from Martin 1999:40)

arrows. Martin explains realization as ‘a scale of abstraction involving the recoding of one level of meaning as another’ (Martin 2008:32). Thus, for example, lexicogrammar realizes the more abstract level of discourse semantics, and the denotative semiotic of language as expression form realizes the connotative semiotic of context, the latter being a more abstract plane. Instantiation is also modelled as a scale, but in this case it is a scale of ‘generalization’ (Martin 2008:32) and can be represented as an ‘instantiation hierarchy’ (see Figure 6.6). System Genre/register Text type Text Reading

(generalized meaning potential) (semantic sub-potential) (generalized actual) (affording instance) (subjectified meaning)

Figure 6.6 Martin’s instantiation hierarchy (Martin 2008:33)

On the right-hand side of Figure 6.6 are several key terms. ‘Generalized meaning potential’ refers to the totality of choices possible within a culture. Within this system are ‘sub-potentials’, such as genres and registers. The term ‘generalized actual’ refers to the text type, which is also a potential, in that various individual texts may be classified as a certain type, but unlike genre and register, which are ‘extra-linguistic’ levels, this term refers to the categorization of a group of actual texts. The term ‘affording instance’ attempts to capture the idea that the semantic and lexicogrammatical choices of a text motivate a certain interpretation or reading of a text, and the notion of ‘subjectified reading’ refers to the way a text is actually read and understood, deriving from the ‘meaning potential afforded by individual texts’ (Martin 2008:33). Thus, the hierarchy represents a narrowing of perspective from potential (most generalized) to instance, the latter being ‘the reading of a particular text’ (Martin 2008:33).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

163

Martin’s work includes comparing and contrasting related genres (genre agnation) to explore the ways in which genres are alike (a typological perspective) and how and to what extent they are similar or different (a topological perspective). That is, a typological analysis categorizes texts as belonging to one generic type or another (Martin 2012a:262). The basis for categorization depends on what is seen as ‘foregrounded’ in a text; it is ‘privileging one dimension of texture over another as more or less critical for categorization’ (Martin and Rose 2008:130). For instance, in developing a typology of history texts, Martin and Rose start with the basic categorical opposition of texts that are ‘field timed’ (following a chronological flow) and those which are ‘text timed’ (following a logogeneric/textual flow). A topological analysis, on the other hand, ‘approaches genre agnation as a matter of degree, arranging texts on clines with respect to their similarities and differences’ (Martin 2012a:262). For instance, in order to ascertain similarities and differences among texts, analysis might focus on such features as choices in mood, expansion, or tense (see Martin 2012a:263–4). Martin has shown the value of this kind of analysis in educational contexts where ‘topological analysis has been used . . . to map learner pathways for generic development’ (Martin 2012a:264), particularly in primary and secondary literacy curricula. Martin’s concept of genre has been widely applied within various fields of education including literacy development in mother-tongue and secondlanguage contexts and cross-disciplinary language and curriculum development (see Martin 2012c; Byrnes, this volume; Mickan, this volume).

6.4

Context: Networks and Scales

Recent work on context includes advances in the development of system networks for the contextual parameters of field, tenor, and mode. For instance, working with networks developed by Butt (2004) and Hasan (1999, 2009), Bowcher (2013, 2014) has developed the field and mode networks in an attempt to ‘focus the Field network more on the nature of the activity in terms of its experiential elements and at the same time . . . [focus] the Mode network more closely on features to do with the modalities of expression and the degree to which these modalities may be deployed in a situation’ (Bowcher 2014:199; also see Figure 6.2 in Section 6.3.1). Part of her work has centred on the argument that the primary choices within the networks ‘should reflect the core defining features of the contextual parameters’ and her system networks reflect this stance (Bowcher 2014:177). She has also considered the contextual choice of material action proposed by Hasan (1999), and whether this choice should be located in the network for field or for mode, with Bowcher (2014) contending that it is better located in mode. Her arguments concerning material action include issues surrounding the contexts of multimodal texts (Bowcher 2007, 2013, 2014),

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

164

WENDY L. BOWCHER

as well as those surrounding the concepts of relevancy (what falls within relevant context) and trace (how do features of different contexts leave their trace (or not) in the language of the text) (Bowcher 2013). These latter issues are also included in Bartlett’s (2013, 2016, 2017) work on context. Bartlett has explored the relationship between context of situation and the ‘environment’ within which a context is embedded, arguing that we need to look ‘beyond the text to the semiotic histories that are embedded in the wider environment’ (Bartlett 2013:346) and thus incorporate into our understanding of the relationship between context and text sociological and ethnographic detail. Part of his argument rests on the observation that context of situation cannot always be ‘read off the text’ (Bartlett 2017:317). Rather, certain choices of language, or twists and turns in a text, may reflect cultural knowledge and understandings rather than ‘illuminate’ any specific contextual feature (Bartlett 2013:348). In order to explore the relations between context and language, Bartlett takes up the concepts of first- and second-order context (see Halliday 1977), where, for example, the first-order field might be a game of football and the second-order field a discussion of that game; first-order tenor relations would be those not defined by language but by the social system, such as mother and child, whereas second-order tenor relations come into being within a given situation, such as questioner-respondent (Halliday 1977:201–2; Matthiessen 2009). Bartlett has also questioned certain relationships said to pertain between context and language. For instance, the relations of ‘construal’ and ‘activation’, as in the following formulation: language construes context and context activates language (see Hasan 2014). He contends that these relations are not truly converse relations for all aspects of context and language. For instance, language does not ‘construe’ the channel of communication (Bartlett 2016). Taking all these issues into account, Bartlett argues for a multiscalar model of context consisting of ‘environment’ – ‘everything that surrounds the situation, including social and individual histories, as well as material features of the setting’ – and ‘semiotic context’, which he defines as ‘the second-order reality that is construed by, and which can be read off, the text itself’, and which he labels ‘sctx’ (Bartlett 2017:385). He also proposes four concepts which he suggests better describe the relations between context and text: ‘activation’ (the influence of any feature of the environment on any aspect of a text whether that be covert or overt); ‘construal’ (the way in which contextual features can be explicitly ‘read off’ the text); ‘correlation’ (the ‘tendencies’ for certain language features to ‘co-occur’ with certain environmental features); and ‘indexicality’ (‘ways of speaking’ associated with socially established activities or groups of people) (see Bartlett 2017:385–6). Berry (2016) is also concerned with the notion of ‘construal’ and with not only features of context that can be read off a text but which may influence linguistic choices in some way. She proposes a model of context that takes

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

165

into account ‘pre-text’ context and ‘via-text’ context features. Pre-text contextual features may be within the general parameters of field, tenor, and mode. For instance, prior to a telephone conversation, there are the pre-text mode features of ‘spoken’ and ‘not-co-present’ already in place. Social distance features might also be in place prior to a text coming into existence, and hence be pre-text tenor features, but during the text, such features may change, such as when a friend who is also one’s boss switches from friendly chat to outlining some job requirements needed within the next day or two. Such changes to the social status quo would be considered via-text contextual features. Berry has developed system networks to account for these kinds of features (see Berry 2016).

6.5

Concluding Remarks

This chapter has provided only an outline of the concepts of context and register, their development, models, and current research. Context and register are cornerstones of SFL theory (Lukin et al. 2011) and offer a rich and varied way into studying language and its relation to society. Readers are encouraged to consult the reference list to gain a deeper understanding of the concepts, their theoretical and research value, and their analytical utility.

References Bartlett, T. 2013. ‘I’ll Manage the Context’: Context, Environment and the Potential for Institutional Change. In L. Fontaine, T. Bartlett, and G. O’Grady, eds., Systemic Functional Linguistics: Exploring Choice. Cambridge: Cambridge University Press. 342–64. Bartlett, T. 2016. Multiscalar Modelling of Context: Some Questions Raised by the Category of Mode. In W. L. Bowcher and J. Y. Liang, eds., Society in Language, Language in Society: Essays in Honour of Ruqaiya Hasan. Basingstoke: Palgrave Macmillan. 166–83. Bartlett, T. 2017. Context in Systemic Functional Linguistics: Towards Scalar Supervenience? In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 375–90. Berry, M. 2016. On Describing Contexts of Situation. In W. L. Bowcher and J. Y. Liang, eds., Society in Language, Language in Society: Essays in Honour of Ruqaiya Hasan. Basingstoke: Palgrave Macmillan. 184–205. Bortoluzzi, M. 2010. Energy and Its Double: A Case Study in Critical Multimodal Discourse Analysis. In W. Swain, ed., Thresholds and Potentialities of Systemic Functional Linguistics: Applications to Other Disciplines, Specialised Discourses on Languages Other than English. Trieste: Edizione Universitarie Trieste. 158–81.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

166

WENDY L. BOWCHER

Bowcher, W. L. 1999. Investigating Institutionalization in Context. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 141–76. Bowcher, W. L. 2007. Field and Multimodal Texts. In R. Hasan, C. M. I. M. Matthiessen, and J. J. Webster, eds., Continuing Discourse on Language, Volume 2. Sheffield: Equinox. 619–46. Bowcher, W. L. 2013. Material Action as Choice in Field. In L. Fontaine, T. Bartlett, and G. O’Grady, eds., Systemic Functional Linguistics: Exploring Choice. Cambridge: Cambridge University Press. 318–41. Bowcher, W. L. 2014. Issues for Developing Unified Systems for Contextual Field and Mode. Functions of Language 21(2): 176–209. Bowcher, W. L. 2015. Structure and Multimodal Texts. In J. Wildfeurer, ed., Building Bridges for Multimodal Research: International Perspectives on Theories and Practices of Multimodal Analysis. Frankfurt am Main: Peter Lang. 167–89. Bowcher, W. L. and J. Y. Liang. 2016. GSP and Multimodal Texts. In W. L. Bowcher and J. Y. Liang, eds., Society in Language, Language in Society: Essays in Honour of Ruqaiya Hasan. Basingstoke: Palgrave Macmillan. 251–74. Butt, D. G. 2001. Firth, Halliday and the Development of Systemic Functional Theory. In S. Auroux, E. F. K. Koerner, H.-J. Niederehe, and K. Versteegh, eds., History of the Language Sciences. Berlin: Walter de Gruyter. 1806–38. Butt, D. G. 2004. Parameters of Context: On Establishing the Similarities and Differences between Social Processes. Unpublished mimeo. Sydney: Macquarie University. Butt, D. G., A. R. Moore, C. Henderson-Brooks, R. Meares, and J. Haliburn. 2010. Dissociation, Relatedness and ‘Cohesive Harmony’: A Linguistic Measure of Degrees of ‘Fragmentation’. Linguistics and the Human Sciences 3(3): 263–93. Carter, R. 1978. Register, Styles and Teaching Some Aspects of the Language of Literature. Educational Review 30(3): 227–36. Catford, J. C. 1965. A Linguistic Theory of Translation. London: Oxford University Press. Cheong, Y. Y. 2004. The Construal of Ideational Meaning in Print Advertisements. In K. O’Halloran, ed., Multimodal Discourse Analysis: Systemicfunctional Perspectives. London: Continuum. 163–95. Cloran, C. 1987. Negotiating New Contexts in Conversation. Occasional Papers in Systemic Linguistics 1: 85–110. Cloran, C. 1994. Rhetorical Units and Decontextualisation: An Enquiry into Some Relations of Context, Meaning and Grammar. Nottingham: University of Nottingham. Cloran, C. 1999 Context, Material Situation and Text. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 177–217. Cloran, C. 2016. Construing Instructional Contexts. In W. L. Bowcher and J. Y. Liang, eds., Society in Language, Language in Society: Essays in Honour of Ruqaiya Hasan. Basingstoke: Palgrave Macmillan. 275–99.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

167

Eggins, S. and J. R. Martin. 2012. Genres and Registers of Discourse. In Z. Wang, ed., Genre Studies: Collected Works of J. R. Martin, Volume 3. Shanghai: Shanghai Jiaotong University Press. Ellis, J. 1965. Linguistic Sociology and Institutional Linguistics. Linguistics 19: 5–20. Ellis, J. 1966. On Contextual Meaning. In C. E. Bazell, J. A. Catford, M. A. K. Halliday, and R. H. Robins, eds., In Memory of J. R. Firth. London: Longmans. 79–95. Ellis, J. and J. Ure. 1974. The Contrastive Analysis of Language Registers. Unpublished mimeo. Firth, J. R. 1950. Personality and Language in Society. The Sociological Review: Journal of the Institute of Sociology 42(2): 37–52. Firth, J. R. 1957. A Synopsis of Linguistic Theory 1930–55. In J. R. Firth et al., eds., Studies in Linguistic Analysis: Special Volume of the Philological Society. Oxford: Blackwell. 1–32. Godley, A. and A. Escher. 2012. Bidialectal African American Adolescents’ Beliefs about Spoken Language Expectations in English Classrooms. Journal of Adolescent and Adult Literacy 55(8): 704–13. Gregory, M. 1967. Aspects of Varieties Differentiation. Journal of Linguistics 3(2): 177–274. Halliday, M. A. K. 1959/1974. The Language of the Chinese ‘Secret History of the Mongols’. Oxford: Basil Blackwell. Halliday, M. A. K. 1961. Categories of the Theory of Grammar. Word 17(3): 241–92. Halliday, M. A. K. 1977. Text as Semantic Choice in Social Contexts. In T. A. van Dijk and J. S. Petofi, eds., Grammars and Descriptions. Berlin: Walter de Gruyter. 176–225. Halliday, M. A. K. 1978. Language as Social Semiotic: The Social Interpretation of Language and Meaning. London: Edward Arnold. Halliday, M. A. K. 1985. Part A. In M. A. K. Halliday and R. Hasan, Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Geelong: Deakin University Press. 1–49. Halliday, M. A. K. 1992. How Do You Mean? In M. Davies and L. Ravelli, eds., Advances in Systemic Linguistics: Recent Theory and Practice. London: Pinter. 20–35. Halliday, M. A. K. 1999. The Notion of ‘Context’ in Language Education. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 1–24. Halliday, M. A. K. and R. Hasan. 1976. Cohesion in English. London: Longman. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Routledge. Halliday, M. A. K., A. McIntosh, and P. Strevens. 1964. The Linguistic Sciences and Language Teaching. London: Longmans. Hasan, R. 1973. Code, Register and Social Dialect. In B. Bernstein, ed., Class, Codes and Control, Volume 2: Applied Studies towards a Sociology of Language. London: Routledge. 253–92.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

168

WENDY L. BOWCHER

Hasan, R. 1978. Text in the Systemic-functional Model. In W. U. Dressler, ed., Current Trends in Textlinguistics. Berlin: Walter de Gruyter. 228–46. Hasan, R. 1981. What’s Going on: A Dynamic View of Context in Language. In J. E. Copeland and P. W. David, eds., The Seventh LACUS Forum 1980. Columbia: Hornbeam Press. 106–21. Hasan, R. 1985. Part B. In M. A. K. Halliday and R. Hasan, Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Geelong: Deakin University Press. 51–118. Hasan, R. 1995. The Conception of Context in Text. In P. H. Fries and M. Gregory, eds., Discourse in Society: Systemic Functional Perspectives. Meaning and Choice in Language: Studies for Michael Halliday. Norwood: Ablex. Hasan, R. 1999. Speaking with Reference to Context. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 219–328. Hasan, R. 2001. Wherefore Context? The Place of Context in the System and Process of Language. In R. Shaozeng, W. Guthrie, and I. W. Ronald Fong, eds., Grammar and Discourse: Proceedings of the International Conference on Discourse Analysis. PRC Macau: Publications Centre, University of Macau. 1–30. Hasan, R. 2005. Language and Society in a Systemic Functional Perspective. In R. Hasan, C. M I. M. Matthiessen, and J. J. Webster, eds., Continuing Discourse on Language: A Functional Perspective, Volume 1. Sheffield: Equinox. 55–80. Hasan, R. 2009. The Place of Context in a Systemic Functional Model. In M. A. K. Halliday and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 166–89. Hasan, R. 2014. Towards a Paradigmatic Description of Context: Systems, Metafunctions, and Semantics. Functional Linguistics 1(9): 1–54. Hasan, R. 2016. In the Nature of Language: Reflections on Permeability and Hybridity. In D. R Miller and P. Bayley, eds., Hybridity in Systemic Functional Linguistics: Grammar, Text and Discursive Context. Sheffield: Equinox. 337–83. Hjelmslev, L. 1961. Prolegomena to a Theory of Language. Madison: University of Wisconsin Press. Khoo, K. 2016. ‘Threads of Continuity’ and Interaction: Coherence, Texture and Cohesive Harmony. In W. L. Bowcher and J. Y. Liang, eds., Society in Language, Language in Society: Essays in Honour of Ruqaiya Hasan. Basingstoke: Palgrave Macmillan. 300–30. Lamb, S. 1966. Outline of Stratificational Grammar. Washington, DC: Georgetown University Press. Lukin, A. 2010. ‘News’ and ‘Register’: A Preliminary Investigation. In A. Mahboob and N. K. Knight, eds., Appliable Linguistics. London: Continuum. 92–113. Lukin, A. 2016. Language and Society, Context and Text: The Contributions of Ruqaiya Hasan. In W. L. Bowcher and J. Y. Liang, eds., Society in

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

Context and Register

169

Language, Language in Society: Essays in Honour of Ruqaiya Hasan. Basingstoke: Palgrave Macmillan. 143–65. Lukin, A., A. R. Moore, M. Herke, R. Wegener, and C. Wu. 2011. Halliday’s Model of Register Revisited and Explored. Linguistics and the Human Sciences 4(2): 187–213. Malinowski, B. 1923. The Problem of Meaning in Primitive Languages. Supplement 1. In C. K. Ogden and I. A. Richards, eds., The Meaning of Meaning. New York: Harcourt, Brace & World. Mann, W. C and S. A. Thompson 1988. Rhetorical Structure Theory: Toward a Functional Theory of Text Organization. Text 8(3): 243–81. Mann, W. C., C. M. I. M. Matthiessen, and S. A. Thompson. 1992. Rhetorical Structure Theory and Text Analysis. In W. C. Mann and S. A. Thompson, eds., Discourse Description: Diverse Linguistic Analyses of a Fund-raising Text. Amsterdam: John Benjamins. 39–78. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, J. R. 1999. Modelling Context: A Crooked Path of Progress in Contextual Linguistics. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 25–61. Martin, J. R. 2008. Tenderness: Realisation and Instantiation in a Botswanan Town. In N. Nørgaard, ed., Systemic Functional Linguistics in Use. Odense: University of Southern Denmark. 30–62. Martin, J. R. 2012a. A Context for Genre: Modelling Social Processes in Functional Linguistics. In Z. Wang, ed., Genre Studies: Collected Works of J. R. Martin, Volume 3. Shanghai: Shanghai Jiaotong University Press. 248–77. Martin, J. R. 2012b. From Little Things Big Things Grow: Ecogenesis in School Geography. In Z. Wang, ed., Genre Studies: Collected Works of J. R. Martin, Volume 3. Shanghai: Shanghai Jiaotong University Press. 278–302. Martin, J. R. 2012c. Language in Education: Collected Works of J. R. Martin, Volume 7. Edited by Z. Wang. Shanghai: Shanghai Jiaotong University Press. Martin, J. R. 2012d. Language, Register and Genre. In Z. Wang, ed., Register Studies: Collected Works of J. R. Martin, Volume 4. Shanghai: Shanghai Jiaotong University Press. 47–68. Martin, J. R. 2012e. A Universe of Meaning: How Many Practices? In Z. Wang, ed., Genre Studies: Collected Works of J. R. Martin, Volume 3. Shanghai: Shanghai Jiaotong University Press. 303–13. Martin, J. R. and D. Rose. 2008. Genre Relations: Mapping Culture. Sheffield: Equinox. Matthiessen, C. M. I. M. 2009. Ideas and New Directions. In M. A. K. Halliday and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 12–58. Matthiessen, C. M. I. M. 2014a. Registerial Cartography: Context-based Mapping of Text Types and Their Rhetorical-relational Organization.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

170

WENDY L. BOWCHER

Proceedings of the 28th Pacific Asia Conference on Language, Information and Computation. Unpublished conference paper. Matthiessen, C. M. I. M. 2014b. Appliable Discourse Analysis. In Y. Fang and J. J. Webster, eds., Developing Systemic Functional Linguistics: Theory and Application. Sheffield: Equinox. 138–208. Matthiessen, C. M. I. M. 2015. Register in the Round: Registerial Cartography. Functional Linguistics 2(9): 1–48. Matthiessen, C. M. I. M. and K. Teruya. 2016. Registerial Hybridity: Indeterminacy among Fields of Activity. In D. R. Miller and P. Bayley, eds., Hybridity in Systemic Functional Linguistics: Grammar, Text and Discursive Context. Sheffield: Equinox. 205–39. McIntosh, A. 1961. ‘Graphology’ and Meaning. Archivum Linguisticum 13: 107–20. Miller, D. R. and P. Bayley, eds. 2016. Hybridity in Systemic Functional Linguistics: Grammar, Text and Discursive Context. Sheffield: Equinox. Ogden, C. K. and I. A. Richards. 1923. The Meaning of Meaning. New York: Harcourt, Brace & World. Reid, T. B. W. 1956. Linguistics, Structuralism and Philology. Archivum Linguisticum 8: 28–37. Thibault, P. J. 1987. An Interview with Michael Halliday. In R. Steele and T. Threadgold, eds., Language Topics: Essays in Honour of Michael Halliday. Amsterdam: John Benjamins. 601–27. Ure, J. 1982. Introduction: Approaches to the Study of Register Range. International Journal of the Sociology of Language 35: 5–33. White, R. 1974. The Concept of Register and TESL. TESOL Quarterly 8(4): 401–16.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:37:13, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.008

7 Intonation Wendy L. Bowcher and Meena Debashish

In memory of William S. Greaves (1935–2014).

7.1

Introduction

Intonation has always been a part of the description of language in Systemic Functional Linguistics (SFL), but it has received less attention than many other areas of the theory. Work that is substantively devoted to the description of intonation primarily includes Halliday’s (1967, 1970) early publications, Tench (1996),1 and Halliday and Greaves (2008). There are also two collections focusing on Systemic Phonology: Tench (1992b) and Bowcher and Smith (2014b). This chapter draws from all these volumes and other individual publications to present an overview of the SFL model of intonation and a detailed description of the English tone groups. Our aim is to provide a usable description of English intonation with occasional reference to recent research, areas under debate, and possible future directions.

7.2

Background

Intonation is the main feature of phonology that is emphasized in SFL theory. The SFL description of intonation derives largely from the work of Firth, who distinguished ‘prosodic systems from phonematic systems’ (Firth 1948:128) thus placing equal emphasis on syntagmatic and paradigmatic features. His concept of ‘prosody’ was different from that of ‘suprasegmental features’ as developed in (mostly) American structuralist traditions in that Firth considered sounds and phonological features in context thus revealing a 1

Although aligning with much of the SFL intonation framework, there are some differences between Tench’s (1996) description of intonation in English and Halliday’s.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

172

WENDY L. BOWCHER AND MEENA DEBASHISH

“top-down” analysis of language’ (Couper-Kuhlen 1986:4). Further, Firth considered the structuralist focus on ‘segmentation’ of sounds as ‘artificial and actually misrepresent[ing] the phonology of the language’ (Tench 1992a:5). Rather, he aimed to integrate into the study of phonology such features as the ‘interrelation of syllables’ in terms of length, stress, tone quality, nasality, and voice quality (Firth 1948:138), and the study of phonology more fully into a theory with the investigation of meaning at its core. Firth’s prosodic analysis was non-universalist and polysystemic: nonuniversalist in that the phonological systems of different languages were described in their own terms, acknowledging that ‘phonetic features which in one language are treated as prosodic may not be so treated, or may be so with reference to different structures, in other languages’ (Robins 1970:194); and polysystemic in that the approach ‘treats language as a complex set of interacting systems, each with its own characteristic properties’ (Ogden 2012: para 3). With regard to his polysystemic approach, Firth (1948:151) says: The phonological structure of the sentence and the words which comprise it are to be expressed as a plurality of systems of interrelated phonematic and prosodic categories. Such systems and categories are not necessarily linear and certainly cannot bear direct relations to successive fractions or segments of the time-track instances of speech. By their very nature they are abstractions from such time-track items. Their order and interrelations are not chronological.

Meaning was central to Firth’s ideas on language (see Butt, this volume) and to the study of phonology: The meaning of any particular instance of everyday speech is intimately interlocked not only with an environment of particular sights and sounds, but deeply embedded in the living processes of persons maintaining themselves in society . . . the dominating interest of the immediate situation, the urge to diffuse or communicate human experience, the intimate sounds, these are the origins of speech. (Firth 1968:13)

Such an approach interprets the sounds of speech as realizing ‘worded meaning within context of situation’ (Bowcher and Smith 2014a:9). Meaning is likewise central to SFL theory. From a phonological viewpoint, spoken language is understood as a ‘succession of melodies’ (Halliday 1970:6) which collaborate with other linguistic choices and resonate with grammatical, semantic, and contextual features during the act of meaning. With regard to the relationship between intonation and grammar, Halliday (1967:10) points out: It is not enough to treat the intonation systems as if they merely carr[y] a set of emotional nuances superimposed on the grammatical and lexical items and categories. . . . English intonation contrasts are grammatical: they are exploited

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

173

in the grammar of the language. The systems expounded by intonation are just as much grammatical as are those, such as tense, number and mood, expounded by other means . . . Therefore, in the description of the grammar of spoken English, ‘intonational’ and ‘non-intonational’ systems figure side by side.

With regard to meaning, intonation is studied from the point of view that ‘[i]f you change the intonation of a sentence you change its meaning . . . Intonation is one of the many kinds of resources that are available in the language for making meaningful distinctions’ (Halliday 1970:21). Moreover, its significance in the act of communicating is indispensable, as Tench (1996:151) comments: Intonation features at every point in the process of communication; it is impossible to account for any kind of linguistic communication without it. From the process of reception by the addressee to interpretation and evaluation, intonation is recognised, processed and taken into account. Even in writing, and then reading, intonation plays a part.

Over the course of the development of SFL theory, there has been a shift in the description of intonation in relation to other linguistic categories. For instance, whereas earlier volumes introducing the SFL model of grammar described intonation as ‘beside the clause’ (see IFG1 and IFG2),2 later volumes (IFG3 and IFG4)3 more fully integrate intonation into descriptions of various features of the grammar, reflecting Halliday’s early assertion that “intonational” systems operate at many different places in the grammar’, they are not independent but are ‘incorporated throughout the description wherever appropriate’ (Halliday 1967:10–11). The different pitch movements – falling, rising, or combinations of these – contribute to melodic variation in language, and each language has its own set of tones, a system of tone choices, which contributes to the meaning-making system of the language. Moreover, as with other systems and levels of analysis in SFL theory, intonation(al) systems are interdependent with other systems. This more encompassing view of the role of intonation in the description of language reflects the SFL multifaceted architecture of language. The next section outlines the place of intonation in the SFL architecture of language.

7.3

Intonation in the SFL Architecture of Language

Within the SFL architecture of language, there are several organizational dimensions from which language is modelled and studied. They include the following:

2

IFG1 and IFG2 refer to Halliday (1985a) and (1994) respectively.

3

IFG3 and IFG4 refer to Halliday and Matthiessen (2004) and (2014) respectively.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

174

WENDY L. BOWCHER AND MEENA DEBASHISH

• • • • •

The hierarchy of strata related through realization; The hierarchy of rank related through constituency; The cline of instantiation which is a relation between system and text; The axes of paradigmatic and syntagmatic, the former displaying relations of delicacy and the latter of rank; The metafunctional organization of language.

Intonation is discussed in relation to each of these dimensions.

7.3.1 The Hierarchy of Strata Related through Realization In SFL theory, language is modelled as systems at several strata. These strata are context, meaning, lexicogrammar, phonology, and phonetics, which are related through realization.4 The stratum of phonetics has a different relationship in terms of realization than the other strata because phonetics refers to the potential resources from which ‘sounded’ meanings are made. A pitch contour, for instance, is a phonetic feature, but when deployed in a meaningful way it becomes a part of the phonological description of a language. In SFL, phonology and lexicogrammar are located at different strata, and often graphology is shown as the ‘written equivalent’ of phonology. This is true only in so far as they are located at the same stratum, but it is important to keep in mind that each of these modes has specific grammatical and semantic consequences and resonances (see Davies 2014; Fawcett 2014). Halliday distinguishes between intonational systems and intonation systems. Intonational systems are located at the stratum of lexicogrammar and include the system of information distribution, and intonation systems at the stratum of phonology and include such systems as tonicity and tone. At the stratum of grammar, spoken English is analyzed in terms of information units, and at the stratum of sound (phonology) in terms of tone units, or tone groups. These units are related to each other through realization: a tone unit realizes an information unit. While realization can be considered between strata – lexicogrammar realizes semantics – it is important to keep in mind that in the act of meaning through languaging, realization is a multi-stratal, simultaneous operation, and not a deterministic or ‘one-by-one’ mechanism that takes place between units in one stratum and those in another, one after the other (see Halliday 1992). In the overall operation of realization, context redounds with the redundancy of the semantics with the redundancy of the lexicogrammar with the phonology, where redound means ‘realizes and is realized by’ (Halliday 1992; see also Taverniers, this volume). The relation of realization is presented in Figure 7.1 with tone unit and information unit shown.

4

The Cardiff model of language is a bi-stratal model. For a discussion of this in relation to intonation, see Fawcett (2014: 331–4).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

Context

(Semancs

(Lexicogrammar informaon unit

175

Phonology )) tone unit

Figure 7.1 The relation of realization among the strata of language with tone unit and information unit shown [note: the downward slanting arrows mean ‘is realised by’] (cf. Halliday 1992:24)

However, to say that a tone unit realizes an information unit appears contradictory to the statement that there is no ‘one-to-one’ relation between units in one stratum and those in another. This apparent contradiction is explained in the following way: The information unit is a unit of the lexicogrammar; so it functions in the construction of meaning – it faces the semantics, so to speak. The tone unit, on the other hand, is a unit of the phonology; it functions in the organization of speech sound. (Halliday and Greaves 2008:41)

Thus in terms of grammar, the information unit organizes speech in terms of the flow of messages, whereas the tone unit organizes speech in terms of the flow of sound or melodies. So the location of the information unit at the stratum of grammar and of the tone unit at the stratum of phonology is a means of modelling these two different perspectives on the same phenomenon, that of speech. However, the resource of sound and the more abstract organizational role of grammar are not the same kind of phenomena. Therefore, the boundaries of the tone unit and the information unit do not always coincide exactly. This is illustrated in the following excerpt, which is analyzed for tone groups. The excerpt also illustrates some of the conventions used for displaying intonation analyses: a double forward slash for the boundary of the tone unit, a single forward slash for a foot boundary, a caret5 used to indicate a silent Ictus. // 4 ᴧ by the /time the /Great /Central was /built the // 1+ trains could / manage the /gradients /much more /easily and the //13 Great /Central /line //. . . (excerpt from Halliday 1970:127) Table 7.1 indicates that the first and second tone units encompass more ‘wording’ than the information units they realize due to the tail of the tone melody extending over a few words belonging to the subsequent grammatical units of clauses or phrases. This wording is shown in italicized text in Table 7.1. When these tone units are represented as information units, the ‘tails’ are shifted into the information unit, which brings the information

5

The use of the caret ᴧ varies across the SFL literature. Some authors prefer to use a raised caret ^ (e.g. Halliday and Matthiessen 2014; O’Grady 2017), while others (Halliday 1970; Halliday and Greaves 2008) prefer to use the low caret ᴧ. In this chapter we use the low caret so as to distinguish this from the raised caret which typically means ‘is followed by’.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

176

WENDY L. BOWCHER AND MEENA DEBASHISH

Table 7.1 Tone units and information units (tone unit boundaries marked with double forward slashes) Type of Unit

Utterance

Tone Unit Information Unit Tone Unit Information Unit Tone Unit Information Unit

//by the time the Great Central was built the// by the time the Great Central was built ___//trains could manage the gradients much more easily and the// the trains could manage the gradients much more easily ______//Great Central line// and the Great Central line

unit more in line with the grammatically discrete elements of clauses, phrases, groups, or words (see Smith 2008:62 note 1, 96, Appendix 3). The coinciding of an information unit with a clause is considered to be an unmarked relation, or ‘neutral tonality’, with deviations from this being marked (Halliday and Greaves 2008: 142). However, in Fawcett’s generative model of intonation, the concept of neutral tonality is not taken up. Rather, the approach is modelled on the lines that where there is a new semantic (syntactic) unit in a sentence, the choice is whether or not that semantic unit should be ‘given a separate information unit’ (Fawcett 2014:328).

7.3.2 The Hierarchy of Rank Related through Constituency The hierarchy of rank is a means of modelling the constituents of each stratum. For intonation, the hierarchy of rank primarily applies to the stratum of phonology. For English, the phonological units are tone unit (melodic line), foot, syllable, and phoneme. These may differ in other languages. For instance, McGregor (1992) suggests that for Gooniyandi they are tone unit, word, syllable, and phoneme. In English, a tone group consists of one or more feet, a foot one or more syllables, and a syllable one or more phonemes. Figure 7.2 shows the constituents of the tone unit. melody

//

TONE UNIT

//

*The tone unit or melodic line is the highest phonological rank. *Tone unit boundary shown by double forward slashes.

//

*A tone unit or melodic line of speech may consist of one or more feet. *A foot boundary is shown by a single forward slash.

melodic line

rhythm

// foot / foot /

//

S2

/

/ S1

foot

/ foot

/ S1 S2

S3

//

arculaon

//

phph/

/phphph /phph phph phph //

*Each foot contains one or more syllables. *Salient syllables are located immediately to the right of the single forward slash. *The first syllable in the tone unit may be silent. *A foot may also be silent. *The caret is used to denote a silent ictus. *The tonic syllable is underlined. *A syllable is constuted by one or more phonemes.

Figure 7.2 The constituents of the tone unit showing some analytical conventions

As already noted, the tone unit at the phonological stratum realizes the information unit in the grammatical stratum. At the grammatical stratum

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

177

the information unit is the only rank recognized in Halliday’s model of systemic phonology and is on par with the clause (see Halliday and Matthiessen 2014:115), albeit of a different realizational phenomenon. Matthiessen (1995:603) suggests that this is because the rank scale ‘is “taken over” from phonology’. This becomes clear in considering the syntagmatic axis in relation to the intonation and intonational systems (see Section 7.3.4). However, Smith (2008) proposes a unit at a lower rank than the information unit which he calls the information group. An information group is realized by a foot, and the Ictus in the foot realizes a Prominent. The associated systems are information grouping, a system at the rank of information group which is below the information unit and located in the system of information distribution, and information prominence, a system at the rank of Prominent, which is below the rank of information unit and located in the system of information focus. Smith identifies instances of rank shift between information unit and information group. One example of this is the choice of a foot for each word in a lexicogrammatical group, such as a nominal group. He explains that this kind of choice construes ‘a textual prosody’, ‘which heightens the attention to the elements of this group’ (Smith 2008:110). This is popularly (i.e. often in the mass media) represented in written form by using a period after each word in a group or a clause: Every. Single. Piece. (word level) I Just. Don’t. Know. (clause level) Smith explains that the selection of a number of Prominents as well as the Focus (realized by the tonic) in a nominal group or clause ‘may be ascribed to the “zooming/focussing” power of this system’ (Smith 2008:111). There is also the suggestion of a unit above the tone unit – a phonological paragraph (a paraphone), that Halliday raised himself in an article on grammatical categories (Halliday 1961; see also Iwamoto 2014; Tench 1996). However, this is not included in the description of rank but is discussed in relation to the textual metafunction (see Section 7.3.5).

7.3.3

The Cline of Instantiation: Relations between System and Instance System networks are a hallmark of SFL representations of choices available at each rank of linguistic analysis and represent sets of choices that are potentially available in any instance of language in use. At the rank of foot are the systems of foot composition and ictus state. The foot is a rhythmic unit, and the choices available range from one or more syllables. Halliday (1970:1) says: The first syllable in the foot is always salient. The salient syllable carries the beat . . . [and is] followed by one or more non-salient, or weak syllables . . . A foot may begin with a silent beat, without the rhythm being disrupted or lost.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

178

WENDY L. BOWCHER AND MEENA DEBASHISH

The first syllable in the foot whether silent or spoken is called the Ictus. Further, this might be the only choice, and thus the foot would be a ‘simple foot’ which could be silent or spoken. On the other hand, there may be any number of syllables in a foot, and those syllables following the Ictus, whether one or more, are known collectively as Remiss. A foot of this latter type is called a compound foot. The system network for choices in foot composition and ictus state is shown in Figure 7.3. At the rank of tone unit, the three main systems of intonation are tonality, tonicity, and tone (see Figure 7.4). Tonality refers to choices available for the organization of a discourse into tone units; tonicity refers to the selection and assignment of prominence within the tone unit; and tone refers to the choices of pitch movement. Halliday (1963a, 1963b, 1967, 1970) based these systems of English intonation on various samples of spoken English, the largest consisting of ‘just under 2,000 tone groups’ (Halliday 1967:9).

simple FOOT COMPOSITION compound +Remiss Remiss: syllable1-n Ictus^Remiss foot +Ictus ICTUS STATE

filled Ictus:syllable1-n empty Ictus:silent

Figure 7.3 System network for choices in and Matthiessen 2014:18)

FOOT COMPOSITION

and ICTUS

STATE

(from Halliday

TONALITY

INTONATION

TONICITY

SYSTEMS

TONE

Figure 7.4 Intonation systems

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

179

Consider, for instance, the following excerpt from a recording of a story read for children: //1 Mr. /Fox//1 ᴧ was /strolling /through the /woods// 3 when he / noticed// 1 ᴧ a /plump /hen// 1 sitting //1 ᴧ on the /branch of a /tree// (extract from Storynory n.d.) In this sample of reading aloud, we can see a high number of tone groups across one or two clauses. For example, the second clause ‘when he noticed a plump hen sitting on the branch of a tree’ is spoken in four tone groups. The choice of reading the sentence into four tone groups represents selections from the system of tonality, and the selection of tonic prominences (shown here by bolded text) reflects choices from the system of tonicity. The choice of tone groups themselves – tones 3, 1, 1, and 1 – represents choices from the system of tone. System networks have also been developed for features at the level of segmental and word phonology, such as Young’s (1992) work on English consonant clusters and Tench’s (2014) work on English word phonology. The concept of system, however, is understood somewhat differently when it is applied at the levels of word, syllable, and phonemes. In discussing his systems of consonant clusters, Young explains that while the concept of system network implies sets of choices of meanings, ‘by the time we get . . . [to] segmental phonology all the meaning choices have been made long ago and everything now to be selected is predetermined’ (Young 1992:58). Tench reiterates this view: ‘System at the level of word (and also at the level of groups/phrases) is rather the specifications of what the speakers of a language recognize as having been established in, or “chosen” by, the language to represent its words’ (Tench 2014:274). Tench has developed system networks outlining the possible syllable structures, peaks, margins, strong and weak vowels, and syllable initial consonants in English (standard southern English pronunciation) (see Tench 2014). An interesting point made by Young is that even though system networks (for consonant clusters) display predetermined sets of choices, ‘they probably have a more positive role to play in decoding, and they certainly need to be built into a model of English which accounts for the ability of speakers to add to their vocabulary (for example, by means of foreign loanwords) words that conform to the phonology of English’ (Young 1992:58). While system networks represent choices, Fawcett (2014) argues that the typical SFL system networks for intonation are largely descriptive frameworks useful for analyzing language instances, or output (2014:325). Fawcett (2014) thus proposes a generative model of intonation and punctuation. He sets out several concepts and realization rules related to ‘intonation components’ and ‘punctuation components’ needed for a comprehensive model of grammar that can be used for generating English text. We do not take up this argument here, but note that Fawcett (2014) provides a careful and detailed proposal for ‘construct[ing] a generative systemic functional grammar of intonation and punctuation for English’ (Fawcett 2014:396).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

180

WENDY L. BOWCHER AND MEENA DEBASHISH

An instance of language represents choices that have been made from various systems (amongst those from the other systems of language). Between system and instance are registers or text types. Different registers display different patterns of choices within the three systems of tonality, tonicity, and tone. These patterns of choices identify an instance as belonging to a specific register and play a role in forming the linguistic basis for comparing and contrasting different registers and text instances. While all options in the linguistic system are available to a speaker on any one occasion, there are constraints on what is or is not ‘at risk’ of being selected when one speaks, and these at risk choices define the various registers. There are many studies scattered throughout the literature indicating the role of intonation in identifying different registers – some of these from an SFL perspective and others from within other theoretical paradigms (e.g. Bowcher 1998; Crystal and Davy 1969; Johns-Lewis 1986; Kuiper 1996; Smith 2008; Tench 1988, 1996, 2014). Along with choices from within the intonation systems, a spoken register is characterized by other systematic and identifiable choices in sound, rhythm, and sound quality. Tench uses the term ‘prosodic composition’ to refer to ‘the choices, preferences, proportions, and omissions of specific features of prosodic substance, including voice quality, pace of utterance, rhythmicality, and loudness, that play an essential role in the distinctive “sound” of a particular genre’ (Tench 2014:273). Speakers of a language can identify a range of registers by their individual prosodic composition and do so quite accurately. That is, they know when they hear a sports commentary, a news report, or an argument whether or not they hear the details of the wording of such spoken registers. Van Leeuwen’s (1992) work on the rhythmic patterns found in different types of radio broadcast registers adds an important dimension to understanding the concept of prosodic composition in relation to registers of spoken language. An interesting argument proposed by van Leeuwen (1992:250) is that specific patterns of accent and juncture are ‘motivated, not by the linguistic system of English, but by the . . . norms and values of [the] social institutions’ in which the speech occurs. There is much scope for developing descriptions of various spoken registers and also for describing the configurations of linguistics features such as grammatical choices, patterns in cohesive ties, and lexical choices alongside phonological patterns. So far, only a few studies focus on the relationship between intonation choices and other linguistic features. These include Lukin’s (2014) multidimensional study of a televised news report; Bowcher’s (2003, 2004) work on radio sports commentary; Bowcher and Zhu’s (2014) study of native and non-native English speakers reading aloud a children’s story; and, of particular note, Smith’s (2008) research into the configuration of a range of features in several register varieties, including casual conversation, talk during a surgical operation, interviews from a television programme, and telephone sales. Other research into intonation within the domain of register studies includes Caldwell’s (2014) analysis of rap and sung performances and Banks’ (2014) analysis of the pronunciation of the past verb form (-ed) in Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

181

classical choral music, such as Handel’s Messiah. These studies broaden the scope of the study of intonation beyond spoken language registers. Thus, while some inroads into identifying the phonological character of different registers have been made, further research using the SFL model of intonation to compare and contrast choices across the various intonation systems in different situations would help to establish a more comprehensive description of different registers in language by adding a phonological dimension to register variation and identification.

