Three Steps

This book is an important step in the engagement with the lexicogrammatical system of language viewed and interpreted ecologically in its semiotic environment (between semantics and phonology or graphology or sign). It is the third step in a series of books focused largely on systemic functional theory of language as a general human semiotic system operating within the domain of lexicogrammar and manifested in the description of the lexicogrammatical systems of particular languages.

The first step was taken by Michael Halliday and me in a short monograph, originally drafted in the 1990s as part of a large-scale publication project planned by our friend and colleague Fred Peng and later published in English with a Chinese translation and additional material in Chinese by Huang Guowen and Wang Hongyang as Reference Matthiessen and HallidayMatthiessen and Halliday (2009). In this monograph, we foregrounded the theoretical aspect of the approach to lexicogrammar in SFL, conceiving of it as a stand-alone complement to Halliday’s Introduction to Functional Grammar (IFG) – which by then had come to be supported by other publications of a more introductory nature (e.g. Thompson’s Introducing Functional Grammar, Martin, Matthiessen and Painter’s Deploying Functional Grammar, and Bloor and Bloor’s Functional Analysis of English), leaving room for IFG to grow in size and coverage, and my own Lexicogrammatical Cartography (LexCart), which is not introductory but which foregrounds systems as the way into the description of lexicogrammar.

These books all interleave theory and description. The descriptions they present are focused on English (although I included typological outlooks in LexCart); but, in our First Step, Halliday and I decided to include examples not only from English but also from Chinese and Japanese in order to illustrate the descriptive application of the theory to languages from different language families and also to languages with different typological profiles. Halliday had, of course, worked extensively on Chinese since the late 1940s, and I had done some work on Japanese (learning from John Bateman, see Reference Matthiessen and BatemanMatthiessen and Bateman, 1991 – although the languages I had engaged with more as a linguist by the early 1990s included (Modern Standard) Arabic and Akan). And since we began the work on the monograph, Kazuhiro Teruya had started his long-term research programme concerned with the development of a comprehensive systemic functional description of Japanese (Reference Teruya1998, Reference 453Teruya2007), which has been used extensively in further work on Japanese and used as a model in the development of descriptions of other languages.

While our First Step was, in a sense, excerpted from Peng’s publication project and was part of a series of steps we had planned, it was J. R. Martin who took the second step with his Systemic Functional Grammar: A Next Step into the Theory – Axial Relations (Reference Martin, Wang and ZhuMartin, Wang and Zhu, 2013). Here it is actually essential to include the full title of his book, because the focus on axial relations is absolutely crucial – as the current volume demonstrates throughout in the engagement with the languages under description. Halliday (e.g. Reference Halliday, Halliday and McIntosh1966) had given priority to paradigmatic relations, deriving syntagmatic relations from them (through realisation statementsFootnote ¹).

This turned out to be crucial, also to the success of the development of a computational model of systemic functional grammar as part of text generation systems (e.g. Reference Matthiessen and BatemanMatthiessen and Bateman, 1991); and I used the paradigmatic axis – that is, systemic organisation – as the principle for organising my account of the resources of English lexicogrammar in LexCart (Reference MatthiessenMatthiessen 1995), and thus as a navigational instrument for readers. (When Halliday and I started on the third edition of his IFG, he wanted to include system networks; but we kept the overall orientation in our presentation to structures that realise meaning in text.)

The focus on paradigmatic organisation also turned out to be absolutely necessary as we developed the theory of a multilingual meaning potential (e.g. Reference Bateman, Matthiessen, Nanri and ZengBateman et al., 1991; Reference Bateman, Matthiessen and ZengBateman, Matthiessen and Zeng, 1999; Reference Matthiessen, Baklouti and FontaineMatthiessen, 2018; see also Reference Matthiessen, Wang, Ma and MwinlaaruMatthiessen et al., 2022) – a theory that is highly relevant to the present book and one which could be explored in another step as a way of comparing and contrasting the lexicogrammatical descriptions in relation to their semantic and contextual environments.

In language description, comparison and typology, the focus on paradigmatic relations can play a fundamental role in identifying cryptogrammatical categories and reactances, as has been shown in the work by Halliday and other systemic functional linguists (with references to B. L. Whorf’s pioneering work). Arguably, most attention has been paid to ideational cryptogrammar and reactances (specifically in relation to the experiential system of transitivity), as in the work by Halliday, Martin and Davidse; but these more hidden aspects of grammar are also relevant when we explore the interpersonal and textual resources of grammar and try to bring out interpersonal cryptogrammar and textual cryptogrammar in our descriptions.