7.3.4 The Paradigmatic Axis and Syntagmatic Structures The paradigmatic axis is concerned with choice: What are the options available or what are the potential choices that can be made within a given system? As already noted, the system of intonation has three sub-systems: tonality, tonicity, and tone. In the process of speaking, selections are made from these systems. Figure 7.4 in Section 7.3.3 shows that these systems are simultaneous systems and hence interdependent, in that choices in one affect each of the other systems (simultaneity of choice is indicated by the use of curly brackets in a system network; either–or choices by square bracketing). For instance, the system of tonality has to do with the range of choices available for chunking information in spoken language, and within each chunk is a tonic syllable (tonicity). Each tone unit is characterized by a specific melodic shape, or tone (tone). The choice of tone type is activated by the contextual and semantic features of the situation. Selections across these systems are made as a discourse unfolds. Further, choices in each of these systems play a role in realizing choices in the system of information at the stratum of lexicogrammar and at the semantic level (for the latter see Section 7.3.5). For instance, the element assigned as tonic (tonicity) realizes the choices in the system of information focus in the information system. The distribution of tone units across a discourse (tonality) realizes choices in the system of information unit and specifically information distribution. And a tone unit (tone) realizes an information unit, although the boundaries of each are not necessarily oneto-one as illustrated in Section 7.3.1. As for the system of tone, according to Halliday, English has five simple tones, i.e. tone units with a single tonic, which are referred to by using the numerals 1 to 5, and two compound tones, i.e. tone units with a major and minor tonic syllable, which are referred to by using the numerals 13 (onethree) and 53 (five-three). Each single tone is realized with a single pitch movement: fall, rise, level rise, falling-rising, or rising-falling, and the compound tones as a major and a minor pitch movement: a fall followed by a slight rise or a rise-fall followed by a slight rise. The pitch movements in the tonic segment(s) define the pitch contour of the tone unit, and are called the primary tones. Thus, the paradigmatic choices available for the utterance But I thought everyone was required to contribute something can be ascertained by asking: How many tonic prominences does the context activate? [Tonicity/ Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

182

WENDY L. BOWCHER AND MEENA DEBASHISH

Tonality]; which syllables are assigned tonic prominence? [Tonicity]; and which tone groups are activated by the context? [Tone] In answer to the first two questions, we can posit that every syllable could potentially be assigned tonic prominence, but the syllables that are actually assigned prominence depend on the contextual environment and certain English pronunciation conventions to do with word accent; for instance, it is highly unlikely that bute in contribute would be given tonic prominence unless, perhaps, someone was trying to point out a pronunciation error of someone else. So, in a context where someone previously thought that everyone was required to contribute something to raising funds for an event but then finds out that only certain people (including themselves) are required to do so, the utterance would likely be spoken in the following way: //5 ᴧ but /I thought /everyone was re/quired to con/tribute /something// The focus in this context must necessarily be on ‘everyone’ (the tonic syllable being ‘ev’ in ‘everyone’), and the element of surprise (or perhaps irritation) is expressed through the use of Tone 5 as shown. On the paradigmatic axis is the relationship of ‘delicacy’. Delicacy is the principle of moving from general to more specific and applies to any system network where there is more than one subsystem. The degree of delicacy is represented by how many subsystems extend to the right of a system network. In the system network of foot composition (see Figure 7.3), for instance, there is little distance between the entry point and the final choice of [simple]. That is, there are few degrees of delicacy involved. In other systems there may be many more degrees of delicacy, such as in the system of tonic composition (see Figure 7.5). In the system of intonation, delicacy relates largely to the choice of tone group and the kinds of meaningful distinctions that are made through height of melodic shape and length of utterance over which the tone group extends, as well as other features such as nasalization or affectations in pitch (i.e. those not related to tone group but to overall voice quality). Within SFL theory, the tone group features that are systematized are the shape and the range in the height/depth of the pitch contour in the tonic and the pretonic segments of the tone group. Whereas the paradigmatic axis is a vertical relation, the syntagmatic axis is a horizontal relation and has to do with the sequencing of structures derived from the paradigmatic choices. Take, for instance, the system of foot composition. This system is located at the rank of foot where there is a salient syllable which functions as ‘Ictus’, and weak syllable(s) (if there is/are any) as ‘Remiss’ (Halliday 1967:12). A simple foot is realized by an Ictus syllable. In a compound foot, the Ictus (whether silent or spoken) is followed by Remiss. Thus, the structure is: Ictus (^ Remiss)

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

183

The terms Ictus and Remiss are functional terms, with the Ictus syllables playing a role in the rhythm of the discourse; these syllables tend to occur at fairly equal intervals of time in continuous spoken English. This affects the syllables which may occur within the feet from one Ictus syllable to another; that is, in order to maintain tempo, the Remiss or non-salient syllables (when present) tend to be ‘squashed’ through contraction or weakening of vowels, particularly when there are multiple syllables. With regard to the tone unit, paradigmatically there are several choices of single and compound tones. The tone groups, as carrying the melodic shape of the language, may be characterized by primary and ‘accompaniment’ pitch movements (Halliday and Greaves 2008). In this sense, the ‘functioning elements’ in the tone group are the Tonic and the Pretonic, ‘which supply the framework within which the speaker’s variations in pitch and loudness are perceived and interpreted by the listener’ (Halliday and Greaves 2008:42). The major pitch movement of the tone unit is initiated on an Ictus syllable. This constitutes the tonic syllable, and the foot in which the tonic syllable is located is called the tonic foot. The tonic foot may be followed by one or more feet, which continue the pitch movement initiated in the tonic syllable. For a Pretonic segment to be present, there must be at least one complete foot prior to the tonic foot and not connected with the previous tone group. The Pretonic has its own set of pitch contour patterns, but the pitch movement in the Tonic is defining as far as tone group choice goes. In other words, the pitch movements in the Pretonic are determined by those in the Tonic, and each Tonic pitch movement has its own set of Pretonic pitch movements. Thus, the tone unit structure is described as having an obligatory Tonic which is optionally preceded by a Pretonic. In terms of syntagmatic structure, when a Pretonic is present, the tone unit is realized as Pretonic ^ Tonic. Figure 7.5 displays the tone group choices and the choices of [with pretonic] or [without pretonic]. Figure 7.5 indicates that a tone unit is realized (&) with a tonic element. The choices in the system of tonic composition are simple or compound. In the system of pretonic are the choices [with pretonic] or [without pretonic]. Moving to the right of the figure we find more delicate choices available for the tones of English. Here we can see the various secondary tones that are possible – the indirect secondary tones are choices deriving from the [with pretonic] system, and the direct secondary tones deriving from the simple and compound tone systems. The system network also indicates that there are choices of both indirect and direct secondary tones for certain tone groups. For example, a narrow Tone 1 [1-] may be spoken on an even pretonic [.1]. These variations in secondary tones are described in Section 7.4.2 of this chapter. We can illustrate the syntagmatic structure of tone units and the paradigmatic choices at a primary level of delicacy using a clause complex from Halliday (1970:120): ‘Not always was the kangaroo as now we do behold him but a different animal with four short legs.’ This is analyzed in the following way:

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

TONIC COMPOSITION

+Tonic 2; Tonic 2: foot1-n Tonic^Tonic 2

compound

simple

SIMPLE PRIMARY TONE

COMPOUND PRIMARY TONE

+ Pretonic: Pretonic: foot1-n Pretonic ^ Tonic

with pretonic

without pretonic

5. high 5- low

tone 5 (fall-) rise-fall

tone 53 (fall-) rise-fall plus low rise

tone 13 fall plus low rise

4. high 4- low

2- broken

2. straight

1- narrow

1+ wide 1. medium

tone 4 (rise-) fall-rise

tone 3 low rise

tone 2 (high rise/high fall-high rise)

tone 1 fall

[tone 3]

[tone 2]

[tone 1]

.3 mid -3 low

.2 high -2 low

.1 even -1 bouncing …1 listing

Figure 7.5 Choices in the TONE U NIT system showing the more delicate choices available in the pretonic and system of TONE [note: in this diagram Tonic 2 refers to a Minor Tonic] (from Halliday and Matthiessen 2014:18)

+Tonic; Tonic: foot1-n

tone group

pretonic

Intonation

185

Table 7.2 Distribution of tone, tonic, and pretonic elements in the excerpt. (NB: MT = minor tonic) Tone choice 13

1 1

Pretonic // ᴧ not

//four / short

Tonic

No. of feet in the tone group

/always was the /kangaroo as /now we do be /hold him but a // MT

5

//different /animal with // /legs //

2 3

// 13 ᴧ not /always was the /Kangaroo as /now we do be /hold him but a //1 different /animal with //1 four /short /legs // The clause complex is spoken on three tone units. The third tone group is realized with both a pretonic (/four /short) and tonic (/ legs//), therefore the choice is ‘+pretonic’. The first and the second tone units are both realized with only tonic segments, i.e. ‘-pretonic’, but with differences between the two. In the first tone unit, the tonic segment is preceded by an incomplete foot (// ˄ not /) which does not constitute a pretonic segment. In the first tone group there are five feet, in the second there are two, and in the third there are three. Each foot begins with an Ictus syllable, or salient syllable, but in the first foot in the first tone unit there is a silent Ictus indicated by the caret symbol. Generally, monosyllabic lexical words and the accented syllables of polysyllabic words tend to take salience in a tone unit while the monosyllabic grammatical or function words tend to be non-salient. We can see that the example reflects this pattern of salience. The rhythm (or beat) is carried by these salient or Ictus syllables. The sequence of Tone Unit Structures in this excerpt is [Tonic^Minor Tonic] ^ [Tonic] ^ [Pretonic ^ Tonic], and at the grammatical level this would mean that there are three quanta of information and thus three information foci, with the first quantum of information comprising a major and minor information focus. The distribution of tone, tonic, and pretonic segments for the excerpt is displayed in Table 7.2. Each tone unit, whether it be of one or more feet, has a distinct pitch contour and contains a point of prominence known as the tonic syllable. Tonic prominence is always assigned to a salient syllable in a foot within the tone group in response to the relevant activating feature(s) of the context of situation and the co-text. This assignment of tonic prominence realizes the functional element of New in an information unit, which is a culmination point or pulse. This relates to the textual function of the information unit; its role in organizing discourse in terms of the status of information – information is presented as ‘Given’ or it is presented as ‘New’. Ascertaining where the New element begins is somewhat indeterminate because New is realized by sound not written words, and New is a culmination point of a range of prosodic features. Furthermore, because the

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

186

WENDY L. BOWCHER AND MEENA DEBASHISH

boundary of the New is indeterminate, ascertaining what might be considered Given – at least in terms of information prior to the culminating point of the New – can also be indeterminate. The definitions of Given and New are typically presented as ‘information that is presented by the speaker as recoverable (Given) or not recoverable (New) to the listener’ (Halliday and Matthiessen 2014:118). However, there is more to these functional elements of the information unit. The determination of what falls within the New and what within the Given involves both phonological and semantic considerations. Drawing on ideas from Prince (1981), O’Grady (2014) discusses the issue of ‘what counts as New’ and suggests the possibility of two types of ‘New’: New1 is where the tonic is assigned to items that are freshly introduced in discourse, and New2 is where the tonic is assigned to items to fulfil a speaker’s ‘communicative goals’ and ‘to direct’ the hearers’ attention to these goals ‘irrespective of whether the particular lexical items are very much in the air’ (O’Grady 2014:49). Although Given typically precedes New, there are times when Given may follow the New. Thus, in terms of syntagmatic structure, the structure of Given and New in the information unit is (Given ^) New (^ Given), where parentheses indicate optionality.

7.3.5 The Metafunctional Organization In SFL ‘function’ is considered to be ‘a fundamental property of language’ (Halliday 1985b:17). That is, the functions of language are not just about the ‘uses’ to which language is put in everyday life, but are ‘the very foundation to the organization of language itself’ (Halliday 1985b:17), and it is for this reason that the functions are labelled ‘metafunctions’ (Halliday and Matthiessen 2014:31). Adult language is modelled as multifunctional – as construing human experience and enacting social relations. These are the ideational and interpersonal functions respectively. Along with these is the textual function which is the ‘enabling’ function of language. This function ‘relates to the construction of text’ in that ‘construing experience and enacting interpersonal relations . . . depend on being able to build up sequences of discourse, organizing the discursive flow, and creating cohesion and continuity as it moves along’ (Halliday and Matthiessen 2014:30–1). The functions of language are the interface for the sociosemiotic domains of experience, and together construe the features of context of situation: Field (the nature of the activity), Tenor (the nature of the participants involved in the activity), and Mode (the nature of the channel of communication). The systems of tone, tonality, and tonicity play a role in the construal of certain kinds of meanings related with different functions of language. For instance, with regard to the interpersonal function, tone choice can accord with or alter the speech function typically associated with certain grammatical structures, such as when a rising tone (Tone 2) is

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

187

Table 7.3 Choices in the system of key and their typical meanings (see Halliday and Matthiessen 2014:169) Grammatical Mood Declarative

WH-interrogative

Yes/no interrogative Imperative clause

Tone Choice Tone 1 unmarked Tone 2 Tone 3 Tone 4 Tone 5 Tone 1 unmarked Tone 2 Tone 2 (tonic on WH-element) Tone 2 unmarked Tone 1 Tone 1 unmarked in positive Tone 3 unmarked in negative Tone 13 marked polarity (tonic is on do/don’t) Tone 4

Typical Meaning certainty/neutral sense of protest or indignation tentativeness sense of reservation insistence neutral tentativeness echo question neutral peremptory question command invitation pleading request plea

used on a declarative clause; instead of giving information, this choice serves to enact the speech function of demanding information. The choices of tone when interpreted in relation to Mood choice may be considered marked or unmarked, and together these represent a system known as key. These are displayed in Table 7.3. The choice of tone, and the more delicate choices within the system of tone, such as degrees in the height of the fall or rise in pitch, play a role in construing other kinds of interpersonal meanings such as intensity of feeling, degrees of involvement, or sense of commitment to what is being said (see Halliday 1970; Halliday and Greaves 2008). With regard to the ideational function, and specifically the logical function, choices in the system of tone play a role in construing the ‘connections’ between bits of information. For example, a relationship of coordination is construed through the sequence of Tone 3 (low rise) followed by Tone 1 (fall), and of subordination by Tone 4 (fall-rise rounded) followed by Tone 1 (fall). When these tone sequence choices parallel the grammatically construed choices, such as where a relation of coordination is grammaticalized by the use of the conjunction and, and that of subordination by the subordinator whereas, the phonological realization is said to be unmarked. However, the two systems do not always parallel each other. A Tone 3 might be used on a grammatically coded subordinate clause, and a Tone 4 on a coordinating clause: such choices are considered marked (see Halliday and Greaves 2008:129–35; Halliday and Matthiessen 2014:553–4). The system of tonality also comes into play in relation to the logical function. For example, a projecting clause complex may not be intonationally construed through the use of two tone groups but through presenting

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

188

WENDY L. BOWCHER AND MEENA DEBASHISH

Figure 7.6 Unmarked Theme/Rheme and Given/New functional elements in a clause

the clause complex as one quantum of information through the use of one tone group only, as in: //1 ᴧ he /said there was /nobody /left// O’Grady (2017:155) suggests that although this choice is considered ‘the unmarked realisation’, he is careful to add that in fact there has been ‘no extensive corpus investigation of this claim’ and thus the claim remains ‘unverified’ (O’Grady 2017:155). This suggests an interesting area for future investigation. In contrast with these kinds of clause complexes, as noted already, one clause might be spoken on several tone groups, as in the example of reading aloud in Section 7.3.3, thus organizing the talk into several quanta of information. Section 7.3.4 included a brief discussion on the information unit in relation to the textual metafunction; its role in organizing discourse in terms of the status of information. There is a complementary relation between the information unit and other textual systems such as cohesion and the clause structures of Theme and Rheme. For instance, often, and indeed in regards to the unmarked case, New falls within the Rheme of the clause and Given within the Theme. Each of these functional elements construes a culmination of information, with Theme being considered speaker-oriented (it is the point of departure of the message), and New as listener-oriented (it is what the speaker wants the listener to attend to in the situation). The textual metafunction, however, enables a variety of ways in which to package experience. For instance, Matthiessen (1995:607) observes that: The assignment of New illustrates how the textual metafunction may work independently of the hierarchic organisation generated by the experiential one. In particular, the element New of the information unit is not restricted to focus (1) on an element of structure selected from within a single clause; nor (2) on an element of clause structure.

Figure 7.6 shows the unmarked mapping of Theme/Rheme and Given/ New functions in a clause, and Figure 7.7, which uses the sentence from the children’s story ‘How Love and Peace Came to the Woods’ presented in

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

189

//1 Mr. / Fox//1 was /strolling /through the / woods// Mr Fox was strolling through the woods Theme Rheme New New

// 3 when he / noticed// 1 when he noticed Theme Rheme New Given

a /plump / hen// //1 sitting //1 on the /branch of a / tree// a plump hen sitting on the branch of a tree New

New

New

Figure 7.7 Analysis of marked information unit distribution

Section 7.3.3 shows several instances of marked New. Bowcher (2003, 2004) describes the relationship between Theme and New choices in excerpts from radio sports commentaries, and specifically in play-by-play talk of a Rugby League game. She finds that in this register there is a complementary relation between information carried by the Theme and that carried by the New, with informational peaks falling predominantly on players either as Participants or as Circumstances (destinations of the ball), and that marked informational prominence is not common in this type of talk. Zhu (2014) focuses on the relationship between Theme and Information choices in a BBC news reading. Lukin (2014) also analyzes the relationship between New and Theme in her study of a news report of the ‘Coalition’s’ war with Iraq in 2003, finding that ‘in the choice of “person” and “place”, the system of IF [information focus] is reinforcing patterns established via the system of theme’ (Lukin 2014:65). Choices from the systems of tonality and tonicity play a role in the degree to which the boundaries of information units coincide with those of clauses and what is assigned focus by the speaker. Various researchers highlight registerial patterns in this regard. For instance, Lukin (2014:63) notes that there is a ‘higher ratio of tone units to the grammatical unit of clause’ in broadcast news (see van Leeuwen 1992; Smith 2008). However, such a claim is not particularly revealing, considering that other registers also exhibit a high number of tone units per clause, such as reading aloud children’s stories (e.g. Bowcher and Zhu 2014; Halliday 1970), and indeed, Lukin effectively acknowledges the too-general nature of her claim when she asks, ‘Is there a metafunctionally significant pattern in what is selected for focus by the location of intonational focus?’ This kind of question is of more value in identifying registerial differences in tonality, tonicity, and information distribution, and Lukin’s analysis indicates that in her news broadcast data choices construe largely textual and interpersonal meanings. Other register-focused findings include Smith’s (2008) work which includes a description of the way that intonation choices (amongst others) play a role in shifting the focus of talk taking place in a surgery from experiential to interpersonal meanings, and Bowcher and Zhu’s (2014:17)

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

190

WENDY L. BOWCHER AND MEENA DEBASHISH

study points out that even though a children’s story ‘has a low lexical density and is grammatically simple, when it is read aloud, it is assigned a high “informational” density through the resources of the intonation system’. Of relevance to intonation choices in relation to the textual function is the higher unit of paraphones, or phonological ‘paragraphs’. Tench (1996:24) argues that ‘phonological paragraphs’ typically begin with a ‘high pitch on the onset syllable of the initial intonation unit’ and that the pitch gradually falls until the final tone unit, wherein ‘the depth of fall in the final unit is the lowest in the whole paragraph’. He also observes that the tempo tends to slow down in the final intonation unit and that ‘there is a longer pause than is normally allowed between intonation units’ between phonological paragraphs (Tench 1996:24; see also Tench 2014:272–3). Iwamoto (2014) picks up the idea of phonological paragraphs, calling them ‘paraphones’ after Halliday (1961:253 note 30) and proposes that ‘paraphoning’ is a textual process across all strata of the language system and is semogenic in nature. He argues that a paraphone is a semantic unit, whose boundaries are realized by specific phonetic cues, such as pitch levels, and he hypothesizes that paraphoning differs across registers in that ‘speaker[s] select one way [of paraphoning] over others according to the context of situation to create distinctions in meaning’ (Iwamoto 2014:143). Iwamoto’s work leaves open an enticing area for further research into register variation. The textual metafunction plays a key role in the construal of the contextual parameter of mode. We noted in Section 2 Tench’s (1996:151) comment on the importance of intonation in reading and writing, and several SFL studies have focused on this very point. These include Davies’ (1989, 1992, 1994a, 1994b, 2014) extensive work on the relation between cohesion, information structure in written and spoken text, punctuation, and intonation systems for effective understanding and reading aloud of written text. There is also Cummings’ (2000, 2001, 2014) research into the interpretation of the intonation patterns relevant to written text, and Bowcher and Zhu’s (2014) study of native and non-native English speakers reading a children’s story. The relationship between spoken and written language features is also of critical importance in Fawcett’s (2014) generative model of English intonation and punctuation. The next section of this chapter describes in more detail the tones in English as modelled in Systemic Functional Linguistics. Our aim is to highlight the basic shapes of the tones of English, as it is the tone group that is the core around which other choices in the system of intonation and the system of information operate.

7.4

The Tones of English

While this section sets out a description of the tones of English, some experience in listening to the different tones should be gained prior to conducting one’s own analysis. A good place to gain experience in hearing the tones is Halliday and Greaves (2008). Additionally, Greaves (2014) is an interactive Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

191

chapter on the SFL system of intonation and includes practice in rhythm and hearing tones, and in analyzing spoken utterances using Praat.6 According to the SFL model, there are seven primary tones in English, within which five are simple tones and two are compound tones. We describe each of the primary tones in turn before turning to the secondary tones.

7.4.1 The Primary Tones 7.4.1.1 Tone 1 Tone 1 is realized with a falling pitch contour in the tonic syllable. This tone begins at a mid or mid high pitch level and continues downward till the end of the tone group.

Figure 7.8 Tone 1 – the tonic and pretonic pitch contours

If there is a pretonic segment, the corresponding pitch contour in the pretonic is level at mid or mid high pitch, as is exemplified in the following examples. I’ve finished my work. Why didn’t you finish your work? Finish your work!

// 1 ᴧ I’ve /finished my /work // // 1 why didn’t you /finish your /work // // 1 finish your /work//

7.4.1.2 Tone 2 Tone 2 is a sharp rising pitch contour from low or mid low, and it covers a wide pitch range. The pretonic pitch contour is either high level (as in Figure 7.9b), or steps down from high or mid high to the point/pitch level from where the tonic pitch movement begins (as in Figure 7.9a).

Figure 7.9a Tone 2 – the sharp rising tonic with a step down pretonic

Figure 7.9b Tone 2 – the sharp rising tonic with a high level pretonic

6

See www.equinoxpub.com/systemic-phonology-files for the accompanying sound files to Greaves (2014).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

192

WENDY L. BOWCHER AND MEENA DEBASHISH

Will you please finish your work? Have you finished your work?

// 2 ᴧ will you / please /finish your / work //(step down pretonic) // 2 ᴧ have you / finished your / work //(high level pretonic)

7.4.1.3 Tone 3 Tone 3 is a level rise from about low or mid low to mid pitch. If there is a pretonic segment, it has a level contour.

Figure 7.10 Tone 3 – the tonic and pretonic pitch contours

perhaps. It (the exam) won’t be so hard.

7.4.1.4

// 3 ᴧ per/haps // (no pretonic) // 3 ᴧ it / won’t be so / hard //

Tone 4

Tone 4 is a fall-rise pitch contour with more force on the falling movement. This tone exhibits a rise-fall hook onset before the main falling-rising pitch movement; the pitch first rises from mid to about mid high before executing the fall-rise movement, which is a key feature of the tone. The fall covers a wide pitch range, and the rise is almost to the same level as the beginning of the fall.

Figure 7.11 Tone 4 – the tonic and pretonic pitch contours

The pretonic contour steps down from high pitch to around mid pitch level, i.e. to the level from where the rise-fall hook onset of Tone 4 begins. He finished his work, but . . . If you don’t finish your work in time

// 4 ᴧ he / finished his / work but // . . . // // 4 ᴧ if you / don’t / finish your / work in / time //. . .

7.4.1.5 Tone 5 Tone 5 exhibits a pitch movement which is in the opposite direction to Tone 4. This is a rise-fall pitch contour with more force on the rising movement. This tone also exhibits a hook onset which has a fall-rise movement. The pitch in the tonic syllable first falls from mid-high to mid and then rises to cover a wide range before executing the fall.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

193

Figure 7.12 Tone 5 – the tonic and pretonic pitch contours

The pretonic steps up from about mid level to mid high pitch level, i.e. almost to the level from where the fall-rise hook onset for Tone 5 begins. I’d never seen such a hullabaloo

// 5 ᴧ I’d /never /seen such a /hullaba/loo//

7.4.1.6 Compound Tones: Tone 13 and Tone 53 In addition to the five simple tones there are two compound tone units, i.e. tone units with two tonic syllables – one major and the other minor. These are considered fusions of two tones rather than two successive tones in that the pretonic, if present, is located before the first tone only. That is, there is no intervening pretonic in the Tone 1!3 or the Tone 5!3 sequence: the pretonic would be before the Tone 1 or Tone 5 only. There is some debate as to whether compound tones should rather be considered as sequences of two tone groups (see O’Grady 2017:152 for discussion of this issue; also see Tench 1990). Of particular interest in this discussion is O’Grady’s (2017:152) comment that ‘the criterion of information structure posits that a tone group equates to an information unit’ and that ‘this is compromised by positing two foci, albeit of different status, within a compound tone group’. However, the issue of what defines a compound tone group is akin to that of compound nouns, viz. where to draw the line as to what constitutes two words or a compound word (e.g. breast feeding, breastfeeding; ready-made, readymade). In such cases, there are two pieces of information drawn together into the compound word. An information unit realized by a compound tone group is essentially doing the same thing. For instance, Halliday (1970:88) explains that the minor element (the rise) in a Tone 13 is typically either (1) an adjunct or dependent clause, or sometimes a co-ordinate clause (i.e. subsidiary new information); (2) a ‘partial utterance’, such as a vocative, speaker’s comment or other ancillary matter; (3) the displaced subject in a substitution clause; or (4) some other element which is not new (e.g. repeating part of a question) but to which the speaker wants to give some prominence.

Thus, it seems reasonable to have an information unit realized by a compound tone group with major and minor tonics. Furthermore, the intonation system and the grammatical system work side by side. There is an interplay among the choices made, with each pointing to the different functions associated with spoken and written language. There is thus an apparent need for more empirical investigations into the issues

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

194

WENDY L. BOWCHER AND MEENA DEBASHISH

surrounding compound tone groups, which is beyond the scope of this chapter. Our description of the compound tones here follows that of Halliday (1970) and Halliday and Greaves (2008).

7.4.1.7 Tone 13 (Tone One Three) Tone 13 has a falling contour from either mid or mid high followed by a low rising contour of Tone 3 to about mid low level.

Figure 7.13 Tone 13 – the tonic and pretonic pitch contours

The pretonic is level at mid or mid-high, and it is defined by the Tone 1 part of this compound tone. //13 Give my re/gards to your /parents /next time you /see them// (example from Halliday 1970:88)

7.4.1.8 Tone 53 (Tone Five Three) This tone exhibits a combination of the rise-fall contour of Tone 5 followed by a low rising contour of Tone 3.

Figure 7.14 Tone 53 – the tonic and pretonic pitch contours

The pretonic pitch steps up from about mid to mid high. //53 ᴧ I’d /rather /like one if you /feel you can /spare it// (example from Halliday 1970:93)

7.4.2 The Secondary Tones As explained in the previous section, the primary tone system indicates a choice among seven tones: five simple and two compound tones. Section 7.4.1 also described the neutral pitch contour for each primary tone, in the tonic segment, and the typical contour in the pretonic segment. However, there are possible variations in the pitch contours of both the tonic segment and the pretonic segment for each tone. While Table 7.3 outlined the typical meanings of the primary tones associated with grammatical mood (the system of key), the variations that are described in this section

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

195

contribute to more subtle meanings within each general meaning of the tones and realize a system of secondary tones, with a distinction between direct secondary tones – those where specific variations occur in the tonic segment – and indirect secondary tones – those in the pretonic segment which follow a specific melodic shape. These more delicate options were mentioned in Section 7.3.4 with the system of primary and secondary tones shown in Figure 7.5. In this section, we describe the possible secondary tones for each primary tone.

7.4.2.1 Direct Secondary Tones of Tone 1 Tone 1 has three systemic pitch choices in the tonic segment, or three direct secondary tones identified conventionally with a period, a plus sign, and a minus sign: [1.], [1+], and [1-]. Tone [1.] is a fall from about mid or mid high level. Tone [1+] is a fall from high to low, and Tone [1-] is a fall from mid low to low.

Figure 7.15 Direct secondary tones of Tone 1

// 1. why don’t you / clean your / room// // 1+ why don’t you / clean your / room// // 1- why don’t you / clean your / room//

7.4.2.2 Indirect Secondary Tones of Tone 1 There are three distinct pitch contours possible in the pretonic segment of Tone 1: [.1] ‘even’, [-1] ‘bouncing’ or ‘uneven’, and [. . .1] ‘listing’. And these express indirect secondary tones of Tone 1 (the fall is only shown once in Figure 7.16).

Figure 7.16 Indirect secondary tones of Tone 1

// .1. why don’t you / get up and / clean your / room// // -1. why don’t you / get up and / clean your / room// // . . .1. one / two / three / four // The neutral type [.1] is also referred to as the ‘even’ type, as the pitch contour is more or less level/even: at mid pitch level for [1.], steps up from mid low to high for tone [1+], and steps down from mid high to mid low for [1-].

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

196

WENDY L. BOWCHER AND MEENA DEBASHISH

Figure 7.17: Indirect (neutral) secondary tone [.1] ([.1.], [.1+], and [.1-] respectively)

The [-1] ‘uneven’ pretonic has a dipping or bouncing contour from around mid to a fairly high pitch in each foot. The [. . .1] pretonic has a level rising contour in each foot.

7.4.2.3 Tone 1: Tone [.1.] vs. Tone [-1.] (Attitudinal) The use of [-1.] instead of [.1.] brings in a sharp attitudinal ‘intensive’ or argumentative meaning, as in // .1. why don’t you / get up and / clean your / room// // -1. why don’t you / get up and / clean your / room// In the second, the swinging movement in each foot adds to the intensity of the meaning of the sentence. This pretonic occurs more naturally with a ‘strong’ [1+] tonic, making it more forceful.

7.4.2.4 Tone 1: Tone [.1.] vs. Tone [. . .1.] (Semantic) The level rising contour in each foot of [. . .1] is used to enumerate the items in a list occurring before a final tonic syllable, or focus of information, as in // . . .1. one / two / three / four // where each pretonic foot ends with a slight rise. The contrast in meaning between a [.1] and a [. . .1] pretonic can be seen in examples such as // .1 red / white and / blue / jackets // (‘jackets with a mixture of red, white and blue colours’) // . . .1 red / white and / blue / jackets // (‘red jackets, white jackets and blue jackets’)

7.4.2.5 Direct Secondary Tones of Tone 2 The direct secondary system of Tone 2 gives two choices: [2.] a sharp rise from low pitch or [2] a sharp fall rise, falling from high to cover a wide pitch range and then rising.

Figure 7.18 Direct secondary tones of Tone 2

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

197

// 2. ᴧ did you /finish your as /signment // // 2 ᴧ did you /finish your as /signment //

7.4.2.6 Indirect Secondary Tones of Tone 2 The indirect secondary system also has two choices: either a high level [.2] or a low level [-2] pitch, as shown in Figure 7.19.

Figure 7.19 Indirect secondary tones of Tone 2

These indirect secondary tones [.2] and [-2] can combine with either of the direct secondary tones, [2.] or [2], but the combination of [-2] with [2] appears to be rare.

7.4.2.7 Tone .2. vs. Tone -2. (Attitudinal) // .2. ᴧ did you /finish your as /signment? //: ‘neutral’ // -2. ᴧ did you /finish your as /signment? //: ‘surprised’ The high level pretonic with a jump down in pitch to a sharp rise in the tonic [.2.] is the most unmarked way of realizing the yes-no interrogative. When a low level pretonic combines with a neutral tonic [-2.], the yes-no interrogative acquires an additional meaning of being more ‘involved’.

7.4.2.8 Tone .2. vs. Tone .2 (Semantic) // .2. ᴧ did you /finish your as /signment? //: ‘neutral’ // .2 ᴧ did you /finish your as /signment? //: ‘specific focus on assignment’ In Tone [2.], there is a straightforward rise in the tonic syllable which is often on the last lexical word, indicating that the query is with reference to the entire information unit. This is the neutral tone for a Yes-No interrogative. On the other hand, the marked fall-rise variant, Tone [2], draws the attention of the listener to a specific point of query in the information unit.

7.4.2.9 Indirect Secondary Tones of Tone 3 Tone 3 has variations only in the pretonic segment, i.e. the indirect type. There are two secondary tones: one is level at mid pitch, [.3], and the other is level at low pitch, [-3]. So, the low level rising tone in the tonic segment can occur with a level pretonic contour either at mid or low pitch.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

198

WENDY L. BOWCHER AND MEENA DEBASHISH

Figure 7.20 Indirect secondary tones of Tone 3 ([.3] and [-3] respectively)

7.4.2.10 Tone .3 and Tone -3 (Attitudinal) While the mid level pretonic [.3] conveys the neutral meaning of a level rising tone, in this case, ‘reassuring’, the low pitched pretonic [-3] gives a marked meaning of being ‘noncommittal’ or ‘indifferent’. // .3 ᴧ it (the exam) / won’t be so /hard // ‘reassuring’ // -3 ᴧ it (the exam) / won’t be so /hard // ‘unconcerned’

7.4.2.11 Direct and Indirect Secondary Tones of Tone 4 There are two direct secondary tones for Tone 4: [4.] and [4]; the indirect secondary tones are determined by the pitch contour of the direct secondary tones.

Figure 7.21 Direct and indirect secondary tones of Tone 4 ([4.] and [4] respectively)

//4. ᴧ it’s a / bit /dangerous// (‘I can’t help being worried; . . .’) (from Halliday 1970:110) Tone [4.] is the neutral one with a fall-rise pitch contour from mid high, and it covers a wide pitch range. The pretonic for this tone is a step down contour from high to about mid pitch level. Tone [4] is the marked variant with the fall-rise pitched lower. The preceding pretonic contour exhibits a fall-rise pitch movement in each foot, which seems to be imitating the fall-rise movement of the tonic segment. This series of low pitched fall-rise movements in the pretonic adds to the intensity of the tone. //4 not unless he’s /willing to a/pologise // (‘I might see him if he does’) (Halliday 1970:111)

7.4.2.12 Tone 4. and Tone 4 (Attitudinal) // 4 ᴧ he’s / finished his as /signment // // 4 ᴧ he’s / finished his as /signment // The Tone [4] variant makes the meaning of fall-rise pitch more intense, and is accompanied by a distinct voice quality. For instance, if, in a Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

199

particular context (such as indicated in the examples), Tone 4. is used to indicate ‘reservation’, the use of Tone [4] instead makes it a more personal or intense reservation.

7.4.2.13 Direct Secondary Tones of Tone 5 Tone 5 has two direct secondary tones: [5.] and [5]. Similar to Tone 4, the pretonic contours are fixed for each direct secondary tone.

Figure 7.22 Direct and indirect secondary tones of Tone 5 ([5.] and [5] respectively)

Tone [5.] is the neutral tone having a rise from about mid to mid high or high and then a fall to a lower pitch level. The pitch in the pretonic segment exhibits a step up from about mid low to about mid high. Tone [5] is a low pitched risefall contour at mid low/mid pitch level. This is preceded by a pretonic, with each foot exhibiting a step down rise-fall movement. //5. ᴧ I /can’t be/lieve they would /ever have /thought that a/bout her// //5 ᴧ I can’t be/lieve they could /be so /stupid//

7.4.2.14 Tone 5. and Tone 5 (Attitudinal) // 5. ᴧ the / soup was / very / tasty // // 5 ᴧ the / soup was / very / tasty // The meaning of the neutral Tone [5.] is related to the rise-fall prosody, i.e. ‘there was some doubt, but all is fine’. On the other hand, Tone [5], which is lower pitched and usually accompanied by a breathy voice quality, is used to indicate ‘awe’, and sometimes, depending on the context, ‘sarcasm’ or ‘disappointment’, as in the examples.

7.4.2.15 Tone 13 and Tone 53: Secondary Tones As mentioned earlier, Tone 13 and Tone 53 are compound tones, and the pretonic, if present, occurs only before the first tonic segment. The pitch contour in the pretonic indicates the direct secondary tones of Tone 1 or Tone 5 in the compound tones.

7.5

Conclusion

This chapter has provided an overview of intonation within the SFL architecture of language with some mention along the way of specific research that has been conducted. The second half of the chapter outlined the melodic shapes and general meanings of the primary and secondary tone groups in Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

200

WENDY L. BOWCHER AND MEENA DEBASHISH

English. Our aim has been to provide a description useful for understanding the place of intonation in the SFL architecture of language and for conducting some basic analyses of spoken English, albeit recognizing that practice in listening to the tone groups would be essential for undertaking such an analysis. Further, while it is impossible to do justice to all of the areas of research that have been developed for intonation within the SFL framework in a chapter of this size, we hope that we have provided sufficient background to offer readers ideas for possible research directions.

References Banks, D. 2014. A Note for –ed: Comments on the Treatment of –ed in Handel’s Messiah. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 221–34. Bowcher, W. L. 1998. Intonation in Radio Sports Commentating: Towards an Analysis and Interpretation. In Y. Nagahara, ed., Kobunnokairuinikansuru Kijutsutekioyobi Rirontekikenkyuu. Japanese Ministry of Education Research Grant no. 07451097. Bowcher, W. L. 2003. Creating Informational Waves: Theme and New Choices in Play-by-play Radio Sports Commentary. In M. Amano, ed., Creation and Practical Use of Language Texts: Proceedings of the Second International Conference for the Integrated Text Science. Nagoya: Graduate School of Letters, Nagoya University. 111–22. Bowcher, W. L. 2004. Theme and New in Play-by-play Radio Sports Commentating. In D. Banks, ed., Text and Texture: Systemic Functional Viewpoints on the Nature and Structure of Text. Paris: L’Harmattan. 455–93. Bowcher, W. L. and B. A. Smith. 2014a. Introduction. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 1–24. Bowcher, W. L. and B. A. Smith, eds. 2014b. Systemic Phonology: Recent Studies in English. Sheffield: Equinox. Bowcher, W. L. and S. Zhu. 2014. Meaningful Reading: Intonation Choices by Native and Non-native English Speakers Reading The Giving Tree. Unpublished manuscript. 1–36. Caldwell, D. 2014. A Comparative Analysis of the Rap and the Sung Voice: Perspectives from Systemic Phonology, Social Semiotics and Music Studies. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 235–63. Couper-Kuhlen, E. 1986. An Introduction to English Prosody. London: Edward Arnold. Crystal, D. and D. Davy. 1969. Investigating English Style. London: Longman. Cummings, M. 2000. The Inference of Given Information in Written Text. In E. Ventola, ed., Discourse and Community: Doing Functional Linguistics. Tübingen: Gunter Narr Verlag. 331–53.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

201

Cummings, M. 2001. Intuitive and Quantitative Analyses of Given/New in Texts. In J. de Villiers and R. J. Stainton, eds., Communication in Linguistics, Volume 1: Papers in Honour of Michael Gregory. Toronto: GREF Publishers. 61–94. Cummings, M. 2014. The Spoken Interpretation of Written Text. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 199–217. Davies, M. 1989. Prosodic and Nonprosodic Cohesion in Speech and Writing. Word 40(1–2): 255–62. Davies, M. 1992. Prosodic Cohesion in a Systemic Perspective. In P. Tench, ed., Studies in Systemic Phonology. London: Pinter. 206–30. Davies, M. 1994a. Intonation IS Visible in Written English. In S. Čmejrková, F. Daneš, and E. Havlová, eds., Writing vs Speaking: Language, Text, Discourse, Communication. Tübingen: Gunter Narr Verlag. 199–204. Davies, M. 1994b. I’m Sorry I’ll Read that Again: Information Structure in Writing. In S. Čmejrková and E. Štícha, eds., The Syntax of Sentence and Text: A Festschrift for Frantisek Daneš. Tübingen: Gunter Narr Verlag. 75–89. Davies, M. 2014. The Black Hole in Graphology. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 153–98. Fawcett, R. 2014. The Meanings and Forms of Intonation and Punctuation in English: The Concepts Required for an Explicit Model. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 324–401. Firth, J. R. 1948. Sounds and Prosodies. Transactions of the Philological Society 47(1): 127–52. Firth, J. R. 1968. Linguistic Analysis as a Study of Meaning. In F. R. Palmer, ed., Selected Papers of J. R. Firth 1952–59. London: Longmans. 12–26. Greaves, W. S. 2014. Locating the Limerick ‘Wall Street Irene’ and the Sonnet ‘On His Blindness’ in the Semiotic Space between the Body as Signal Generator/Receiver and the Body as Social Interactant. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 405–38. Halliday, M. A. K. 1961. Categories of the Theory of Grammar. Word 17(3): 242–92. Halliday, M. A. K. 1963a. Intonation in English Grammar. Transactions of the Philological Society 62(1): 143–69. Halliday, M. A. K. 1963b. The Tones of English. Archivum Linguisticum 15(1): 1–28. Halliday, M. A. K. 1967. Intonation and Grammar in British English. The Hague: Mouton de Gruyter. Halliday, M. A. K. 1970. A Course in Spoken English: Intonation. London: Oxford University Press. Halliday, M. A. K. 1985a. Introduction to Functional Grammar. London: Edward Arnold.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

202

WENDY L. BOWCHER AND MEENA DEBASHISH

Halliday, M. A. K. 1985b. Part A. In M. A. K. Halliday and R. Hasan, Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Geelong: Deakin University Press. Halliday, M. A. K. 1992. How Do You Mean? In M. Davies and L. Ravelli, eds., Advances in Systemic Linguistics: Recent Theory and Practice. London: Pinter. 20–35. Halliday, M. A. K. 1994. Introduction to Functional Grammar. 2nd ed. London: Edward Arnold. Halliday, M. A. K. and W. S. Greaves. 2008. Intonation in the Grammar of English. Sheffield: Equinox. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2004. Introduction to Functional Grammar. 3rd ed. London: Arnold. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Routledge. Iwamoto, K. 2014. A Multistratal Approach to Paragraph-like Organisation in Lectures. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 116–49. Johns-Lewis, C. 1986. Prosodic Differentiation of Discourse Modes. In C. Johns-Lewis, ed., Intonation in Discourse. London: Croom Helm. 199–219. Kuiper, K. 1996. Smooth Talkers: The Linguistic Performance of Auctioneers and Sportscasters. Mahwah: Lawrence Erlbaum. Lukin, A. 2014. Creating a Parallel Universe: Mode and the Textual Metafunction in the Study of One News Story. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology. Sheffield: Equinox. 53–90. Matthiessen, C. M. I. M. 1995. Lexicogrammatical Cartography: English Systems. Tokyo: International Language Science Publishers. McGregor, W. B. 1992. Towards a Systemic Account of Gooniyandi Segmental Phonology. In P. Tench, ed., Studies in Systemic Phonology. London: Pinter. 19–43. Ogden, R. 2012. Firthian Prosodic Analysis. Firthian Phonology Archive. Available online at: https://sites.google.com/site/firthianarchive/fpa. (Last accessed 27/07/2017.) O’Grady, G. 2014. An Investigation of How Intonation Helps Signal Information Structure. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology. Sheffield: Equinox. 27–52. O’Grady, G. 2017. Intonation and Systemic Functional Linguistics: The Way Forward. In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 146–62. Prince, E. 1981. Towards a Taxonomy of Given-New Information. In P. Cole, ed., Radical Pragmatics. New York: Academic Press. 223–56. Robins, R. H. 1970. Aspects of Prosodic Analysis. In F. R. Palmer, ed., Prosodic Analysis. London: Oxford University Press. 104–11. Smith, B. A. 2008. Intonational Systems and Register: A Multidimensional Exploration. PhD Thesis, Macquarie University. Available online at: www.isfla.org/Systemics/Print/Theses/SmithBradPhD.pdf. (Last accessed 27/07/2017.)

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

Intonation

203

Storynory, n.d. How Love and Peace Came to the Woods. Available online at: www.storynory.com/2017/04/22/love-peace-came-woods. (Last accessed 27/07/2017.) Tench, P. 1988. The Stylistic Potential of Intonation. In N. Coupland, ed., Styles of Discourse. London: Croom Helm. 50–84. Tench, P. 1990. The Roles of Intonation in English Discourse. Frankfurt am Main: Peter Lang. Tench, P. 1992a. From Prosodic Analysis to Systemic Phonology. In P. Tench, ed., Studies in Systemic Phonology. London: Pinter. 1–17. Tench, P., ed. 1992b. Studies in Systemic Phonology. London: Pinter. Tench, P. 1996. The Intonation Systems of English. London: Cassell. Tench, P. 2014. Towards a Systemic Presentation of the Word Phonology of English. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 267–93. van Leeuwen, T. 1992. Rhythm and Social Context: Accent and Juncture in the Speech of Professional Radio Announcers. In P. Tench, ed., Studies in Systemic Phonology. London: Pinter. 231–62. Young, D. 1992. English Consonant Clusters: A Systemic Approach. In P. Tench, ed., Studies in Systemic Phonology. London: Pinter. 44–69. Zhu, S. 2014. Intonation: Signal of Information Peaks. In W. L. Bowcher and B. A. Smith, eds., Systemic Phonology: Recent Studies in English. Sheffield: Equinox. 91–115.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 08:26:40, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.009

8 Continuing Issues in SFL Mick O’Donnell

8.1

Introduction

This chapter will discuss various issues that are not fully resolved within Systemic Functional Linguistics (SFL). Two main issues will be addressed: •

•

To what degree are the grammatical categories of SFL determined on notional grounds (mirroring extra-linguistic organization), rather than on regularities of form? Where does genre belong in relation to other components of the model?

Both of these issues are still under debate within the community, often leading to divergent approaches, and, if we are not aware of the underlying differences, result in misunderstanding of the arguments others are making. This chapter will refer extensively to the four editions of ‘Introduction to Functional Grammar’, the first two by Halliday (Halliday 1985, 1994), and the last two revised by Matthiessen (Halliday and Matthiessen 2004, 2014). To simplify references, I will refer to these as IFG1, IFG2, IFG3, and IFG4.

8.2

How ‘Semantic’ is the Grammar?

In this section, I will explore the criteria used to define grammatical categories in SFL, in particular, the degree to which grammatical categories are based on notional grounds rather than in terms of grammatical reactance. Prior discussions related to this point can be found in Butler (2003),

My thanks to Tom Bartlett, Margaret Berry, Lise Fontaine, Jim Martin, and Geoff Thompson for comments on this work. While Butler (2003) is only lightly cited in this work, it has strongly influenced my way of thinking about functional grammar discussed in Section 8.2.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

205

but see also Hudson (1971); O’Donnell et al. (2008); Fawcett (2009); Tucker (2014); Gwilliams and Fontaine (2015).

8.2.1 Linguistic Levels as Arbitrary Points of Abstraction Linguistic descriptions of discourse can range from descriptions of the physical manifestation of the discourse (the marks placed on the page, the sounds produced by the speaker) up to descriptions of the cultural, social, and material context in which the discourse is produced. The realm of most linguistic work is, however, in the space between these two more manifested end-points. Between these points, linguists typically choose to produce a number of levels of analysis, each more abstract than the last. On the physical manifestation side, linguists move away from the purely phonetic description of the discourse, focusing instead on a more abstract phonemic description. More abstractly, words are identified, and within them, morphemic elements. Even more abstractly, the grouping of words into larger units is explored. Moving further, some linguists will explore the way in which these groupings of words relate to what the speaker/writer is trying to achieve through the production of these words. In constructing these levels of analysis, the linguist is to a degree free to determine where in the space between physical manifestation and context the levels may go. Some choose to traverse the space in small steps, positing eight or so levels of linguistic analysis. Others may choose to skip across the divide in large steps. Others still, afraid of deep water, ignore the task of spanning the divide, and instead focus on describing the physical manifestation, ignoring context- and use-related aspects of language. The most typical levels of analysis in linguistic models focus on firstly, the organization of sounds (phonetics or phonemics) and secondly, the organization of words into larger units (syntax or grammar). Often, a number of levels above this will be posited, exploring the deep water far from the physical manifestation (semantics, pragmatics, rhetoric, etc.). This section is concerned with the grammatical level of description, looking to the variation in abstraction that is possible between different approaches. A basic mechanism for determining grammatical categories involves grouping together units of text that fill the same textual context, and assigning them a class. Using an over-simplified example, all words that can fill the gap in ‘a ______ is needed’ could be classed as ‘noun’, and more specifically, ‘singular noun’. Grammars can then be written describing the valid classes of words and word groups, the sequences of these classes, and their interrelation in terms of dependency or constituency. Classes are based on the ‘mutual substitutability’ of elements. Many approaches, particularly those following Bloomfield, believed in the autonomy of syntax, that ‘syntactic phenomena are essentially independent of the conventional semantic, pragmatic and discoursal functions

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

206

MICK O’DONNELL

of those phenomena’ (Butler 2003:6). In these approaches, one would construct the grammar of a language by exploring the patterns of words independent of their meaning. One could take a body of text, replace each word by a number, and work towards discovering the grammar, free of the confusing distraction of the meanings of the words.

8.2.2 Functional Grammars This however is not the only approach to building grammars. Functional linguists work on the assumption that the system of grammar is ‘so intimately bound up with external motivating factors that it makes no sense to try to describe it without reference to those factors’ (Butler 2003:12). In Halliday’s words: The particular form taken by the grammatical system of language is closely related to the social and personal needs that language is required to serve. (Halliday 1970:142) The relation between the meaning and the wording is not, however, an arbitrary one; the form of the grammar relates naturally to the meanings that are being encoded. A functional grammar is designed to bring this out; it is a study of wording, but one that interprets the wording by reference to what it means. (IFG1:xvii) A systemic grammar is one of the class of functional grammars, which means (among other things) that it is semantically motivated, or ‘natural’. In contradistinction to formal grammars, which are autonomous, and therefore semantically arbitrary, in a systemic grammar every category (and ‘category’ is used here in the general sense of an organizing theoretical concept, not in the narrower sense of ‘class’ as in formal grammars) is based on meaning: it has a semantic, as well as a formal, lexicogrammatical reactance. (Halliday and Matthiessen 2006:3–4)

Halliday argues that this intimate relation between grammar and its context of use is not accidental, but rather the result of language (including grammar) having evolved in its use: Language has evolved to satisfy human needs; and the way it is organized is functional with respect to these needs – it is not arbitrary. A functional grammar is essentially a ‘natural’ grammar, in the sense that everything in it can be explained, ultimately, by reference to how language is used. (IFG2:xiii) The concept of the social function of language is central to the interpretation of language as a system. The internal organisation of language is not accidental; it embodies the functions that language has evolved to serve in the life of social man. (Halliday 1973:44)

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

207

Or in Martin’s words: [Functional grammar] explores grammar as being shaped by, at the same time as playing a significant role in shaping, the way we get on with our lives. (Martin et al. 2010:1)

This is not to say that grammars can be defined entirely in terms of how words are used. This would lead to what is called a ‘notional grammar’: grammar defined entirely in terms of meaning.1 In such an approach, a ‘verb’ might be defined as ‘a word that expresses an event’, and ‘noun’ as ‘a word that expresses an entity’. For Halliday, a functional grammar has to relate outwards to the meanings it realizes, and also account generatively for the range of forms that realize these meanings. He stresses that our grammatical organization cannot be divorced from the need to account for structural patterning: All the categories employed must be clearly ‘there’ in the grammar of the language. They are not set up simply to label differences in meaning. In other words, we do not argue: ‘these two sets of examples differ in meaning; therefore they must be systematically distinct in the grammar’. They may be; but if there is no lexicogrammatical reflex of the distinction, they are not. (IFG1:xx)

Thus, in a systemic grammar, categories are determined both externally (to capture similarities in meaning/use) and structurally (to capture similarities in lexicogrammatical realization). By IFG3, these criteria were further developed into the idea of ‘trinocularity’, that grammatical concepts are defined from above (the semantics), from the same level (grammatical form), and also from below (phonology in this case): We cannot expect to understand the grammar just by looking at it from its own level; we also look into it ‘from above’ and ‘from below’, taking a trinocular perspective. But since the view from these different angles is often conflicting, the description will inevitably be a form of compromise. All linguistic description involves such compromise; the difference between a systemic description and one in terms of traditional school grammar is that in the school grammars the compromise was random and unprincipled, whereas in a systemic grammar it is systematic and theoretically motivated. (IFG3:31)

Halliday and Matthiessen raise the point here that, when building a grammatical description, evidence from the three viewpoints may conflict, and the model builder needs to choose what importance they give to each

1

David Rose, in the Sysfling discussion list (09/04/2011), suggested that the term ‘notional grammar’ is appropriate for grammars which ignore grammatical reactance totally, while ‘functional grammar’ is appropriate for a grammar based on the identification of recurring grammatical structures which have distinct semantic functions.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

208

MICK O’DONNELL

Table 8.1 Two approaches to structural analysis Structural analysis Premod. seven a a

Head

Postmod.

apples handful man

of apples of means

Functional analysis Quantifier

...

Thing

Qualifier

a

apples apples man

of means

seven a handful of

source of evidence. To exemplify this problem, Table 8.1 shows three nominal groups, and presents two alternative ways of structurally analyzing them (also see Fontaine and Schönthal, this volume). The leftmost analysis, a more structural approach, takes structural similarity as the most important principle (‘from its own level’). In addition, if these phrases were Subject in a sentence, the finite verb would usually agree with the noun denominated as ‘Head’. The functional approach places lower emphasis on the seeming structural similarity, focusing instead on what the elements are doing semantically (‘from above’): both of the first two examples are talking about apples, and seven and a handful of are functioning to specify how many apples are involved. In a similar vein, functionalists would claim that with a cup of water, we are not talking about a cup, but rather about a quantity of water. So when we say he threw a cup of water over his brother, we usually understand that the cup was not in fact thrown, just the water.

8.2.3 Weighing Formal and Notional Evidence So far in this paper, functional grammars have been defined just as those which take into account evidence from both meaning (above) and from form (aside/below). Halliday and Matthiessen, in the quotation above, point out that functional grammars need to reach compromises between these different sources of evidence. One point that needs to be made clear is that, even within SFL, the principles for making these compromises can vary: SFL linguists vary in the degree of importance they give to structural and meaningful criteria in determining grammatical categories. Halliday gives a clue as to where he places priority: The fact that this is a ‘functional’ grammar means that it is based on meaning; but the fact that it is a ‘grammar’ means that it is an interpretation of linguistic forms. Every distinction that is recognized in the grammar . . . makes some contribution to the form of the wording. Often it will be a very indirect one, but it will be somewhere in the picture. (IFG1:xx)

The fact that the required grammatical reactance can be very indirect shows that priority is given to similarity of meaning when deciding on

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

209

grammatical categories, rather than to structural regularity. This is made explicit in IFG3: Being a ‘functional grammar’ means that priority is given to the view ‘from above’; that is, grammar is seen as a resource for making meaning – it is a ‘semanticky’ kind of grammar. But the focus of attention is still on the grammar itself. (IFG3:31)

Rather than talking about the relative importance of the criteria, it may be better to talk of starting point. In IFG2 (xiv), Halliday contrasts the traditional approach of starting with modelling the word forms (morphology), building a syntax on top of that, and only then asking what these forms mean, with the SFL approach, which starts by interpreting a language as a system of meanings, and then explores how those meanings can be realized as forms. A possible critique to this approach is that there are many ways to organize a language in terms of meaning, and only a subset of these will allow a simple mapping onto forms. What we ideally want is a meaningful organization of language which has the strongest correlation with regularities of form. We cannot do this by exploring meaning in isolation from the forms that realize them. This is the reverse of the criticism levelled by functionalists and cognitivists against Bloomfield and his successors, who tried to construct an autonomous syntax without considering meaning (Tomlin 1990; Newmeyer 1991:62; Halliday in Martin 2013:164). The full answer is that, in the construction of a language model, we need to consider meanings and forms at the same time. Structuralists often apply the principle of Occam’s Razor: the best grammatical description is that which uses the least rules to describe the phenomena at hand. The principle can also be applied to the construction of functional grammars: the best description is the briefest which represents the meaningful aspects of language use and from which all forms can be generated (with the mapping of meaning onto form included in the size of the description). Unfortunately, SFL grammars (or semantic specifications) are rarely presented with both system network and realization statements (Hudson 1971 and Matthiessen 1995 being good exceptions), and thus commonality of meaning may play a bigger role in grammar construction than it should. Halliday’s verbal processes (see Berry, this volume) offer a good area through which to demonstrate the problems of grammatical classification. Exactly what constitutes a verbal process is often debated within the community, and the four versions of IFG have shifted on the issue over time. There is a common belief within the community that the test to identify a verbal process is that there must be projection in clausal form (or at least potential for clausal projection). This test would result in He said he was going being classified as verbal, while He talked about the weather would not.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

210

MICK O’DONNELL

However, all versions of IFG include at least two classes of verbal process: the projecting kind just mentioned, and one which involves a Target (I’m always praising you to my friends). As there is no structural similarity between these two types of verbal process which would motivate their grouping, we must assume they are being grouped totally on notional grounds: they both express a verbal action.2 We might propose putting targeted verbal processes aside, and say that the clausal projection criterion applies to the remaining verbal processes. However, it seems that the SFL community as a whole is divided as to what to do with processes which involve verbal action, but where clausal projection is not involved. This involves verbs such as talk, and grumble, and includes cases where no Matter is specified (We talked for hours), and also where Matter is specified (He talked about his hometown). An online survey was conducted in 2004 to test how a range of SFL practitioners coded various clauses in terms of process types (O’Donnell et al. 2008). Seventy-five respondents coded thirty-two difficult clauses. The survey revealed a spread of coding styles, ranging from heavy dependence on structural criteria, to those who coded largely on the semantics of the clause. In respect to We talked for hours, 60 per cent of coders placed it as behavioural, and 35 per cent followed notional criteria, coding it as verbal (the remaining 5 per cent coded it as material, some indicating they did not use the behavioural category). In regards to He talked about his hometown, a surprising result was that 15 per cent of those who had coded the previous sentence as behavioural swung over to verbal for this case. This suggests that these coders do require presence of the verbal product to code as verbal, but do not go so far as to require clausal projection (a similar pattern was shown in the coding of mental processes). This variation in the coding community demonstrates that the nature and degree of structural reactance needed varies across the community. A further study reported in Gwilliams and Fontaine (2015) confirmed these results. Part of the disparity in coding verbal processes may stem from the treatment of this area in the four versions of IFG. In IFG1, verbs like talk were not covered explicitly in the section on verbal processes, although a later section on Range classified She speaks German and Don’t talk nonsense! as verbal processes (IFG1:133). Behavioural processes are said to be intermediate between material and mental processes, which seems to exclude the talk verbs from this category. IFG2 however expands behavioural processes to include a ‘near verbal’ category, which includes talk, grumble, and chatter (IFG2:139). The possibility of Matter with these verbs is explicitly mentioned, so He talked about the 2

Tom Bartlett (personal communication) prefers to phrase the semantic label for verbal processes as ‘transfer of information from one person (or semiotic object) to another’. This would leave We talked about the weather out of verbal processes, while leaving He called me a bastard in.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

211

Table 8.2 Types of verbal processes (IFG4:305) TYPE activity

semiosis

targeting talking (neutral quoting) indicating

imperating

Examples of verbs praise, flatter, commend, compliment, congratulate; insult, abuse, slander, blame, criticize, chide, censure, pillory, rebuke speak, talk say, tell; go, be like tell (sb that), report, announce, notify, explain, argue, convince (that), persuade (sb that), promise (that) ask (sb whether), question, enquire (whether) tell (sb to do), ask (sb to do), order, command, require, promise, threaten, persuade (sb to do), convince (sb to do), entreat, implore, beg

weather would be classified as behavioural. The talk example was removed from the Range examples. In terms of verbal processes, IFG2 says that verbal processes do display distinctive patterns of their own. Besides being able to project . . ., they accommodate three further participant functions in addition to Sayer. (IFG2:141)

The targeting type of process is again mentioned, with a statement that this subtype of verbal process does not easily project reported speech. A list of verbs taking a (nominal) Verbiage includes some which cannot easily project: He described the apartment, or He outlined his plan. In IFG3, representing a revision of IFG2 by Matthiessen, we see a change back towards more notional coding. To talk to that priest about Kukal is said to be verbal (IFG3:252). However, there seems to be some inconsistency here, as grumbled about the food (IFG3:251) is said to be behavioural. I believe this was a state of transition from Halliday’s original more structural orientation towards Matthiessen’s more notional orientation. In IFG4, the talk processes are fully instantiated as a subtype of verbal process, as shown in Table 8.2. However, behavioural processes still include verbs such as chatter, grumble, and talk, which appears to be an inconsistency, with grumbled about the food explicitly mentioned as behavioural (IFG4:302). Thompson (2015) believes this is not just a problem for verbal processes, but general across transitivity classification: Halliday (1994: xix) has consistently argued that ‘all the categories employed must be clearly “there” in the grammar of the language. They are not set up simply to label differences in meaning’; and in the case of transitivity certain key grammatical criteria for categorization (such as preferred tense/aspect, and the potential to project) have been elaborated (e.g. Halliday 1994: 115–16). However, it has proved difficult to implement the principle of ‘clearly “there” in the grammar’ in all cases: the grammatical criteria by

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

212

MICK O’DONNELL

which one process type can be differentiated from another are not always precisely definable. As a result, analysts may, implicitly or explicitly, find themselves forced to fall back on purely semantic criteria. (Thompson 2015:21–2)

The above discussion has tried to show that in Halliday’s grammar, the notional criteria dominate over secondary, indirect, grammatical reactance. Some in the community however stress the importance of the grammatical: Process types are entirely grammatical categories. The names of the process types are just aide-memoires that capture only their most common notional features; they are not useable as criteria for defining on notional grounds. (Tom Bartlett 2011, sys-func discussion list) I would strongly encourage holding onto grammatical reactances when reasoning about process type. These are the grounding strength of our SFL approach to case relations, compared with work in other models. We should be enriching our argumentation based on reactances . . . We need to push on to tackle the challenge of finding distinctive reactances for process types as we move from language to language – and NOT abandon the criterial argumentation the reactances afford. It is very worrying to think that the power of the SFL approach (its revelation of the meaningful ways in which languages construe reality) might become its undoing via a collapse into notionalism, or accommodation of notionalist ‘reasoning’ (sic) alongside reasoning based on grammatical criteria. (Jim Martin 2011, sys-func discussion list)

To summarize the discussion so far, we can distinguish three types of grammars: • • •

notional grammars, based on semantic concerns, ignoring syntactic reactance; formal grammars, based on structural concerns, ignoring semantic factors; functional grammars, which take both meaning and structural issues into account.

Approaches to functional grammars, however, can vary widely, from, on one side, structurally oriented functional grammars which favour structural criteria over semantic ones, to notionally oriented functional grammars, which favour the semantic over the structural.

8.2.4

Complex Relations between Notional Situation and Grammatical Form One problem for notionally oriented functional grammars is that there is often a complex relationship between semantic representations and the forms that express them. One example is in the relation between the speech function systems and their grammatical expression. The ‘demand information’ speech

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

213

function can be realized grammatically in terms of interrogative syntax (Did you like the movie?) or via declarative syntax with interrogative intonation (You liked the movie?). In this area of the grammar (Mood) at least, Halliday’s grammar has given more weight to the structural organization of language, grouping clauses based on similarity of form, rather than on similarity of speech function (a semantic concern). However, when dealing with Transitivity, things are not so clear. Fawcett (2009:214) points out that ‘there is NOT a simple one-to-one relationship between the realms of experience and the types of Process’ (his ‘realms of experience’ correspond to a language-external experiential representation). He points out that different notional situations can be realized through distinct grammatical realizations. Using my own examples in (1) to (4), a situation of a man being preoccupied about his tomatoes could be expressed in various ways (the process categories given are arguable, but based on the syntactic similarity to less contentious clauses): (1)

He is very worried about his tomatoes (relational attributive)

(2)

He worries about his tomatoes (behavioural)

(3)

He has concerns about his tomatoes (relational possessive)

(4)

He thinks his tomatoes might die (mental)

The point here is that, given the different possible expressions of a notional situation, using notional criteria to classify process types does not seem promising. Gwilliams and Fontaine (2015) look at the problem from the other side: because of ideational grammatical metaphor, similar clausal expressions can be used for distinct notional situations, for instance, ‘Ivy touched Fred with a stick’ (representing a notionally material action) and ‘Ivy touched Fred with her words’ (representing a notionally mental action). They note that such ambiguous cases give rise to two distinct problems. Firstly, because these examples allow for two analyses, they introduce the potential for inconsistent coding amongst analysts, given that some may favour semantic criteria, and others, syntactic criteria. They point out however that this can be avoided by explicit direction as to the coding criteria. The second problem they think is more important: If a process can be interpreted in more than one way, being constrained to a single classification may lead to an analytic interpretation that does not truly reflect the semiotics of the message, going against the primary objective of SFL. (Gwilliams and Fontaine 2015:3)

The solution they propose is to always allow for two analyses of clauses: a surface analysis based on syntactic tests, and a deep analysis based on notional grounds (although in most cases, these would be the same). They

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

214

MICK O’DONNELL

suggest that most of the contentious clauses encountered in process type coding are exactly those where the deep coding differs from the surface. There are other cases of ambiguity not due to grammatical metaphor, but rather the result of words covering a range of notional space. For instance, look at the clauses in examples (5) and (6): (5)

We agreed with each other that I was right.

(6)

I agree with the Prime Minister that something should be done.

The first example implies verbal discussion, so would be coded notionally as a verbal process. The second example is a little more difficult: it looks like a statement of a mental state rather than of an explicit verbalization, so would be coded, in many contexts, as a mental process (as the agreement is not a verbalization but rather a statement of concord of ideas). So, two clauses with essentially the same sentence structure are coded differently with reference to the type of situation being referred to. Structural criteria cannot help us decide on process type, and in these cases, we need to ask ourselves what kind of (notional) activity is being represented. Gwilliams and Fontaine (2015:8) argue that ‘semantic information is a kind of subjective distractor’ from proper coding of transitivity, and that one should instead follow Fawcett’s approach, basing process type coding on the presence of structural elements (Participant roles = PRs): In analyzing Process types and PRs, it doesn’t help to use the realm of experience as a guide. And the analyst who has been forewarned of this problem is less likely to fall into the trap of skipping the stage of applying the tests for the Participant Roles, when trying to establish the Process type of a clause. (Fawcett 2009:215)

The point of this discussion has been to show that the issue of relative importance of notional vs. structural criteria in the grammar is still an open issue. Surveys of coding practice show practitioners range from more to less notional in their coding of process type, and the four versions of IFG are themselves in flux as to the importance of notional vs. structural criteria.

8.3

Current Issues Related to Genre

The issues of what ‘genre’ is, and where it belongs in relation to the other components of the linguistic model, have long been debated within SFL, and the debate continues today between different parts of the community. This section will discuss some of the main issues in this area. One note on terminology: Halliday uses the term ‘register’ to refer to the set of linguistic features that realizes a particular configuration of situational features. Martin, on the other hand, uses the term ‘register’ to refer

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

215

to the configuration of situational features (e.g. Martin 1992). In the discussion below, I will follow Halliday’s use rather than Martin’s, even when discussing Martin’s work.

8.3.1 How Do We Define Genres? A first issue involves exactly how genres are defined. Paltridge (1996) put forward two principal approaches to defining genres: • •

In terms external to the text, generally the purpose of the text, e.g. to persuade, to educate, to entertain, etc. In terms internal to the text, most typically in terms of common schematic structures, or linguistic styles, e.g. editorial, narrative, anecdote, report, etc.

‘Genre’ has been used in various places in both of these senses. Lee (2001:38) describes the first approach, which makes a distinction between ‘genres’ (defined on external criteria) and ‘text types’ (defined on internal criteria): A genre, in this view, is defined as a category assigned on the basis of external criteria such as intended audience, purpose, and activity type, that is, it refers to a conventional, culturally recognised grouping of texts based on properties other than lexical or grammatical (co-)occurrence features, which are, instead, the internal (linguistic) criteria forming the basis of text type categories.

Biber (1988:170) is a prominent proponent of this approach: Genre categories are determined on the basis of external criteria relating to the speaker’s purpose and topic; they are assigned on the basis of use rather than on the basis of form.

Swales (1990:58) also follows this approach: A genre comprises a class of communicative events, the members of which share some set of communicative purposes. These purposes are recognised by the expert members of the parent discourse community, and thereby constitute the rationale for the genre.

In earlier works within SFL, ‘genre’ was not directly covered by the model, although aspects relatable to external definitions of genre were mentioned, placed within Context of Situation. This includes Halliday’s inclusion of ‘purpose’ and ‘rhetorical mode’, Ure and Ellis’ ‘role’, and Gregory and Carroll’s ‘functional tenor’. These aspects will be discussed further below. More recently, the internal definition of genre has been more prominent within SFL: genres being defined as groupings of text with common text structures. Hasan (1978:229), for instance, clearly takes this approach: The generic membership of the text is determined by reference to the structural formula to which the actual structure can be shown to belong.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

216

MICK O’DONNELL

Martin (and Martin and Rose) also indicate the use of internal criteria: Genre networks would thus be formulated on the basis of similarities and differences between text structures. (Martin 1992:505) Recurrent global patterns were recognized as genres, and given names. (Martin and Rose 2008:7)

Martin (personal communication) notes that the external/internal distinction is complicated when applied to his model, as genre is realized through constraints on register, and register could be seen as external to the linguistic system. One could thus say that genres are recognized in terms of patterns external to language. He comments however that as register is itself a connotative semiotic system, it may be extra-linguistic, but it is not extra-semiotic. Paltridge (1996:238) argues that Martin often labelled genres with categories more frequently associated with external, purpose-based definitions, such as poems, lectures, seminars, recipes, etc. Martin however responds that he avoids such labels precisely because they are common sense everyday labels which aren’t names of recurrent patterns of meaning (e.g. a poem can be almost any genre in my terms – anecdote, report, description, narrative, procedure etc. – where genre is a recurrent pattern of meaning). (Martin 2015, personal communication)

Martin’s key definition of genre as ‘a staged, goal-oriented, purposeful activity in which speakers engage as members of our culture’ (Martin 1984:25) includes internal aspects (‘staged’, ‘activity’) but also includes external aspects (‘goal-oriented’, ‘purposeful’). The internal/external definition of genre is also reflected in the following formulation from Martin and Rose (2003:7): We use the term genre in this book to refer to different types of texts that enact various types of social contexts.

One possible interpretation parallels the arguments in Section 2 in relation to the form/function co-evolution: our social needs and the activity sequences we use to satisfy them have evolved in tandem, with the consequence that each text type is an inseparable fusion between an externally defined purpose and a linguistic means of achieving that purpose. Taking this approach, one could argue that if one discovers a set of texts which share the same generic structuring, then the functional assumption would be that these texts also reflect externally defined commonalities: similarity of purpose, etc. There is no explicit discussion of this assumption in the literature, but such an underlying assumption might explain the mix of internal and external elements in Martin’s approach.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

217

In summary of this section, while the definition of genre in terms of shared text structure seems to dominate within SFL today, it is not the only approach.

8.3.2 How are Genres Realized? In some approaches, the linguistic realization of a ‘genre’ is explored in terms of the recurrent stylistic qualities of the texts that belong to the genre. In other words, a genre can be defined as a set of expected linguistic patterns, whether lexical, grammatical, or semantic. Genre in this approach is very relatable to Halliday’s notion of Register. Alternatively (or additionally), a genre can be realized in terms of requiring recurrent schematic staging of the texts in the genre. Within SFL, this is often called ‘generic structure’. SFL models are perhaps most differentiated in the degree to which they explicitly handle the notion of generic structure. In many early approaches, the term ‘genre’ was avoided as a technical term, and instead, terms such as ‘purpose’, ‘purposive role’, ‘language use’, or ‘functional tenor’ were used, all of which were categories of the Context of Situation, and were realized in terms of stylistic patterning of the text, via Register. In these approaches, generic staging was not usually mentioned. In later work, Martin and Hasan do explicitly address genre and generic structure, and in their works, genre is defined in terms of commonality of text structure. It remains to be discussed whether genre in their models is also realized through overall generic styling of the text. In the work of Hasan, it seems that the stylistic realization of a genre is treated as an aspect of register. For example, Hasan (1978:241) says that ‘[t]he terms “register” and “genre” as used here are then interchangeable’. In Martin’s model, since genre selections are realized as constraints on register, the overall stylistic commonality of a genre can also be seen as handled via register, although genre, where genre choices are made, is placed stratally above register, where the stylistic consequences are managed.

8.3.3 Where Does ‘Genre’ Belong in the SFL Model In this section, we consider where SFL approaches place ‘genre’ in relation to the level of Context of Situation (which is known as ‘Register’ in Martin’s model). 8.3.3.1 Genre as a Component of Context of Situation As mentioned above, in earlier SFL models (and still sometimes today), concerns similar to genre were placed as components of the Context of Situation. Different practitioners placed these concerns in different components of the situation, whether Field, Tenor, or Mode (see Martin 1999).

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

218

MICK O’DONNELL

As a component of Field: Halliday’s earliest model of Context of Situation included ‘purpose’ as an aspect of Field: THE FIELD: Here we include the subject matter; and also the type of situation in which language is used, including the purpose – e.g., didactic or explanatory, for information, for action, consolation or self-satisfaction. (Halliday 1965:14)

Matthiessen (2015) reports on unpublished work by Jean Ure, where she is said to include under field the category ‘field of activity’, which was subdivided into eight areas: expounding, reporting, recreating, sharing, doing, enabling, recommending, and exploring. Matthiessen (2015:6) also follows this approach, although renaming the parameter as ‘socio-semiotic process’: The field of activity is ‘what’s going on’ in context . . . The activity is either primarily social or primarily semiotic – i.e. either primarily a process of interactive behaviour or one of exchanging meaning. To capture this, I have called this parameter SOCIO-SEMIOTIC PROCESS.

Later in the paper, he relates his socio-semiotic processes to the concept of ‘genre’ as used by Martin and others. Hasan (1999) includes ‘verbal action’ under Field, which is further specified with features such as ‘informing’, ‘narrating’, ‘instructing’, etc. (her earlier works had placed much of this under ‘rhetorical mode’, discussed below). As a component of Tenor: the model of Gregory and Carroll (1978) includes two subcomponents under Tenor: ‘functional tenor’ and ‘personal tenor’. ‘Functional tenor’ is close to ‘purpose’, being described as the category used to describe what language is being used for in the situation. Is the speaker trying to persuade? to exhort? to discipline? (Gregory and Carroll 1978:53)

Functional tenor was placed under tenor because it relates in effect to how the interactants relate to each other (as the persuader and persuaded, etc.), very external criteria. Note however that Gregory and Carroll state that ‘genre’ covers more than just functional tenor, involving also field, personal tenor, and mode: We prefer to characterise genre in terms of all the dimensions of language variety. Most significant literary genres, such as epic, ode, lyric, sonnet, tragedy, farce and comedy, have author/reader expectations as regards not only the medium relationship involved but also as regards the purposive roles/on-going social activities, and the personal and functional addressee relationships which are at risk, and so field and personal and functional tenors are likewise relevant to their description. Literary genres can be seen as individual kinds of marked registers within literature. (Gregory and Carroll 1978:44–5)

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

219

As a component of Mode: under Halliday’s later model, the closest thing to ‘genre’ is included as an element of Mode, called ‘rhetorical mode’ (Halliday 1978:63; Halliday and Hasan 1989:143; Halliday and Matthiessen 2006:332). IFG4 defines rhetorical mode as follows: Rhetorical mode encompasses a number of rhetorical categories concerned with the contribution of the text to the situation it operates in: informative, didactic, persuasive, exhortatory, pragmatic, and so on. (IFG4:38)

IFG4 notes that some rhetorical modes are oriented towards the Field of the text (e.g. informative, didactic, explanatory and explicatory contexts), and others towards the Tenor of the text (e.g. persuasive, exhortatory, hortatory, polemic contexts). As a fourth component: Ellis and Ure (Ellis 1965; Ellis and Ure 1969; Ure and Ellis 1977) have two distinct components in place of Halliday’s Tenor: ‘formality’ and ‘role’. Role corresponds roughly to genre, being defined as ‘the dimension correlating with the social or other role of the utterance or text, e.g., conversation, literature, technical writing, etc.’ (Ellis 1965:13).

8.3.3.2 ‘Genre’ as Realization of Context of Situation Hasan, like Halliday, does not often use the term ‘genre’ directly. Where she does, she states that, for her, ‘the term “genre” is a short form for the more elaborate phrase “genre-specific semantic potential” (Halliday and Hasan 1989:108). Here she is talking about the stylistic realization of genres. She considers the category of genre as superfluous, given that the association between a genre and the semantic patterns that realize it is already covered by the notion of register: ‘For most material purposes register and genre are synonymous’ (Hasan 1978:230) and ‘The terms “register” and “genre” as used here are then interchangeable’ (Hasan 1978:241). Since, in her model, register stratally realizes Context of Situation, Hasan thus places genre as below the Context of Situation. Here though, she is referring to genre as stylistic realization in terms of patterns of linguistic selection, and not to generic structure. For her, ‘generic structure’ is distinct from ‘genre’. She places generic structure as a realization of selections of the context of situation. More details of her approach will be given below. 8.3.3.3 ‘Genre’ Stratified above Context of Situation It should be apparent from the discussion above that ‘genre’ in the SFL model does not clearly sit within any one of the situational categories, and has been, at different times, placed in all three of the components. Additionally, even when placed in one component, it is still shown to have influence on the other situational components. When Martin was first teaching a ‘Functional Varieties’ course at the University of Sydney, he started off using Gregory’s ‘Functional Tenor’

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

220

MICK O’DONNELL

approach, but students found this difficult, as they had earlier been exposed to Halliday’s Rhetorical Mode approach (see Martin 2014). Class discussions explored the cross-component implications of genre, and two class members, Guenter Plum and Joan Rothery, suggested ‘positioning functional tenor as a deeper variable, since the purpose of a text influenced all of interpersonal, ideational and textual meaning’ (Martin 2014:12). The eventual result of this discussion, under Martin’s leadership, was to rename the deeper variable as ‘Genre’, recognizing it as a stratum separate from Context of Situation. ‘Personal Tenor’ could thus be renamed as simply ‘Tenor’. Genre in Martin’s approach is described in terms of both a system network (defining genres and their variants), and a layer of structure, such that choices in the genre network determine which schematic elements are realized in the text. In his model, the register of each stage is determined via interstratal realization: As part of the realisation process, generic choices would preselect field, mode and tenor options associated with particular elements of text structure. (Martin 1992:505)

For Martin it seems, Genre only interfaces with Context of Situation (which he calls Register), and does not interface directly with lower strata: Genre is a pattern of register patterns, register a pattern of discourse semantics ones, which are in turn a pattern of lexicogrammatical ones, in turn a pattern of phonological ones. (Martin 2014:14) Register [is] the expression plane of genre. (Martin 2014:13)

If this is so, then there must be some variables in the register which can pass on the linguistic constraints of the genre to the lower stratum. The linguistic patterns which were previously activated by functional tenor still need to be activated by some variables in the Context of Situation (Martin’s Register layer), and these variables need to be activated by the choice of Genre. The alternative is to allow selections in the Genre stratum to directly interface with each of the strata below: limiting the allowable contextual configurations, in addition to activating linguistic possibilities. Both approaches are viable, although each one has strong consequences for linguistic modelling, and there should be a clear statement as to which approach is being followed. Martin’s model of Genre over Register is fairly widely accepted within the educational side of SFL. His approach is not uncritically accepted however, particularly in regards to Hasan and those who follow her model (e.g. see Hasan 1995). She argues that by putting Genre outside the semiotic space, human interaction is de-humanized: My own view is that the stratification of genre and register, the collapsing of the social and the verbal, at both these planes, . . . has a highly deleterious

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

221

effect: It moves the whole issue of text structure and its activation from active, feeling, reacting participants co-engaged in some interaction to given forms of talk that represent the ways things are done in our culture, as if the culture is unchanging and as if the participants are simply pre-programmed. (Hasan 1995:283)

Martin responds that his Genre is in fact a semiotic system, open to human choice: A culture for me is a system of genres . . . genres do meaning, just like everything else in semiosis does . . . genres don’t realise the social, they are my theory of the social: we live genres. (Martin 2015, personal communication)

Lukin et al. (2011:189) put forward a different argument, that placing genre and register together provides a simpler analytical tool: As a central conceptual tool that does not stratify the relation of genre and register, Halliday’s notion of register helps us recognize – or at least frame and test – the idea that recognized social situations might sometimes be the same register, or identify and evaluate the register differences in what are normally counted as ‘the same’ social activities: it is a model well suited to calibrating the shuffling and reshuffling of cultural space-time and its boundaries.

Martin (1999:505), on the other hand, argues that placing genre within context of situation is just not feasible: It seems to us impossible to associate the accomplishment of genres as stated goal-oriented social process with any one metafunction (ideational, interpersonal or textual) or correlating register variable (field, tenor or mode). For us genre redounds simultaneously with field, tenor and mode, and thus with ideational, interpersonal and textual meanings.

8.3.4 Where Does Generic Structure Belong in the Model? For Martin, the issue of where generic structure belongs in his model is clear: selections from the Genre network are realized (in part) by creating a schematic structure at the Genre level (Martin 1992:505). Each schematic element then constrains the Context of Situation (his Register) by setting appropriate Field, Tenor, and Mode selections. For him then, generic structure determines contextual features, while for Halliday and Hasan, contextual features determine generic structure. For Halliday (1978:134), ‘Generic structure is outside the linguistic system; it is language as the projection of a higher level semiotic structure’, but ‘it can be brought within the general framework of the concept of register.’ Register for Halliday is not a stratum of the model, but rather a relationship between the Context of Situation and the linguistic strata: a

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

222

MICK O’DONNELL

register is the set of linguistic choices that recurrently occur in the texts produced in a given situation type. So, by placing generic structure within the framework of register, generic structures are thus linguistic patterns that realize particular situation types. One of the most likely places for generic structure in the linguistic model is on the Semantic stratum. The closest thing to genre in Halliday’s later model is ‘rhetorical mode’, which is a component of Mode. And because Halliday often states that Mode is in most cases realized through the Textual metafunction, we might infer that, for Halliday, generic structure is a part of the Textual component of the Semantic stratum, along with Information Structure and Thematic Structure. For Hasan also, generic structure is a realization of the Context of Situation: the structure of a text is determined by the selection of features from the Context of Situation network, which can predict the obligatory and the optional elements of a text’s structure as well as their sequence vis-à-vis each other and the possibility of their iteration. (Halliday and Hasan 1989:56)

She stresses that one cannot expect elements of the text structure to be determined by individual situational features, but that they are determined by the configuration of features selected from Field, Tenor, and Mode (what she calls a ‘contextual configuration’, or a CC): We need the notion of CC for talking about the structure of the text because it is the specific features of a CC . . . that permit statements about the text’s structure. We cannot work from the general notion of, say, ‘field’ since it is not possible to claim, for example, that field always leads to the appearance of this or that element. (Halliday and Hasan 1989:56)

Thus, in her model, generic structure is the realization of the Context of Situation. What is not clear from her work is whether this is intra-stratal or inter-stratal realization. If the first, then text structure would be seen as structure at the Context of Situation stratum. If the second, text structure would be stratally below Context, which would then place text structure on the Semantic stratum (as the Hallidayan model does not posit any stratum between Context and Semantics). There is evidence (although not very strong) that Hasan places generic structure stratally below Context of Situation, with register (and thus genre) linking them: In the SF model the concept of register is a ready-made link between context and generic structure. (Halliday and Hasan 1989:230)

Matthiessen (2015:10) follows Hasan, saying that generic structure is a realization of situational features:

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

223

Certain systemic terms in the systems of the socio-semiotic process (field of activity) network . . . have realization statements associated with them. These realization statements specify fragments of the structure of the situation type . . . For example, ‘non-sequential’ is realized by the presence of the contextual element of Phenomenon Identification (‘non-sequential’ & + Phenomenon Identification) . . . In other words, the kind of potential that is built into Hasan (1978, 1984) specifications of generic structure potential (GSP) is represented systemically here – which is, of course, in line with her theory of context.

The phrase ‘structure of the situation type’ implies that he takes generic structure to be a structure level of the context of situation (context of situation thus having system and structure specifications). He confirms this two pages later: The realization statements in Figure 6 are inter-axial but intra-stratal: that is, they relate paradigmatic order to syntagmatic order within the stratum of context. (Matthiessen 2015:12)

Contextual features can alternatively be realized directly as registerial constraints: But contextual realization statements can, of course, also refer to patterns below the stratum of context – inter-stratal realization statements. If contextual elements are realized linguistically, these patterns are semantic; that is, contextual elements are realized by patterns of meaning, as shown by Hasan, Ruqaiya (1984). (Matthiessen 2015:12)

8.3.5

How are Variations in Language across Generic Structure Explained? One problem for these approaches is to explain how patterns of linguistic choices are not constant across a text as a whole, but rather, change as the text shifts from stage to stage. Bateman (2008:185) explains this point: when we describe the linguistic details of texts in close detail, it is rare that an entire text exhibits precisely the same range of stylistic options. More often we can locate particular phases or segments of a text showing a relatively homogenous range of stylistic options and other segments of the same text that show different options being taken up. Therefore, a single linguistic text, or linguistic event, may appeal to several distinct registers while it is unfolding and yet still be seen as a coherent example of a single ‘type’ of text. . . . Since texts need not be homogeneous, simple ‘labels’ for registers or genres are rarely appropriate. . . . Each stage can take on a distinctive register.

For Martin’s model, this is not a problem, as each schematic element can be related directly to the register selections appropriate for that section. He says:

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

224

MICK O’DONNELL

Making genre rather than register variables responsible for generating schematic structure makes it easier to handle changes in experiential, interpersonal and textual meanings from [one] stage to another in a text . . . Underlying register, genre can be used to predict these changes, stage by stage, while at the same time accounting for a text’s overall coherence. (Martin 1992:506)

For the Halliday and Hasan models, this is more of a problem. For them, generic structure is created by the selected contextual configuration. But there is no mechanism to allow for the element of the generic structure to turn back and change the contextual configuration which would allow variation in linguistic selection, although Martin (personal communication) suggests that Hasan’s (2015) ‘ITERATION’ systems, which allow re-entry into the Field network using a recursive system, might work for this. However, in that article, the recursive system was applied to modelling situations with multiple fields, not to modelling the staging of texts/interactions.

8.3.6

Alternative Approaches: Phasal Analysis and Dynamic Context As discussed just above, in models where context is seen as constant over the text as a whole, it is difficult to account for the differing linguistic choices that occur over stages of a text. A solution to this problem is to allow for the context of situation to shift as the text unfolds. There has been various discussions of dynamicity in SFL over the years, mostly in respect to dialogic interaction, (e.g. see Hasan 1981; Ventola 1983, 1987; Halliday 1984; Martin 1985; Cloran 1987; O’Donnell 1990, 1999). Hasan (1981:118), for instance, says: When the context is co-operatively negotiated, the text and context evolve approximately concurrently, each successive message functioning as an input to the interactants’ definition of what is being achieved.

These words suggest that the Context of Situation is not constant over a given text or interaction, but can change as the text unfolds. Generic staging in a text can thus be seen as the result of a sequence of shifts in the context of situation, a change in what the participants are trying to achieve at each point of the text (e.g. from motivating a study to detailing that study). Each shift in Context of Situation is associated with a shift in the register of the text. One approach that takes this assumption most seriously is that of Phasal Analysis (Gregory and Malcom 1981; Malcolm 2010), which allows for phasal shifts in register throughout a text: Phase characterizes those stretches of text where there is a significant measure of consistency in what is being selected ideationally, interpersonally and textually . . . phase can be thought of as a delicate statement of register. (Gregory and Malcolm 1981:8)

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

225

Stillar (1992:105), another proponent of Phasal Analysis, expresses the idea that the apparent staging of a text can be viewed as the product of shifts in the context of situation: Discourse is necessarily the linguistic reflection of dynamic shifts in our recurrent relations to experience, interaction and medium – that is, interdependent ‘parts’ of a discourse can be viewed as the micro-registerial realization of instantial situations.

Phases differ from stages in that phases can overlap; for instance, we might have a phase of consistent ideational selections (e.g. talking about Lego), containing shifts in the interpersonal selections (change from monologue to dialogue). My own work on dynamic modelling of interaction (O’Donnell 1990, 1999; O’Donnell and Sefton 1995) takes a similar approach to the idea of context as dynamically mutable, but that work was not trying to explain the registerial shifts over stages. O’Donnell (2012) looked more deeply at dynamic shifts in tenor over a text. In dynamic approaches like these, we might do away with modelling generic structure as such, seeing the apparent staging as the result of the dynamic shifts in the Context of Situation, the interactants’ notion of what is going on. As the context shifts throughout the interaction/text, as a result, the register shifts as well. We do not need to posit text structure intermediate between context and text. We are left however with the problem of modelling the process of how shifts in the context of situation take place, both in dialogue and in written text. Cloran (1987) offers an interesting discussion of how contextual shifts in interaction can be negotiated by the participants. Much work is however needed to apply this dynamic context perspective to describe registerial shifts in written texts, such as are usually explained by generic structure.

8.3.7 Summary The problem of where Genre belongs in the SFL model stems from the seemingly circular relation between Genre and Context of Situation. On the one hand, we can say that the Context of Situation determines whether a given Genre (or generic structure) is appropriate or not, and more deeply, particular variants of a generic structure potential may be activated or deactivated by particular contextual features. On the other hand, each element of a generic structure is associated with distinct language patterns, so, to this extent, the stages of the genre determine the register used within them. In the traditional SFL approach, the Context of Situation, and thus Register, is seen as something constant over a text as a whole. And given the need to account for staging of language patterns over a text, the traditional approach thus needs to account for this staging outside of context. Hasan

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

226

MICK O’DONNELL

has taken one approach, placing the staging as a realization of the Context of Situation. But taking that approach makes it difficult to account for the distinct micro-registers of language use in each stage. Martin has chosen to place the notion of genre, and thus of staging, above that of Context of Situation, and thus allows each generic stage to reflect distinct registerial patterns. An alternative approach avoids the problem by rejecting the assumption that the Context of Situation is static, allowing for micro-shifts in register to result from a dynamically shifting context. This chapter has explored two areas of interest within SFL that are still unresolved, that of the degree of notionalism in determining grammatical categories, and also the exact role and nature of genre within the model. Both of these areas lead to active discussions on the SFL discussion lists, at conferences, and in publications. Often these discussions contain confusions where participants see the issue through the lens of their own assumptions, not aware of the underlying issues that lead their very words to mean different things to different readers. It is the hope of the author that, by bringing these underlying issues to the surface, future discussion will be less distracted by mistaken interpretations, and driven more through mutual understanding of the different sides of the issues.

References Bateman, J. 2008. Multimodality and Genre: A Foundation for the Systematic Analysis of Multimodal Documents. Basingstoke: Palgrave Macmillan. Biber, D. 1988. Variation across Speech and Writing. Cambridge: Cambridge University Press. Butler, C. 2003. Structure and Function: A Guide to Three Major StructuralFunctional Theories, Part 1: Approaches to the Simplex Clause. Amsterdam: John Benjamins. Cloran, C. 1987. Negotiating New Contexts in Conversation. Occasional Papers in Systemic Linguistics 1: 85–110. Ellis, J. 1965. Linguistic Society and Institutional Linguistics. Linguistics 3(19): 5–20. Ellis, J. and J. Ure. 1969. Language Varieties: Register. In A. R. Meetham, ed., Encyclopedia of Linguistics: Information and Control. Oxford: Pergamon. 251–9. Fawcett, R. 2009. Seven Problems to Beware of When Analyzing Processes and Participant Roles in Texts. In S. Slembrouck, M. Taverniers, and M. van Herreweghe, eds., Will to Well: Studies in Linguistics, Offered to Anne-Marie Simon-Vandenbergen. Ghent: Academia Press. 209–24. Gregory, M. and S. Carroll. 1978. Language and Situation: Language Varieties and Their Social Contexts. London: Routledge. Gregory, M. and K. Malcolm. 1981. Generic Situation and Discourse Phase: An Approach to the Analysis of Children’s Talk. Unpublished mimeo. Toronto.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

227

Gwilliams, L. and L. Fontaine. 2015. Indeterminacy in Process Type Classification. Functional Linguistics 2(8): 1–19. Halliday, M. A. K. 1965. Speech and Situation. English in Education 2(A2): 14–17. Halliday, M. A. K. 1970. Language Structure and Language Function. In J. Lyons, ed., New Horizons in Linguistics. Harmondsworth: Penguin. 140–65. Halliday, M. A. K. 1973. Explorations in the Functions of Language. London: Edward Arnold. Halliday, M. A. K. 1978. Language as Social Semiotic: The Social Interpretation of Language and Meaning. London: Edward Arnold. Halliday, M. A. K. 1984. Language as Code and Language as Behaviour: A Systemic Functional Interpretation of the Nature and Ontogenesis of Dialogue. In R. Fawcett, M. A. K. Halliday, S. Lamb, and A. Makkai, eds., The Semiotics of Culture and Language, Vol 2: Language and Other Semiotic Systems of Culture. London: Pinter. 3–35. Halliday, M. A. K. 1985. Introduction to Functional Grammar. London: Edward Arnold. Halliday, M. A. K. 1994. Introduction to Functional Grammar. 2nd ed. London: Edward Arnold. Halliday, M. A. K. and R. Hasan. 1989. Language, Context and Text: Aspects of Language in a Social-semiotic Perspective. 2nd ed. Oxford: Oxford University Press. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2004. Introduction to Functional Grammar. 3rd ed. London: Edward Arnold. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2006. Construing Experience through Meaning: A Language-based Approach to Cognition. London: Continuum. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Routledge. Hasan, R. 1978. Text in the Systemic-functional Model. In W. Dressler, ed., Current Trends in Text Linguistics. Berlin: Mouton de Gruyter. 228–46. Hasan, R. 1981. What’s Going On: A Dynamic View of Context. Seventh LACUS Forum. Columbia: Hornbeam Press. Hasan, R. 1995. The Conception of Context in Text. In P. Fries and M. Gregory, eds., Discourse in Society: Systemic Functional Perspectives. New York: Ablex. 183–284. Hasan, R. 1999. Speaking with Reference to Context. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 219–328. Hasan, R. 2015. Towards a Paradigmatic Description of Context: Systems, Metafunctions, and Semantics. Functional Linguistics 2(9): 1–54. Hudson, R. A. 1971. English Complex Sentences. Amsterdam: North Holland. Lee, D. Y. W. 2001. Genres, Registers, Text Types, Domains, and Styles Clarifying the Concepts and Navigating a Path through the BNC Jungle. Language Learning and Technology 5(3): 37–72. Available online at: https:// llt.msu.edu/vol5num3/lee. (Last accessed 15/05/2017.)

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

228

MICK O’DONNELL

Lukin, A., A. R. Moore, M. Herke, R. Wegener, and C. Wu. 2011. Halliday’s Model of Register Revisited and Explored. Linguistics and the Human Sciences 4(2): 187–213. Malcolm, K. 2010. Phasal Analysis: Analyzing Discourse through Communication Linguistics. London: Bloomsbury. Martin, J. R. 1984. Types of Writing in Infants and Primary School. In L. Unsworth, ed., Reading, Writing, Spelling: Proceedings of the Fifth Macarthur Reading/Language Symposium. Sydney: Macarthur Institute of Higher Education. 34–55. Martin, J. R. 1985. Process and Text: Two Aspects of Human Semiosis. In J. Benson and W. Greaves, eds., Systemic Perspectives on Discourse, Vol. 1. Norwood: Ablex. 248–74. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, J. R. 1999. Modelling Context: A Crooked Path of Progress in Contextual Linguistics. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 25–61. Martin, J. R., ed. 2013. Interviews with M. A. K. Halliday: Language Turned Back on Himself. London: Bloomsbury. Martin, J. R. 2014. Evolving Systemic Functional Linguistics: Beyond the Clause. Functional Linguistics 1(3): 1–24. Martin, J. R., C. M. I. M. Matthiessen, and C. Painter. 2010. Deploying Functional Grammar. Beijing: Commercial Press. Martin J. R. and D. Rose. 2003. Working with Discourse: Meaning beyond the Clause. London: Continuum. Martin J. R. and D. Rose. 2008. Genre Relations: Mapping Culture. Sheffield: Equinox. Matthiessen, C. M. I. M. 1995. Lexicogrammatical Cartography: English Systems. Tokyo: International Language Sciences Publishers. Matthiessen, C. M. I. M. 2015. Register in the Round: Registerial Cartography. Functional Linguistics 2(9): 1–49. Newmeyer, F. J. 1991. Functional Explanation in Linguistics and the Origins of Language. Language and Communication 11(1–2): 3–28. O’Donnell, M. 1990. A Dynamic Model of Exchange. Word 41(3): 293–328. O’Donnell, M. 1999. Context in Dynamic Modelling. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 63–99. O’Donnell, M. 2012. Tenor in a Dynamic Model of Context. Paper presented at Register and Context 2012. Macquarie University, 6–8 February 2012. Available online at: www.wagsoft.com/Presentations/ODONNELL-MAcquarieTenor2012.pdf. (Last accessed 15/05/2017.) O’Donnell, M. and P. Sefton. 1995. Modelling Telephonic Interaction: A Dynamic Approach. Interface: Journal of Applied Linguistics 10(1): 63–78. O’Donnell, M., M. Zappavigna, and C. Whitelaw. 2008. A Survey of Process Type Classification over Difficult Cases. In C. Jones and E. Ventola, eds.,

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

Continuing Issues in SFL

229

From Language to Multimodality: New Developments in the Study of Ideational Meaning. London: Continuum. 47–64. Paltridge, B. 1996. Genre, Text Type, and the Language Learning Classroom. ELT Journal 50(3): 237–43. Stillar, G. 1992. Phasal Analysis and Multiple Inheritance: An Appeal for Clarity. Carlton Papers in Applied Language Studies 9: 104–28. Swales, J. M. 1990. Genre Analysis: English in Academic and Research Settings. Cambridge: Cambridge University Press. Thompson, G. 2015. Pattern Grammar and Transitivity Analysis. In N. Groom, M. Charles, and S. John, eds., Corpora, Grammar and Discourse. Amsterdam: John Benjamins. 21–41. Tomlin, R. 1990. Functionalism in Second Language Acquisition. Studies in Second Language Acquisition 11: 155–77. Tucker, G. 2014. Process Types and Their Classification. In K. Kunz, E. Teich, S. Hansen-Schirra, S. Neumann, and P. Daut, eds., Caught in the Middle: Language Use and Translation. A Festschrift for Erich Steiner on the Occasion of his 60th Birthday. Saarbrücken: universaar, University of the Saarland. 401–16. Ure, J. N. and J. Ellis. 1977. Register in Descriptive Linguistics and Linguistic Sociology. In O. Uribe-Villegas, ed., Issues in Sociolinguistics. The Hague: Mouton de Gruyter. 197–243. Ventola, E. 1983. The Dynamics of Genre. Nottingham Linguistic Circular 13:103–23. Ventola, E. 1987. The Structure of Social Interaction: A Systemic Approach to the Semiotics of Service Encounters. London: Pinter.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:59, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.010

9 The Cardiff Model of Functional Syntax Anke Schulz and Lise Fontaine

9.1

Introduction

Within the theory of Systemic Functional Linguistics (SFL), there is, what Halliday refers to as the ‘powerhouse’ of the theory, the ‘central processing unit’ where ‘meanings are created’ (Halliday 1994:15). Despite this central role in the theory, there have been relatively few developments specifically related to the lexicogrammar since the 1980s and 1990s. There has been, however, one concerted effort to promote debate in this area and to suggest theoretical developments to the grammar by a team of scholars at Cardiff University (e.g. Fawcett 1980, 2000a, 2008a, 2017; Tucker 1998, 2017; Tench 1990, 1996, 2017). This chapter provides an overview of the Cardiff approach to syntax within the SFL framework. What we might refer to as the ‘standard’ model of SFL1 is described in various chapters in this volume, notably, Berry, Butt, Fontaine and Schönthal, Taverniers, and Webster, and is represented in various other chapters throughout the rest of the volume. There is also a very useful comparison between the more ‘standard’ model and the Cardiff model in Butler (this volume), but see also Butler (2003a, 2003b) and Butler and Gonzálvez-García (2014). While such comparisons are valid and important, the focus in this chapter is not on comparing the two models, although where appropriate, important issues are mentioned. The Cardiff model has its basis in SFL theory and, in particular, in Halliday’s earlier work. While many of the principles are shared between the two models, Butler (2003a:153) points out that ‘there are in the Cardiff account some important differences in the underlying goals, as well as extensions and simplifications of the grammar itself’. There have been several key concerns that have driven the model, and while a full 1

For ease of reference, the term ‘standard model’ will be used to refer to the more widely known model of grammar, e.g. Halliday and Matthiessen (2014) or any of the editions of IFG, Introduction to Functional Grammar (e.g. Halliday 1994 or Halliday and Matthiessen 2004), also sometimes referred to by some as the Sydney grammar.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:03:22, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.011

The Cardiff Model of Functional Syntax

231

consideration of all of them is beyond the scope of this chapter, we will focus on two in particular. One relates to how an SFL theory can or should account for a single representation (e.g. a clause) which expresses various different structures (e.g. mood, transitivity); this concerns the relationship between meaning and form, or the concept of instantiation in SFL. Another relates to the nature of choice and the place of semantics in the system network (see Fawcett 2013). The aim of this chapter is to examine these two concerns in terms of how they contribute to the Cardiff model of functional syntax as it applies to the English language. The chapter is organized as follows. In Section 9.2, we situate the Cardiff model in its historical roots, describing briefly how it developed. Section 9.3 provides the foundation for the functional syntax developed in the Cardiff model by outlining its main features, including the role of planners, system networks, and probabilities. Following this, in Section 9.4, we give a brief overview of the main grammatical units along with a discussion of the key concepts of filling, componence, and exponence. This also includes examples of the way in which transitivity and participant roles work in the Cardiff model and how the various strands of meaning are expressed in a single representation. In Section 9.5, we outline the Cardiff approach to clause analysis using a simple example. Finally, the chapter is summarized in Section 9.6.

9.2

A Brief History of the Cardiff Model

All SFL theory can be said to stem primarily from Halliday’s (1961) most important and influential article, ‘Categories of the Theory of Grammar’. Butler (2003a:153) acknowledges, as Fawcett himself does, that there are more similarities than differences between the two models. While the Cardiff model is very clearly rooted in Halliday’s early work, there are points of divergence that have shaped the path leading to the Cardiff model. Fawcett (2010:93) explains the different pathways as follows: developments in SFL theory, from early work in the 1960s, were shaped by Halliday’s involvement with the Penman project (see Matthiessen and Bateman 1991) and all the work that Halliday has done since then, whereas the Cardiff model shares the same roots but diverges slightly, being influenced by Hudson (1971) and work by Fawcett (1973, 1980). Fawcett considers Halliday’s 1970 paper, ‘Language as Choice in Social Contexts’, as resembling most the Cardiff Grammar (Fawcett 2010:93). The Cardiff model was also shaped by a computational implementation, the COMMUNAL project (e.g. Fawcett et al. 1993). Butler and GonzálvezGarcía (2014:49) suggest that, because of this, the model ‘offers a high level of explicitness’. However, as Fawcett states (2008a:13), both models share ‘the same historical roots and they still share essentially the same basic concepts’.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:03:22, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.011

232

ANKE SCHULZ AND LISE FONTAINE

Over the course of its development, Halliday’s language model has developed with an increasing focus on semantic phenomena (see O’Donnell, this volume). In fact, Fawcett (2010:10) claims that the more recent developments give no detailed account of the syntax and that the article ‘Categories of the Theory of Grammar’ (Halliday 1961) is the best and the only account of syntax given by Halliday or his colleagues. Fawcett’s first series of publications on this topic, ‘Some Proposals for Systemic Syntax, Parts 1–3’ (1974, 1975, 1976), was developed by trying to apply Halliday’s model to the analysis of text. The title of Fawcett’s 1980 book, Cognitive Linguistics and Social Interaction: Towards an Integrated Model of a Systemic Functional Grammar and the Other Components of an Interacting Mind, is an explicit indication of the different direction Fawcett was taking in developing the theory, i.e. that both social interaction and cognition were important to linguistic modelling.

9.3

Overview of the Model/Features of the Model

The theoretical framework of the Cardiff model has been described in many different publications but notably in Fawcett (1980, 2000a, 2008a), Tucker (1998), and Neale (2002), as well as in many articles and book chapters. It has been and is being developed around the world and in different languages, for example, in work on Chinese (He 2014), German (Schulz 2008, 2015), and Japanese (Funamoto 2014). For Fawcett (2000a:34), the basic relationship between meaning and form in any sign system can be described in terms of realization (i.e. meaning is realized by form). The relationship between meaning and form is illustrated in Figure 9.1, where the system networks are a components of the grammar, representing the semantic options available to speakers. The output of the networks is a set of selection expressions which then becomes the input to the realization rules; the realization rules and the potential structures are another component of the grammar, also expressed as potential.2 In this sense, as Butler (2003a:185) explains, ‘The level of form is also regarded as having a potential, consisting of realisation rules’. The output from this component is a layer of richly labelled tree structures. As shown in the diagram in Figure 9.1, there is a loop enabling this process to continue, where it is possible for a realization rule to state a re-entry rule, for example, when an element of a unit is ‘filled’ by another unit (this will be made clear in Section 9.4).

2

The relationship between meaning potential and instance is described by the concept of instantiation, which operates in a different dimension from that of realization. Although this is an important distinction, it will not be discussed in this chapter (see e.g. Wegener 2011).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:03:22, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.011

The Cardiff Model of Functional Syntax

233

Figure 9.1 The main components of a systemic functional grammar (e.g. Fawcett 2000a:36)

9.3.1 Cognitive and Social Aspects of the Cardiff Model For Halliday, language is primarily a social semiotic system; speakers use language to interact with others in culturally determined ways. In the Cardiff Grammar, although this view is shared, the theory is developed within a cognitive approach to language. Butler and Gonzálvez-García (2014) found that both models were very close together as compared to other theories in the wider functional-cognitive context since they were readily distinguishable from all other theories in their study, but they also found that the differences between them were largely due to the more cognitive orientation of the Cardiff model, e.g. it attempts in some way to model language production and understanding processes.

9.3.2 The Main Components of the Model The main components of the model are those that have been developed for the computational implementation of the theory as part of the COMMUNAL project. The computational model has been crucial in testing the language model, and although it is in some way attempting to model human cognition, it does not claim to provide an accurate description of cognitive behaviour. The main components of the generative (language production) aspect of the model are given in Figure 9.2. This diagram has been described in differing degrees of detail in the following publications: Fawcett (1980, 1993, 2013) and Fawcett et al. (1993). For the purposes of this presentation, only the production components are being considered, and so this simplified diagram leaves out detail concerning the language understanding components (see Fawcett (2013) for the complete diagram). The topmost components of the model concern the cognitive and sociolinguistic processes: the processes required to plan what to say and to guide the selection of lexical and grammatical units so that the output (e.g. sentence or clause) matches the speaker’s goals and intentions. The lower components, specifically the sentence planner, relate most directly to traditional linguistics (e.g. syntax and morphology). With input from the higher

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:03:22, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.011

234

ANKE SCHULZ AND LISE FONTAINE

Figure 9.2 Partial view of the main generative components of the COMMUNAL computer model

components, the sentence planner produces the best formal and semantic representation.

9.3.2.1 The Overall Planner The overall planner, the first and highest component, is a kind of ‘overseer’ which directs discourse planning and guides the system as a whole. This is the component that plans the propositional content in consultation with the belief system; in other words, this is where the earliest decisions are made in terms of what is going to be said or what is going to be talked about for a given proposition. As we can see from Figure 9.2, there is a two-way flow of information between the two; i.e. a relationship of consultancy between the overall planner and the belief system. The main difference between a planning component and the belief system is that the latter is static (although updatable); it contains objects. Planning components have a role to play in the entire process of language generation. The planners consult the belief system for various reasons to assist in the planning process, i.e. a kind of decision-making. The belief system is at the heart of both language generation and understanding. Although this is not shown here, various components of language understanding also need to consult the belief system. According to Fawcett (1994:78), it is object-oriented ‘in the sense that it consists of a vast number of specific objects and generic objects’. It is not a system in the dynamic sense of something that is operational; however, it is assumed that the

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:03:22, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.011

The Cardiff Model of Functional Syntax

235

objects it contains are organized systemically. As Fawcett explains (1990:164), the belief system includes ‘general and specific beliefs about (“knowledge of”) situations and things in some domain; specific beliefs about the content of the preceding discourse, about various aspects of the current social situation, about the addressee and his beliefs of all types, his attitudes, his goals and plans’. No theoretical distinction is being made here between knowledge, belief, and information, including general and specific beliefs about social situations, about the addressee, about the performeraddressee relationship. It also contains beliefs about register and the content of any relevant prior discourse. Finally, it holds ontological relations (e.g. knowledge about lexical relations). The output from the first stages of planning is a basic logical form (Lin 1993), which would specify, for example, the type of process involved, how many objects are involved, and the participant roles involved. For example, the logical form for a clause such as (1), taken from Fontaine (2008), would be event(event1[agent=object1, process=work]), which is basically describing the speaker’s intention to say something about something working. (1)

The chemo pills she takes are working

This basic logical form is then the input into the next stage of planning, which is called the microplanner. The microplanner is a component that handles various algorithms that guide the choices (the selection of options) in the system networks. The system networks represent the networks of systems of semantic options, not decision trees. This is a very important component, yet very little attention has been given to this area in SFL. It is broadly accepted in natural language generation that such a component is necessary. However, there are still many unanswered questions as to how it should work and what parts of the generation process belong in the microplanner and what parts belong elsewhere. The microplanner is simply a set of algorithms that determine the selection of options in the system networks; for example, the selection of Theme or verb tense.

9.3.2.2 The Discourse Planner The output from the microplanner feeds into the Discourse Planner, and, in through an enriched logical form, it feeds into the component for predetermination rules, which in turn is the input to the system networks in the lexicogrammar. With the output of the algorithms in the microplanner, it modifies planning from a higher level into plans that fit the more local discourse constraints of genre and exchange structure grammars. In a sense, it is this component that ensures the clause being generated will make sense or fit in with the ongoing discourse (Lin et al. 1993). The output is a discourse structure representation which, together with the enriched logical form, serves as the input into the system networks. At this point then, a good number of systemic selections have been made, for example, Theme, time/tense, the semantic requirements of any referring expressions.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:03:22, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.011

236

ANKE SCHULZ AND LISE FONTAINE

9.3.2.3 The Sentence Planner The final component considered here is the sentence planner. It contains a component for the lexicogrammar, which in turn contains two further components: the semantic system networks component and the realization component (see Figure 9.1). The semantic system networks are first traversed, and the output from this is a semantic representation (a collection of the semantic features selected, which is called a selection expression). This in turn results in a set of relevant realization rules. The output of the realization component is a rich formal representation: a fully labelled tree diagram representing semantic and formal elements (ignoring here the actual process of realization as speech). The lexicogrammar is therefore the component that generates the clause and produces the formal representation. The role of the system networks is to integrate the various decisions or selections made in the various higher components, for example, the microplanner. It is not a decision-making component but rather an integration and production mechanism. 9.3.2.4 The System Networks Both the Sydney and Cardiff system networks rely on the same basic notation and presentation. However, in the Cardiff model, there is ‘only one level of networks in the specification of the potential of a language at the level of meaning, rather than two (semantic, lexicogrammatical) in the Sydney grammar’ (Butler and Gonzálvez-García 2014:49). The realization rules are integral to the system network since they determine the relationship between the semantic options selected by the speaker and the formal representation realized (as speech or text, etc.). They are an essential part of the lexicogrammar even if they are rarely made explicit in most SFL writing. These rules or statements are basically instructions on how a particular meaning is realized. Tucker (1998:47) identifies four components of the realization rules: 1. 2. 3. 4.

rule number; network feature(s); any conditions on the rules; rule operations.

For example, in the small system network presented in Figure 9.3 below, if the feature [situation] is selected (as it would need to be in order to generate a clause), then as Fawcett (2008a:100) explains, the corresponding realization rule is to insert a clause and within the clause to insert a main verb (see Section 9.4). If the feature [information giver] is selected then the realization rule specifies that the Subject must be positioned before the Operator. Realization rules may be simple or complex. For example, the realization rule for the feature [thing] in the system network for thing, or referent-asthing, is given by Fawcett (1998) as the following:

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:03:22, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.011

The Cardiff Model of Functional Syntax

237

Figure 9.3 A very small systemic functional grammar for the English clause (Fawcett 2008a:93)

60: thing: if congruent_thing then ngp, if minor_relationship_with_thing then pgp. This example shows the rule number (60), the system feature ([thing]), and the operation (insert unit, i.e. ‘unit insertion rule’). This rule handles the difference between examples such as the woman and to the man as in the woman gave the ticket to the man. In the first case, the woman, rule 60 would insert a nominal group. In the second case, it would insert a prepositional group for to the man. A more complex rule will have conditions such as in the following example, which covers the realization of the system feature of [prediction] and [future time reference point] (Fawcett 2008b): 5: prediction or future-trp (time reference point): if not

negative then O < ‘will’,

if

negative then O < ‘wo’.

Rule (5) applies when either [prediction] or [future time reference point] has been selected. It also describes the conditions of realization dependent on whether or not the system feature [negative] has also been selected. If it has not been selected, the Operator will be expounded by

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 11 Aug 2019 at 10:03:22, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.011

238

ANKE SCHULZ AND LISE FONTAINE

will, and if it has, the Operator will be expounded by wo. This will be combined with the realization rule for [negative], which is expressed as follows (and also applies when the feature [confirmation seeker] is selected): 21: negative or confirmation-seeker: O ) can be included or interspersed in the following element. The superscript arrow signals that the element can be repeated indefinitely. Square brackets indicate the limits of any rule so that in the above notation the Placement (if present) can either precede the Initiating Event or be included within it (and no later element). The terms Initiating Event and Sequent Event in the GSP sketched out above may seem rather vague, but they are only as vague as is necessary to capture the range of variation possible at this point in superordinate terms, and this apparent vagueness is overcome in two possible ways. Firstly, in different genres the different elements may be specified with greater precision so as to reflect the rather more constrained meaning potential at that point, so that GSP of a shopping transaction, for example, would appear as in Figure 11.2 (Hasan 1996:56).

Figure 11.2 GSP of a Shopping Transaction (Hasan 1996:56)

This GSP has the added complication that while Sale Request and Sale Compliance necessarily appear in that order, they interact as a whole unit with other elements of structure (as signalled by the curly brackets). Secondly, in all cases the meaning potential for each stage is specified in terms of ‘its crucial semantic attributes’ (Hasan 1996:58) and from there to the range of lexicogrammatical patterns which potentially realize these. Returning to the Nursery Tale, one of the crucial semantic attributes of the element Placement is ‘person particularisation’, which will be realized lexicogrammatically by ‘indefinite modification . . . of an animate/quasi animate noun as Thing’ (Hasan 1996:62) – in other words, the Placement will include a nominal group such as three little pigs or a beautiful princess.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

293

The full description of a genre thus includes both the structure potential of the necessary and optional elements and the further specification of the semantic and lexicogrammatical attributes of these. And with regard to the context of situation which these genres realize, it can be stated that it is the least delicate options within the context that generate the overall GSP, while more delicate features will generate the semantic details of each element and hence the cohesive texture that runs throughout the discrete elements like the coloured threads in a tapestry. Thus, the contextual features shopkeeper and customer in a transaction would generate the structure in Figure 11.2, which would hold for all sales transactions (as generically defined), while the inclusion of further details within the contextual variable of field might specify that there will be continued reference to clothes, fabrics, sizes, and colouring across the text, and more delicate features within the contextual variable of tenor, such as familiarity or status, would generate more specific ways of making enquiries and requests. In this way the realization relationship between context and text is able to capture both important generalizations across instances of a genre type and the specific features of individual instances of that genre. For a fuller discussion of texture and structure, see Halliday and Hasan (1989: Part B). One potential problem with this approach is Hasan’s (1995:219) concept of relevant context, glossed as those aspects of the non-linguistic environment that are made relevant through language. In this formulation there is by definition a correlation between text and context, and this has led to discussions as to whether this pairing is the essence of a supervenient and non-essentialist conception of context or a circularity which ignores, or at least marginalizes, how extra-textual features affect text in less visible ways than direct inscription.7 I will return to these issues in Section 11.7, but see Bartlett (2017) and Moore (2017) for exchanges on this point.

11.3.5 Rhetorical Units At roughly the same time as research into the language of repeated and recognizable social activities was motivating the development of the concepts of register and genre, a rather different approach to the analysis of units above the clause was emerging from the applied work of Hasan (1989) and her collaborator Carmel Cloran (1994, 2010), focusing on regular variation in semantic patterning in mother–child talk in families from different socioeconomic backgrounds. This research was firmly rooted in Bernstein’s work on socialization and sought answers to a specific social problem (see my point in Section 11.1 regarding the contingent nature of 7

Recent work on multimodality opens this question up a little, as the borderline between material context and semiotic system becomes increasingly fuzzy. This is an area which would merit further theoretical discussion.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

294

TOM BARTLETT

developments): why children from working-class families were less successful at school than their middle-class counterparts. Hasan and Cloran’s work recategorized the two socioeconomic groups as low autonomy professions (LAP) and high autonomy professions (HAP) according to their degree of self-regulation in the workplace, and Cloran’s research considered the degree to which talk within the different families was either directly related to the immediate task in hand or moved more freely across a range of less immediate or even hypothetical topics. This categorization was based on Bernstein’s much misunderstood notion of restricted and elaborated codes and was seen as a cline from more to less heavily contextualized language. The different categories were based on the degree of remove of the talk from the material context in terms of both space and time, and stretches of talk representing each category, or Rhetorical Unit (RU), could extend for an indefinite number of clauses. Rhetorical Units are thus seen as textual units comprising one or more messages (see Section 11.3.1). At the most contextualized (ancillary) end of the cline of contextualization are Actions,8 talk which regulates the ongoing activity of the participants, and Commentaries, talk which describes this activity, as in examples (1) and (2) respectively: (1)

Here, you sit in Nana’s seat.

(2)

Do you want some passionfruit?

At the most decontextualized (constitutive) end of the cline are Generalizations and Conjectures, as in examples (3) and (4) respectively: (3)

Passion fruits usually come when it’s warm.

(4)

You might fall over, if you do, and spill my tea.

The full schema of RU categories and an indication of their place on the cline of contextualization are shown in Table 11.1 and Figure 11.3. Based on this categorization, Cloran (2010) concludes that families from LAP backgrounds tended to use more contextualized language than their HAP counterparts. Taking the analysis a step further, and drawing on Halliday and Hasan’s concept of a text as a cohesive stretch of talk or writing of any length, Cloran (2010) goes on to describe the linkage between RUs not only in terms of their semantic continuity, but also in terms of whether new RUs are either embedded in the previous ones or expansions of them. The distinction depends on whether the semantic content of a previous RU is picked up on in the Theme or Rheme of the first clause of the new RU. If it is picked up in the Theme, the relationship is said to be one of embedding, as the second RU would seem to serve a function within the overall purpose of the matrix RU, as in example (5). 8

Note the capital letters to show that these are technical metalinguistic terms, the names of which refer to similar lay metalinguistic categories.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

295

Table 11.1 Classification of rhetorical units (Cloran 2010) now here event proposal concurrent non-habitual habitual central enty interactant

acon

co-present person/object absent person/object generalised person/object

commentary

reflecon observaon

report

proposion prior forecast non-hypothecal hypothecal

account

volional recount

plan

nonvolional predicon

conjecture

predicon

generalisaon

Figure 11.3 The cline of (de)contextualization (Cloran 2010)

(5)

Mother: Stephen: Mother:

There aren’t many passionfruit out there at the moment Report Why? Because passionfruit come when it’s warm Generalization

In contrast, when the semantic content is picked up on in the Rheme, there appears to be a more significant switch of angle, as a new rhetorical purpose is introduced as an expansion upon a topic without a loss of cohesion, as in example (6): (6)

Mother:

It’s too cold for passionfruit now Generalization They don’t like the cold weather Do you think we should plant a passionfruit vine at our new Plan house?

What Cloran’s research showed is that while LAP families stayed on task with their talk, HAP mothers very often used highly contextualized talk as prompts for discussing less immediate or more abstract matters, as in example (5) (with example (6), conversely, taking the conversation back closer to home, though not the immediate task in hand). Cloran’s results, in terms of both the extent and the nature of decontextualized talk, therefore supported Bernstein’s claim that working-class children were less prepared for the decontextualized language of the classroom and that their relative lack of educational success was a result of socialization practices rather than a difference in intelligence.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

296

TOM BARTLETT

Hasan and Cloran’s work in this area is a departure from previous analytical approaches within SFL in a number of significant ways. Firstly, in relating the variation revealed to extra-linguistic social factors and the differences in socialization practices according to the parents’ position in the production of capital, RU analysis posits a causal relationship between the circumvenient context and the supervenient context. In contrast, where register theory, for example, accounts for variation, this is largely in terms of the function of the text as text, and so a linguistic-internal concept of context is in some way maintained. Secondly, while explorations of GSP looked for repeated patterns in language-activity pairings where the activity was strictly defined and the role of incidental surrounding language largely downplayed, RU analysis looks at all the linguistic behaviour accompanying non-linguistic activity and, in Cloran’s research at least, highlights the language that strays from the obvious task at hand. Thirdly, and relatedly, the structural relations between RUs are not hierarchical, not ‘predetermined’ by the ultimate goal of the activity and combining to realize distinct stages along the way to this goal; they are, rather, immanent, arising spontaneously from the ongoing talk and the non-linguistic activity and not necessarily leading in any particular direction. These are all points I shall return to below.

11.4

Berry’s Exchange Structure

In the development of both GSP and RU analysis, the importance of elements of structure above the clause but below the text is clear (leaving open the question of whether the relationship between the text and such elements is hierarchical or immanent). Developing such intermediate units was also the concern of work in Birmingham, where Sinclair and Coulthard (1975) were working on teacher–pupil interaction in the classroom, and Nottingham, where Margaret Berry (1981a, 1981b) was developing her ideas on exchange structure. Focusing on stretches of speech in terms of their function in interaction, Sinclair and Coulthard adopted a rather more structured and hierarchical approach than that developed for RUs, with discrete exchanges between speakers comprising turns from each speaker, and with each turn comprising one or more moves. Individual moves were labelled as initiating (I), responding (R), or feedback (F), for example: (7)

1 2 3 4

A: Have you finished your homework yet? B: No. What time is it? A: Ten o’clock. B: Ok, I’d better hurry up.

I R;I R F

This invented dialogue comprises a single exchange which is played out in four turns, two for each speaker. Turn 1 is a single initiating move,

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

297

whereas B’s turn in 2 is made up of two moves, a response to A and an initiating question. Turn 3 comprises a single move, a response to the question in Turn 2; and Turn 4 is an acknowledgment of that response, or feedback, which signals the closure of the exchange. Writing of her own work in developing exchange structure at this time, Berry (forthcoming) stresses the need for a complementary approach to the synoptic mapping of conventional systems networks in SFL, which are based on norms and tendencies across population sets, and describes exchange structure as a means of analyzing the dynamic and potentially unexpected movements of real-time discourse. Taking a metafunctional perspective, Berry developed interpersonal and experiential descriptions of exchanges to complement what she considered the textual labelling of Sinclair and Coulthard. As Berry (forthcoming) explains, interpersonal exchange structure is concerned with the roles that the interactants adopt in relation to the transmission of information or negotiating action (as with mood types), irrespective of who is initiating the exchange as text. The interpersonal roles are ‘primary knower’ (k1), the interactant assumed to know the information, and ‘secondary knower’ (k2), the interactant assumed not to know the information (though these roles may not reflect the reality of who knows what). Using Berry’s illustration below, we can see how the following examples share the same textual structure but different interpersonal structures, as signalled by the intonation patterns across the moves and the wording of the finishing move. In example (8), the quizmaster (primary knower) already knows the answer and is testing the knowledge of the contestant (secondary knower), while in example (9), the son (secondary knower) does not know the information and is seeking an answer from his father (who he assumes is primary knower). In example (8) the quizmaster defers transmitting the information, so their initial move is labelled dk1 and their second as k1. In providing the definitive information, the final move in example (8) contrasts with the final move in (9), which is merely a follow-up move by the newly enlightened secondary knower and so marked as k2f: (8)

Quizmaster: In England, which cathedral has the tallest spire Contestant: Salisbury Quizmaster: Yes

(9)

Son (doing crossword): Which English cathedral has the tallest spire Father: Salisbury Son: Oh, right

dk1 k2 k1

k2 k1 k2f

Experiential exchange structure concerns the establishing and development of propositional content, and individual moves include the propositional base (pb), the propositional completion (pc), and the propositional

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

298

TOM BARTLETT

support (ps). Adding these labels to example (9) above provides the complete metafunctional analysis of the exchange: (10)

Son (doing crossword): Which English cathedral has the tallest spire? Father: Salisbury Son: Oh, right

I k2 pb R k1 pc F k2f ps

Berry (forthcoming) sets out more delicate options within the three metafunctions, exploring how exchanges can depart from the smooth and narrow, and discussing the function and variability of side sequences. The complete framework developed therefore incorporates important insights from Conversation Analysis and Gricean pragmatics into an SFL framework, though there is not room here to expand on these ideas in depth. In this way Exchange Structure provides analysts with a more dynamic and emergent view of discourse as something that happens, and Berry (2016) provides a systems network that accounts for the different possibilities that arise at different points in the discourse. While the formalisms of Exchange Structure all relate to in-text relations, they provide a useful tool for discourse analysis. As Berry (2016) explains, labelling for all three metafunctions provides richer analysis of exchanges, often distinguishing superficially similar structures. Though the approach was developed within and is particularly suited to educational settings, it can be applied to a range of contexts to analyze, for example, how power is played out in various settings in terms of who introduces the propositional bases that delimit the scope of the conversation and who act as primary and secondary knowers within these contexts. And while the approach is based at one level on structure and hierarchy, the localized range of these hierarchies, which flow from one to the next without developing into superordinate structures, means the approach can also be used to analyze less structured genres, including casual conversation. In Sections 11.5 and 11.7 we will see contrasting approaches to discourse analysis in these terms: Rhetorical Structure Analysis, which works very much within the tradition of hierarchicization, combining units of everincreasing size to analyze whole texts as single structures, and Phase Analysis, which emphasizes the flow of texts across the metafunctions and their emergent properties.

11.5

Rhetorical Structure Theory

While Berry’s approach to Exchange Structure allows for turns of more than one move, this only occurs in a limited number of cases. As the name suggests, this framework focuses on the dynamics of interaction rather than the internal analysis of single-speaker stretches. This complementary approach is provided by Rhetorical Structure Theory (RST; Mann and

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

299

Thompson 1988; Mann and Matthiessen 1991; Mann et al. 1992), a method for analyzing the logical relations, both syntactic and functional, between individual messages within what might be classed as a single prolonged move in Berry’s terms. And while this approach could be extended to cover dialogic speech, it is therefore most appropriate to monologue. Mann et al. (1992:43–6) list the basic assumptions underlying RST, which are paraphrased and condensed here: a. Texts consist of functionally significant parts, combining as elements within patterns of text; b. As with Halliday and Hasan, ‘to be recognised as a text, the writing must create a sense of overall unity to which every part contributes’ (Mann et al. 1992:43); c. In contrast to Halliday and Hasan’s emphasis on cohesion in creating textuality, unity and coherence are a result of the imputed function of the individual elements and their contribution to ‘a single purpose of the writer’; d. Texts are hierarchically organized ‘such that elementary parts are composed into larger parts, which in turn are composed into yet larger parts’ (Mann et al. 1992:43). This contrasts with the more linear though equally goal-oriented approach within GSP (and Martin, see Section 11.6); e. The relational structure between elements is the same at every scale; f. The principal structural pattern is relational, with a small set of highly recurrent relations linking pairs of elements at all scales; g. In most cases the structural relationship between pairs of elements is asymmetrical, comprising a nucleus and a satellite; h. Relations are functional and can be stated in terms of the effects that they produce, in other words ‘the purposes of the writer, the writer’s assumptions about the reader, and certain propositional patterns in the subject matter of the text’ (Mann et al. 1992:45). In these terms, the relations are not between the words of the text (in contrast to the early cohesion approach), but the meanings and intentions behind each element, of which the wording is the realization; i. The number of relations is open and additional relations can be identified through investigation. Relations commonly used in RST include the following: (i) nucleus-satellite relations (i.e. where two elements at any scale are in an asymmetrical relation) such as evidence (for the proposition in the nucleus), concession, elaboration, condition, evaluation, antithesis, purpose, and summary; and (ii) multinuclear relations (i.e. where two or more elements at any scale share nuclear status) such as sequences and contrasts. Note that elements functioning as satellites at one scale have their own nucleus with potential satellites at the scale below. Relations between a nucleus and a satellite are defined in terms of the constraints on each and the effect ‘that plausibly the writer was trying to produce in employing the relation’ (Mann et al. 1992:48). For example, for

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

300

TOM BARTLETT

the evidence relation to be said to hold, the constraints are that the reader may not believe the proposition in the nucleus; that the reader is likely to find the satellite credible; and that the reader’s comprehension of the satellite will increase their belief in the nucleus. The effect therefore is that the reader’s belief of the nucleus is increased. For full examples of analyzed texts, see Mann and Matthiessen (1991) and Mann et al. (1992). Given the focus on the recurrent functional relationships between elements of a text at different scales, Hasan and Fries (1995:xxxiii) suggest that RST would be better labelled as ‘logical structure theory’. If we take this conclusion at face value, it could be claimed that RST provides the missing metafunctional link, adding a description of logical relations across texts to the experiential (lexical cohesion and cohesive harmony), the textual (RU analysis; Cohesion in English), and the interpersonal (message semantics and exchange structure, though these also consider aspects from the other metafunctions). In an extension of the original descriptions of RST that overlays the basic analysis with interpersonal and ideational features to provide a fuller representation of extended text that is reminiscent of Hasan’s conception of register as the accumulation of message semantics across whole texts, Mann and Thompson (1991) go as far as to suggest that a correlation between the relations of RST and all the metafunctions is robust. In terms of the text/discourse distinction proposed at the beginning of this chapter, while RST relations are based on predictions of authorial intention and favoured reading, these are all features that can be read off from the decontextualized text alone so that, despite the additional analytical features afforded by RST, it would have to be considered text rather than discourse analysis according to this definition. Webster et al. (2013) is an example of RST in analyzing political speeches, while Bateman (2008) utilizes a somewhat extended version of RST as one of the layers of description for multimodal static page-based documents.

11.6

Martin’s Discourse Semantics and Genre

Whereas the metafunctional correlates of RST were elaborated almost as an afterthought, the work of Jim Martin over almost four decades has focused on developing not just a model but a theory of text that incorporates metafunctional diversity as a defining principle within Martin’s conceptualization of Genre and Discourse Semantics (see Martin, this volume). As noted above, Martin (Martin in Anderson et al. 2015:53) critiques Halliday and Hasan’s approach to textuality and cohesion as being a ‘grammar and glue approach’ in which text is a by-product of cohesive relations between messages. Turning things around, and focusing on a process-based analysis rather than a productbased one,9 Martin sees the semantics of the text as a whole as paramount 9

Though Martin sees both perspectives as useful.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

301

(as one dimension of the realization of a genre), with meanings realized within the text in different ways: periodically and cumulatively across the text, as well as structurally within the confines of the individual clause or clause complex. Thus, rather than focusing on realizations of experiential, logical, interpersonal, and textual meanings at clause rank, Martin (1992) proposes the following metafunctional categories at the discourse semantic stratum: ideation, conjunction and continuity, negotiation (based largely on Berry’s exchange structure), and identification, with the further category of texture ‘interleaving’ the features of each together to create texts. In extended collaborations with Peter White and David Rose, Martin produced two further landmark books, Evaluation in English (Martin and White 2005) and Working with Discourse (Martin and Rose 2003, 2007). The first of these extends the discourse semantic treatment of interpersonal (including subjective) meaning to comprise three major systems – Attitude, Engagement, and Graduation – which consider attitudinal language, interspeaker/intertextual relations, and amplification/moderation respectively. There is no room here to discuss these categories in depth (but see Martin, this volume). In the second of these books, Martin and Rose update the analytical framework of Martin’s English Text (1992), re-presenting the approach from a more text-analytical-up and less theory-down perspective. The categories presented have now evolved into ideation, conjunction, appraisal (based on Martin and White’s work on attitude, engagement, and graduation), negotiation, identification, and periodicity (a revised version of texture). An interesting point worth dwelling on here is Martin’s (1992:249–64) critique of the hierarchical representation foundational to RST, which he sees as too product-oriented and especially at variance with ‘the dynamics of text as process, particularly in the spoken mode’ (Martin 1992:258). In contrast he proposes a linear dynamic which resonates with his characterization of genre as ‘a staged goal-oriented social process’ (Martin 1992:505). From this perspective, genres unfold as a sequence of necessary stages, semantically driven and realized through the metafunctionally diversified resources of discourse semantics, lexicogrammar, and phonology/graphology, to fulfil recognized and recognizable social activities. Much of the work of what has come to be termed the Sydney School, after Martin’s university, has been produced within action research projects aimed at extending the literacy skills of disadvantaged groups through a visible pedagogy that focuses on producing highly valued written work across a range of disciplines at all levels of the curriculum (though Martin is quick to emphasize this is just one application of the theory, which has a broader genesis). Within such a framework, a genre can be defined as situation/language pairings which have become socially recognizable through repeated association (a point I return to in Section 11.7), and Martin and Rose (2008) exemplifies this approach through the analysis into

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

302

TOM BARTLETT

named stages of genres across a range of disciplines. While the goal of this work is to make visible and hence replicable what is hidden to students and teachers alike, this apprenticeship approach has regularly invited criticisms of prescriptivism and acculturation within the oft-heated debate on minority rights in education. More recently, the work of Martin and his collaborators has moved into new fields, particularly restorative justice, and this has led not only to further socially motivated applications of the approach, but also to a continuing enrichment of the descriptive and theoretical power of their work. Bednarek and Martin (2010) is a strong testimony to the achievements of this prolonged collaborative labour between Martin and his research students, many of whom are now recognized figures in SFL. A particularly fruitful collaboration has been with the Bernsteinian sociologist Karl Maton, resulting in the 2014 volume Knowledge and Knowers: Towards a Realist Sociology of Education, which analyzes a range of disciplines across the sciences, social sciences, and humanities in terms of the kinds of knowledge and epistemic stances favoured and their transparency and hence their potential for transmission through visible pedagogies (see Maton 2014). One continuous thread in Martin’s work is his insistence on a supervenient model of context and his occasionally scathing rejection of ethnography, or at least what often passes as such in applied linguistic work, which he labels ‘ethNOgraphy’. Martin consistently takes the line that, within a social realist ontology such as SFL, culture is realized in text, via genres, semantics, lexicogrammar, and phonology/graphology and that non-linguistic information, including observation and the views of insiders, adds nothing to and potentially distorts our understanding of ‘what’s going on here’.10 In the following section, I take issue with Martin’s stance, drawing on the work of Michael Gregory and my own heavily contextualized approach, before suggesting a potential reconciliation.

11.7

YESnography: Gregory and Bartlett

The ‘dialect’ of SFL developed by Michael Gregory and colleagues, known as Communication Linguistics, takes a radically different stand from Martin with regard to ethnography and the status of the supervenient context in an overall architecture of discourse. In the opening paragraph of his 1995 paper, Gregory (1995:67) states that the goal of communication linguistics is to develop a model ‘from the systemic-functional tradition in particular, 10

Martin (personal communication) suggests the following alternative wording: ‘do not provide an answer to questions addressed in social semiotics analysis informed by SFL (including the multimodal analysis evolving out of SFL).’ I have decided to keep both my original wording and Martin’s alternative.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

303

ethnographic linguistics in general, and in terms of a dialectical materialist theory of knowledge’. The model of discourse presented (Gregory 1995:72) follows other SFL approaches in having a metafunctionally diverse semantic stratum of register (which would correspond to Discourse Semantics in Martin’s model) which is realized through specialized features of the lexicogrammar. Where it differs most significantly is that the stratum above the semantics is not the tripartite division of context into field, tenor, and mode as in Hasan’s and Martin’s models, which seek to characterize ‘what is going on through language here’ and therefore tend towards a supervenient approach, but rather a stratum of knowledge which represents the resources brought to bear in determining what discourses are possible within a particular environment, so tending towards a circumvenient approach (though see comments later in this section). The knowledge brought to bear on the situation is represented as a conjunctive binary system, with knowledge of the Community Communicating Context informing what generic activities are potentially available in the present situation, and the knowledge of Language: Dialect Configuration informing what registerial variables are available as resources in performing these. There is thus a crucial difference between Gregory’s model and both Hasan’s and Martin’s. In Hasan’s model, the contextual configuration of field, tenor, and mode activates certain meanings as at risk (i.e. potentially available) within a situation; in Martin’s model, the genre being performed carries out the same role. For both these models, therefore, description starts with the activity underway. Gregory’s model goes one stage further back, however, in stating that there is a range of contextual configurations or generic situations at risk in any given environment as a function of the knowledge the participants bring to bear within that environment (Gregory 1995:71). In order to describe and account for discourses, then, it is necessary to account not only for ‘the linguistic items that occur in them, but also the relations they enter into with each other and with the knowledge of users and receivers of the language’ (Gregory 1995:69, emphasis added). Important to note is that the knowledge brought to bear is not considered in any mentalist or idealist sense, but ‘as viable knowledge, knowledge as function, as social fact’ (Gregory 1995:69), a generalized knowledge of recurrent situations and the potential for action within them that has been internalized in the Vygotskian sense. I take this to mean that the knowledge brought to bear in a communicating environment can be considered as the distillation of the semiotic histories of the various participants, histories which act on the text but cannot be read off it. Discourse analysis therefore demands an ethnographic approach in order to provide an understanding of the rich backdrop of past meaning-making that is woven into the text as action, and Gregory’s approach aims to account for texts as instances of discourse which orient to yet manipulate general tendencies (note the title of

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

304

TOM BARTLETT

Gregory’s paper), rather than accounting solely for tendencies in terms of regular language-internal features.11 One important feature which Communication Linguistics does share with the Martinian approach to genre is that texts are seen as dynamic and linear rather than hierarchically ordered. However, where in Martin’s model stages are primarily characterized in terms of their discourse function, for Gregory a discourse progresses through phases displaying metafunctional consistency, with a change in meaning in any of the metafunctions resulting in a new phase. Ultimately, this difference between the two approaches might prove to be no more than a difference in perspective, given the basic SFL tenet that form and function are two sides of the same coin, as captured when Gregory (1995:71) describes textual progression across distinct phases as reflecting alternatively ‘the dynamic instantiation of microregisters’ (i.e. real-time shifts in meaning patterns) or the ‘microinstantial situations of the discourse’ (i.e. real-time shifts in activity). However, while both Martin’s approach and Gregory’s Communication Linguistics emphasize the dynamics of discourse, both also focus on generic situations and their realization as texts, thereby suggesting a closure that is in some way predetermined. Relatedly, both models rely at some level on a concept of shared histories within a single overarching context of culture: Martin (Martin and Rose 2008) describes a culture as the sum of genres available within it, while Gregory (1995:71) states that ‘[a]ny particular communicating community context is characterizable in terms of the generic situations which are potential within it’. My own work (Bartlett 2012; 2017) takes a rather different perspective in not taking generic histories for granted, deriving as it does from novel contexts of intercultural discourse between the Amerindian communities of the North Rupununi Savannahs in Guyana and the international development workers with whom they negotiate. Adopting Gregory’s notion of phase, I base my analysis on the different metafunctional configurations within and across phases and the ways in which these instantiate situation/language pairings that can be related to the cultural background of the different groups involved and the distinct semiotic histories and ways of speaking they bring to bear on the situation. Rather than looking to identify genres, therefore, I was keen to see the following: the extent to which the different voices of the groups involved, as represented by these situation/language pairings, were realized in the intercultural discourse; the degree to which these different voices were legitimated; and the resultant emergence of new ways of talking – or hybrid voices. In such novel and evolving instances, including institutional talk, it would be wrong to talk of genres, a term which assumes social recognizability through repeated association, and better to focus on recognizable voices which are intertwined in novel ways according to the 11

More recent work in the Martinian tradition has acknowledged the emphasis on the system rather than the instance in textual analysis and is working to develop better descriptions of individual texts as instances.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

305

dynamics of the situation. If these novel ways of speaking become standardized within this particular social environment, only then will it be possible to talk of new genres. Given this focus, an emphasis on ethnography is essential both to relate the different ways of speaking of the different groups to their distinct semiotic histories and to explain under what conditions a particular discourse was actualized from the potential afforded by the environment. However, rather than characterizing such an approach as circumvenient, with the suggestion of a dualist approach to context and semiotics, this view is better described as scalar supervenience in that the nontextual background features that motivate and inflect language in use are themselves the result of past semiosis which are reactivated as relevant to specific contexts in different ways for different speakers. My own approach, therefore, differs from those of Hasan, Martin, and Gregory, in taking neither a single context of culture nor the idea that all discourse is ‘genred’ for granted, while adhering to the core SFL conception of language as a stratified, metafunctionally diverse social semiotic system.

11.8

Eggins and Slade on Casual Conversation

If my own work questions whether institutional talk can always be considered generic, Eggins and Slade’s (1997:6) landmark Analysing Casual Conversation appears to come from the opposite direction in suggesting that, ‘despite its sometimes aimless appearance and apparently trivial content, casual conversation is, in fact, a highly structured, functionally motivated, semantic activity’. Eggins and Slade (1997:7) argue that such functional structure is lost in other work on casual conversation, in which ‘analysis has frequently been fragmentary, dealing only with selected features’ and ‘has not sought to explore the connections between the “social work” achieved through the micro-interactions of everyday life and the macrosocial world within which conversations take place’. Combining the approaches of Halliday and Hasan and those developed by Martin and his colleagues with other traditions such as conversation analysis and narrative analysis from outside SFL, the authors identify four types of patterning that occur at different levels of language and that ‘interact to produce the meanings of casual talk’ (Eggins and Slade 1997:7): • • • •

grammatical patterns at the clause level which indicate power and subordination within interaction; semantic patterns which indicate frequency of contact and familiarity between interactants; conversational structural patterns which indicate affective involvement and shifting alignments within conversation; the use of text types which give some indication of shared worldviews about normality and predictability. (Eggins and Slade 1997:18)

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

306

TOM BARTLETT

These different patterns occur in ‘chunk and chat segments’ in casual conversation (Eggins and Slade 1997:227), characterized by frequent exchanges of turn and domination by a single speaker respectively. Both of these structural types have a social function to play, with the generically recognizable chunks tending to index shared or opposing group values at a fairly broad level, while the more fluid chat segments do the hard interpersonal labour of signalling and negotiating status and familiarity between individuals. Taking these two aspects together, Eggins and Slade (1997:22) conclude that the primary function of casual conversation is to negotiate the ‘tension between, on the one hand, establishing solidarity through the confirmation of similarities, and, on the other, asserting autonomy through the exploration of differences’. Casual conversation thus stands in contrast to ‘pragmatic conversation’ (Eggins and Slade 1997:19), which is motivated by a clear pragmatic purpose through which the different interactants achieve complementary goals (such as buying or selling). Following Hasan’s distinction, in Section 11.3.4, it would appear that pragmatic conversation is primarily structured according to the recognizable social roles of the different interactants at a fairly indelicate degree of differentiation, while casual conversation is about negotiating the more delicate aspects of the context in the absence of, or temporary suspension of, such defining roles. Returning to the comparison made with my own approach at the beginning of this section, I would suggest that my own research has focused on contexts in which such defining roles, and the generically structured talk that accompanies them, are yet to be firmly established, so that much of the work that goes on in embryonic institutional contexts relies on the patternings of meaning displayed by casual conversation.

11.9

Matthiessen’s Appliable Linguistic Analysis and Concluding Remarks

Whereas many of the approaches to discourse analysis outlined so far have focused on specific areas of language and/or society as meriting particular attention, Matthiessen (2014) calls for an ‘appliable discourse analysis’ (ADA) that can serve as a universal resource in responding to social problems and issues where language in use is a significant element. The use of this label is an explicit reference to Halliday’s notion of an ‘appliable linguistics’ as a functionally based theory of language that can be applied across contexts rather than being applicable in certain cases. For Matthiessen (2014:147), ‘ADA corresponds to Appliable Language Description, the two being complementary aspects of appliable linguistics’. As such, ADA depends on full functional descriptions of the language under study in order that the correspondence between specific contexts of situation and the features of language in use within them can be located within the system as a whole

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

307

and hence related in a motivated way to other instances of use. This is of course a huge undertaking, and one that Matthiessen sees as best solved through ‘long-term collective discourse analysis designed to address problems that are beyond the scope of individual research projects’ (Matthiessen 2014:148). There is not the space to discuss the details of this approach in the present chapter, but the essential properties outlined by Matthiessen (2014:147–8) provide a flavour of the scope, objectives, and logic of ADA: • • • • • •

It must be grounded in a holistic theory of language in context; It must reference a comprehensive description of the particular language in culture that is in focus; It must be reasonably explicit so that manual analysis can easily be related to automated analysis; It must be multilingually and multisemiotically oriented; It must provide an account of the context in which the analysis was undertaken in order to reason about the analytical choices made; It must be geared towards data sharing and reuse and the goal of longterm collaborative discourse analysis.

From my own perspective, Matthiessen’s approach is problematical in that, despite the huge coverage called for, the description of context for each language event is still linguistically determined (2014:168). It could, however, be argued that in calling for the analysis of instances to be situated within a holistic description of the entire discourse system, then the whole of context as remembered and partially shared semiosis is accounted for. In this way, ADA shares many of the goals of Communication Linguistics and can be related to my own perspective of scalar supervenience. However, I think that the level of description called for in ADA is neither possible in practice nor representative of what individual speakers bring to bear to individual instances of real life. As with Hasan’s work on register and Martin’s work on genre, it is an approach best suited to uncovering tendencies across instances rather than accounting for the specifics of those instances: not only what is made relevant by text, and what is not, but also the locally contingent reasons for and effects of such ‘choices’. It is the purview of ethnography to explore techniques for considering what is relevant to whom and when, and there is surely room for integrating ethnographic techniques in a happy compromise with the level of systematic description proposed for ADA.

References Andersen, T. H., M. Boeris, E. Maagerø, and E. S. Tønnessen. 2015. Social Semiotics: Key Figures, New Directions. London: Routledge. Bartlett, T. 2012. Hybrid Voices and Collaborative Change: Contextualising Positive Discourse Analysis. London: Routledge.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

308

TOM BARTLETT

Bartlett, T. 2013. ‘I’ll Manage the Context’: Context, Environment and the Potential for Institutional Change. In L. Fontaine, T. Bartlett, and G. O’Grady, eds., Systemic Functional Linguistics: Exploring Choice. Cambridge: Cambridge University Press. 342–64. Bartlett, T. 2017. Context in Systemic Functional Linguistics: Towards Scalar Supervenience? In T. Bartlett and G. O’Grady, eds., The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 375–90. Bateman, J. A. 2008. Multimodality and Genre: A Foundation for the Systematic Analysis of Multimodal Documents. London: Palgrave Macmillan. Bednarek, M. and J. R. Martin. 2010. New Discourse on Language: Functional Perspectives on Multimodality, Identity, and Affiliation. London: Continuum. Berry, M. 1981a. Systemic Linguistics and Discourse Analysis: A Multilayered Approach to Exchange Structure. In M. Coulthard and M. Montgomery, eds., Studies in Discourse Analysis. London: Routledge. 120–45. Berry, M. 1981b. Towards Layers of Exchange Structure for Directive Exchanges. Network 2: 23–32. Berry, M. 2016. Dynamism in Exchange Structure. English Text Construction 9(1): 33–55. Butt, D. G., A. R. Moore, C. Henderson-Brooks, R. Meares, and J. Haliburn. 2007. Dissociation, Relatedness and ‘Cohesive Harmony’: A Linguistic Measure of Degrees of ‘Fragmentation’? Linguistics and the Human Sciences 3(3): 263–93. Cloran, C. 1994. Rhetorical Units and Decontextualisation: An Enquiry into Some Relations of Context, Meaning and Grammar. Monographs in Systemic Linguistics 6. Nottingham: University of Nottingham. Cloran, C. 2010. Rhetorical Unit Analysis and Bakhtin’s Chronotope. Functions of Language 17(1): 29–70. Eggins, S. and D. Slade. 1997. Analysing Casual Conversation. London: Cassell. Gregory, M. 1995. Generic Expectancies and Discoursal Surprises: John Donne’s The Good Morrow. In P. H. Fries and M. Gregory, eds., Discourse in Society: Systemic Functional Perspectives. Meaning and Choice in Language: Studies for Michael Halliday. London: Ablex. 67–84. Halliday, M. A. K. 1964. Descriptive Linguistics in Literary Studies. In A. Duthie, ed., English Studies Today: Third Series. Edinburgh: Edinburgh University Press. 23–9. Halliday, M. A. K. 1975. Learning How to Mean: Explorations in the Development of Language. London: Edward Arnold. Halliday, M. A. K. 1978. Language as Social Semiotic: The Social Interpretation of Language and Meaning. Baltimore: University Park Press. Halliday, M. A. K. and R. Hasan. 1976. Cohesion in English. London: Longman. Halliday, M. A. K. and R. Hasan. 1989. Language, Context and Text. 2nd ed. Oxford: Oxford University Press. Hasan, R. 1984. Coherence and Cohesive Harmony. In J. Flood, ed., Understanding Reading Comprehension: Cognition, Language and the Structure of Prose. Newark: International Reading Association. 181–219.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

Models of Discourse in SFL

309

Hasan, R. 1989. Semantic Variation and Sociolinguistics. Australian Journal of Linguistics 9: 221–75. Hasan, R. 1995. The Conception of Context in Text. In P. Fries and M. Gregory, eds., Discourse in Society: Systemic Functional Perspectives. Meaning and Choice in Language: Studies for Michael Halliday. London: Ablex. 183–284. Hasan, R. 1996. The Nursery Tale as Genre. In C. Cloran, D. Butt, and G. Williams, eds., Ways of Saying: Ways of Meaning. London: Cassell. 51–72. Hasan, R. 2009. Wanted: A Theory for Integrated Sociolinguistics. In J. J. Webster, ed., Collected Works of Ruqaiya Hasan,Volume 2: Semantic Variation: Meaning in Society and in Sociolinguistics. Sheffield: Equinox. 5–40. Hasan, R. 2014. Towards a Paradigmatic Description of Context: Systems, Metafunctions and Semantics. Functional Linguistics 1(9): 1–54. Hasan, R. 2016. Collected Works of Ruqaiya Hasan, Volume 2: Context in the System and Process of Language. Edited by J. J. Webster. Sheffield: Equinox. Hasan, R. and P. Fries. 1995. Reflections on Subject and Theme: An Introduction. In R. Hasan and P. Fries, eds., On Subject and Theme: A Discourse Functional Perspective. Amsterdam: John Benjamins. xiii–xlv. Hasan, R., C. Cloran, G. Williams, and A. Lukin. 2007. Semantic Networks: The Description of Linguistic Meaning in SFL. In R. Hasan, C. M. I. M. Matthiessen, and J. J. Webster, eds., Continuing Discourse on Language: A Functional Perspective, Volume 2. Sheffield: Equinox. 697–738. Lukin, A. 2012. Meanings in Questions: A Case Study of the ABC’s Current Affairs Coverage of the 2003 Invasion of Iraq. In T. Bartlett and H. Chen, eds., Special Issue of Journal of Applied Linguistics and Professional Practice 9(1): 424–44. Lukin, A. 2015. Language and Society, Context and Text: The Contributions of Ruqaiya Hasan. In W. Bowcher and J. Y. Liang, eds., Essays in Honour of Ruqaiya Hasan: Society in Language, Language in Society. London: Palgrave. 143–65. Lukin, A., A. R. Moore, M. Herke, R. Wegener, and C. Wu. 2011. Halliday’s Model of Register Revisited and Explored. Linguistics and the Human Sciences 4(2): 187–213. Mann, W. C. and C. M. I. M. Matthiessen. 1991. Functions of Language in Two Frameworks. Word 42(3): 231–49. Mann, W. C. and S. A. Thompson. 1988. Rhetorical Structure Theory: Towards a Functional Theory of Text Organisation. Text (8)3: 243–81. Mann, W. C., C. M. I. M. Matthiessen, and S. A. Thompson. 1992. Rhetorical Structure Theory and Text Analysis. In W. C. Mann and S. A. Thompson, eds., Discourse Description: Diverse Linguistic Analyses of a Fund-raising Text. Amsterdam: John Benjamins. 39–78. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, J. R. 1999. Modelling Context: A Crooked Path of Progress in Contextual Linguistics. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 25–61.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

310

TOM BARTLETT

Martin, J. R. 2010. Semantic Variation: Modelling Realisation, Instantiation and Individuation in Social Semiosis. In M. Bednarek and J. R. Martin, eds., New Discourse on Language: Functional Perspectives on Multimodality, Identity and Affiliation. London: Continuum. 1–34. Martin, J. R. and D. Rose. 2003. Working with Discourse: Meaning beyond the Clause. London: Bloomsbury. Martin, J. R. and D. Rose. 2007. Working with Discourse: Meaning beyond the Clause. 2nd ed. London: Bloomsbury. Martin, J. R. and D. Rose. 2008. Genre Relations: Mapping Culture. Sheffield: Equinox. Martin, J. R. and P. R. R. White. 2005. The Language of Evaluation: Appraisal in English. London: Palgrave Macmillan. Maton, K. 2014. Knowledge and Knowers: Towards a Realist Sociology of Education. London: Routledge. Matthiessen, C. M. I. M. 2014. Appliable Discourse Analysis. In F. Yan and J. J. Webster, eds., Developing Systemic Functional Linguistics: Theory and Application. Sheffield: Equinox. 138–208. Moore, A. R. 2017. Register Analysis in Systemic Functional Linguistics. In T. Bartlett and G. O’Grady, eds., 2017. The Routledge Handbook of Systemic Functional Linguistics. London: Routledge. 418–37. Scott, C. 2010. Peace and Cohesive Harmony: A Diachronic Investigation of Structure and Texture in ‘End of War’ News Reports in The Sydney Morning Herald. In F. Yan and C. Wu, eds., Challenges to Systemic Functional Linguistics: Theory and Practice. Proceedings of the 36th ISFC. Beijing: ISFC Organising Committee. 89–96. Sinclair, J. and R. M. Coulthard. 1975. Towards an Analysis of Discourse: The English Used by Teachers and Pupils. London: Oxford University Press. Vygotsky, L. S. 1978. The Mind in Society: The Development of Higher Psychological Processes. Edited by M. Cole, V. John-Steiner, S. Scribner, and E. Souberman. Cambridge, MA: Harvard University Press. Webster, J. J., J. Chan, V. Yan, and K. Wong. 2013. Visualizing the Architecture and Texture of a Text: A Case Study of Selected Speeches of US President Barack Obama. In F. Shi and G. Peng, eds., Festschrift in Honour of Prof. William S-Y. Wang’s 80th birthday. Hong Kong: City University of Hong Kong Press. 301–24. Widdowson, H. G. 2007. Text, Context, Pretext: Critical Issues in Discourse Analysis. Malden: Blackwell.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 20 Aug 2019 at 13:23:01, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.013

12 Cohesion and Conjunction Maite Taboada

12.1

Texture as the Weaving of Cohesion and Conjunction

Most texts we encounter on an everyday basis are coherent. They make sense in the situation in which they are presented, and their meaning and intention can usually be understood. Even in cases where the communication may be more difficult (a ‘different’ accent; spelling mistakes; complex argumentative structure), we tend to accept texts as being coherent, and make an effort to grasp their meaning. Coherence is such a fundamental property of texts and of our communication that it is difficult to conceive of completely incoherent texts. Consider the two invented examples in (1) and (2). In the first case, we have a set of sentences, each connected to the previous one through a lexical item. This is an instance of sets of cohesive links, with items such as last night – at night, which may have a semantic relation of repetition in most texts, but which do not here. The passage, however, does not seem to have a common thread; it is not coherent. Conversely, the two sentences in (2) are coherent in terms of a thread (dark clouds – rained), but the sentences are not well related, because the conjunction however does not usually relate two units in this way. It sets up an unfulfilled expectation, or one contrary to expectation, but rain following dark clouds is actually not contrary to expectation. Example (2) is coherent, but fails in the way that coherence is made explicit through the conjunction however. (1)

I went home very late last night. At night, owls come out and hunt. Harry Potter uses an owl to have his mail delivered. The mail was very erratic over the Christmas holidays. The holidays were too short, and short indeed is this paragraph.

(2)

There were dark clouds in the sky today. However, it rained.

I would like to thank Geoff Thompson, for inviting me to contribute this chapter, and for providing very insightful comments shortly before his untimely death.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

312

MAITE TABOADA

Cohesive and conjunctive elements constitute the fundamental property of texts, the property of texture (Halliday and Hasan 1976; Hasan 1985). Texture is the quality that makes a particular set of words and sentences a text; what holds them together to give them unity, in the same way that the weaving of warp and weft create a piece of cloth, a textile (which has the same root as text and texture). Texture is created through relationships of choice: the speaker or writer chose to use those words and sentences over other choices, to make the text meaningful in context. Traditionally, and since Halliday and Hasan (1976), texture has been characterized as the interaction between coherence, how the text relates to the context outside the text, and cohesion, how the elements in the text itself contribute to making it a unified whole. Coherence, then, enters into the realm of intentions (what we want to achieve with the text), and the representation of the world through propositions and their connections. Cohesion is more local to the text, and includes links among words (such as dark clouds – rain in example (2) above). I suggest that coherence and conjunction form a separate system, distinct from cohesion proper (see Section 12.5). But before that, let us examine the traditional organization of cohesion, and examples of cohesive elements.

12.2

Cohesive Devices

A ‘cohesive device’ is an element in the text that requires another element for its interpretation. The relation between the two is a ‘cohesive tie’. The first classification of cohesive devices, in Halliday and Hasan (1976), proposed a classification along the following lines: 1. 2. 3. 4. 5.

Reference Substitution Ellipsis Conjunction Lexical cohesion

A more abstract classification lists the first three types as instances of grammatical cohesion, i.e. types which are realized through the grammatical system of the language, as opposed to conjunction and lexical cohesion, which rely on the lexis to achieve cohesion. The most recent Introduction to Functional Grammar (Halliday and Matthiessen 2014: Chapter 9) lists four systems of cohesion: 1. 2. 3. 4.

Conjunction Reference Substitution and ellipsis Lexical organization

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

313

Conjunction includes conjunction and continuity (Theme-Rheme relations, in particular, textual Theme, see Berry, this volume). The difference between conjunction and the other systems is that conjunction links whole clauses or combinations of clauses, whereas reference creates cohesion by creating links between elements. Elements may be referents (persons, things) or facts, including a proposition or a whole passage of text captured as a fact (Halliday and Matthiessen 2004: Section 9.2, Paragraph 7). Substitution and ellipsis create links between referents by replacing or leaving out referents that can be easily recovered in the context. Lexical organization also creates cohesion between referents, but by using exclusively lexical resources, as opposed to reference, substitution, and ellipsis, which enlist all of the resources of the lexicogrammar to do so. The rest of this section briefly examines the types proposed by Halliday and Matthiessen, with illustrative examples. Unless otherwise indicated, examples are extracted from informal reviews, posted online on the Internet Movie Database website.1 They are reviews of Spirited Away, a 2001 animated film by Japanese filmmaker Hayao Miyazaki. The film has been characterized as one of the top 100 movies of all time, and online reviewers in IMDB rate it very highly. The reviews are reproduced verbatim, and may include typos and unorthodox grammatical constructions.

12.2.1 Conjunction Halliday and Hasan (1976) initially divided cohesion into two types: grammatical and lexical. Conjunction was listed under the lexical label, because it makes use of lexical items, i.e. conjunctions such as and, but, or if. Halliday and Matthiessen (2014) move it out of the lexical realm, and present it as the first element in the list of cohesive resources. This is understandable, because, although it deploys lexical items, the particular items are function words, i.e. not open-class items such as nouns or verbs. In example (3), we see how the two clauses in the first sentence are joined by the conjunction but. Conjunction also links sentences, as we can see towards the end of example (3), which uses the conjunction and to relate the two parallel sentences (I laughed. I cried) to the last sentence.2 (3)

I can rarely say that a movie made me laugh and cry without feeling like an idiot, but the caliber of this picture is so high that I don’t even feel embarrassed. I laughed. I cried. And you will too.

1

See www.imdb.com

2

Halliday and Hasan restrict cohesion to links across, not within, sentences (i.e., beyond the clause and clause complex level). In example (3), then, the connection signalled with and would be cohesive, but not the one indicated by but in the first clause. Here, I will consider both of these as instances of cohesion, because I believe that cohesion and conjunction occur across clauses, whether in different sentences or not.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

314

MAITE TABOADA

In addition to grammatical conjunctions, the conjunction system deploys a few other grammatical devices which elaborate, extend, or enhance the meaning of previous discourse. These are often adverbials (actually, alternatively) or phrasal units (in any case, in the same manner). The class of such elements is collectively referred to as ‘conjunctive Adjuncts’, and they partially overlap with what, in other approaches, are referred to as discourse markers, filled pauses, or backchannels (Schiffrin 1987; Stenström 1994). We can see one example in (4), where the argument develops around how different viewers interpret the film differently. This point is interpreted and summarized with the help of I mean at the beginning of the last sentence. (4)

I remember talking to friends after seeing it and we all had interesting points. I felt the film focuses on Chihiro’s innocence as compared with the other characters she encounters, but her child like views are so carefree (and naive at times) and her youthful exuberance really makes it endearing. Another friend said it was a coming of age and how Chihiro herself progresses throughout the film. I mean, if you can find so much insight in a film, you know you have a great film.

Conjunction has an uneasy status as a member of the cohesion set of resources. In fact, many researchers separate it from cohesion, as has been done in the title of this chapter, presented as a coordinated item with cohesion, rather than as subordinate to it. This is because conjunction relates to a separate system, that of the clause complex, since it enables and signals relations between clauses. Scott and Thompson (2001) characterize the distinction as repetition vs. conjunction. Repetition (i.e. cohesion in the sense used in this paper) handles continuity. Conjunction, on the other hand, deals with discontinuity and transitions across units of discourse, mostly clauses and clause complexes, but it can extend beyond the clause complex, relating entire sentences and portions of a text or even chapters in a book. I discuss this issue in Section 12.5.

12.2.2 Reference Reference is achieved mostly through relations between a pronoun and an antecedent, forming a referential chain in the text, which is characterized as anaphoric reference. Reference links to elements outside the text constitute ‘exophoric reference’, whereas links within the text are ‘endophoric’. Some reference chains may include both exophoric and anaphoric reference. For instance, first- and second-person personal pronouns refer to relationships defined outside the text, but often also create text-internal relations. In example (5), the personal pronoun I in the first sentence is exophoric, in that it refers to the writer as somebody outside the text. The reference chain, however, continues inside the text, with another I in the second sentence, and the possessive my.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

(5)

315

Spirited Away is one of the most perfect movies I have ever seen. The least I can say about it is that there was not a single moment during it that my attention wasn’t completely focused.

Reference is often ‘personal’, establishing co-reference between the elements. The other two main forms of reference are demonstrative and comparative. ‘Demonstrative reference’ is probably deictic and exophoric in origin, pointing to referents outside the text, but can be used endophorically, as text-internal anaphora. In example (6), this world in the second sentence refers to the world first introduced in the preceding sentence, the world seen through the main character’s eyes. (6)

We discover the world as Chihiro does and it’s truly amazing to watch. But Miyazaki doesn’t seem to treat this world as something amazing.

‘Comparative reference’ is different from personal and demonstrative in that identity of reference is not established; the link is rather to a comparison class. Example (7) is a shortened excerpt where different qualities of the movie are discussed. The last sentence contains two instances of comparative reference: another great point and the best (part of it). Another establishes a comparative reference of difference to the other aspects already discussed, and then goes on to correct that reference to make it particular, stating that the score is not just another good aspect, but in fact the best. (7)

The story is imaginative and the characters and animations endlessly unique and strange. . . . What I also loved in this film is that the animation gives it a real sense of cinematography, . . . Another great point in fact the best part of it, is the fantastic score.

12.2.3 Substitution and Ellipsis Substitution and ellipsis are two forms of the same phenomenon, as ellipsis can be described as substitution by zero. In substitution, a cohesive device belonging to a closed class in the language is used to replace an open-class lexical item. Common substitution devices in English are one, so, or do so. In example (8), we see one in this one as a substitute term for cartoon movies. The repetition of the term would have probably made the sentence heavier, with more lexical items than necessary. The pronoun one helps make it cohesive. (8)

Even if you don’t normally like ‘cartoon movies’, you might give this one a chance.

Ellipsis makes coherent text possible, and even more so in spoken language. The simple ‘yes’ and ‘no’ answers that we furnish as answers to questions would be cumbersome if they were always accompanied by a full answer which repeats information already present in the question (although

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

316

MAITE TABOADA

repetition is also possible, and sometimes desired). Example (9) shows an elliptical verb in the answer to the rhetorical question what does it matter? The answer is it didn’t [matter]. (9)

I can’t imagine seeing a live action foreign language film dubbed into another language, but hey, this is a kids cartoon, what does it matter? Up to a point it didn’t, because I loved the film.

Ellipsis seems to display different frequencies across languages, with some languages allowing ellipsis of Given Subjects, and sometimes Objects, but ellipsis is common even in English; such is the case in (10), where the object of eat (presumably, the food in the market) is not mentioned, because it is so clearly recoverable from the context. (10)

When Chihiro’s parents see the food in the market, they just sit down and eat but they turn out to turn into pigs.

12.2.4 Lexical Organization At a very general level, Hasan (1985) classified all cohesive elements (or cohesive ties; see below) as belonging to one of three types: co-reference, coclassification, and co-extension. Co-reference is the relation between a pronoun and its proper name antecedent. In a co-classification relation between two cohesive elements, A and B, ‘the things, processes, or circumstances to which A and B refer belong to an identical class, but each end of the cohesive tie refers to a distinct member of this class’ (Hasan 1985: 74). In (11) both Princess Mononoke and Spirited Away belong to the class movies directed by Miyazaki. (11)

Much like Miyazaki’s previous feature Princess Mononoke, Spirited Away is an epic fairytale fantasy that deserves no better medium than the stunning animation work of Studio Ghibli.

Finally, in co-extension, the relation holds between two cohesive elements that are related to each other by virtue of belonging to the same general class. Such is the relation between story, animation, and score in example (7) above, which are all members, i.e. different aspects, of the class film. This general taxonomy (co-reference, co-classification, co-extension) can be made more fine-grained by labelling the different types, in particular of co-classification and co-extension. Hasan (1985) referred to these as sense relations, and included synonymy, antonymy, hyponymy, and meronymy. To that, one may also add repetition, which is not a sense relation per se, but likewise contributes to cohesion. In Halliday and Hasan (1976), lexical cohesion is classified along slightly different lines, with lexical cohesion and referential relations as the main categories (see Table 12.1). Referential relations are included because they also contribute to creating cohesion across cohesive devices.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

317

Table 12.1 Types of lexical cohesion according to Halliday and Hasan (1976) Type of lexical cohesion 1. Reiteration Same word (repetition) Synonym Superordinate General word 2. Collocation

Referential relation Same referent Inclusive Exclusive External

Lexical cohesion may be the hardest to define because different researchers have proposed different categorizations of the various phenomena that can be included as part of lexical cohesion. The superordinate category in Table 12.1, for instance, seems to call for a subordinate category. And general nouns (people, thing, stuff) can be considered a class of superordinate terms. There are multiple reinterpretations and categorizations of the semantic relations that language, and ultimately, thought, allows (Cruse 2000; Hasan 1985; Martin 1992; Morris and Hirst 1991; Tanskanen 2006). Perhaps the one that has created the most trouble is the concept of collocation. Halliday and Hasan (1976:284) define it as ‘the association of lexical items that regularly co-occur’. This includes pairs such as boy and girl, which are complementary and not easily defined as either antonyms or meronyms. It likewise includes words from an ordered series, such as days of the week, but also a more general category of words that are related to each other by virtue of a connection, often in the real world: basement . . . roof; road . . . rail; and box . . . lid are some of the examples suggested by Halliday and Hasan. More generally, collocation captures the relationship between lexical items that tend to occur together in certain text types. In a sense, characterizing collocation involves creating taxonomies of ideas, concepts, and the world, of the type represented in thesauri and in WordNet (Fellbaum 1998). Eggins (2004) addresses the difficulty of classifying lexical relations by dividing them into two groups: •

•

‘Taxonomic’ relations include the types of relations listed under ‘reiteration’ in Table 12.1, such as synonymy, hyponymy, and meronymy, or part-whole relations ‘Expectancy’ relations hold when there is a predictable relation between the process and the participants in the process (e.g. mouse – squeak, or play – a musical instrument).

Eggins defines expectancy relations as those related through transitivity (which leads to cohesive harmony; see Section 12.4). If we were to make this category broader, then it could account for collocation, because

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

318

MAITE TABOADA

Table 12.2 Types of lexical cohesion (Halliday and Matthiessen 2014:644) Nature of relation

Type of expansion

paradigmatic [lexical set]

elaborating

identity

attribution

syntagmatic [collocation]

Type of lexical relation repetition synonymy hyponymy

extending

meronymy

(enhancing)

collocation

Examples bear – bear sound – noise sound – silence [antonymy] tree – oak, pine, elm… oak – pine – elm… [co-hyponyms] tree – trunk, branch, leaf… trunk – branch – leaf… [comeronyms] fire – smoke

collocation arises out of expectancy, that in a text we will find words in certain relations to other words (same semantic field, doer-process, adjective-noun, etc.). A different view of the class of lexical items is provided in Halliday and Matthiessen (2014), where the taxonomic vs. expectancy relation is characterized as paradigmatic vs. syntagmatic. Paradigmatic relations are those of choice among alternatives. Halliday and Matthiessen describe them as the types of relations contained in a thesaurus, and they would be equivalent to Eggins’ taxonomic relations. Syntagmatic relations are such by virtue of linear relation, otherwise described as collocation. The classification is further refined in terms of the type of expansion that the tie provides, as either elaborating, extending, or enhancing. We will see in Section 12.4 that the same three-way organizational principle applies to conjunctive relations. The entire classification from Halliday and Matthiessen (2014) is reproduced in Table 12.2, where we see how the general list from 1976 which included all types of lexical relations plus collocation has seen a new level superimposed, to take it to a more abstract level. This does not mean, however, that collocation has been made easier to understand. Halliday and Matthiessen note that collocation relations may or may not be enhancing. Collocation, in their view, includes the type of expectancy relations mentioned as expectancy above, but also collocation relations that happen in particular registers, such as those in technical fields.

12.3

Cohesive Chains

Elements in the text related through cohesion establish a ‘cohesive tie’: the interpretation of one element in the discourse depends on the interpretation of another, whether preceding (anaphoric relation) or following (cataphoric). The fact that the interpretation is successfully established creates

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

319

the cohesive tie between the two elements. As more elements are related to each other, a ‘cohesive chain’ is created, a series of elements related to one another. Chains are frequently created through identity of referent (I-I-my in example (5)), but Hasan (1985:73) clearly states that the relation between elements in a cohesive tie is semantic, i.e. the elements are tied together through some meaning relation, even if the relation is expressed through grammatical resources, such as personal pronouns. One way of exploring the relationship in a cohesive tie, and ultimately in cohesive chains, is to measure the distance between components. The relationship may be immediate (the cohesive element refers to an immediately preceding one); remote (the referent is more than one clause away); or mediated (where the ultimate referent is a few clauses earlier in the discourse, but has been recaptured in some other element). Although I often refer to ‘preceding discourse’, this naturally applies only to anaphoric relations; cataphora works in the opposite direction, by establishing links that look forward to an element that completes the interpretation. Cataphora seems to be extremely rare: in a study of over 11,000 instances of third-person pronouns, my colleague Radoslava Trnavac and I found only fifty-seven instances of cataphora (Trnavac and Taboada 2016). Halliday and Matthiessen (2014:625) also state that cataphora is rare, with the exception of ‘structural cataphora’, where the reference to a pronoun is resolved immediately afterwards, in the same clause. Such is the case in (12), where the referent for those is in the relative clause immediately following (who are just looking . . .). (12)

Highly recommended to anime fans and those who are just looking for a film that is unique and interesting.

Cohesive chains run through texts, and provide them with the links to create texture. Most texts (spoken or written) contain more than one chain. The short excerpt in (13) contains at least five interrelated chains, as shown in Table 12.3. The table breaks down the text into units, which are somewhat arbitrary, for ease of presentation, and do not necessarily correspond to clauses or independent units. We can see that there is a chain relating to the film under discussion, which includes Spirited Away as the first element in the chain. The noun group Spirited Away is not technically a cohesive element yet, as this is the beginning of the text, and the noun group does not refer to anything preceding, although it naturally establishes links outside the text proper, on the web page where this review appeared,3 and the web page from which the review is linked.4 The next element in this chain, the latest, relates to Spirited Away through ellipsis of the Head noun film. The chain continues with repetition of the noun film. A second 3

See www.imdb.com/title/tt0245429/reviews-25

4

Some of these ties are multimodal, because they relate different modalities. Multimodal cohesion and conjunction are beyond the scope of this paper, but see Bateman (2008) for an overview of multimodality in discourse.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

320

MAITE TABOADA

Table 12.3 Cohesive chains in example (13) Unit Spirited Away is the latest in a string of incredible animated films by Hayao Miyazaki, the most renowned animator in Japanese history and most say in the best in world. He takes a couple of steps close to the best in the world title with this film.

Film

Animation

Miyazaki

Spirited Away the latest films

animated

Hayao Miyazaki

animator

this film

He

Excellence

Reference frame

the most renowned

Japanese history

the best (in world) best (in the world) title

in world in the world

chain, related to the first, establishes that this is an animation film, and the director then naturally an animator. The third chain refers to the director, through reference in the form of the personal pronoun he in the second sentence. Finally, two related chains establish Miyazaki as an excellent director, first in a narrower frame of reference (Japan), and then more broadly, as the best in the entire world. (13)

12.4

Spirited Away is the latest in a string of incredible animated films by Hayao Miyazaki, the most renowned animator in Japanese history and most say in the best in world. He takes a couple steps close to the best in the world title with this film.

Cohesive Harmony

Cohesive chains and chain interaction are some of the most interesting constructs for describing cohesion in text, and how texture is achieved. Hasan (1984) proposed the idea of ‘cohesive harmony’, a measure of how well-integrated cohesive chains are (see also Hoey 1991; Khoo 2016; Parsons 1996). Chains do not occur in isolation, but alongside other chains. However, the mere presence of two or more chains in a text does not guarantee a cohesive effort. Although chains contribute to cohesion in a text, they need to be related to each other somehow. This relationship is called ‘chain interaction’. The relationships are mostly grammatical, as part of the transitivity structure of the clause, such as the relationship between Processes and Participants. Hasan establishes a minimum requirement for chain interaction: at least two members of one chain should stand in the same relation to two members of another chain. For a better definition of the interactions, she divides the tokens in a text into three categories:

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

•

•

321

Relevant tokens: all tokens that enter into chains, further divided into: ○ Central tokens: relevant tokens that interact. ○ Non-central tokens: relevant tokens that do not interact. Peripheral tokens: Tokens that do not enter into any kind of chain.

We are, then, in a position to define cohesive harmony as the function of three phenomena: • • •

Low proportion of peripheral tokens to the relevant ones. High proportion of central tokens to non-central ones. Few breaks in the interaction.

Hasan affirms that coherence is a function of cohesive harmony. Our perception that a text is coherent, that it somehow makes sense, is dependent on its cohesive harmony. This explains why example (1) is not coherent, even though it contains cohesive ties: the ties do not form long chains, and they do not interact with each other very much, outside of the sentence where they appear. Example (13) above, on the other hand, shows a high degree of chain interaction: the chain that contains words such as Spirited Away and films is related to the chain that contains Hayao Miyazaki through transitivity: by Hayao Miyazaki is the Postmodifier of films in the first sentence, and the Subject in the last sentence, which contains the group this film also as a Circumstance. Many scholars have pointed out that cohesion and coherence are not allor-nothing categories, but rather a matter of degree. Parsons (1996) stated that, in any given text type, there is a gradation dependent on the extent to which a text relies on cohesion to provide coherence. Thus, texts belonging to different text types and registers will show different degrees of cohesive harmony. In a study of cohesion in task-oriented dialogue, I found very little chain interaction (Taboada 2000, 2004). Breaks in cohesive chains, on the other hand, were indicative of breaks in the text, where one stage of the genre finished and another one started.

12.5

Conjunction as a Separate System

The system of conjunction has had an uneasy status as a member of the general cohesion class. All other cohesive relations (reference, substitution, lexical cohesion, and their cognates) are relations among entities in the discourse, or propositions presented as entities (such as using that to refer to a previous sentence). Conjunction, on the other hand, marks relations between clauses in a clause complex, between text segments realized by clause complexes, or between longer text segments, named rhetorical paragraphs by Halliday and Matthiessen (2014:605). Hasan (1985:81) makes a clear distinction between ‘componential’ and ‘organic’ cohesive devices. Componential devices are items that form part

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

322

MAITE TABOADA

of a cohesive tie, and are at the same time members of the transitivity structure of the clause. It is the linking of those devices that creates cohesion across clauses. Organic devices, i.e. conjunctive devices, tie whole clauses rather than clause components. Hasan points out that other devices of language organization, such as adjacency pairs, are also organic. Conjunctive devices (conjunctions, some continuatives, and some Adjuncts) link clauses in logico-semantic relations such as cause, concession, or condition. They also serve to indicate temporal and additive relations. In this sense, conjunction is closely related to the clause complexing system, and moves cohesion outside of the clause proper. Halliday and Hasan would always have cohesion act outside of the clause, but the view taken here is more inclusive, whereby cohesive and conjunctive links occur both within the clause and across clauses. Conjunction serves to indicate that a relation exists between clauses (clause complexes, sentences, or entire text passages), and sometimes it provides an indication of the nature of the relation. This indication can be quite clear, such as the relation signalled by because, or it can be underspecified, as is the case with and, which can indicate a variety of relations. This general idea, that conjunction relates clauses, or propositions, has been made specific and instantiated under different theories, and different taxonomies. Halliday and Hasan (1976) proposed a top-level classification into additive, adversative, causal, and temporal. The classification is exclusively based on the presence of a conjunctive item (e.g. so, consequently, for this reason, or it follows for cause). In the first edition of Introduction to Functional Grammar (Halliday 1985), and in subsequent editions, this semantic classification (based on how the conjunction specifies the semantic content of the linkage) is made more abstract, with a higher-level classification based on how one clause adds to another, and using three types of connection: elaboration, extension, or enhancement. This classification is based on the form of the contribution, rather than the semantic meaning that is contributed, but typically relies on the presence of a conjunctive item.5 Martin (1992) proposed a slightly different classification of what he named conjunctive relations, which are outside of cohesion proper. Additionally, he expands on the internal/external distinction, relating to whether relations refer to external relations, in the real world, or to the internal organization of the events in the text. The latter are more ‘rhetorical’ in nature, in that they have to do with how arguments are presented. The distinction is quite clear with temporal relations. An external temporal relation describes sequences of activities as they occur in the world. Internal temporal relations, on the other hand, capture time within the text, i.e. in relation to what is being said and how the text is organized. The two following examples are from Martin (1992:182), with the words in 5

Halliday does mention non-finite clauses as examples of conjunctive relations which do not have an explicit marker for the relation.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

323

bold, from the original text, indicating conjunctive items. Example (14) presents an external temporal relation, where the events are presented as they unfolded in the world. In (15), an internal relation, the sequence firstsecond could have well been presented in the reverse order, depending on the effect that the writer or speaker wanted to create. (14)

Ben came in and then had a drink.

(15)

Ben wasn’t ready. First he hadn’t studied; and second, he’d been up all night.

The internal/external distinction is cross-categorized with a semantic classification, much along the lines of Halliday’s (1985): additive, comparative, temporal, consequential. What makes Martin’s description different, however, is the fact that he does not exclusively rely on conjunctions as signals of a particular relation. For instance, one example given of similarity relations is that they are signalled by so followed by a Finite verb. That is, it is not the conjunction alone, but a combination of the conjunction and lexicogrammatical features that makes it clear which relation the writer/speaker intends. Martin suggests that there is congruent and metaphorical signalling of relations. Congruent signalling involves conjunctions, but metaphorical signalling takes many other forms. For instance, the verbs enable, cause, and follow may be signals of relations, equivalent to the conjunctions by, because, or before. There is, then, a range of lexical and grammatical options available to signal conjunctive relations. Scott and Thompson (2001) characterize the range of possibilities as explicit signals (conjunctions such as although or conjuncts such as on the other hand), large-scale signposts (We can draw three main conclusions from this . . .), or no signal, leaving it to the reader or hearer to establish the link, as well as the type of link. Let us examine the issue of signalling through two invented examples. We can describe the link between the two clauses in example (16) as a ‘classical’ example of cohesive conjunction, given the fact that it contains the conjunction because.6 The case would not be so clear-cut, however, in (17). There, the causal connection is presumably still active, but there is no signal, and the relation occurs across sentence boundaries. There are, of course, other cohesive devices in the two clauses/sentences that link them, such as the reference Dominique – he and the lexical connection between job and long hours (maybe long hours are attributes of some jobs). Regardless of the other cohesive links, the reader still has to infer that there is a relation, as well as what specific relation it is. (16)

Dominique quit his job because he was tired of the long hours.

(17)

Dominique quit his job. He was tired of the long hours.

6

Not strictly ‘classical’ in the sense of Halliday and Hasan, since for them cohesion only takes place across, not within sentences. For them, this example would be accounted for within clause complexing, not cohesive conjunction.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

324

MAITE TABOADA

In examples (18) and (19) we see instances of explicitly signalled relations, within and across sentences. In the first example, the concessive relation between the two clauses is clearly marked by while, and the relation takes place within the confines of the sentence. In (19), a similar concessive relation occurs across sentence boundaries. In this case, it is signalled by but at the beginning of the sentence. (18)

While these two themes are very much current in Japan, they are also universal themes.

(19)

Sometimes in real life the most grim moments contain honest elements of comedy that do not seem out-of-place. But trying to put that sort of convoluted emotion into a film creates a very thin line that too many have fallen off of.

The broadening of the scope of conjunction, beyond the boundaries of the sentence, and beyond the confines of signalling by conjunctions, is the realm of the phenomenon explored in theories of discourse, coherence, or rhetorical relations. The emphasis shifts from the conjunctions themselves to the relation, whether signalled or not. These relations have received multiple names: rhetorical predicates (Grimes 1975); combinations of predications (Longacre 1976); coherence relations (Hobbs 1979; Sanders et al. 1992, 1993); rhetorical relations (Mann and Thompson 1988); or discourse relations, a label for three different approaches (Polanyi 1988; Renkema 2009; Webber et al. 2003), and including work in Segmented Discourse Representation Theory (Asher and Lascarides 2003). Despite distinct theoretical differences, and diverging treatment of the phenomenon, all the approaches share an interest in explaining how discourse is coherent through the combination of ideas or propositions. Among these, one of the theories better connected to Systemic Functional Linguistics is Rhetorical Structure Theory (Mann and Thompson 1988; Mann et al. 1992). Rhetorical Structure Theory (RST) was developed by Bill Mann, Christian Matthiessen, and Sandra Thompson in an effort to represent text for computational purposes. It is closely aligned with SFL because Christian Matthiessen was one of the developers of the theory, and the overall computational project had Michael Halliday as a consultant. In RST, the view of conjunctive/rhetorical relations is top-down, that is, from the intention of the speaker or writer, rather than from the lexicogrammatical realization, and the signalling through conjunctions. Naturally, SFL always takes meaning-making as the point of departure, but some of the descriptions, including those of cohesion, are more bound to the lexicogrammar. In RST, on the other hand, intentions are key, in part because RST evolved in the context of language planning and Natural Language Generation (Mann 1983a, 1983b). Rhetorical relations are equivalent to relations among clauses and clause complexes at the most basic level. The difference is that relations can be recursive, and apply at all levels of discourse. Thus, units of analysis are no

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

325

Figure 12.1 RST representation of example (20)

longer clauses or clause complexes, but ‘spans’. Unlike in the conjunction system of Halliday and Hasan, or the conjunctive relations of Martin, in RST there is no overarching taxonomy of relations, and relation lists are quite flat, i.e. they have no hierarchy, although there is a distinction between Subject Matter and Presentational relations. Subject Matter relations (Circumstance, Condition, Elaboration, Cause, or Result) have the intended effect that the reader recognize the relation. Presentational relations, on the other hand, attempt to increase some inclination in the reader, such as the desire to act, or the degree of positive regard for the nucleus of the relation. Examples are Concession, Enablement, Evidence, Motivation, or Summary. This is somewhat similar to the external/internal distinction of Martin, discussed above. RST is perhaps best explained through an example. In (20), we see a short excerpt, this time of a review of the book Hot Six by Janet Evanovich, posted online.7 The excerpt has been divided into clausal units, in the example marked with square brackets. An RST analysis of this text is presented in Figure 12.1. In the figure, each of the units of analysis (spans) has a horizontal line on top. Additionally, they are connected to other units through either straight or curved lines. For instance, the connection between Spans 2 and 3 shows that Span 3 has a straight line above it, marking it as the nucleus, whereas Span 2 has a curved line, an arrow pointing to Span 3. This indicates that Span 3 is the satellite. The two clauses are connected in a Condition relation: the action in the nucleus (do not start the series by reading this book) has a condition attached to it, that the reader should not have read any of the books in the series. The terms ‘nucleus’ and ‘satellite’ refer to the relative importance of spans in the organization of discourse. At the clause level, they are equivalent to the concepts of main and subordinate clause in traditional grammar. The difference is that this relative importance reaches across clause and sentence boundaries. We see this in the unit that joins Spans 2–3 to Span 1,

7

www.epinions.com/review/Hot_Six_by_Janet_Evanovich_and_narrated_by_Debi_Mazar/2004218900/963557

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

326

MAITE TABOADA

which are related through a Background relation: the fact that the book under review is the sixth in the series is background information necessary for understanding the Condition conveyed in Spans 2–3. Here, Spans 2–3, combined, are the nucleus, whereas Span 1 is the satellite. The excerpt has a second part, composed of a relation which does not contain a nucleus and a satellite, like the ones we have seen so far, but which instead is made up of three nuclei, three units of equal importance (Spans 3–6). Such relations are referred to as ‘multinuclear’ relations. Together, these three units constitute a Sequence relation, outlining the steps that the author thinks the reader should take. Finally, these two macro-units or sequences of spans (1–3 and 3–6) are joined together into another multinuclear relation, this time of Contrast. The author establishes a contrast between two possibilities, not reading the book under review and reading the previous ones in order. (20)

[This is the sixth book in the Stephanie Plum series.] [If you have never had the fun of reading a book in this series,] [do not start with this one.] [Go to the library] [and start with One For The Money] [and work your way up to Hot Six.]

This short example illustrates the principles of nuclearity and recursion that are fundamental to RST analyses. The important aspect with regard to cohesion is that these relations are postulated to exist even in the absence of conjunctive devices to signal them. The Condition relation between Spans 2 and 3 has a nice if to indicate its presence. But the higher-level Contrast relation does not have any conjunction to guide the reader. It is, in fact, the presence of other lexicogrammatical items, including other cohesive devices than conjunction, that gives clues to the relation, like the lexical chain linking books and libraries, the chain start – work your way up, or the repeated use of imperatives, first a negated one in Span 3, and then three consecutive imperative clauses in Spans 4–6. I return now to the concept of texture. The texture of a text, the way in which it makes sense, is brought about by this interaction between cohesive devices proper (reference, substitution and ellipsis, lexical organization) and a wider view of conjunction in the form of rhetorical relations. Together, cohesion and conjunction/rhetorical organization create the weft and warp that weave together a text.

12.6

Computational and Other Applications

The general concept of cohesion has found favour in different areas of knowledge, because of how elegantly it accounts for why a text may or may not be seen as coherent, and how adjacent portions of a text are connected. Thus, it has been used in computational applications to measure coherence of generated text or to detect breaks where a new topic is being

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

327

introduced; in educational contexts to measure coherence of students’ texts; and in multimodal studies to describe the connection among modalities. Perhaps the most extensive area of application has been in computational linguistics. Simplified treatments of the concept of lexical chains, involving only semantic relations, are frequently used to segment text in chunks (Purver 2011). This is the idea behind TextTiling, one of the most popular discourse-oriented computational algorithms (Hearst 1994). In TextTiling, text is divided into topic units, that is, chunks with a common topic each, by examining relations of lexical cohesion, and positing a break between chunks when lexical relations seem to fall below a certain threshold. Morris and Hirst (1991) showed that it is possible to compute lexical chains, and that the chains are useful in determining text structure. Cohesive chains are sometimes computed using only lexical cohesion (in which case they are referred to as lexical chains), and sometimes include (pronominal) reference. Cohesive or lexical chains have turned out to be quite useful in multiple computational applications. For instance, in text summarization Alonso i Alemany and Fuentes Fort (2003) compute lexical chains and use the characteristics of chains to decide which parts of the text to use in an automatic summary. Characteristics of chains which are good candidates include their length, the kinds of cohesive relations in them, and the point of the text where they start. This lexical information is also enhanced with features derived from the rhetorical structure of the text. Similar linkages between lexical cohesive devices and rhetorical or coherence relations are proposed in other computational work (Cristea et al. 1998; Harabagiu 1999). Textual coherence is the goal of anyone who desires to master a foreign language, a new register, or a form of language for specific purposes. Therefore, an accurate measure of cohesion can show how far from or close to that goal one is. In educational contexts, cohesion helps establish how non-native language differs from native-like texts. For instance, Schleppegrell (1996) examines and classifies the use of the conjunction because in native and non-native writing, and finds an overuse by non-native speakers. Native speakers employ additional constructions, and thus show more variety in their conjunctions and other connectives. McNamara et al. (2014) propose Coh-Metrix, a cohesion measure that includes many factors (such as lexical diversity and syntactic complexity), but also uses the types of cohesive links established throughout the text as a way to indicate how coherent the text is. An area where cohesion has been consistently applied is in the automatic scoring of essays, in particular those produced by foreign language learners for tests such as TOEFL (Burstein et al. 2010; Rahimi et al. 2015; Somasundaran et al. 2014). A related application is in machine translation, where cohesion is used to measure the coherence of machine-translated text (Wong and Kit 2012). Outside of the computational arena, interesting treatments of cohesion include the connection between coherence/rhetorical relations and genre.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

328

MAITE TABOADA

Taboada (2000, 2004) showed that breaks in lexical chains correspond to breaks in the genre structure of task-oriented conversations. This is the same intuition behind automatic text segmentation methods: a new stage or a new chunk of the text involves some sort of break in the set of cohesive chains in the text. Most of the research described in this chapter has been carried out in English. Analyses of cohesion in other languages exist, however, and they show subtle and interesting contrasts with the available descriptions of English. To take just two examples, Berzlánovich and Redeker (2012) study the interaction between genre, cohesion, and coherence in an annotated corpus of Dutch texts, as do Kunz and Lapshinova-Koltunski (2014) for German.8

12.7

Conclusion

Cohesion and conjunction (in the form of rhetorical relations) constitute the two fundamental properties of text, the properties that give it coherence, or texture. Cohesion establishes links across entities in the text, whereas conjunction links propositions at all levels (clause, clause complex, and paragraph). Rhetorical relations capture what we want to achieve with the text, in terms of the intention of the text creator. The exact distinction between cohesion, conjunction, coherence, and texture has sometimes been criticized as unclear (Brown and Yule 1983). Flowerdew (2013) points out that there is ambiguity in the description, especially because Halliday and Hasan (1976), in addition to cohesion, include register and thematic development as sources of texture. The most likely answer is that all aspects of context contribute to texture, but we tend to include cohesion and conjunction as the most clear-cut phenomena to account for the perception of coherence in text. We can summarize cohesion as links by means of referential, lexical, and logical ties (Eggins 2004: 53).

8

In addition, for further reading on cohesion, consider the following. The original description of cohesion (in English) is Halliday and Hasan (1976), and it still remains the most detailed account of the phenomenon, with plenty of examples. The theory was refined and developed in a book by Halliday and Hasan (1985), and in particular a chapter by Hasan in that book (Hasan 1985). Several introductions to Systemic Functional Linguistics explain cohesion in very concise terms, often in a single chapter (Eggins 2004; Thompson 2014), but perhaps the most clear and concise is Flowerdew’s introduction to SFL (Flowerdew 2013), which emphasizes applications of Systemic Functional Linguistics to language education. And, of course, Halliday and Matthiessen’s (2014) Halliday’s Introduction to Functional Grammar remains the ‘official’ version of the theory, with a chapter on cohesion and its place in the functional analysis of language. More specialized descriptions focus on cohesion, often with an introduction that then leads to an in-depth study, typically corpus-based. Tanskanen (2006) does not strictly follow Halliday and Hasan’s classification, but hers is a thorough corpus-based study. Fox (1987) presents an analysis of both cohesion and coherence in conversational speech.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

329

References Alonso i Alemany, L. and M. Fuentes Fort. 2003. Integrating Cohesion and Coherence for Automatic Summarization. Proceedings of EACL’03 Student Research Workshop. Budapest, Hungary. 1–8. Asher, N. and A. Lascarides. 2003. Logics of Conversation. Cambridge: Cambridge University Press. Bateman, J. 2008. Multimodality and Genre: A Foundation for the Systematic Analysis of Multimodal Documents. London: Palgrave Macmillan. Berzlánovich, I. and G. Redeker. 2012. Genre-dependent Interaction of Coherence and Lexical Cohesion in Written Discourse. Corpus Linguistics and Linguistic Theory 8(1): 183–208. Brown, G. and G. Yule. 1983. Discourse Analysis. Cambridge: Cambridge University Press. Burstein, J., J. R. Tetreault, and S. Andreyev. 2010. Using Entity-based Features to Model Coherence in Student Essays. Proceedings of Human Language Technologies: The 11th Annual Conference of the North American Chapter of the Association for Computational Linguistics. Los Angeles. 681–4. Cristea, D., N. Ide, and L. Romary. 1998. Veins Theory: A Model of Global Discourse Cohesion and Coherence. Proceedings of the 36th Annual Meeting of the Association for Computational Linguistics and the 17th International Conference on Computational Linguistics (ACL-98/COLING-98). Montreal, Canada. 281–5. Cruse, D. A. 2000. Meaning in Language: An Introduction to Semantics and Pragmatics. Oxford: Oxford University Press. Eggins, S. 2004. Introduction to Systemic Functional Linguistics. 2nd ed. London: Continuum. Fellbaum, C., ed. 1998. WordNet: An Electronic Lexical Database. Cambridge, MA: MIT Press. Flowerdew, J. 2013. Discourse in English Language Education. New York: Routledge. Fox, B. A. 1987. Discourse Structure and Anaphora: Written and Conversational English. Cambridge: Cambridge University Press. Grimes, J. E. 1975. The Thread of Discourse. The Hague: Mouton. Halliday, M. A. K. 1985. An Introduction to Functional Grammar. London: Arnold. Halliday, M. A. K. and R. Hasan. 1976. Cohesion in English. London: Longman. Halliday, M. A. K. and R. Hasan. 1985. Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Oxford: Oxford University Press. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2004. An Introduction to Functional Grammar. 3rd ed. London: Arnold. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Arnold.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

330

MAITE TABOADA

Harabagiu, S. 1999. From Lexical Cohesion to Textual Coherence: A Data Driven Perspective. International Journal of Pattern Recognition and Artificial Intelligence 13(2): 247–65. Hasan, R. 1984. Coherence and Cohesive Harmony. In J. Flood, ed., Understanding Reading Comprehension. Newark: International Reading Association. 181–219. Hasan, R. 1985. The Texture of a Text. In M. A. K. Halliday and R. Hasan, Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Oxford: Oxford University Press. 70–96. Hearst, M. 1994. Multi-Paragraph Segmentation of Expository Text. Proceedings of 32nd Annual Meeting of the Association for Computational Linguistics (ACL’94). Las Cruces, New Mexico. 9–16. Hobbs, J. 1979. Coherence and Coreference. Cognitive Science 6: 67–90. Hoey, M. 1991. Another Perspective on Coherence and Cohesive Harmony. In E. Ventola, ed., Functional and Systemic Linguistics: Approaches and Uses. Berlin: Mouton de Gruyter. 385–414. Khoo, K. M. 2016. ‘Threads of Continuity’ and Interaction: Coherence, Texture and Cohesive Harmony. In W. L. Bowcher and J. Y. Liang, eds., Society in Language, Language in Society: Essays in Honour of Ruqaiya Hasan. New York: Palgrave Macmillan. 300–30. Kunz, K. and E. Lapshinova-Koltunski. 2014. Cohesive Conjunctions in English and German: Systemic Contrasts and Textual Differences. In L. Vandelanotte, K. Davidse, C. Gentens, and D. Kimps, eds., Recent Advances in Corpus Linguistics: Developing and Exploiting Corpora. Amsterdam: Rodopi. 229–62. Longacre, R. E. 1976. An Anatomy of Speech Notions. Lisse: The Peter de Ridder Press. Mann, W. C. 1983a. An Overview of the Nigel Text Generation Grammar: ISI/RR83–113. Information Sciences Institute, University of Southern California. Mann, W. C. 1983b. An Overview of the Penman Text Generation Grammar: ISI/RR83–114. Information Sciences Institute, University of Southern California. Mann, W. C., C. M. I. M. Matthiessen, and S. A. Thompson. 1992. Rhetorical Structure Theory and Text Analysis. In W. C. Mann and S. A. Thompson, eds., Discourse Description: Diverse Linguistic Analyses of a Fund-raising Text. Amsterdam: John Benjamins. 39–78. Mann, W. C. and S. A. Thompson. 1988. Rhetorical Structure Theory: Toward a Functional Theory of Text Organization. Text 8(3): 243–81. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. McNamara, D. S., A. C. Graesser, P. M. McCarthy, and Z. Cai. 2014. Automatic Evaluation of Text and Discourse with Coh-Metrix. Cambridge: Cambridge University Press.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

Cohesion and Conjunction

331

Morris, J. and G. Hirst. 1991. Lexical Cohesion Computed by Thesaural Relations as an Indicator of the Structure of Text. Computational Linguistics 17(1): 21–48. Parsons, G. 1996. The Development of the Concept of Cohesive Harmony. In M. Berry, C. S. Butler, R. Fawcett, and G. Huang, eds., Meaning and Form: Systemic Functional Interpretations (Meaning and Choice in Language: Studies for Michael Halliday). Norwood: Ablex. 585–99. Polanyi, L. 1988. A Formal Model of the Structure of Discourse. Journal of Pragmatics 12: 601–38. Purver, M. 2011. Topic Segmentation. In G. Tur and R. De Mori, eds., Spoken Language Understanding: Systems for Extracting Semantic Information from Speech. Hoboken: Wiley. 291–317. Rahimi, Z., D. Litman, E. Wang, and R. Correnti. 2015. Incorporating Coherence of Topics as a Criterion in Automatic Response-to-Text Assessment of the Organization of Writing. Proceedings of the Tenth Workshop on Innovative Use of NLP for Building Educational Applications. Denver. 20–30. Renkema, J. 2009. The Texture of Discourse. Amsterdam: John Benjamins. Sanders, T., W. Spooren, and L. Noordman. 1992. Toward a Taxonomy of Coherence Relations. Discourse Processes 15(1): 1–35. Sanders, T., W. Spooren, and L. Noordman. 1993. Coherence Relations in a Cognitive Theory of Discourse Representation. Cognitive Linguistics 4(2): 93–133. Schiffrin, D. 1987. Discourse Markers. Cambridge: Cambridge University Press. Schleppegrell, M. J. 1996. Strategies for Discourse Cohesion: Because in ESL Writing. Functions of Language 3(2): 235–54. Scott, M. and G. Thompson. 2001. Introduction: Why ‘Patterns of Text’? In M. Scott and G. Thompson, eds., Patterns of Text: In Honour of Michael Hoey. Amsterdam: John Benjamins. 1–11. Somasundaran, S., J. Burstein, and M. Chodorow. 2014. Lexical Chaining for Measuring Discourse Coherence Quality in Test-taker Essays. Proceedings of the 25th International Conference on Computational Linguistics (COLING). Dublin, Ireland. Stenström, A.-B. 1994. An Introduction to Spoken Interaction. London: Longman. Taboada, M. 2000. Cohesion as a Measure in Generic Analysis. In A. Melby and A. Lommel, eds., The 26th LACUS Forum. Chapel Hill: The Linguistic Association of Canada and the United States. 35–49. Taboada, M. 2004. Building Coherence and Cohesion: Task-oriented Dialogue in English and Spanish. Amsterdam: John Benjamins. Tanskanen, S.-K. 2006. Collaborating towards Coherence: Lexical Cohesion in English Discourse. Amsterdam: John Benjamins. Thompson, G. 2014. Introducing Functional Grammar. New York: Routledge.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

332

MAITE TABOADA

Trnavac, R. and M. Taboada. 2016. Cataphora, Backgrounding and Accessibility in Discourse. Journal of Pragmatics 93: 68–84. Webber, B., M. Stone, A. K. Joshi, and A. Knott. 2003. Anaphora and Discourse Structure. Computational Linguistics 29(4): 545–87. Wong, B. T. M. and C. Kit. 2012. Extending Machine Translation Evaluation Metrics with Lexical Cohesion to Document Level. Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning. Jeju Island, Korea. 1060–8.

Downloaded from https://www.cambridge.org/core. Newcastle University, on 27 Aug 2019 at 06:51:51, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.014

13 Semantic Networks Andy Fung and Francis Robert Low

13.1

Introduction

Discourse studies featuring Systemic Functional Linguistics (SFL) as the framework of analysis have gained increasing momentum in the past decades (see Hyland and Paltridge 2011; Gee and Handford 2012; Hyland 2013). It is perhaps not surprising, because this theory of language is essentially ‘appliable’, attaching fundamental importance to ‘social accountability’ (Halliday 2006b, 2006a; Mahboob and Knight 2010a, 2010b; Matthiessen 2012, 2014). That is to say, it is a linguistic theory committed to solving the problems encountered by language users in their daily social practices (Halliday 2008:189). From a systemic point of view, the emphasis on ‘appliability’ is particularly relevant to discourse analysis because, without theory, the analysis of discourse would be ad hoc, inconsistent, and ineffective (Halliday 2006b:19), and by the same token, without the analysis of discourse, there would be no raison d’être of SFL since scholars working within SFL model language use as their starting point (Mahboob and Knight 2010b:1). Given the strong orientation to discourse in SFL, a number of approaches and discourse tools have been proposed and developed, enabling discourse analysts to interpret and make sense of the meaning of ‘what people say and write and listen to and read’ (Halliday 1994:xxii).1 Take semantics as an illustration. In SFL, language is conceptualized as a meaning potential, and semantics is a stratum in the SFL model of language. This stratum, together with lexicogrammar, phonology, and phonetics, constitute the language-internal strata (for a recent review of stratification of language, see Halliday and Matthiessen 2014; also see Hasan 2013, 2014). Above these language–internal strata is the stratum of 1

While recent years have witnessed a growing trend in analyses of multimodal discourse, this chapter, following Hasan (2014:3), regards language as the central object of enquiry in SFL.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

334

AN DY F U NG AN D FR ANCI S ROB ERT LOW

context — the extra-linguistic universe where the text is functioning. Being a ‘linguistic inter-level to context’ (Matthiessen 1993:227; see also Halliday 2009; Hasan et al. 2007; Hasan 2009c, 2010), semantics serves as the point of departure in describing and accounting for context and lexicogrammar. More importantly, it enables analysts to make sense of human life, since most of our daily social practices are essentially ‘acts of meanings’ (Hasan 2010:267). Given the significance of semantics, various studies have been undertaken since the late 1960s, each of which has produced a number of semantic descriptions (for a historical development of semantics in SFL, see Matthiessen 2007, 2009, 2015a). So far, there are at least seven orientations in SFL semantic descriptions, each of which has developed its own set of analytical tools. Though they are slightly different in their own right, all of them enable discourse analysts to study the ‘meanings we give language and the actions we carry out when we use language in specific contexts’ (Gee and Handford 2012:1). Table 13.1 summarizes these research approaches and semantic descriptions. The semantic description discussed in this chapter is based on Ruqaiya Hasan’s message semantics system networks. The first publication of message semantics networks appeared in 1983 in a mimeo prepared for her project investigating the different ways of meaning construed in mother– child talk. Remaining unpublished and circulated only among her team members, this pioneering work served as the foundation of message Table 13.1 Semantic descriptions in SFL (building on Cloran et al. 2007) Orientations

Semantic descriptions

1) Text texture

– COHESION (e.g. Halliday and Hasan 1976) – COHESIVE HARMONY ANALYSIS (e.g. Hasan 1984) – SOCIOLOGICAL SEMANTIC NETWORKS (e.g. Halliday 1973; Turner 1973) – MESSAGE SEMANTICS NETWORKS (e.g. Hasan 1983, 1996, 2009e; Hasan et al. 2007) – RHETORICAL UNIT ANALYSIS (e.g. Cloran 1994, 1999) – RHETORICAL STRUCTURE THEORY (e.g. Matthiessen 1988a, 2004; Halliday and Matthiessen 1999) – PHASAL ANALYSIS (e.g. Gregory 1985)

2) Sociolinguistics and semantic variation

3) Discourse structure in constituency terms 4) Discourse structure in dependency terms 5) Discourse structure in phasal terms 6) Collaborative and interactive exchange of dialogue 7) Discourse semantics

– SPEECH FUNCTION NETWORKS (e.g. Halliday 1984; Martin 1992; Matthiessen 1995; Eggins 1990; Eggins and Slade 2004) – IDEATION, – CONJUNCTION, – NEGOTIATION, – INVOLVEMENT, – APPRAISAL, – IDENTIFICATION, and – PERIODICITY (see Martin 1992, 2000; Martin and White 2005; Martin and Rose 2007; Martin 2014)

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

335

semantics, semantic variation, and sociolinguistic research. Despite the fact that her semantic description was developed specifically for semantic variation research (e.g. Cloran 1994; Williams 1995; Hasan 2009e), the past years have witnessed an increasing number of discourse studies featuring message semantics networks as the tool for the analysis of meaning (e.g. Hall 2004; Wake 2006; Hasan et al. 2007; Wong 2009; Chu 2011; Lukin et al. 2011; Lukin 2012, 2013). Having benefited from Hasan’s outstanding work for over three decades, the time is ripe to survey the application of message semantics. This chapter is thus organized as follows. Section 13.2 briefly describes the theoretical foundation of Hasan’s message semantics networks. Elaborating Halliday’s theoretical concepts of sociological semantic networks to a large extent, Hasan has developed her own position in terms of semantic description. Such elaborations and subsequent developments constitute the basis of what is understood as ‘message semantics networks’. Given the close association between Hasan’s and Halliday’s semantic networks, we first discuss the concepts postulated by Halliday (see Section 13.2.1), followed by a discussion on the theoretical constructs which led to Hasan’s message semantics networks (see Section 13.2.2). Having discussed the theoretical constructs of message semantics networks, in Section 13.3 we then move to the primary use of Hasan’s networks, with a particular focus on the notion of semantic variation. Section 13.4 discusses the extended uses of semantic networks in discourse studies. Here, we report on two domains of investigations which employ message semantics networks as the primary analytical tool, thereby illustrating how the networks are used and presenting the associated research implications.

13.2

The Theoretical Concepts of Semantic Networks

13.2.1 Halliday’s Sociological Semantic Networks Early in the 1970s, Halliday published the first paper on semantic networks, entitled ‘Towards a Sociological Semantics’. As the title suggests, a key point in Halliday’s work is that the semantic description is grounded in Bernstein’s theories of socialization and social learning, and attaches fundamental importance to the connection between social context and linguistic meanings. In theorizing the meanings accessible to speakers, Halliday recognizes that verbal behaviour is essentially a phenomenon which can be described sociologically and linguistically. These descriptions, however, could not be directly related because the social system is ‘wholly outside language’ and the grammatical system is ‘wholly within language’ (Halliday 1973:88). To relate these descriptions and illustrate how ‘social meanings are organised into linguistic meanings’ (Halliday 1973:72), Halliday proposed the idea of ‘semantic network’, defining it as a ‘hypothesis about patterns of meaning’ which forms a bridge between the ‘behavioural

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

336

AN DY F U NG AN D FR ANCI S ROB ERT LOW

patterns and linguistic forms’ (Halliday 1973:75). The semantic description is important in that it draws on the system network as representation, that is, meanings are represented as options within systems, and each option is systemically related to one another (Halliday 1973:68). These semantic options, as maintained by Halliday (1973:68), are context-dependent in the sense that they reflect only ‘what the speaker can do, linguistically, in a given context’. Such an approach to semantic description is not only ‘context-based’,2 but also ‘strategic’ in nature. As Matthiessen (1990:324–5, emphasis added) writes: When we approach semantics from above it is the interface between context and language that is highlighted. The role of semantics can be stated with respect to context as follows: semantics is the set of strategies for construing contextual meanings as linguistic meanings and thus moving into the linguistic system. Or if we focus on the notion of goal in particular, semantics is the set of strategies for achieving some goal through symbolic activity. This is a functional approach to semantics: it interprets semantics in terms of the uses it has evolved to serve in different communicative contexts.

For example, Halliday (1973) postulates a semantic network of parental control, illustrating the sets of strategies that a mother could employ in the regulatory context. Such strategies, or more specially, the goal-oriented symbolic activities, are semanticized as various semantic options. As shown in Figure 13.1, in regulating the behaviour of a child, a mother could either select the option [threat] or [warning] – the former denotes the actions that are threatened to be undertaken by care-givers, whereas the latter refers to the possible undesirable consequences of a child doing something that he or she is being told not to do. Both [threat] and [warning] serve as the point of entry to further sub-options. Important in this network representation is that not only are the options clearly identified and related, but they are also specified in terms of lexicogrammatical realization statements. That is to say, each semantic option is viewed from the lexicogrammatical stratum. Take [physical punishment] as an example. Halliday suggests that [physical punishment] is a subcategory of [threat], defining it as follows: The ‘threat’ may be a threat of physical punishment. Here the clause is of the action type, and, within this, of intentional or voluntary action, not supervention (i.e. the verb is of the do type, not the happen type). The process is a two-participant process, with the verb from a lexical set expressing ‘punishment by physical violence’, roughly that of § 972 (PUNISHMENT) in Roget’s Thesaurus, or perhaps the intersection of this with § 276 (IMPULSE). The tense is simple future. The Goal, as already noted, is you; and the clause may be

2

This contrasts with the description of semantics from below, or chooser and inquiry semantics (in Matthiessen’s 1990 terminology), which is typically employed in the model of text generation. Examples include Matthiessen (1988b), Patten (1988), and Matthiessen and Bateman (1991), to name but a few. See Matthiessen (1990) for details.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

337

Figure 13.1 The semantic network of warning and threat (Halliday 1973:89)

either active, in which case the agency of the punishment is likely to be the speaker (I as Actor), or passive, which has the purpose of leaving the agency unspecified. (Halliday 1973:78, emphases in original)

Given this orientation, the semantic option [physical punishment] is lexicogrammatically realized as ‘clause: action: voluntary (do type); effective (two-participant): Goal = you; future tense; positive; verb from Roget § 972 (or 972, 276)’, as in I will smack you, Daddy will smack you, or You’ll get smacked. The total set of semantic options, together with their lexicogrammatical realization statements, constitutes the ‘register-specific semantic potential’ (Hasan 1996:114).

13.2.2 Hasan’s Message Semantics Networks The semantic network postulated by Halliday in the 1970s was still in a nascent form, and it is not surprising that there remained much room for further development. With the subsequent advancement of SFL, the

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

338

AN DY F U NG AN D FR ANCI S ROB ERT LOW

conceptions of semantic networks have been greatly refined and elaborated. The major features of Hasan’s message semantics include (1) open context, (2) unit of analysis, (3) trinocularity, and (4) metafunctional regulation (see also Williams 1995; Hasan 1996, 2009e, 2013, 2014).

13.2.2.1 Open Context While Halliday’s network was ‘strategic’ in nature, inviting a description of the meaning of a given situation (Matthiessen 1990, 2015b), Hasan’s network has been developed to be contextually open. In Hasan’s view, a situationspecific semantic network is less desirable, if not impractical, when investigating the meanings at play in her mother–child talk research since the talk per se entails extensive contexts (Hasan et al. 2007).3 In this sense, rather than perceiving the context specificity as a ‘categorical one’, she argues that one should view it as a ‘relative matter’ so that the semantic networks could serve as a ‘heuristic device for the definition of a specific class of context of situation’ (Hasan 2009a:151). In so doing, she relocates the networks from the mid region of the cline of instantiation to the potential end of the cline, aiming at a description of general semantic systems, or more precisely, an account of the ‘meaning potential of English’4 (Hasan et al. 2007:712; see also Fung 2016). One important consequence is that semantic networks are no longer strategic, but essentially social in the sense that the approach focuses on the nature of the linguistic meanings in general. 13.2.2.2 Unit of Analysis: Message In presenting a more general conception of semantic networks, Hasan postulates a hierarchy of units, or more specifically, a four-unit rank scale in English, moving from the highest to the lowest: text ~ rhetorical unit ~ message ~ seme (Hasan 2013).5 As in other language-internal strata, these units stand in a relation of constituency, that is, a text is made up of rhetorical units, a rhetorical unit of messages, and a message of semes. Among the four units, Hasan regards message as the ‘ultimate descriptum’ in semantics (Hasan 2014:10), defining it as ‘the smallest semantic unit which is capable of realising an element of the structure of a text’ (Hasan 1996:117). For Hasan, it is this descriptum which serves as the object of enquiry in semantics, and is described exhaustively in system networks in terms of semantic options (for a recent account, see Hasan 2013, 2014). As pointed out by Hasan (1989:245), a message can be further categorized in terms of its productivity. A message

3

As illustrated in Hasan’s subsequent work in contextual modelling, mother–child talk is essentially registerially/ contextually inconsistent, entailing frequent reclassifications of con/text as the talk develops (see Cloran 1999; Hasan 1999, 2000 for a detailed discussion on con/textual shift).

4

Though the account of meaning potential is robust, such descriptions, as noted by Hasan, are not yet exhaustive. Further tests and applications are thus needed.

5

It should be emphasized that the term ‘message’ is also used in another sense in SFL, denoting the textual unit of meaning (Halliday and Matthiessen 2004:588–9). In this chapter, ‘message’, following Hasan, refers to the semantic rank scale only.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

339

which is productive and forms structural paradigms is termed a progressive message. That is, it concerns proposition/proposal exchange and is typically realized in a major clause. By contrast, a message which is non-productive and serves to manage the ongoing interaction is termed a punctuative message. The default realization of punctuative messages is a minor clause, such as Hello, Sorry, or Hey. According to Hasan, it is only progressive messages which are metafunctionally regulated (for detail, see Section 13.2.2.4; see also Hasan 2013 for a recent review).

13.2.2.3 Trinocularity Following Halliday, each semantic option in Hasan’s message semantics network attaches fundamental importance to the ‘concept of trinocularity’ (see Halliday 2009:79–80). That is, Hasan’s semantic networks not only concern the interrelations among semantic options (i.e. whether the options postulated are internally duplicate or contradictory (Hasan 1996:110)), but also emphasize the relations with context (i.e. what contextual features are construed) and lexicogrammar (i.e. what lexicogrammatical patterns are activated). In other words, the analysis of meaning through utilizing semantic networks not only illustrates the meanings at risk, but also enables analysts to explain ‘why and how something is said’ and ‘why these patterns of wordings appear rather than any other’ (Hasan 2009c:170). 13.2.2.4 Metafunctional Regulation In Hasan’s message semantics networks, semantic options are ‘multi-focal’ (Hasan 1996:111) in the sense that they relate not only inter-stratally, but also metafunctionally, i.e. each semantic option pertains to the highly generalized functions of language, which are known as metafunctions. Early in the 1970s, Halliday identified three metafunctions: the ideational, the interpersonal, and the textual, where the ideational metafunction is further categorized into the experiential and the logical subtypes (for details, see Halliday and Hasan 1985: Chapter 2). Central to this metafunctional hypothesis is that language is functional in the sense that the functions of a language are the ‘fundamental principle of language’ and are ‘basic to the evolution of the semantic systems’. Hasan incorporates Halliday’s metafunctional hypothesis into her semantic descriptions, arguing that a progressive message entails four simultaneous systems (from Hasan 1989:224): (a)

interpersonal meanings, for example, options in message function (questioning, informing, commanding . . .), options in personal evaluation, point of view; (b) experiential meaning, for example, the ascription of actional, evolutional, etc. roles, identification, definition; construction of time; (c) logical meaning, for example, cause, condition, meta-textual relations; (d) textual meanings, for example, options in topic maintenance, topic changes.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

340

AN DY F U NG AN D FR ANCI S ROB ERT LOW

Figure 13.2 The overall organization of semantic networks of ‘progressive message’ (Hasan 2013:286)6 Gloss: (1) the arrow indicates the point of entry to the sub-systems; (2) the dots indicate the name of the sub-systems.

These four systems of meanings are termed relation enactment, continuation, amplification,7 and classification respectively (Hasan 2013).8 It should be emphasized that these four systems are simultaneous, that is, they are of equal status and no cluster of meaning is more powerful or more important than the others. As one moves through the networks and chooses options, choices are made from most primary (at the left-hand end of the systems) to most delicate (choices at the right-hand end of the systems). The increase of degree of delicacy yields a full account of meanings within a single message. Figure 13.2 illustrates the overall organization of message semantics networks. Hasan’s proposal for metafunctionally regulated semantic networks, compared with Halliday (1973), constitutes a significant advance in semantic description because not only can subtle meaning differences be captured, but it also enables analysts to explore the calibration of context, semantics, and lexicogrammar (see Section 13.4).

6

Due to space constraints, Figure 13.2 only includes the primary sub-system for a message with the feature [progressive] under the four metafunctions. For example, the topic indicates the sub-system of CONTINUATION, selecting between [turn-maintaining] and [turn-changing]. Unfortunately, lack of space precludes a detailed discussion of each semantic feature. For detail, see Williams (1995); Lukin (2012, 2013).

7

It should be noted that the term AMPLIFICATION was previously used in Martin’s earlier accounts of APPRAISAL (see Martin 2000). However, it has been re-labelled GRADUATION in Martin and White (2005). Following Hasan, the term AMPLIFICATION used here refers to the semantic system of logical meanings.

8

Contra Halliday, Hasan separates the experiential metafunction from the logical metafunction, leading to four systems of meanings.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

341

In summary, these four primary systems constitute a major advance in semantic description, leading to our current conception of ‘message semantics networks’ (e.g. Hasan 1996, 2009e; Hasan et al. 2007). Such networks, as shown in the following sections, have a strong descriptive power which enables a variety of research problems to be tackled.

13.3

Application of Semantic Networks: An Investigation of Semantic Variation

As noted in our introductory section, the genesis of Hasan’s semantic networks lies in the need to investigate semantic variation, the central research agenda of Hasan and her colleagues in the early 1980s (see also Cloran 2000; Williams 1995, 2005; Hasan et al. 2007). Prior to illustrating how Hasan’s semantic networks contribute to the understanding of semantic variation, we briefly revisit the conception of semantic variation (see Section 13.3.1), followed by a review of her research into mother–child interaction (see Section 13.3.2). Section 13.3.3 then illustrates how analysis utilizing semantic networks serves as the linguistic evidence of semantic variation.

13.3.1 The Conception of Semantic Variation Sociolinguistic studies focusing on linguistic variation are not in themselves novel; they have a long history, with rich descriptive accounts focusing on phonological or morphosyntactic variation (e.g. Labov 1972, 1978). While the Labovian framework of variation has gained widespread acceptance in sociolinguistic research, Hasan’s work on linguistic variation is unique and innovative in the sense that her approach is meaning– and sociologically– oriented, and situated within the Systemic Functional model of language. Such a pioneering approach, as remarked by Hasan, reflects her dissatisfaction with sociolinguistic variation studies conducted in the 1970s, in terms of (i) analytical framework and (ii) variation explanations. With regard to the former, Hasan recognizes that variation frameworks which focus almost exclusively on phonology and lexicogrammar are essentially ‘meaning preserving’, thereby giving no place to meaning variations. For Hasan, neither phonology nor lexicogrammar is the ‘site of socially significant variation’ (Hasan 2011:xxxvii) – it is the level of semantics which entails ‘all the necessary characteristics of language varieties’ (Hasan 2009a:144). Thus, rather than perceiving semantics as ‘immune to variation’ (Hasan 1989:269), Hasan takes semantics as the point of departure in her study, with a particular focus on the ‘systematic differences in selection and organisation of linguistic meanings’ (Hasan 2009a:144, see also Hasan 1989, 2009b, 2009d, 2009f). Hasan postulates that approaches to linguistic variation which feature no social theory in explaining variation

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

342

AN DY F U NG AN D FR ANCI S ROB ERT LOW

are undesirable. In so doing, she turns to British sociologist Basil Bernstein, who postulates a coding orientation concerning the legitimacy and appropriateness of meanings.9 According to Bernstein, meanings do not exist independently from social realities. More specifically, it is social class, the ‘fundamental dominant cultural category’ (Bernstein 1975:175), which exerts ‘the most formative influence upon the procedures of socialisation’ (Bernstein 1987:37; also see Bernstein 1971, 1990, 2000 for a detailed discussion).

13.3.2 The Research Design of Hasan’s Mother–Child Research Bringing these two perspectives together, Hasan argues that if linguistic meanings vary from one social class to another, it follows that the distinctive meaning patterns implicated by speakers will not be merely ‘expressive, stylistic matter’ which are ‘totally empty of cognitive content’ (Hasan 2009f:116) but will correlate with the speakers’ social class. To investigate the extent of correlation, Hasan conducted an investigation focusing on twenty-four mother–child dyads (children aged between 3.6 to 4.2 years). These mothers and children were categorized into two contrasting social class groups termed High Autonomy Professionals (HAP) and Low Autonomy Professionals (LAP). According to Hasan, the distinction between HAP and LAP lies in the degree of professional autonomy of the breadwinners of the participating families. Breadwinners who exerted a high degree of autonomy in their workplace were categorized as HAP, while those who imposed little control over their working life and practices were classified as LAP. In Phase 1 of her mother–child talk research, Hasan (2009a:144) asks: ‘Does the selection and organization of linguistic meanings vary in correlation with variation in social class? If yes, then how can that variation be interpreted?’ The most pressing issue in answering these questions concerns the way in which linguistic meanings are conceptualized and analyzed so that viable claims can be made about the correlation between social class and meaning patterns. As discussed in Section 13.2.2, the very conceptualization of meaning adopted in Hasan’s network is a functional one. Central to the model of language is that it places much emphasis on society and language, and more importantly, the dialectic relation of realization functioning across social organization, social context, and language (Hasan 1989:271). In this sense, one could estimate the meaning orientations based on the social class of speakers, and by the same token, one could predict the social class of speakers based on the ways in which meanings are construed.

9

‘Meanings’ refers to all modalities of semiosis in Bernstein’s coding orientation. Hasan, by contrast, takes a restricted view of meaning, with a particular focus on the modality of language. Such a restricted view on coding orientation, in Hasan’s word, is termed semantic orientation.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

343

13.3.3

The Selection and Organization of Linguistic Meanings in Mother–Child Talk To exemplify Hasan’s approach in analyzing linguistic meanings, we reproduce a fragment of the semantic network of relation enactment, viz., the network of demanding information, i.e. of questions (see Figure 13.3), and use a segment of interaction between a mother and her daughter, Donna, for illustration (see Text 13.1).10 We first describe the semantic options in the question network, followed by an illustration of how such a network could enable analysts to illuminate the selection and organization of meanings in dialogue. As shown in Figure 13.3, question is the result of the simultaneous selection of two semantic options, viz., [demand] and [information]. The combination of these two options [demand; information] permits an entry to more delicate systems of question, entailing a selection between two mutually exclusive options: [confirm] or [apprize].11 According to Hasan, [confirm] is interpreted as a question which aims to elicit a ‘yes-no response’, whereas [apprize] aims to elicit ‘some specific element of information’ (Hasan 2009d:243). The system labelled G in the network in Figure 13.3 provides a further specification of the choice [confirm], selecting either [verify] or [enquire]. Questions selecting the option [verify] constitute ‘tagged questions’ in

Figure 13.3 Options in expressing questions: a simplified fragment (see Hasan 1989:246; Hasan et al. 2007:713) Gloss: Each system in the QU ESTION network is labelled. For example, the primary options are G and H in Figure 13.1, and each of the successive systems is labelled a, b, c . . ., and finally each of the terms is labelled 1, 2, 3 . . .

10 11

This short excerpt of interaction is taken from Hasan (2009e). Following Hasan’s recent (2014:17) account, distinctions between the terms (i) option, (ii) choice, and (iii) feature deserve to be noted. Briefly, the term option is ‘choose-able’ in the sense that it refers to the ‘as-yet-unexplored property of potential’ in the system; the term choice, by contrast, denotes ‘the option selected for further exploration’; and the term feature refers to the ‘properties of unit under description’ (see also Hasan 2013 for a distinction between choice and feature).

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

344

AN DY F U NG AN D FR ANCI S ROB ERT LOW

Text 13.1 Extract of interaction between mother and Donna Message ID 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99

Speakers

Message

Mother: Donna: Mother: Donna: Mother:

you know who that picture is of? pardon? do you know who that picture is of? my Daddy No looks like your Dad, doesn’t it? but it’s not him who is it? [?Daddy]? it is his Daddy he’d be your grandfather, if he were alive today pardon? he would be your grandfather if he were alive today your Daddy’s Daddy [?his] Daddy Daddy? your Daddy’s father my [?grandma] your grandfather, mm but you don’t know him grandma’s husband, he’s now Martin, isn’t he? but he’s - she’s remarried, I mean she’s remarried your father’s .. father .. died remember we went up to see his grave a little while ago? .. remember? .. where he’s buried in the cemetery? no .. you’d remember if we took you back there, I think

Donna: Mother:

Donna: Mother:

Donna: Mother: Donna: Mother:

Donna: Mother:

English, which entail a further selection of one of two options: [probe] or [reassure]. According to Hasan, the feature [probe] functions to ‘probe the veracity of a presented thesis’ whereas the feature [reassure] seeks ‘to be reassured about its veracity between the interactants’ (Hasan 2009d:246). While both options imply a need for verification, there is a subtle semantic difference in the sense that the latter enacts a ‘minimum social distance between interactants’, whereas the former does not (Hasan 2010:293). The semantic choice [enquire] permits an entry to both system c and d, selecting options either [ask] or [check] as well as [assumptive] or [non-assumptive]. Questions with the feature [ask], in Hasan’s view, are the most neutral way of eliciting a yes/no response in English – what Hasan refers to as ‘nonattitudinal questions’. Questions selecting the feature [check], in many cases, are ‘attitudinally marked’, or attempt to ‘draw an attestation from the addressee’ (Hasan, 2009d:246). These questions will simultaneously enter system d, or more precisely, the system of assumptiveness, selecting the option [assumptive] or [non-assumptive]. An [assumptive] question is a question where an ‘unvoiced assumption’ is made by the speakers (Hasan 2009d:249). Lexicogrammatically, the feature [assumptive] is realized by a

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

345

Table 13.2 Realization of options from system G in Figure 13.3 Lexicogrammatical realization Semantic option G a1 a2 b1

confirm verify enquire probe

b2

reassure

c1 c2 d1 d2

ask check assumptive non-assumptive

Systemic realization

Structural realization

major: indicative major: indicative: declarative: tagged major: indicative major: indicative: declarative: tagged: constant

S-F S^F . . . F^S

(i) S^Fneg . . . Fneg^S (ii) S^pos . . . Fpos^S major: indicative: declarative: tagged: reserved (i) S^Fneg . . . Fpos^S (ii) S^Fpos . . . Fneg^S major: indicative: interrogative: polar F^S^P . . . major: indicative: declarative: untagged tone 2 S^F^P . . . / Tone 2 F preselects negative polarity S ^ Fneg . . . preselects positive polarity S^Fpos . . .

Table 13.3 Selection expressions and examples of some choices in system G Selection expressions

Examples

[confirm: verify: reassure]

(i) You love Uncle Matt, don’t you? (ii) You don’t love Uncle Matt, do you? (i) You love Uncle Matt, do you? (ii) You don’t love Uncle Matt, don’t you?* Do you love Uncle Matt? Don’t you love Uncle Matt? You love Uncle Matt?

[confirm: verify: probe] [confirm: enquire: ask: non-assumptive] [confirm: enquire: ask: assumptive] [confirm: enquire: check: nonassumptive] [confirm: enquire: check: assumptive]

You didn’t love Uncle Matt?

* As noted in Hasan (2013), this is normal usage in some varieties of Australian English.

clause with a negative polarity. The lexicogrammatical realizations of options under system G and their examples are summarized in Table 13.2 and Table 13.3 respectively. In contrast to system G, system H presents another semantic environment of questions, which is traditionally known as wh-question in English. Questions with the option [apprize] permit entry into a more delicate system, selecting either [precise] or [vague]. According to Hasan, a [vague] question is ‘vague’ in the sense that the item of information sought has not been specified clearly and is only interpretable with reference to another message. A [precise] question, by contrast, states precisely what item of information is being sought. Depending on the type of information, the choice [precise] can be further categorized into two options – [explain] and [specify] – where the former concerns why and how, and the latter focuses on the identity of participant or the specification of circumstance such as when and where, etc. (Cloran 2000:164). Like [enquire: confirm], the choice [explain] permits entry into the system of assumptiveness, selecting the option [assumptive] or [non-assumptive]. The lexicogrammatical

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

346

AN DY F U NG AN D FR ANCI S ROB ERT LOW

Table 13.4 Realization of options from system H in Figure 13.3 Lexicogrammatical realization

Semantic option H

Apprize

a1

Precise

a2

Vague

b1

Explain

b2

Specify

d1 d2

Systemic realization

Structural realization

major: indicative: interrogative: nonpolar major: indicative: interrogative: nonpolar major: indicative: non-polar: Rinterrogative elliptical: maximum: formal As a1; wh-conflated with Theme And circumstance of cause why or how ^ F^S^P (or equivalent expression e.g. what for) As a1; wh-conflated with Theme wh- outclassifies why and how

wh- S^F^P

(i)

(i)

major: indicative: non-polar: Residue-interrogative (ii) major: indicative: non-polar: Subject-interrogative Assumptive as b1; F preselects negative polarity non-assumptive as b1; F preselects positive polarity

wh- S^F^P preselects as Adjunct with what about + nom group wh-conflated with Residue Adjunct or Complement/wh^F^S^P

wh-conflated with Residue Adjunct or Complement/wh^F^S^P (ii) wh-conflated with Subject Subject/ wh - ^F^P Why ^ Fneg^ S ^ P . . . Why ^ Fpos^ S ^ P . . .

Table 13.5 Selection expressions and examples of some choices in system H Selection expressions [apprize: vague] [apprize: precise: explain: assumptive] [apprize: precise: explain: non-assumptive] [apprize: precise: specify]

Examples What about swimming? Why didn’t you stay? Why did you leave? (i) When is it now? (ii) Where did you go?

realization of options under system H and their examples are summarized in Table 13.4 and Table 13.5 respectively. Though Figure 13.3 is a highly simplified network, it is sufficient to show that the construal of English questions, in Hasan’s view, is essentially a selection and organization of linguistic meaning. That is, the speaker is making choices in meanings from the semantic network of expressing questions. In identifying semantic features, shunting across strata is required. For example, viewing from the co-text, the points of enquiry of message 71 and message 73 in Text 1 lie in the person in the photograph. Lexicogrammatically, they are realized in a subject-interrogative where who conflates with the subject. In this sense, the selection expressions of messages 71 and 73 are [demand; information: apprize: precise: specify]. Messages 76 and 91, by contrast, select another set of semantic options. For example, scrutiny of the linguistic context suggests that in message 76, Donna’s mother is verifying the correctness of her propositional content.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

347

Compared with [probe], the choice [reassure] sounds appropriate in this situation, considering that the social distance between mother and child is minimal. Viewed from below, message 76 is realized by a clause preselecting [declarative: tagged: reversed]. Bringing these two perspectives together, message 76 can thus be analyzed as [demand; information: confirm: verify: reassure] in the sense that Donna’s mother not only performs a verification, but also an elicitation of reassurance from Donna. Message 76 can thus be best paraphrased as ‘I think this guy looks like your Dad and I believe you do, too; is that so?’ Messages 95 and 96, on the other hand, present another semantic aspect of questioning, with a selection expression [demand; information: confirm: enquire: check: non-assumptive]. The use of the feature [check] in message 95 can be related to Donna’s response in message 86, where she implies some reservations in conceding that the person in the picture is her grandfather (i.e. Daddy Daddy?). Recognizing that her daughter’s reservation is contradictory to the reality, Donna’s mother thus ‘challenges’ such reservations by asking her daughter if she remembers that they had recently gone to see her grandfather’s grave. Message 95 is thus a question loaded with ‘additional information about the attitude of the speaker’ (Hasan 2010:293). This attitudinally marked question, as remarked by Hasan, is realized both lexicogrammatically (i.e. an elliptical clause preselecting [declarative: untagged]) and phonologically (i.e. a high rising tone).12 Though the transcription precludes a phonological annotation, the clause realizing message 95 preselects [declarative: untagged], which serves as key evidence of [check]. It should be emphasized that the semantic features identified typically do not exist in isolation but relate to others forming identifiable meaning clusters. As maintained by Hasan, language is not a set of rules but behaviour, which can be measured and calculated through a principal components technique. Through measuring and calculating the principal components, one could reveal the patterns in speakers’ ways of meanings. The statistical calculation of principal components further suggests that meaning variations exist between HAP and LAP families, in terms of the mothers’ style of control (Hasan 2009b) and questioning and answering behaviours (Hasan 2009a, 2009d). Subsequent research adopting Hasan’s approach in studying semantic variation also yields similar results. For instance, in the exploration of semantic variation in joint book-reading between families and schools, Williams (1995) has found that the types of supplementation and its configuration with speech functions differ significantly between HAP and LAP families.

12

Since the transcription is not phonologically annotated, one could equally interpret Message 95 as an elliptical interrogative, with an ellipted Mood element Do you. In this case, Message 95 would be analyzed as [demand; information: confirm: enquire: ask].

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

348

AN DY F U NG AN D FR ANCI S ROB ERT LOW

Presenting only the relation enactment, of course, does not provide an exhaustive analysis. The perspectives of classification, continuation, and amplification are also needed. However, the relation enactment analysis, even at this preliminary stage of development, is already sufficient to suggest that the message semantics framework is essentially robust and analytical, taking into account a trinocular perspective. Hasan’s comprehensive ‘mapping’ of semantic features thus serves as a powerful tool in discriminating meaningful choices in dialogue, and more specifically, serves as linguistic evidence of social variation.

13.4

The Application of Semantic Networks in Discourse Studies

The strong descriptive power of message semantics networks has attracted the attention of discourse analysts who are working on aspects other than semantic variation. This is perhaps not surprising because the primary objective of discourse analysis is the ‘study of language in use’ (Gee and Handford 2012:1). The concern of ‘language in process’ in message semantics networks (Hasan 1996:124) thus fits into this research agenda. Table 13.6 summarizes the most relevant discourse studies featuring message semantics networks as the research tool.

13.4.1

Illustration: Semantic Networks in Pedagogical and Journalistic Discourses Due to space constraints, this section reports only two extended uses of message semantics networks in discourse analysis, and discusses the ways in which the networks are used and what their research implications are in discourse studies.

Table 13.6 Applications of message semantics networks Domain

Foci

Discourse studies

Legal

Court room Police interview Online chat

Maley and Fahey (1991) Hall (2004) Wong (2009)

Classroom teaching Early childhood education

Wake (2006); Chu (2011); Williams (1995); Torr (2004); Kim (2014) Hasan et al. (2007) Lukin (2012, 2013) Lukin et al. (2011) Fung (2016); Moore (2016)

Computer-mediated communication Education

Business Journalism Health and medicine

Service encounter News interview Surgical interaction Doctor–patient communication

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

349

Hasan’s message semantics networks have been applied in pedagogical discourse (e.g. Wake 2006; Chu 2011). Wake’s (2006) study, to a certain extent, shades into Hasan’s (2009d) work, being concerned with how learning is achieved dialogically. While Hasan is concerned with mode of learning in mother–child talk, Wake focuses on dialogic learning in a group of second-language international students, examining its effectiveness in the context of a university curriculum. Her case study analysis reveals an interesting phenomenon: students ask more questions than the tutor in the tutorial talk, with a frequent selection of [explain] in apprize questions and [ask], [check], and [validate] in confirm questions.13 Central to this question distribution is that dialogic learning entails a shift of classroom dynamics. That is, contrary to traditional classroom learning contexts where tutors enact the majority of the questions, it is students who frequently pose questions to seek explanation and confirmation in the university context, thereby unwittingly changing ‘the focus and direction of the lecturer’s explanation’ (Wake 2006:199). Similar to Wake (2006), Chu (2011) applies semantic networks to the classroom context of a New Arrival Programme (NAP), with a particular focus on the newly arrived students in South Australia. Offered by the Department of Education and Children’s Services, NAP aims to prepare the newly arrived students for learning the English needed for living and studying in South Australia. Chu aims to investigate the exploitation of meanings of visual and verbal modes in multimodal picture books, as well as the ways in which teachers engage with students during teacher–student interaction through picture books. Important in Chu’s (2011) work is the reconceptualization of the interpersonal functions of questions in the context of picture book reading, drawing on Hasan’s message semantics networks. Chu argues that even though teachers pose the same type of questions during picture book reading, the communicative functions vary in accordance with student literacy levels. For instance, while [apprize: precise: specify] questions are widely employed in both higher and lower literate students, their degree of interaction and points of enquiry differ. In higher-literate groups, the interaction between teachers and students is less restrictive, in the sense that teachers aim to invite students to contribute their ‘personal experience and ideas for interpretation’. In other words, questions selecting [apprize: precise: specify] in higher-literate groups aim to ‘probe further into students’ views or opinions’, as in What thoughts do you have?. By contrast, the degree of interaction between teachers and students in lower-literate groups is more restrictive, and the questions posed by teachers aim ‘to retrieve and to verbalise the found information’ of the multimodal texts, as in And what are the pictures we can see? (Chu 2011:228). 13

Wake (2006) uses the feature [validate] to refer to questions which are realized by clauses preselecting [declarative: Adjunct right?], as in The price is part of the world price, right?

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

350

AN DY F U NG AN D FR ANCI S ROB ERT LOW

Another illuminating use of message semantics networks in discourse studies is Lukin (2012, 2013). Lukin’s primary concern lies in journalistic discourse, or, more specifically, the professional performance of journalists in current affairs interviews. To investigate and characterize the mode of interviewing of Kerry O’Brien, the Australian senior political journalist of the Australian Broadcasting Corporation (ABC), Lukin (2013) adopts Hasan’s message semantics networks, with a particular focus on the choices of meaning pertaining to questions. Focusing on Kerry O’Brien as a case study, Lukin argues that current affairs programmes deserve particular attention because journalists in news interviews might not be performing the ‘core democratic functions’ of journalism (Clayman and Heritage 2002:2),14 but working in the service of the interviewees, allowing them to construe the affairs in accordance with their own purposes. Lukin’s primary use of message semantics is to discriminate the meaningful choices enacted by the speaker. She analyzes O’Brien’s questions from a multidimensional perspective, discriminating among the choices of meaning in the systems of relation enactment, continuation, amplification, and classification. She finds that O’Brien’s questions frequently select the features [confirm], [topicchanging], and [non-prefaced]. According to Lukin, the feature [topicchanging] denotes a change of topicality in play, whereas [non-prefaced] refers to messages which concern ‘what the world is like’ rather than inquiring about ‘someone’s . . . mental representation of the world’ (Hasan 2009d). The combination of these features suggests that O’Brien only touches on the issue in a general sense, with fewer follow-up questions (i.e. [topic-changing]), and his questions fail to invite the mental representations of interviewees (i.e. [non-prefaced]). In other words, rather than encouraging the interviewees to account for their views concerning the Iraqi invasion, Lukin argues that O’Brien’s questioning is essentially following the interviewee’s ideological direction, and his news interviews serve as the platform for those ‘military experts’, opening the floor to them to cast their messages in their own ways. Lukin (2013) demonstrates that Hasan’s message semantics functions not only as a tool in discriminating the meanings enacted by speakers, but also as a tool in revealing invisible ideologies in professional practices, or in Bartlett and Chen’s (2012:10) words, making ‘visible key features and functions of professional practice that are, or have become, invisible to the practitioners themselves and so to those being apprenticed into their practices’.

14

Examples of ‘core democratic functions’, as stated by Clayman and Heritage (2002:2) include ‘soliciting statements of official policy, holding officials accountable for their actions, and managing the parameters of public debate, all of this under the immediate scrutiny of the citizenry’.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

13.5

351

Concluding Remarks

This chapter has demonstrated one claim, namely, that Hasan’s message semantics network is essentially a discourse analytical tool enabling discourse analysts to study language in use in various contexts. In Section 13.2, we highlighted the theoretical foundations of semantic networks, and more specifically, the key advancements that Hasan has made, which contribute to our current understanding of message semantics. Since it is not our intention to repeat all the ideas which have been discussed in the previous literature, we have deliberately kept the discussion short and precise. However, the issues that we have highlighted are sufficient to demonstrate that Hasan’s message semantics is a powerful tool for discriminating among the meaningful choices enacted by interlocutors in a dialogue. Given this significance, it is therefore not surprising that such a paradigmatic description of semantics has been extensively applied in various discourse studies. Hasan’s semantic networks are thus not just a tool for semantic variation or integrated sociolinguistic research (see Section 13.3), but essentially a discourse analytical tool for meaning analysis. We have also surveyed the use of semantic networks in discourse studies, presenting an up-to-date review of the different uses of message semantics (see Table 13.4). Space precludes a detailed discussion of all the domains of application of semantic networks; the illustrations of pedagogical and journalistic discourses are, we hope, sufficient to exemplify in what ways message semantic networks are employed by discourse analysts to tackle various research problems. For Hasan (2005:56), ‘why and how language works’ and ‘the nature of the relationship between language and society’ are two sides of the same coin. While one of the primary tasks of discourse analysts is to interpret and make sense of the meaning of what people say and write and listen to and read in context, the ability to calibrate context to the semantics and to lexicogrammar in Hasan’s contextually open semantic networks thus serves as a powerful analytical tool in analyzing meanings in dialogue, enabling discourse analysts to understand ‘why and how language works’.15

References Bartlett, T. and H. Chen. 2012. Applying Linguistics in Making Professional Practice Re-Visible. Journal of Applied Linguistics and Professional Practice 9(1): 1–12. 15

Recent years have witnessed an increase in studies describing context through system networks (e.g. Butt 2004; Hasan 1999, 2009c, 2014; Bowcher 2007, 2014). Such descriptions enable analysts to integrate network-based descriptions from context to semantics to lexicogrammar.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

352

AN DY F U NG AN D FR ANCI S ROB ERT LOW

Bernstein, B. 1971. Class, Codes and Control: Theoretical Studies towards a Sociology of Language, Volume 1. London: Routledge. Bernstein, B. 1975 Class, Codes and Control: Applied Studies towards a Theory of Educational Transmission, Volume 3. London: Routledge. Bernstein, B. 1987. Social Class, Codes and Communication. In V. Ammon, N. Dittmar, and K. J. Matthier, eds., Sociolinguistics: An International Handbook of the Science of Society. Berlin: Walter de Gruyter. 536–79. Bernstein, B. 1990. Class, Codes and Control: The Structure of Pedagogic Discourse, Volume 4. London: Routledge. Bernstein, B. 2000. Pedagogy, Symbolic Control and Identity: Theory, Research, Critique. 2nd ed. Oxford: Rowman and Littlefield. Bowcher, W. L. 2007. Field and Multimodal Texts. In R. Hasan, C. M. I. M. Matthiessen, and J. J. Webster, eds., Continuing Discourse on Language, Volume 2. Sheffield: Equinox. 619–46. Bowcher, W. L. 2014. Issues in Developing Unified Systems for Contextual Field and Mode. Functions of Language 21(2): 176–209. Butt, D. 2004. Parameters of Context: On Establishing the Similarities and Differences between Social Processes. Unpublished mimeo, Macquarie University. Chu, P. Y. 2011. Picture Book Reading in a New Arrival Context: A Multimodal Perspective on Teaching Reading. PhD Thesis, University of Adelaide. Clayman, S. and J. Heritage. 2002. The News Interview. Cambridge: Cambridge University Press. Cloran, C. 1994. Rhetorical Units and Decontextualisation: An Enquiry into Some Relations of Context, Meaning and Grammar. Monographs in Systemic Linguistics 6. Nottingham: University of Nottingham. Cloran, C. 1999. Context, Material Situation and Text. In M. Ghadessy, ed., Text and Context in Functional Linguistics: Systemic Perspectives. Amsterdam: John Benjamins. 219–328. Cloran, C. 2000. Socio-semantic Variation: Different Wordings, Different Meanings. In L. Unsworth, ed., Researching Language in Schools and Communities: Functional Linguistic Perspectives. London: Cassell. 152–83. Cloran, C., V. Stuart-Smith, and Y. Young. 2007. Models of Discourse. In R. Hasan, C. M. I. M. Matthiessen, and J. J. Webster, eds., Continuing Discourse on Language: A Functional Perspective, Volume 2. Sheffield: Equinox. 647–70. Eggins, S. 1990. Keeping the Conversation Going: A Systemic-functional Analysis of Conversational Structure in Casual Sustained Talk. PhD Thesis, University of Sydney. Eggins, S. and D. Slade. 2004. Analyzing Casual Conversation. Sheffield: Equinox. Fung, A. 2016. Hasan’s Semantic Networks Revisited: A Cantonese Systemic Functional Approach. In W. L. Bowcher and J. Y. Liang, eds., Society in Language, Language in Society: Essays in Honour of Ruqaiya Hasan. London: Palgrave Macmillan.115–40.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

353

Gee, J. P. and M. Handford, eds. 2012. The Routledge Handbook of Discourse Analysis. Abingdon: Routledge. Gregory, M. 1985. Towards ‘Communication’ Linguistics: A Framework. In J. Benson and W. Greaves, eds., Systemic Perspectives on Discourse, Volume 1: Selected Theoretical Papers from the Ninth International Systemic Workshop. Norwood: Ablex. 119–34. Hall, P. 2004. Prone to Distortion? Undue Reliance on Unreliable Records in NSW Police Service Formal Interview Model. In J. Gibbons, V. Prakasam, K. V. Tirumalesh, and H. Nagrajan, eds., Language in the Law. New Delhi: Orient Longman Private Limited. 44–81. Halliday, M. A. K. 1973. Towards a Sociological Semantics. In M. A. K. Halliday, Explorations in the Functions of Language. London: Edward Arnold. 72–102. Halliday, M. A. K. 1984. Language as Code and Language as Behaviour: A Systemic-functional Interpretation of the Nature and Ontogenesis of Dialogue. In R. P. Fawcett et al., eds., The Semiotics of Culture and Language, Volume 1: Language as Social Semiotic. London: Pinter. 3–35. Halliday, M. A. K. 1994. An Introduction to Functional Grammar. 2nd ed. London: Arnold. Halliday, M. A. K. 2006a. Some Theoretical Considerations Underlying the Teaching of English in China. The Journal of English Studies (Sichuan International Studies University) 4: 7–20. Halliday, M. A. K. 2006b. Working with Meaning: Towards an Appliable Linguistics. In J. J. Webster. ed., Meaning in Context: Implementing Intelligent Applications of Language Studies. London: Continuum. 7–23. Halliday, M. A. K. 2008. Complementarities in Language. Beijing: The Commercial Press. Halliday, M. A. K. 2009. Methods – Techniques – Problems. In M. A. K. Halliday and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 59–86. Halliday, M. A. K. and R. Hasan. 1976. Cohesion in English. London: Longman. Halliday, M. A. K. and R. Hasan. 1985. Language, Context and Text: Aspects of Language in a Social-semiotic Perspective. Geelong: Deakin University Press. Halliday, M. A. K. and C. M. I. M. Matthiessen. 1999. Construing Experience through Meaning: A Language-based Approach to Cognition. New York: Continuum. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2004. An Introduction to Functional Grammar. 3rd ed. London: Edward Arnold. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Routledge. Hasan, R. 1983. A Semantic Network for the Analysis of Messages in Everyday Talk between Mothers and Their Children. Unpublished mimeo, Macquarie University.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

354

AN DY F U NG AN D FR ANCI S ROB ERT LOW

Hasan, R. 1984. Coherence and Cohesive Harmony. In J. Flood, ed., Understanding Reading Comprehension. Newark: International Reading Association. 181–219. Hasan, R. 1989. Semantic Variation and Sociolinguistics. Australian Journal of Linguistics 9(2): 221–76. Hasan, R. 1996. Semantic Networks: A Tool for the Analysis of Meaning. In C. Cloran, D. Butt, and G. Williams, eds., Ways of Saying, Ways of Meaning: Selected Papers of Ruqaiya Hasan. London: Cassell. 104–30. Hasan, R. 1999. Speaking with Reference to Context. In M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 219–328. Hasan, R. 2000. The Uses of Talk. In S. Sarangi and M. Coulthard, eds., Discourse and Social Life. London: Longman. 30–81. Hasan, R. 2005. Language in Society in a Systemic Functional Perspective. In R. Hasan, C. M. I. M. Matthiessen, and J. J. Webster, eds., Continuing Discourse on Language, Volume 1. Sheffield: Equinox. 55–80. Hasan, R. 2009a. Language in the Processes of Socialization: Home and School. In J. J. Webster, ed., Semantic Variation: Meaning in Society and in Sociolinguistics: The Collected Works of Ruqaiya Hasan, Volume 2. Sheffield: Equinox. 119–79. Hasan, R. 2009b. Meaning in Sociolinguistic Theory. In J. J. Webster, ed., Semantic Variation: Meaning in Society and in Sociolinguistics: The Collected Works of Ruqaiya Hasan, Volume 2. Sheffield: Equinox. 271–308. Hasan, R. 2009c. The Place of Context in a Systemic Functional Model. In M. A. K. Halliday and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 165–89. Hasan, R. 2009d. Questions as a Mode of Learning in Everyday Talk. In J. J. Webster, ed., Semantic Variation: Meaning in Society and in Sociolinguistics: The Collected Works of Ruqaiya Hasan, Volume 2. Sheffield: Equinox. 231–68. Hasan, R. 2009e. Semantic Variation: Meaning in Society and in Sociolinguistics: The Collected Works of Ruqaiya Hasan, Volume 2. Edited by J. J. Webster. Sheffield: Equinox. Hasan, R. 2009f. A Sociolinguistic Interpretation of Everyday Talk between Mothers and Children. In J. J. Webster, ed., Semantic Variation: Meaning in Society and in Sociolinguistics: The Collected Works of Ruqaiya Hasan, Volume 2. Sheffield: Equinox. 75–118. Hasan, R. 2010. The Meaning of ‘Not’ Is Not in ‘Not’. In A. Mahboob and N. K. Knight, eds., Appliable Linguistics: Texts, Contexts and Meanings. London: Continuum. 267–305. Hasan, R. 2011. A Timeless Journey: On the Past and Future of Present Knowledge. In D. N. Cheng, ed., Selected Papers of Ruqaiya Hasan on Applied Linguistics. Beijing: Foreign Language Teaching and Research Press. xiv–xliii.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

355

Hasan, R. 2013. Choice, System, Realization: Describing Language as Meaning Potential. In L. Fontaine, T. Bartlett, and G. O’Grady, eds., Systemic Functional Linguistics: Exploring Choice. Cambridge: Cambridge University Press. 269–99. Hasan, R. 2014. Towards a Paradigmatic Description of Context: Systems, Metafunctions, and Semantics. Functional Linguistics 2(9): 1–54. Hasan, R., C. Cloran, G. Williams, and A. Lukin. 2007. Semantic networks: The Description of Meaning in SFL. In R. Hasan, C. M. I. M. Matthiessen, and J. J. Webster, eds., Continuing Discourse on Language, Volume 2. Sheffield: Equinox. 697–738. Hyland, K., ed. 2013. Discourse Studies Reader: Essential Excerpts. London: Bloomsbury. Hyland, K. and B. Paltridge, eds. 2011. Continuum Companion to Discourse Analysis. London: Continuum. Kim, J. E. 2014. Parent Child Sharing Reading: The Affordances of Print, Digital and Hand-held Electronic Storybooks. PhD Thesis, The University of British Columbia. Labov, W. 1972. Sociolinguistic Patterns. Oxford: Basil Blackwell. Labov, W. 1978. Where Does the Sociolinguistic Variable Stop? A response to Beatriz Lavandera. Working Papers in Sociolinguistics Paper #44. Working Papers in Sociolinguistics. Austin: Southwest Educational Development Laboratory. 1–16. Lukin, A. 2012. Hasan’s Semantic Networks as a Tool in Discourse Analysis. In J. S. Knox, ed., Papers from the 39th International Systemic Functional Congress. Sydney: Organising Committee of the 39th International Systemic Functional Congress. 141–6. Lukin, A. 2013. Evaluating Questions in Journalism: A Case Study of the Australian Public Broadcaster’s Coverage of the 2003 Invasion of Iraq. Journal of Applied Linguistics and Professional Practice 9(1): 127–47. Lukin, A., A. Moore, M. Herke, C. Wu, and R. Wegener. 2011. Halliday’s Model of Register Revisited and Explored. Linguistics and the Human Sciences 4(2): 187–213. Mahboob, A. and N. Knight. 2010a. Appliable Linguistics. London: Continuum. Mahboob, A. and N. Knight. 2010b. Appliable Linguistics: An Introduction. In A. Mahboob and N. Knight, Appliable Linguistics. London: Continuum. 1–12. Maley, Y. and R. Fahey. 1991. Presenting the Evidence: Constructions of Reality in the Court. International Journal for the Science of Law 4(1): 3–17. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, J. R. 2000. Beyond Exchange: Appraisal Systems in English. In S. Hunston and G. Thompson, eds., Evaluation in Text: Authorial Stance and the Construction of Discourse. Oxford: Oxford University Press. 142–75. Martin, J. R. 2014. Evolving Systemic Functional Linguistics: Beyond the Clause. Functional Linguistics 1(1): 3.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

356

AN DY F U NG AN D FR ANCI S ROB ERT LOW

Martin, J. R. and D. Rose. 2007. Working with Discourse: Meaning beyond the Clause. 2nd ed. London: Continuum. Martin, J. R. and P. R. R. White. 2005. The Language of Evaluation: Appraisal in English. London: Palgrave. Matthiessen, C. M. I. M. 1988a. Representational Issues in Systemic Functional Grammar. In J. D. Benson and W. S. Greaves, eds., Systemic Functional Perspectives on Discourse. Norwood: Ablex. 136–75. Matthiessen, C. M. I. M. 1988b. Semantics for a Systemic Grammar: The Chooser and Inquiry Framework. In J. D Benson, M. J. Cummings, and W. S. Greaves, eds., Linguistics in a Systemic Perspective. Amsterdam: John Benjamins. 221–42. Matthiessen, C. M. I. M. 1990. Two Approaches to Semantic Interfaces in Text Generation. COLING 90: 322–9. Matthiessen, C. M. I. M. 1993. Register in the Round: Diversity in a Unified Theory of Register Analysis. In M. Ghadessy, ed., Register Analysis: Theory and Practice. London: Painter. 221–92. Matthiessen, C. M. I. M. 1995. Lexicogrammatical Cartography: English Systems. Tokyo: International Language Sciences Publishers. Matthiessen, C. M. I. M. 2004. The Semantic System of Relational Expansion: Rhetorical Structure Theory Revised. Unpublished draft, Macquarie University. Matthiessen, C. M. I. M. 2007. The Architecture of Language According to Systemic Functional Theory: Developments since the 1970s. In R. Hasan, C. M. I. M. Matthiessen, and J. J. Webster, eds., Continuing Discourse on Language: A Functional Perspective, Volume 2. Sheffield: Equinox. 505–61. Matthiessen, C. M. I. M. 2009. Ideas & New Directions. In M. A. K. Halliday and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 12–58. Matthiessen, C. M. I. M. 2012. Systemic Functional Linguistics as Applicable Linguistics: Social Accountability and Critical Approaches. D.E.L.T.A. 28: 435–71. Matthiessen, C. M. I. M. 2014. Appliable Discourse Analysis. In Y. Fang and J. J. Webster, eds., Developing Systemic Functional Linguistics: Theory and Application. Sheffield: Equinox. 138–208. Matthiessen, C. M. I. M. 2015a. Halliday on Language. In J. J. Webster, ed., The Bloomsbury Companion to M. A. K. Halliday. London: Bloomsbury. 137–202. Matthiessen, C. M. I. M. 2015b. Modelling Context and Register: The Longterm Project of Registerial Cartography. Letras, Santa Maria 25(50): 15–90. Matthiessen, C. M. I. M. and J. Bateman. 1991. Systemic Linguistics and Text Generation: Experiences from Japanese and English. London: Pinter. Moore, A. R. 2016. Can Semantic Networks Capture Intra- and InterRegisterial Variation? Palliative Care Discourse Interrogates Hasan’s Message Semantics. In W. L. Bowcher and J. Y. Liang, eds., Society in Language,

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

Semantic Networks

357

Language in Society: Essays in Honour of Ruqaiya Hasan. London: Palgrave Macmillan. 83–114. Patten, T. 1988. Systemic Text Generation as Problem Solving. Cambridge: Cambridge University Press. Torr, J. 2004. Talking about Picture Books: The Influence of Maternal Education on Four-year-old Children’s Talk with Mothers and Pre-school Teachers. Journal of Early Childhood Literacy 4(2): 181–210. Turner, G. J. 1973. Social Class and Children’s Language of Control at Age Five and Age Seven. In B. Bernstein, ed., Class, Codes and Control, Volume 2: Applied Studies towards a Sociology of Language. London: Routledge. 121–79. Wake, B. J. 2006. Dialogic Learning in Tutorial Talk: A Case Study of Semiotic Mediation as a Learning Resource for Second Language International Students. PhD Thesis, University of Adelaide. Williams, G. 1995. Joint Book-Reading and Literacy Pedagogy: A Sociosemantic Examination. PhD Thesis, Macquarie University. Williams, G. 2005. Semantic Variation. In R. Hasan, C. M. I. M. Matthiessen, and J. J. Webster. eds., Continuing Discourse on Language, Volume 1. Sheffield: Equinox. 457–80. Wong, Y. T. 2009. The Linguistic Function of Cantonese Discourse Particles in the English Medium Online Chat of Cantonese Speakers. MA Dissertation, University of Wollongong.

Downloaded from https://www.cambridge.org/core. Nottingham Trent University, on 27 Aug 2019 at 06:51:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.015

14 Discourse Semantics J. R. Martin

14.1

Discourse Semantics

Discourse semantics is the term used by Martin and his colleagues (after Martin 1992; Martin and Rose 2003) to refer to the stratum of meaning interfacing lexicogrammar with context (register and genre) in SFL. It comprises six major systems, organized into three metafunctions: ideation and conjunction (ideational); negotiation and appraisal (interpersonal); and identification and periodicity (textual). As reviewed in Martin (2014), this work reinterprets Halliday and Hasan’s model of cohesion (e.g. Halliday and Hasan 1976) from the perspective of Gleason’s work on text semantics (e.g. Gleason 1968) – as a set of text-forming resources realized through grammar, lexis, and intonation. From this perspective, Halliday and Hasan’s cohesive ties are reinterpreted as discourse structures; and the resources which Halliday (e.g. 2009:85) positions as non-structural components of the textual metafunction in grammar are reinterpreted at a deeper level of abstraction as discourse semantic systems. This reinterpretation foregrounds meaning beyond the clause as fundamental to semantic analysis in SFL and can be usefully compared with the clause semantics research foci inspired by Halliday and Matthiessen (1999), Hasan (2009), or Fawcett (2008). Halliday and Hasan (1985:82 in particular) can be read as developing work on cohesion in a similar direction.

14.2

Discourse Semantic Systems

As noted above, discourse semantic systems are organized by metafunction; this contrasts with Halliday and Hasan’s treatment of all cohesive devices as textual. As one would expect in SFL, they are modelled from the complementary perspectives of system and structure; but the relation of system to structure is not as tightly bound as in most SFL work on grammar and

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

359

phonology – where a particular selection of features prescribes a particular structural configuration, and where by the same token a particular structural configuration presumes a particular selection of features (see the discussion of axis in Martin 2013a). Following Bateman’s (2014) deployment of Peirce, it is more appropriate to think of discourse structures as semantic relations abduced between parts of a text, and of discourse systems as constraining the set of possible relations that can be abduced. For example, the identification relation abduced between discourse semantic systems in the first sentence of this paragraph and they in the second is presuming anaphora; and the relation abduced between the first sentence and the second in the previous ranking clause is one of presuming nominal ellipsis. Describing the discourse semantics of a given language involves establishing the range of discourse structures which can be abduced from discourse systems. Since we are abducing discourse relations rather than deriving them, text analysis necessarily involves some degree of ‘play’ in interpretation. At a given point in a text there may be more than one analysis available; there may be a need to revise analysis as the text unfolds; and there may in the end be multiple analyses available that cannot be resolved. Some analysts experience this difference between discourse interpretation and grammar analysis as a source of frustration; others experience it with a sense of liberation. As usual, coming to appreciate the complementarities involved in analyzing a clause and interpreting a text is the key to understanding how language has evolved so we can use it to live. Discourse semantic systems are briefly summarized below, drawing principally on Martin (1992), Martin and Rose (2003), and Martin and White (2005).

14.2.1 Ideational Systems: Ideation and Conjunction Two systems are involved here, ideation and conjunction. Their main function is to construe the register variable field, where field is defined as a set of activity sequences oriented to some global institutional purpose, including the taxonomies of entities involved in these sequences (Martin 1992). In Halliday and Matthiessen’s (1999) terms, ideation is concerned with the semantics of figures, and conjunction with the semantics of relations between figures. Ideation extends earlier work on lexical cohesion (Halliday and Hasan 1976). Drawing in part on Halliday and Hasan (1985), a more detailed account of possible relations among semantic units realized through lexical items is proposed – including repetition, synonym, antonymy, hyponymy, and meronymy. In addition a model of nuclear relations is proposed for semantic units typically realized through nominal groups, verbal groups, and clauses – drawing on Halliday’s notion of logical-semantic relations (i.e. elaboration, extension, and enhancement; Halliday and Matthiessen 2014); the purpose of this extension is to capture semantic relations grouped

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

360

J. R. MARTIN

together under the heading collocation in Halliday and Hasan (1976). The discourse structures afforded by these ideation relations are termed lexical strings. Conjunction integrates earlier work on cohesion (Halliday and Hasan 1976) and expanding clause complexes (Halliday and Matthiessen 2014) through consideration of semantic relations between figures, whether realized within or between clause complexes – i.e. cohesively between clause complexes, paratactically or hypotactically between clauses in a clause complex, or via logical metaphor (Halliday 1998) as a participant, process, or circumstance within a clause. Four main types of conjunctive relation are recognized: additive, comparative, temporal, and consequential. Halliday and Hasan’s important distinction between external and internal conjunctive relations is sustained. And the model recognizes the possibility of implicit conjunctive relations abduced in the interpretation of adjacent figures (with the constraint that any relation so abduced could be made explicit). The discourse structures afforded by these conjunctive relations are modelled as reticula, elaborating on Gleason’s (1968) notion of an ‘event-line’.

14.2.2 Interpersonal Systems: Negotiation and Appraisal The two systems involved here are negotiation and appraisal. Their main function is to enact the register variable tenor, where tenor is concerned with the relations of power and solidarity whereby speakers position themselves as interlocutors in discourse. In general terms we can think of negotiation as focusing attention on the inter-(personal) dimension of interpersonal meaning and appraisal as focusing attention on the (inter)personal. Negotiation draws on earlier work on exchange structure (Sinclair and Coulthard 1975) and speech function (e.g. Halliday 1984) and attends to the relation of one move to another in conversation. Following Berry (1981) and Ventola (1987), exchanges are modelled as consisting of between one and five basic moves, including the possibility of expansion through locally contingent tracking and challenging moves. Exchange rank systems and structures are realized at a lower rank through move systems, which are in turn realized inter-stratally through lexicogrammar and phonology. These inter-stratal relations can be direct or indirect, depending on whether interpersonal grammatical metaphor is involved. Negotiation analysis provides a useful scaffolding for consideration of the role in dialogue of substitution and ellipsis (Halliday and Hasan 1976) and of rhythm and intonation (Halliday 1967, 1970; Halliday and Greaves 2008). Appraisal provides a model of resources for evaluation, including types of attitude (affect, judgement, and appreciation), graduation (the strength and prototypicality of feelings), and engagement (the range of voices in play). Affect is concerned with emotion in relation to a trigger; judgement deals

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

361

with opinions about targeted behaviour; and appreciation focuses on the value of targeted semiotic and natural phenomena. All three types of attitude can be either positive or negative (in the sense of good feelings or bad ones), and all can be inscribed in a text through explicitly evaluative lexis or invoked through a choice in ideation which affords a reaction. Knight (2010, 2013) refers to shared feelings as bonds, inspiring recent SFL work on identity and affiliation (e.g. Martin et al. 2013).

14.2.3 Textual Systems: Identification and Periodicity The two systems involved here are identification and periodicity. Their main function is to compose discourse as waves of information texturing the register variable mode, where mode deals with the affordances of various media of communication (speaking, phoning, tweeting, texting, emailing, posting, writing etc.) in relation to turn-taking, aural/visual feedback, abstraction, and inter-modality. Identification extends the cohesion variable reference to include consideration of both how entities are introduced into a text and how they are kept track of once there (including consideration of how they are related to other entities through comparative reference). As explored in Gleason (1968) and Martin (1983), languages vary in terms of how they introduce and track entities through nominal deixis, Theme selection, and clause complex (‘switch-reference’) systems. Phoric entities presume information which has to be recovered from the co-text or nonverbal context, for which a range of recovery strategies is proposed (anaphora, cataphora, exophora, homophora, etc.). The discourse structures afforded by presuming endophoric (i.e. co-textual) identification relations are termed reference chains. Periodicity develops SFL work on theme and information structure (Halliday 1967, 1970; Halliday and Greaves 2008; Halliday and Matthiessen 2014). Drawing on work by Fries (1981) in particular, it extends Halliday’s description of information flow in the clause to higher levels of discourse. In this model Theme is interpreted as a clause resource for focusing our attention on a region of a field, and New as a complementary resource for developing what we want to say in that region of meaning. A hyper-Theme can then be interpreted as predicting the orientation to the field that will be composed through an ensuing pattern of Theme selections in a phase of discourse (what Fries refers to as the ‘method of development’ of a phase). And hyperNew can be interpreted as consolidating, often with some evaluative interpretation, the information composed through a preceding pattern of News (what Fries refers to the ‘point’ of a phase). Depending on the degree of planning and editing afforded by the mode, additional layers of thematic prediction and news consolidation may be found (i.e. macro-Themen and/or macro-Newn). Higher-level themes may be graphologically foregrounded as headings and perhaps a table of contents, where such occur.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

362

J. R. MARTIN

14.3

Working with Discourse Semantics: Text Synthesis

In this section I will try and illustrate the discourse semantic resources introduced above and address the practical concerns of this handbook by engaging in an exercise I will refer to as text synthesis. Text synthesis proceeds by building up a coherent text and thus contrasts with text analysis which starts with a finished product and breaks it down. Synthesis is rhetorical in orientation; it aims to demonstrate how linguistic resources can be deployed. As such it reflects a long-standing but until recently underutilized tradition in language teaching – one which features in my grandfather’s language textbooks in late nineteenth-century rural Canada, but had to be re-introduced in the genre-based literacy programmes designed by the ‘Sydney School’ (Rose and Martin 2012). Synthesis disappeared, one has to presume, because the knowledge about language available to students and teachers across sectors in education became so impoverished that it could not sustain rhetorically oriented text construction. Now, thanks particularly to SFL, the knowledge about language we need is readily available. So let us put it to use. As our starting point, consider the following string of alphabetically listed clauses. How might we begin to compose a text from these? [1]

Available finance has not been managed. Damaged paths and walls have not been repaired. Frescoes have not been preserved. In some places ancient lead water pipes have been exposed. Inexpensive mortar has also been used. Mangy dogs roam the site. No conservation and interpretation program has been put in place. No walkways for viewing platforms have been constructed. Over thirty different varieties of weed have been identified, including ivy, fennel, and fig. Over time this mortar has cracked. The roof collapsed. The roof could not support the weight of the tiles. The roots grow. The roots open up further cracks. The timber roof on the House of Meleager was designed for lightweight roofing. This allowed water and vegetation to penetrate. This allows even more weeds in. This was to protect ancient stonework. Tourists enter buildings that are not roped off. Tourists walk along ancient paths.

Let us begin with ideation. Nuclear relations have been provided for us, in order to make this exercise fit into a short chapter of this kind (for a

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

363

synthesis which builds nuclear relations, see Martin 2013b). So we can simply accept relations such as timber = roof, poorly x designed, damaged + paths and so on. As far taxonomic relations among entities are concerned, let me suggest the following re-organization: [2]

The timber roof on the House of Meleager was designed for lightweight roofing. The roof could not support the weight of the tiles. The roof collapsed. Inexpensive mortar has also been used. This was to protect ancient stonework. Over time this mortar has cracked. This allowed water and vegetation to penetrate. çè Over thirty different varieties of weed have been identified, including ivy, fennel, and fig. The roots grow. The roots open up further cracks. This allows even more weeds in. çè No walkways for viewing platforms have been constructed. Tourists walk along ancient paths. Tourists enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. çè Damaged paths and walls have not been repaired. Frescoes have not been preserved. Mangy dogs roam the site. Available finance has not been managed. No conservation and interpretation programme has been put in place.

The principal lexical strings underpinning this construal of the field are outlined in Table 14.1 – with headings highlighting the kinds of relation involved as mostly based on composition (co/meronymy)1 or mostly based on classification (co/hyponymy). In general terms then our gaze shifts from construction to vegetation to vantage points to aspects of the site overall. Table 14.1 Selected lexical relations in Text 2 ‘co/meronymy’

‘co/hyponymy’

‘co-meronymy’

‘co/meronymy’

timber = roof roof roof tiles mortar stonework mortar

(vegetation) varieties = weeds ivy fennel fig roots weeds

walkways viewing = platforms paths buildings lead = water = pipes

paths walls frescoes site

As a next step we can draw on graphology to scaffold the four phases of discourse as paragraphs. [3]

1

The timber roof on the House of Meleager was designed for lightweight roofing. The roof could not support the weight of the tiles. The roof collapsed. Inexpensive mortar has also been used. This was

In Table 14.1 the ‘co/meronymy’ headings indicate that lexical relations involving both meronymy (part/whole relations) and co-meronymy (part/part relations) are found; the ‘co/hyponymy’ heading indicates that both hyponymy (class/subclass relations) and co-hyponymy (subclass/subclass relations) are found.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

364

J. R. MARTIN

to protect ancient stonework. Over time this mortar has cracked. This allowed water and vegetation to penetrate. Over thirty different varieties of weed have been identified, including ivy, fennel, and fig. The roots grow. The roots open up further cracks. This allows even more weeds in. No walkways for viewing platforms have been constructed. Tourists walk along ancient paths. Tourists enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. Damaged paths and walls have not been repaired. Frescoes have not been preserved. Mangy dogs roam the site. Available finance has not been managed. No conservation and interpretation programme has been put in place. At this point we can bring identification into the picture and consider anaphoric identity chains introducing (technically speaking ‘presenting’) and tracking (technically speaking ‘presuming’) the entities introduced as the timber roof on the House of Meleager, poor quality mortar, over thirty different varieties of weed, and tourists. Each of these entities was fully lexicalized in Text 3; this redundancy is adjusted as in Text 4 below. [4]

The timber roof on the House of Meleager was designed for light-weight roofing. It could not support the weight of the tiles. It collapsed. Inexpensive mortar has also been used. This was to protect ancient stonework. Over time this mortar has cracked. This allowed water and vegetation to penetrate. Over thirty different varieties of weed have been identified, including ivy, fennel, and fig. The roots grow. They open up further cracks. This allows even more weeds in. No walkways for viewing platforms have been constructed. Tourists walk along ancient paths. They enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. Damaged paths and walls have not been repaired. Frescoes have not been preserved. Mangy dogs roam the site. Available finance has not been managed. No conservation and interpretation programme has been put in place.

The identity chains at issue here are outlined in Table 14.2. The mortar, weeds, and tourists chains are initiated non-phorically (via non-specific Table 14.2 Identity chains in Text 4 ‘roof’

‘mortar’

‘weeds’

‘tourists’

the timber roof . . . it it

inexpensive mortar this mortar

over thirty different varieties . . . the roots they even more weeds

tourists they

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

365

nominal deixis); the roof chain on the other hand involves both esphora (the timber roof pointing forward in its nominal group to its Qualifier on the House of Meleager) and homophora, since the House of Meleager is treated as assumed knowledge in this field. Thereafter pronouns and specific deixis are used to track entities (it, it; this mortar; the roots; they); and a ‘sub-entity’, the roots of the weeds, is introduced via bridging (taking advantage of the part/whole ideation relation between weeds and roots). The weeds chain also includes a comparative reference, introducing an additional set of weeds beyond the weeds initiating the chain. As we can see, even in a few phases of discourse of this kind, as far as identification is concerned there is a lot going on. This can be especially challenging for speakers coming from a language that manages identification differently from English. The lack of an obligatory presenting/presuming reference distinction in many languages is especially troubling (as teachers and supervisors of academic writing well know); and this problem may be exacerbated by the fact that in such languages the distinction between specific and generic reference is not explicitly grammaticalized. We should also note at this point three instances of text reference (technically ‘extended reference’) in Text 4. With text reference indefinitely long phases of meaning can be presumed (one or two sentences worth in the examples below). As the term implies, what is identified is phases of unfolding discourse rather than specific discourse semantic entities that are introduced and tracked as part of the construal of a field. Inexpensive mortar has also been used. This was to protect ancient stonework. Over time this mortar has cracked. This allowed water and vegetation to penetrate. The roots grow. They open up further cracks. This allows even more weeds in.

As an alternative to this text reference we can draw on conjunction to relate figures to one another. Each instance is reworked as a clause complex below (with the conjunctive relation connecting the figures specified between clauses and explicit connectors in italics). Note the contrast with the examples above, where conjunctive relations were realized inside the clause. There the text reference incorporated the relevant figure as a participant, which could then be connected to another figure through Process and Participant transitivity relations (for which see Halliday and Matthiessen 2014). Inexpensive mortar has also been used (explicit purpose) to protect ancient stonework. Over time this mortar has cracked, (implicit manner) allowing water and vegetation to penetrate. Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

366

J. R. MARTIN

As the roots grow (explicit simultaneity) they open up further cracks, (implicit manner) allowing even more weeds in. It is perhaps appropriate at this point to adjust the rest of the external conjunctive relations at stake here as in Text 5a below. [5a]

The timber roof on the House of Meleager was not designed to support the weight of the tiles and collapsed. Inexpensive mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. Over thirty different varieties of weed have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more weeds in. No walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has not been managed, and no conservation and interpretation programme has been put in place.

These relations are specified below; the specification (-) indicates that a comparative, temporal or consequential conjunctive relation cannot be made explicit between these two clauses (by convention, implicit additive relations are left unspecified, by way of lightening the workload for text analysts). Note that the use of non-finite clauses (e.g. allowing water and vegetation to penetrate) and branched paratactic clause complexes (e.g. so tourists walk along ancient paths and enter buildings that have not been roped off) means that certain entities are ideationally implicit, and are thus tracked here via ellipsis as far as identity chains are concerned. This further reduces the ideational redundancy apparent in Text 2 above. [5b]

The timber roof on the House of Meleager was not designed (explicit purpose) to support the weight of the tiles (explicit additive, implicit causal) and collapsed. (-) Inexpensive mortar has also been used (explicit purpose) to protect ancient stonework. (implicit succession)

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

367

Over time this mortar has cracked,2 (implicit manner) allowing water and vegetation to penetrate. Over thirty different varieties of weed have been identified, including ivy, fennel, and fig.3 (-) As the roots grow (explicit simultaneous) they open up further cracks, (implicit manner) allowing even more weeds in. No walkways for viewing platforms have been constructed, (explicit cause) so tourists walk along ancient paths (explicit addition) and enter buildings that are not roped off. (-) In some places ancient lead water pipes have been exposed. Damaged paths and walls have not been repaired, (implicit addition) frescoes have not been preserved, (explicit addition) and mangy dogs roam the site. (-) Available finance has not been managed, (explicit addition) and no conservation and interpretation program has been put in place. At this point we can turn to periodicity and consider how to scaffold the waves of information in the text. Text 5’s topical Themes are highlighted in bold below for each of its finite ranking clauses;4 (“) there indicates that the orientation to the field is being sustained through ellipsis (a more common pattern in many languages than in English, both within and between clause complexes). [5c]

2

The timber roof on the House of Meleager was not designed to support the weight of the tiles

Over time is interpreted here as a circumstance of Location, ideationally construing the time frame over which the stonework and mortar problem arose, rather than as a temporal connector realizing conjunction.

3

The dependent non-finite clause, including ivy, fennel, and fig, grammatically elaborates the varieties of weed that have invaded; accordingly it is not interpreted here as a distinct figure conjunctively related to the identification of the weeds.

4

Theme has not been analyzed in non-finite ranking clauses, with their non-finiteness interpreted as downgrading the figure as far as textual and interpersonal meaning are concerned.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

368

J. R. MARTIN

and (“) collapsed. Inexpensive mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks,5 allowing even more weeds in. No walkways for viewing platforms have been constructed, so tourists walk along ancient paths and (“) enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has not been managed, and no conservation and interpretation programme has been put in place. Theme’s complementary pulse of informational prominence, New, is highlighted below – assuming unmarked tonicity for each clause (Halliday and Greaves 2008). Minimal New is highlighted in bold (i.e. the Process, Participant, or Circumstance containing the tonic syllable); italics highlights my reading of the left-ward domain of New as far as including further information extending the field of each phase is concerned.6 [5d]

The timber roof on the House of Meleager was not designed to support the weight of the tiles and collapsed. Inexpensive mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more weeds in.

5

Following Martin and Rose (2003) Topical Theme in declarative clauses is analyzed up to and including the Subject.

6

Since intonation does not specify how much of the clause beyond the constituent containing the tonic syllable is involved, this reading has to be undertaken in relation to the ‘point’ of each phase – by including information relevant to expanding the field.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

369

No walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has not been managed, and no conservation and interpretation programme has been put in place. At this point we can consider adding a higher layer of periodicity, either prospective (hyper-Themes) or retrospective (hyper-News). We will lean toward layers of prediction here,7 adding hyper-Themes, highlighted in Text 6 below. [6]

Much of the restoration work on Pompeii has been done by local firms with no knowledge of restoration techniques. For example, the timber roof on the House of Meleager was not designed to support the weight of the tiles and collapsed. Inexpensive mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. The incursion of uncontrolled weeds has hastened the decay of the ruins. Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more weeds in. Pompeii’s position as an international tourist attraction brings half a million visitors each year. No walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. There seems to be no overall management plan for the site. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has not been managed, and no conservation and interpretation programme has been put in place.

7

Writers who proceed from a carefully constructed plan lean towards front-loading of this kind, since they know where they are going as they write; writers who figure out what they want to say as they work through successive drafts may find periodic summarizing, via hyper-News, more appropriate.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

370

J. R. MARTIN

These scaffolded phases of discourse can be explicitly connected to one another by drawing on the resources of internal conjunction. In Text 7 below, the first phase is thus connected to the second via internal succession (to begin, second), the second to the third via internal addition (in addition), and the fourth to the rest via internal culminative succession (finally). [7]

To begin, much of the restoration work on Pompeii has been done by local firms with no knowledge of restoration techniques. For example, the timber roof on the House of Meleager was not designed to support the weight of the tiles and collapsed. Inexpensive mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. Second, the incursion of uncontrolled weeds has hastened the decay of the ruins. Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more weeds in. In addition, Pompeii’s position as an international tourist attraction brings half a million visitors each year. No walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. Finally, there seems to be no overall management plan for the site. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has not been managed, and no proper conservation and interpretation programme has been put in place.

As noted above, identification (text reference) can function as an alternative to conjunction. So we might have connected up the third and fourth phase circumstantially in Text 7 above (e.g. in addition to this, on top of all this). Ideation can also be brought into play if we draw on metadiscourse to name phases, as factors, for example. This makes it possible to order factors by bringing comparative reference into the picture (e.g. a second factor, another factor, a final factor to consider). A range of conjunction (internal), identification (text reference and comparison), and ideation (metadiscourse) resources are used to link phases in Text 8 below. [8]

To begin, much of the restoration work on Pompeii has been done by local firms with no knowledge of restoration techniques. For example, the timber roof on the House of Meleager was not designed to support the weight of the tiles and collapsed. Inexpensive mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. A second factor is the incursion of uncontrolled weeds which have hastened the decay of the ruins. Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more weeds in.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

371

In addition to this, Pompeii’s position as an international tourist attraction brings half a million visitors each year. No walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. Finally, there seems to be no overall management plan for the site. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has not been managed, and no conservation and interpretation programme has been put in place. Let us now bring periodicity back into the picture, and cook up a rhetorical sandwich for the phases as a whole – introducing a macro-Theme to predict the hyper-Themes of each phase and a macro-New to consolidate their News. macro-Theme Since its discovery, Pompeii has been damaged as an archaeological site, affected by the quality of restoration work, vegetation, tourism, and site management issues. [phases 1–4] Damage to the site remains a key issue for archaeologists and administrators. There are ongoing concerns arising in relation to restoration work, uncontrolled vegetation, tourism, and management which need to be addressed if the site is to be preserved for future generations of research and public access. macro-New

To bring out the sense in which lower layers of periodicity reformulate higher level ones, the symbol ‘=’ (Halliday and Matthiessen’s 2014 elaboration) has been used to annotate the relevant relations in Text 9. In addition I have used metadiscourse (a number of factors) to name ensuing phases in the macro-Theme, and I have specified an internal consequential connection between the macro-New and the rest of the text by drawing on text reference, metadiscourse, and a circumstantial realization of cause (as a result of these factors).8 Note in passing that higher levels of periodicity regularly draw on grammatical metaphor and abstract ideation to ‘generalize’ the meaning specified in lower layers (e.g. the quality of restoration work, vegetation, tourism, site management issues in the Macro-Theme below). [9]

8

Since its discovery, Pompeii has been damaged as an archaeological site, affected by a number of factors, including the quality of restoration work, vegetation, tourism, and site management issues. =

We can compare this selection with some other possibilities: thus (conjunction only); as a result (circumstance of cause now lexicalized as a cohesive conjunction); as a result of this (conjunction plus text reference); as a result of these factors (conjunction plus text reference plus metadiscourse).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

372

J. R. MARTIN

To begin, much of the restoration work on Pompeii has been done by local firms with no knowledge of restoration techniques. = For example, the timber roof on the House of Meleager was not designed to support the weight of the tiles and collapsed. Inexpensive mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. A second factor is the incursion of uncontrolled weeds which have hastened the decay of the ruins. = Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more weeds in. In addition to this, Pompeii’s position as an international tourist attraction brings half a million visitors each year. = No walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been exposed. Finally, there seems to be no overall management plan for the site. = Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has not been managed, and no conservation and interpretation programme has been put in place. = As a result of these factors, damage to the site remains an issue for archaeologists and administrators. There are decisions which need to be made in relation to restoration work, uncontrolled vegetation, tourism, and management which need to be addressed if the site is to be preserved for future generations of research and public access. This bring us to appraisal, and the question of how and how far to make explicit the evaluative stance appropriate to a history text of this kind. One fundamental value in historical axiology is the preservation of archaeological sites, and so propagating a negative prosody is relevant here – including negative judgements of individuals, agents, and agencies failing to preserve the site, and negative appreciations of phenomena involved. Propagation of this prosody is highlighted in Text 10. [10]

To begin, much of the restoration work on Pompeii has been done by local firms with no specialized knowledge of restoration techniques. For example, the timber roof on the House of Meleager was so poorly designed it could not support the weight of the tiles and collapsed. Poor quality mortar has also been used to protect ancient

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

373

stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. A second factor is the incursion of dangerous weeds which have hastened the decay of the ruins. Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more threatening weeds in. In addition to this, Pompeii’s position as an international tourist attraction brings half a million visitors each year. No special walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been carelessly exposed. Finally, there seems to be no overall management plan for the site. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has been poorly managed, and no proper conservation and interpretation programme has been put in place. Here, three negative judgements have been inscribed, the first of which is intensified (so poorly designed, have been carelessly exposed, has been poorly managed); and there are six negative appreciations, the fourth quantified (no specialized9 knowledge of restoration techniques, poor quality mortar, dangerous weeds, even more threatening weeds in, no special walkways, no proper conservation and interpretation programme). This puts us in a position to predict the negative prosody of conservation problems in the text’s macro-Theme (adversely affected by a number of conservation problems) . . . Since its discovery, Pompeii has been damaged as an archaeological site, adversely affected by a number of conservation problems, including the quality of restoration work, vegetation, tourism, and site management issues.

. . . and to amplify the prosody in its macro-New – reiterating the concerns raised and appreciating their significance (a key issue, important decisions): As a result of these problems, damage to the site remains a key issue for archaeologists and administrators. There are important decisions which need to be made in relation to poor restoration work, invasive vegetation, insensitive tourism, and neglectful management which need to be addressed if the site is to be preserved for future generations of research and public access.

Through these successive iterations we arrive at the factorial explanation (Coffin 2006; Martin and Rose 2008) presented as Text 11a, with its generic structure labelled stage by stage as a list of factors explaining an outcome

9

Strictly speaking there is an interaction of engagement (no realizing contract: deny) and attitude (specialized realizing appreciation: valuation) enacting the negative evaluation here (as for no special walkways, no proper conservation and interpretation programme as well).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

374

J. R. MARTIN

and culminating in re-appreciation of that outcome in the Reinforcement of Factors stage. [11a] Outcome Since its discovery, Pompeii has been damaged as an archaeological site, adversely affected by a number of conservation problems, including the quality of restoration work, vegetation, tourism, and site management issues. Factor 1

To begin, much of the restoration work on Pompeii has been done by local firms with no specialized knowledge of restoration techniques. For example, the timber roof on the House of Meleager was so poorly designed it could not support the weight of the tiles and collapsed. Poor quality mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. Factor 2

A second factor is the incursion of dangerous weeds which have hastened the decay of the ruins. Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more threatening weeds in. Factor 3

In addition to this, Pompeii’s position as an international tourist attraction brings half a million visitors each year. No special walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been carelessly exposed. Factor 4

Finally, there seems to be no overall management plan for the site. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has been poorly managed, and no proper conservation and interpretation programme has been put in place. Reinforcement of factors

As a result of these problems, damage to the site remains a key issue for archaeologists and administrators. There are important decisions which need to be made in relation to poor restoration work, invasive vegetation, insensitive tourism, and neglectful management which need to be addressed if the site is to be preserved for future generations of research and public access.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

375

A factorial explanation of this kind can be topologically related to an exposition, where the point is not to explain historically how something came to be the case, but rather to argue for a particular interpretation of events. An appropriate macro-Theme for an argumentative text of this kind would deploy appraisal to create a debatable issue,10 with implications across all discourse semantic systems for choices promoting one position or another in the stages which follow: Pompeii has been described as a victim of state neglect and indifference and an archaeological catastrophe of the first order. Its ongoing destruction since its discovery in the 1590s has arguably resulted in a greater disaster than its initial destruction by the eruption of Mt Vesuvius one and a half millennia earlier.

From another perspective, the ideation at risk here could have been used to construe a classifying report listing the ways in which archaeological sites are potentially at risk. A suitable macro-Theme might be the following: There are several types of damage that can affect archaeological sites. These include the quality of restoration work, vegetation, tourism, and site management issues.

In a genre of this kind, appraisal has a much quieter role to play since the focus is on classifying and describing types of phenomena (in this case types of activities that are nominalized as abstract things). The macro-Theme classifies the types of damage, and the remainder of the text describes each type in detail. For relevant discussion of genre typology and topology, see Martin and Rose (2008). Returning to our factorial explanation, it is salutary to keep in mind that most of the scaffolding we have introduced whereby Text 11a in effect announces its genre is invisible to an untrained eye. In everyday terms, only the paragraphing is visible (Text 11b below) as a reflection of all that is going on. Bringing the discourse semantics of genres to consciousness, and using this knowledge to design and inform teaching practice, has been an ongoing concern in SFL since the inception of the genre-based literacy pedagogy of the ‘Sydney School’ (Rose and Martin 2012). [11b]

10

Since its discovery, Pompeii has been damaged as an archaeological site, adversely affected by a number of conservation problems, including the quality of restoration work, vegetation, tourism, and site management issues. To begin, much of the restoration work on Pompeii has been done by local firms with no specialised knowledge of restoration techniques. For example, the timber roof on the House of Meleager was so poorly designed it could not support the weight of the tiles and

Note how expanding engagement resources have been used here to position the inscribed evaluation as contestable: has been described as, arguably.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

376

J. R. MARTIN

collapsed. Poor quality mortar has also been used to protect ancient stonework. Over time this mortar has cracked, allowing water and vegetation to penetrate. A second factor is the incursion of dangerous weeds which have hastened the decay of the ruins. Over thirty different varieties have been identified, including ivy, fennel, and fig. As the roots grow they open up further cracks, allowing even more threatening weeds in. In addition to this, Pompeii’s position as an international tourist attraction brings half a million visitors each year. No special walkways for viewing platforms have been constructed, so tourists walk along ancient paths and enter buildings that are not roped off. In some places ancient lead water pipes have been carelessly exposed. Finally, there seems to be no overall management plan for the site. Damaged paths and walls have not been repaired, frescoes have not been preserved, and mangy dogs roam the site. Available finance has been poorly managed, and no proper conservation and interpretation programme has been put in place. As a result of these problems, damage to the site remains a key issue for archaeologists and administrators. There are important decisions which need to be made in relation to poor restoration work, invasive vegetation, insensitive tourism, and neglectful management which need to be addressed if the site is to be preserved for future generations of research and public access. Because we have chosen to exemplify discourse semantic systems by synthesizing a piece of writing, negotiation resources have not as yet entered into our discussion here. We might however imagine situating Text 11a as a response to an exam question, as in the following exchange: Explain how Pompeii has been affected since its discovery as an archaeological site. — Since its discovery, Pompeii has been damaged as an archaeological site, adversely affected by a number of conservation problems, including the quality of restoration work, vegetation, tourism, and site management issues . . .

Or we might imagine using Text 11a in a Reading to Learn programme11 and synthesizing a set of detailed reading exchanges that guide students to recognize features of the genre, such as appraisal, metadiscourse, and genre staging, in the following: The first sentence tells us what’s happened to Pompeii since it became an archaeological site, and previews the factors that have damaged it.

The sentence would then be read aloud, and its elements identified and elaborated as follows:

11

For access to this pedagogy, see Rose and Martin (2012) and Rose’s R2L website: www.readingtolearn.com.au.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

Teacher

Students Teacher

Teacher Students Teacher

Teacher Students Teacher Teacher

Teacher Students Teacher

377

It says Pompeii has been damaged as an archaeological site, then there are two words that mean badly affected. Can you see those two words? – adversely affected – Exactly Let us highlight adversely affected. Adversely means ‘negatively’, so it’s been affected for the worse. Then it says what’s affected it. Can you see what it’s been affected by? – a number of conservation problems – Right. Let us just highlight conservation problems. So these problems are the factors that have damaged it. They have made it hard to conserve Pompeii, to look after it. Then it tells us four factors. What’s the first factor? – the quality of restoration work Yep. ... So this paragraph is called the Outcome of the explanation. It tells us what the outcome is – Pompeii has been damaged – and then it previews the factors that caused this outcome. Now each of these factors is explained in more detail. ... How do we know we’re up to the first factor? – To begin, – Exactly right.

I will leave it to our educational linguists to synthesize possible extensions to classroom interaction of this kind. The only comment I will add here is to foreground the ways in which text-focused exchanges of this kind make it possible for teachers and students to attend explicitly to the discourse semantic patterns which construe, enact, and compose a genre – the very patterns I have been exemplifying as Texts 1 to 10 unfold in this brief tour.

14.4

Meaning beyond the Clause

Is this chapter, I have outlined and exemplified discourse semantic resources, drawing on the description of English resources presented in Martin (1992), Martin and Rose (2003), and Martin and White (2005). As noted above, this model reinterprets cohesion as discourse semantic system and structure – on a deeper stratum realized through lexicogrammar. This

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

378

J. R. MARTIN

Figure 14.1 An outline of discourse semantic systems (by metafunction)

makes it possible to interpret discourse semantic systems metafunctionally, as outlined in Figure 14.1 above (in the proportions ideation and conjunction are to ideational meaning, as negotiation and appraisal are to interpersonal meaning, as identification and periodicity are to textual meaning). This also makes it possible to interpret discourse semantic systems as realized congruently or incongruently in lexicogrammar and phonology (i.e. as involving grammatical metaphor or not), as realized congruently across one or more lexicogrammatical and phonological systems (e.g. affect realized through Comment Adjuncts, mental processes of reaction and desire, or attitudinal Epithets), and as connecting phases of discourse of indefinite extent. This last point perhaps needs some elaboration. What we are saying here is that discourse semantic structures involve both ‘local’ and ‘extended’ realizations. Conjunction, for example, especially external conjunction, regularly relates one figure to another (as detailed in Text 5b above); but in writing internal conjunction regularly relates more than one figure to another (as illustrated in Text 7 above). Similarly, identification regularly relates one entity to another, via anaphoric reference (as highlighted in Text 4 above); but with text reference, indefinitely long passages of discourse can be related to one another (as exemplified in Texts 9 and 11 above: as a result of these factors, as a result of these problems respectively). The same holds true for other discourse semantic systems (see the scope of clause level vs. higher level periodicity; lexical cohesion vs. metadiscourse; negotiation of the exchange of goods and services and information vs. regulatory linguistic services; attitude triggered by or targeted to entities vs. evaluations of discourse). Discourse semantic relations are abduced, in other words, between stretches of unfolding discourse of indefinite extent. It is perhaps in this respect that discourse semantics can be seen as most strongly complementary to

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

379

Figure 14.2 Discourse semantics in relation to register and lexicogrammar

lexicogrammatical and phonological resources,12 where units of structure (e.g. clause, group/phrase, word, and morpheme; tone group, foot, syllable, and phoneme) bound the scope of the meaning to be made. Reinterpretation of cohesion as metafunctionally organized discourse semantic systems also has important implications for analyzing context, in models which treat context as a higher level of meaning. As outlined in Figure 14.2 above, discourse semantics is positioned here as realized through lexicogrammar and as realizing register – where register is a cover term for field, tenor, and mode systems in a model of context stratifying register and genre. As outlined in Martin (1999, 2014), this model was developed to salvage Halliday’s proposal that field is by and large construed through ideational meaning, tenor by and large enacted through interpersonal meaning, and mode by and large composed through textual meaning. Models in which the whole of cohesion is interpreted as textual meaning confound this picture,13 almost as badly, it might be argued, as it is confounded by failing to position field, tenor, and mode as realizing genre. But this unfortunately takes us well beyond the remit of this chapter. I will have to stop here.

References Bateman, J. A. 2014. Text and Image: A Critical Introduction to the Verbal/Visual Divide. London: Routledge.

12

Lemke (1985:287–8) refers to these semantic relations as covariate relations, contrasting them with the multivariate

13

Equally confounding is the practice in much SFL research of treating field, tenor, and mode as realized directly by

and univariate relations constituting lexicogrammatical structure. lexicogrammatical systems, bypassing discourse semantic systems altogether in text analysis. The final chapter of Martin et al. (2010) explores an alternative rite of passage.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

380

J. R. MARTIN

Berry, M. 1981. Systemic Linguistics and Discourse Analysis: A Multilayered Approach to Exchange Structure. In M. Coulthard and M. Montgomery, eds., Studies in Discourse Analysis. London: Routledge. 120–45. Coffin, C. 2006. Historical Discourse: The Language of Time, Cause and Evaluation. London: Continuum. Fawcett, R. 2008. Invitation to Systemic Functional Linguistics through the Cardiff Grammar. Sheffield: Equinox. Fries, P. H. 1981. On the Status of Theme in English: Arguments from Discourse. Forum Linguisticum 6(1): 1–38. Gleason, H. A., Jr. 1968. Contrastive Analysis in Discourse Structure. In J. E. Alatis, ed., Contrastive Linguistics and Its Pedagogical Implications, Report of the Nineteenth Annual Round Table Meeting on Linguistics and Language Studies. Georgetown: Georgetown University Press. 39–63. Halliday, M. A. K. 1967. Intonation and Grammar in British English. The Hague: Mouton de Gruyter. Halliday, M. A. K. 1970. A Course in Spoken English: Intonation. London: Oxford University Press. Halliday, M. A. K. 1984. Language as Code and Language as Behaviour: A Systemic-functional Interpretation of the Nature and Ontogenesis of Dialogue. In R. Fawcett, M. A. K. Halliday, S. M. Lamb, and A. Makkai, eds., The Semiotics of Language and Culture, Volume 1: Language as Social Semiotic. London: Pinter. 3–35. Halliday, M. A. K. 1998. Things and Relations: Regrammaticising Experience as Technical Knowledge. In J. R. Martin and R. Veel, eds., Reading Science: Critical and Functional Perspectives on Discourses of Science. London: Routledge. 185–235. Halliday, M. A. K. 2009. Methods – Techniques – Problems. In M. A. K. Halliday and J. J. Webster, eds., Continuum Companion to Systemic Functional Linguistics. London: Continuum. 59–86. Halliday, M. A. K. and W. S. Greaves. 2008. Intonation in the Grammar of English. Sheffield: Equinox. Halliday, M. A. K. and R. Hasan. 1976. Cohesion in English. London: Longman. Halliday, M. A. K. and R. Hasan. 1985. Language, Context, and Text: Aspects of Language in a Social-semiotic Perspective. Geelong: Deakin University Press. Halliday, M. A. K. and C. M. I. M. Matthiessen. 1999. Construing Experience through Language: A Language-based Approach to Cognition. London: Cassell. Halliday, M. A. K. and C. M. I. M. Matthiessen. 2014. Halliday’s Introduction to Functional Grammar. 4th ed. London: Arnold. Hasan, R. 2009. Semantic Variation: Meaning in Society and Sociolinguistics. Sheffield: Equinox. Knight, N. 2010 Wrinkling Complexity: Concepts of Identity and Affiliation in Humour. In M. Bednarek and J. R. Martin, eds., New Discourse on Language: Functional Perspectives on Multimodality, Identity and Affiliation. London: Continuum. 35–58.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

Discourse Semantics

381

Knight, N. 2013. Evaluating Experience in Funny Ways: How Friends Bond through Conversational Humour. Text & Talk 33(4–5): 553–74. Lemke, J. L. 1985. Ideology, Intertextuality and the Notion of Register. In J. D. Benson and W. S. Greaves, eds., Systemic Perspectives on Discourse, Volume 1: Selected Theoretical Papers from the 9th International Systemic Workshop. Norwood: Ablex. 275–94. Martin, J. R. 1983. Participant Identification in English, Tagalog and Kâte. Australian Journal of Linguistics 3(1): 45–74. Martin, J. R. 1992. English Text: System and Structure. Amsterdam: John Benjamins. Martin, J. R. 1999. Modelling Context: The Crooked Path of Progress in Contextual Linguistics. In. M. Ghadessy, ed., Text and Context in Functional Linguistics. Amsterdam: John Benjamins. 25–61. Martin, J. R. 2013a. Systemic Functional Grammar: A Next Step into the Theory – Axial Relations. Beijing: Higher Education Press. Martin, J. R. 2013b. Modelling Context: Matter as Meaning. In C. Gouveia and M. Alexandre, eds., Languages, Metalanguages, Modalities, Cultures: Functional and Socio-discursive Perspectives. Lisbon: BonD & ILTEC. 10–64. Martin, J. R. 2014. Evolving Systemic Functional Linguistics: Beyond the Clause. Functional Linguistics 1(3): 1–24. Martin, J. R. and D. Rose. 2003. Working with Discourse: Meaning beyond the Clause. London: Continuum. 2nd Revised Edition. Martin, J. R. and D. Rose. 2008. Genre Relations: Mapping Culture. Sheffield: Equinox. Martin, J. R. and D. Rose. 2012. Learning to Write, Reading to Learn: Genre, Knowledge and Pedagogy in the Sydney School. Sheffield: Equinox. Martin, J. R. and P. R. R. White. 2005. The Language of Evaluation: Appraisal in English. London: Palgrave. Martin, J. R., C. M. I. M. Matthiessen, and C. Painter. 2010. Deploying Functional Grammar. Beijing: Commercial Press. Martin, J. R., M. Zappavigna, P. Dwyer, and C. Cleírigh. 2013. Users in Uses of Language: Embodied Identity in Youth Justice Conferencing. Text & Talk 33(4–5): 467–96. Sinclair, J. and M. Coulthard. 1975. Towards an Analysis of Discourse: The English Used by Teachers and Pupils. London: Oxford University Press. Ventola, E. 1987. The Structure of Social Interaction: A Systemic Approach to the Semiotics of Service Encounters. London: Pinter.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:24:47, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.016

15 Appraisal Susan Hood

15.1

Introduction

This chapter is offered in support of educators and scholars interested in becoming more familiar with the theorization and application of system choices in appraisal. The chapter moves from theory to practice. First, appraisal is introduced and located within the broader theoretical framework of Systemic Functional Linguistics (SFL). This is necessarily an abbreviated description. For more comprehensive theoretical accounts of appraisal, readers are referred to sources such as Martin (2000), Martin and White (2005), and Martin and Rose (2007). Other contributions to the development of the theory and to its application in research are referenced throughout the chapter. Descriptions of applied studies address diverse objects of study and methods of research.

15.2

Appraisal as a System of Interpersonal Meaning

Important to the initial and evolving theorization of appraisal is Martin’s (1992a) stratification of context as register and genre, and of the content plane of language as discourse semantics and lexicogrammar. In relation to context, early work from the 1980s on story genres after Labov (e.g. Rothery 1984; Martin and Plum 1997) prompted closer attention to the place of evaluation in SFL. Martin (2014:17) reflects on these contributions, noting ‘the importance of the type of evaluation used as well as its placement in genre structure’ as a means of distinguishing among different types of story genres and that the ‘point of a story depends on the interaction of evaluative language with ideational meaning’. In an anecdote, for example, the point is to share an affective response to an unexpected event. In an exemplum, it is to share judgement of an incident. Stratification of the content plane of language as discourse semantics and lexicogrammar (Martin 1992a; Martin and Rose 2007) provided the

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

Appraisal

383

foundation for a view of evaluation beyond grammar. From the perspective of lexicogrammar, evaluative meanings are realized across a range of systems including, for example, affective mental processes, epithets in nominal groups, adverbial groups realizing comment adjuncts, and circumstances of manner. However, to explore rhetorical expressions at the level of whole texts, a system was required that could ‘generalise across these diverse lexicogrammaticalizations, bringing feelings together in relation to one another’ (Martin 2014:17). Comprehensive accounts of appraisal as a system of interpersonal meaning in discourse include Martin (1992a), Martin (2000), and Martin and White (2005). The final publication offers the single most elaborated account to date. However, the diversity of contexts in which the tools of appraisal have been applied has generated multiple field-specific accounts, and in some cases the particular challenges those contexts pose have continued to push the theory forward. Notable early contributions include Rothery and Stenglin’s (1997, 2000) work in educational linguistics. They were motivated to better understand the linguistic challenges for school students in the effective management of evaluative discourse in writing literary response texts. Fuller’s (1998) study of the language of popular science was foundational to the later development by White (1998) of the appraisal system of engagement, which presents options for the negotiation of values in text. White’s work has predominantly focused on media discourse, and other significant early publications in that field include Feez et al. (1994) and White (2003). The analysis of conversational talk in Eggins and Slade (1997) was also influential. Many others have followed in the footsteps of these early studies, expanding contexts of exploration and deepening understanding.

15.2.1 Contextualizing Appraisal in SFL Theory The contextualization of appraisal within the broader theory of SFL is critical to understanding its potential as a research tool. If appraisal is amputated from its place in its broader theoretical structure, the interpretations it affords and its value as a framework for empirical study are necessarily diminished. Lost are the critical relations to other dimensions of SFL theory, not only within the metafunctional realm of interpersonal meaning, but also in relation to the other metafunctions. To clarify, I begin by situating appraisal in relation to some key dimensions of the architecture of an SFL model of language. Meaning in SFL is theorized as metafunctional, that is, as always and simultaneously ideational, interpersonal, and textual (Halliday 1985, 1994; Martin 1992a; Halliday and Matthiessen 2004). Ideationally, language functions to construe kinds of experience: it is about something. Interpersonally, language functions to enact roles and share values in the negotiation of social relations. Textually, language functions to compose the flow of

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

384

SUSAN HOOD

information into messages that make sense in relation to a mode of interaction and co-textual setting. A discussion of appraisal as a model of evaluative language primarily locates us in the realm of interpersonal meaning – primarily, as in all instances of text there are interactions across metafunctions. For example, in any analysis of evaluation in discourse we need to consider both what is being evaluated, which implicates ideational meaning, and how evaluation is textured, which implicates textual meaning. This issue is returned to later in the chapter. SFL models language as tri-stratal, as composed of meaning systems at the levels of phonology/graphology (Halliday and Greaves 2008), lexicogrammar (Halliday 1985, 1994; Halliday and Matthiessen 1999), and discourse semantics (Martin 1992a; Martin and Rose 2007). The three strata are in a relationship of realization across relative levels of abstraction. In other words, patterns in language at the level of text (discourse semantics) are realized in patterns of language at the level of clause (lexicogrammar), which are in turn realized as expression in sound or writing systems (phonology/graphology). We can explore interpersonal meaning across all strata, but reference to appraisal means we are approaching evaluation from the stratum of discourse semantics (Martin and White 2005). A further important note is that appraisal is one of two discourse semantic systems of interpersonal meaning. Where appraisal models the -personal in interpersonal meaning, a system of negotiation models the inter- of the interpersonal. The implication is that in analyzing texts, appraisal is not loaded with the full responsibility to account for the potential to mean interpersonally. The differentiated responsibilities of the two systems concern the complementarity of relations of solidarity and power.1 The brief theoretical contextualization provided here establishes a general framework, which will be extended in descriptions of how appraisal has been applied in research. Figure 15.1 presents the three sub-systems of appraisal that comprehensively account for evaluative meaning potential in discourse in English. Attitude models the general categories of values that can be expressed: affect, appreciation, and judgement. Graduation models options for scaling attitudinal meanings by degree. Engagement models options for introducing and managing space for alternate propositions or proposals. Each dimension of appraisal opens up more delicate system networks of meaning choices.

15.2.2 The System of Appraisal as Attitude Figure 15.2 presents the system of attitude as a skeletal network of choices. Instances of expression select for the category of feeling as affect,

1

For further insights into the complementary system of

N EGOTIATION ,

see Martin and Rose (2007: Chapter 7).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

Appraisal

385

Figure 15.1 An outline of the system of appraisal (from Martin and White 2005:38)

affect category

judgement appreciation positive

ATTITUDE

‘vibe’ negative inscribe mode of realization invoke

Figure 15.2 Simultaneous choices in identifying instances of 2017:74)

ATTITUDE

(adapted from Liu

appreciation, or judgement. Additionally, we select for positive or negative value or ‘vibe’ (suggested by Martin in person correspondence, to differentiate from the grammatical system of polarity) and for the mode of realization. The mode of realization refers to options for expressing attitude either explicitly as inscribed attitude or implicitly as invoked attitude (Martin 2000; Martin and White 2005). While this choice is represented in Figure 15.2 as a simple dichotomy, options for invoking attitude can be positioned on a cline with some approximating inscription and others more distanced and relatively more implicit. These more delicate options are

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

386

SUSAN HOOD

discussed later under graduation. I elaborate first on the categorical choices in expression in attitude.

15.2.3 Attitude as Affect An expression of evaluation as affect is one of positive or negative emotion. It may be realis or irrealis phenomena that trigger the emotional response, or alternatively the emotion may constitute a general state or disposition. The semantic space of affect is differentiated as follows: • • • •

un/happiness, e.g. miserable, ecstatic dis/satisfaction, e.g. disappointed, content in/security, e.g. suspicious, assured dis/inclination, e.g. fearful, longing for

Affect as dis/inclination expresses an irrealis choice in which ‘feelings relate to future, as yet unrealized, states rather than present existing ones’ (Martin 2000:150), a distinction reflecting that between desiderative and emotive mental processes, as in I wanted them to win/I like them winning (Martin 2017:31). Each kind of affect opens to more delicate meaning choices. In/security, for example, is glossed in Martin and White (2005:49) as to do with ‘our feelings of peace and anxiety in relation to our environs’. Finer categories of in/security include ‘confidence’ or ‘trust’, and ‘disquiet’ or ‘surprise’. Bednarek (2008), approaching a study of attitude from a corpus-based perspective, suggests that ‘surprise’ (as triggered insecurity) ought to constitute a separate category of affect, given that the appearance of the word ‘surprise’ frequently co-occurs with both positive and negative evaluative expressions, as in It was a lovely/nasty surprise. Martin (2017) emphasizes the important role of corpus studies in providing supportive evidence for classification schemes for attitude. However, he points to some important theoretical features that need to be accounted for in applying this mode of analysis. First, it is important that the name of specific categories of emotion should not be confused with the name of a specific feeling. This can lead to problems in analyses and claims. The category named ‘surprise’ refers to a realm of emotion as response to a sudden disruption of expectancy, and hence security. Lexical realizations, other than surprise itself, include feelings of being ‘disturbed, shocked, unsettled, stunned, staggered, thrown, taken aback . . . jolted’ (Martin 2017:37), all of which express a sense of negative security. It is also important to note that the positive/negative value of the affect does not have to match that of the trigger. So in a lovely surprise, the trigger is negatively appreciated as disruptive, while the affectual response is positive. Martin also notes that while available corpora might appear massive, they are not yet of a size that can generate adequate numbers of instances

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

Appraisal

387

for other than a few core lexical expressions. For a more detailed linguistic justification for the categorizations in Martin and White (2005), see Martin (2017:33–6). In the interest of clarification, Martin (2017) does propose a renaming of the sub-category of triggered insecurity, from ‘surprise’ to ‘perturbance’. Martin (2017) explains the development of this typological perspective on affect as an evolutionary process: I was parenting a small child at the time and suggested categories based on my reading of his emotional repertoire in relation to his parents coping (or not) with his moments of distress – basically asking whether he was unhappy because he wanted his mother or father (contented sociability), or because he wanted the comfort of his security blanket (which he called ‘baggy’), or because he wanted the satisfaction of his bottle (‘bopple’). This gave us the [unhappiness/happiness], [insecurity/security] and [dissatisfaction /satisfaction] oppositions. (Martin 2017:31)

He continues: In retrospect, if work on space grammar had already been available at the time, I might equally well have drawn on Stenglin’s (e.g. 2009) notions of bonding (in relation to [un/happiness]) and binding (in relation to [in/security]), and McMurtrie’s (e.g. 2013 [2017]) concept of promenade (in relation to the telos oriented notion of [dis/satisfaction]). (Martin 2017:31)

This is illustrated in Table 15.1.

15.2.4 Attitude as Judgement judgement refers to the evaluation of people or their behaviour; it is appraisal as praise or criticism. Expressions of judgement may relate to either social esteem or social sanction (Martin and White 2005:52). Social esteem has to do with judgement as ‘normality’ (how un/usual a person is); ‘capacity’ (how in/capable they are); and ‘tenacity’ (how ir/resolute they are). Social sanction has to do with judgement as ‘veracity’ (how un/ truthful a person is) or ‘propriety’ (how un/ethical they are) (Martin and White 2005:52). Table 15.1 Additional AFFECT parameters (Martin 2017:31) ‘parenting’ un/happiness in/security dis/satisfaction

Mummy/Daddy baggy bopple

‘space grammar’ bonding binding promenade

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

388

SUSAN HOOD

15.2.5 Attitude as Appreciation Expressions of appreciation evaluate entities and events. At a finer level of delicacy, instances of appreciation are differentiated as follows: • • •

reaction, e.g. fascinating, boring, beautiful, ugly composition, e.g. complex, simple, harmonious, flawed valuation, e.g. exceptional, insignificant

Martin and White (2005:57) offer metafunctional associations for each of the domains of appreciation, ‘with reaction oriented to interpersonal significance, composition to textual organization and valuation to ideational worth’. Appreciation expressed as ‘reaction’ describes the degree to which a reader or audience notices or is made aware of the phenomenon described. In other words, our evaluations of phenomena are expressed in terms of how we react to them. Here a further distinction can be made. Reaction as ‘impact’ is glossed in Martin and White (2005:56) as addressing the question Did it grab me?, as in ‘a fascinating read’, or ‘an unremarkable performance’. Reaction as ‘quality’ is glossed as Did I like it?, as in ‘an interesting artefact’, or ‘an ugly building’. Because we are talking about appreciation of phenomena in terms of their affectual impact on the appreciator, we approximate the realm of affect. This approximation can often prove challenging for learners of English as another language, where slippage can readily occur between expressions such as It is interesting and I am interested – resulting in inadvertent exclamations such as I am so boring!. More delicate options are also noted for appreciation as ‘composition’, more delicately as ‘balance’ (Did it hang together?), or ‘complexity’ (Was it hard to follow?). The point has been made, by Macken-Horarik and Isaac (2014) and others, that attitude as appreciation ‘is the most sensitive to context’, resulting in a potential need to modify the categorical dimensions in relation to ‘specific discourse contexts, texts and topics’ (Macken-Horarik and Isaac 2014:74). For more comprehensive accounts of categorical distinctions in attitude, see Martin and White (2005:48–57) and Martin and Rose (2007:65–71). The idea of being able to reconfigure typological distinctions as topological relations was touched on above, and is presented diagrammatically in Figure 15.3, in which relative proximities of semantic realms of attitude are indicated.

15.3

The System of Appraisal as Graduation

A second semantic domain of appraisal is graduation. This sub-system concerns the potential available in language for ‘meaning by degree’ (Martin 1992b). Meanings can be adjusted or graduated along dimensions of force or focus.

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

Appraisal

389

Figure 15.3 A topological perspective on ATTITUDE resources (from Martin 2000:165)

15.3.1 Graduation as Force Force refers to the up-scaling or down-scaling that adjusts the ‘intensification’ of a value, for example, from important to very important or to not so important. This kind of adjustment has been analogized as turning up or down the volume control on a sound system (Martin 1992a:533; Martin and White 2005:37). In the examples above, attitude is realized as a quality, and the intensity is adjusted up or down. Expressions of attitude can also be realized in a circumstance of manner around a process, and by such means the process is adjusted in intensity, as in rigorously investigated. This option is referred to as ‘vigour’ (Hood 2010; Martin and White 2005). A third means of ‘intensification’ associates with lexicalized modulation, which adjusts a whole proposal in terms of degree of obligation, as in It is vital that you follow my instructions. When a quality is expressed as an entity, as in importance or satisfaction, degrees of force express +/- ‘quantification’ (rather than ‘intensification’). This is evident in expressions such as more importance, much happiness, or little satisfaction. ‘Quantification’ may be in terms of number (amount, volume, or mass) or extent, with extent as either proximity or distribution in Martin and White (2005), and as scope or distance in Hood (2010). In both cases the options apply to time and space. A process may also be quantified as frequency, as in He was very often anxious. As an aside, some years ago when observing the teen talk of my stepson and his friends, I noted their creative play with evaluative language involved a flipping of the expected association of intensification + quality and quantification + entity, to arrive at so fun and heaps cool.

15.3.2 Graduation as Focus graduation as focus refers to the relative sharpening or blurring of categorical boundaries. As such it can be analogized to the focus function

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

390

SUSAN HOOD

on a camera. This is exemplified in expressions such as absolute misery, real research, a vaguely relevant study, a sort of pleasure, pure joy. In Martin and White (2005) graduation is explored in relation to the adjustment of options in both attitude and engagement (e.g. Martin and White 2005:136). In Hood (2010) and Hood and Martin (2007) it is explored primarily in relation to attitude and particularly as a resource for invoking rather than inscribing attitude. In written academic texts, the play of inscribed and invoked attitude has been shown to be of strategic importance (Hood 2010). Where writers frequently express inscribed attitude in reporting on their object of study, more restraint is typically exercised in reporting on other contributions to knowledge. A preferred strategy is to adjust objective (experiential) meanings by relativizing or subjectifying them and by so doing to invoke or flag attitude. Hood (2010) suggests that an avoidance of dichotomizing inscriptions of attitude enables the academic writer to maintain a veneer of objectivity while implying stance.

15.3.3 Graduation Invoking Attitude Force as intensification and quantification are exemplified above with instances of inscribed attitude. However, force can also function to grade experiential (non-attitudinal) meanings. For example, the underlined in the nominal group an action-oriented study is an experiential classifier. If we adjust this meaning by degree to a more action-oriented study, the process of ‘intensification’ shifts the function to an epithet and implies an attitudinal position. The grading of experiential meaning in a process may also invoke an attitudinal interpretation. For example, in choosing to express an investigative activity as explore rather than look into, we can imply intensified vigour and again an evaluative potential. The grading of modulation (e.g. ought to/need to/must) may also invoke an interpretation of a proposal as relatively more significant. As noted above, this can move to inscription when lexicalized as in, for example, essential or vital. Quantifying entities as number (amount, volume, or mass) adjusts experiential meaning by degree, as in many studies, a sizeable volume of literature, the weight of research. The adjustment of quantity in each case encodes a subjective orientation, and opens a space for it to be read evaluatively. Martin and White (2005) refer to this as ‘flagging’ an attitudinal interpretation. There are similar implications in expressions of frequency and extent as scope or distance, as exemplified in the following: • • • •

graduation graduation graduation graduation

as scope: in time – it’s been like this for decades as scope: in space – it’s a pervasive phenomenon as distance: in time – in a more recent occurrence as distance: in space – a further removed example

The study of graduation in the context of academic research papers has led to some proposed extensions to the system of graduation (Hood 2010).

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

Appraisal

391

This is particularly so with respect to focus. The model of focus in Martin and White (2005) concerns the adjustment of boundaries around entities (e.g. a true father, an apology of sorts). Hood (2010) suggests an extension to what is considered a category of ideational meaning beyond that of entity to include process and figure (as a proposition). The adjustment of focus here is generalized as ‘fulfilment’. The boundary around a process can be sharpened or softened by adjusting the degree of ‘completion’ (e.g. tried to show, managed to show). Implicated here are resources of phase: ir/realis in the verbal group. The boundary around a figure (as proposition) can be sharpened or softened by adjusting the degree of ‘actualization’ (e.g. seems to show, definitely shows). Implicated here are resources of conation and modalization in the verbal group. Again, the grading of non-attitudinal (experiential) meanings encodes a subjective positioning on behalf of the speaker/writer, and hence an indication to the audience to interpret attitudinally (Hood 2010; Hood and Martin 2007). Graduation options are shown in Figure 15.4.

15.3.4

The Function of Concessive Contractors in Negotiating Graduated Attitude In a study of problematic in-bound customer service calls to a call-centre, Hood and Forey (2008) noted the significant role that concessive contractors such as just, already, once, yet, and actually played in adjusting and managing the attitudinal intensity of caller emotions in the flow of interaction. They

Figure 15.4 A network of graduation options (adapted from Hood 2010:105)

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 06 Aug 2019 at 14:29:28, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316337936.017

392

SUSAN HOOD

Figure 15.5 A network of modes of realization for 2007:746)

ATTITUDE

(from Hood and Martin

found that rather than mirroring the amplified levels of emotive intensity on the part of customers, the agents made frequent use of concessive contractors (and sometimes silence) to rein in or defuse the attitudinal intensity. For example, an expression such as just a few days on the part of an agent could defuse an amplification of extended time by the caller.

15.3.5 Degrees of Invocation in Implicit Expressions of Attitude Graduation is discussed above as a significant resource by which users imply rather than directly encode an attitudinal position in discourse. There are also other means by which users can achieve this end. A network of options for invoking attitude is in Figure 15.5. The options in the system can also be interpreted as a cline of relative implicitness, or relative commitment of an attitudinal meaning. The closest we come to direct inscription of attitude is in the use of idiom or lexical metaphor. Both are said to provoke an attitudinal interpretation. Chang (2017:324) notes that most idioms are ‘figurative fixed expressions’, such as cool as a cucumber, too hot to handle, as sharp as a tack. As idioms are taken up in common usage in a culture, they are essentially divested of the remnants of an imported field. While not gradable in themselves, they tend to become overtly positive or negative and can be readily substituted with an inscription of attitude (Chang 2017). Lexical metaphor is interpreted as marginally more implicit. The appropriated reference to the literal field, as in He’s a prisoner of his own volition, brings an associated value to its metaphoric usage. Yet more implicit choices are referred to in Figure 15.5 as ‘inviting’ an attitudinal interpretation. That interpretation becomes more reliant on the co-text and/or on the field of the discourse. The option of inviting an evaluative reading as ‘flagging’ refers to the deployment of resources of graduation to adjust the force or fo