Here one example of a challenge is to return to Subject as a function long mistakenly assumed to be present in the description of any language due to the inertia of tradition and considerations from ‘below’ such as concord, case and sequence, in a helical move where it can then be viewed and reinterpreted as part of the more cryptic patterns of the interpersonal grammar of the clause as a move in dialogic exchanges in terms of interpersonal elevation of one element taking interpersonal reactances into consideration (see Reference Matthiessen, Caffarel, Martin and MatthiessenMatthiessen, 2004, and, for Japanese, see Reference 453TeruyaTeruya, 2007). One important general point is that there will be variation across languages in what categories are more overt and what categories are more covert or cryptic, and this variation may be related to metafunction (which may show up when one explores which metafunctions ‘control’ marking of ‘dependents’ by adpositions or cases or of ‘head’ marking).

The Second Step, Reference Martin, Wang and ZhuMartin, Wang and Zhu (2013), is well documented in the present book since it provides an essential part of the foundation for engaging with different languages not only syntagmatically but also paradigmatically – in fact, paradigmatically in the first instance. Nevertheless, the importance of this approach needs to be emphasised clearly and repeatedly because while there is a well-recognised extended family of functional approaches to language covering a wide range of traditions and concerns, SFL is still the only member of this family giving clear priority to the paradigmatic axis.Footnote ² This fact highlights the unique value of the present volume: there is nothing comparable in the linguistic literature.

The Third Step

Martin, Wang and Zhu’s Second Step was thus concerned with foregrounding the importance of axial relations and the priority given in SFL to the paradigmatic axis, using English as the language of illustration. This feeds directly into the present volume, which extends the descriptive base; it brings out a way of reasoning in the development of descriptions of the lexicogrammatical systems of the languages spoken around the world that has typically remained in the background even when leading linguists have devoted enormous energy to syntagmatically oriented descriptions of a growing range of languages.

This Third Step is certainly important in that it sheds light on paradigmatic reasoning (including centrally the use of proportionalities); but it does much more than that: it shows how Halliday’s trinocular vision can – and actually must – inform the development of descriptions of languages around the world. His original articulation of trinocular vision (e.g. Reference HallidayHalliday, 1978) was focused on the hierarchy of stratification: lexicogrammar can be viewed ‘from above’ – most immediately from the vantage point of semantics, and by another step from the vantage point of context, and ‘from below’ – from the vantage point of phonology (or alternatively, graphology or sign); and it can be viewed ‘from roundabout’, from the vantage point of its own stratum of lexicogrammar (the stratum of meaning created as wording). But this trinocular vision extends to the other semiotic dimensions defining the ‘architecture’ of language in context (see Reference Matthiessen, Hasan, Matthiessen and WebsterMatthiessen, 2007, forthcoming; Reference Matthiessen and TeruyaMatthiessen and Teruya, forthcoming) – the hierarchies of rank and axis (organising lexicogrammar locally), the spectrum of metafunction (permeating the content plane and the highest ranks of the expression plane, recognised as ‘prosody’), the cline of delicacy, and the cline of instantiation (like stratification a completely global semiotic dimension).

As Martin, Quiroz and Wang present the descriptions of English, Spanish and Chinese, they make extensive use of Halliday’s trinocular vision, showing again and again how we can reason about phenomena in different languages ‘from above’, ‘from below’ and ‘from roundabout’. Traditionally, the view ‘from below’ – in terms of rank, and also in terms of stratification – has tended to be foregrounded because it includes the most ‘exposed’ parts of any language, i.e. the parts that are the easiest to observe and engage with as overt categories (see Reference Halliday and WebsterHalliday, 1977/2003, Reference Halliday and Webster1984/2002). The nature of the most exposed part of language will, obviously, vary from one language to another (as does the nature of the more cryptic, covert aspects of the language); in Ancient Greek, in Latin and in Sanskrit, it was essentially the grammar of the word since it is easy to observe that words have different forms, like the case forms of nouns – but the situation was very different in Classical Chinese; so these languages ‘invited’ the construction of different linguistic traditions focused on partly different aspects of the total system of language.

In SFL, Halliday’s trinocular vision ensures that we always keep shunting (see Reference Halliday and WebsterHalliday, 1961/2002) along semiotic dimensions from one vantage point to another in order to ensure that we do not get locked into a single point of view. The methodology of shunting is illustrated throughout this volume. Very often, the three views adopted along any one semiotic dimension will resonate with one another (at the risk of overusing the term, we might say that they are congruent with one another); but sometimes they do not – and this is significant, partly because this lack of resonance (or ‘incongruence’) may indicate an internal tension in the system suggesting change in progress, which is of course an integral feature of all linguistic change (as shown by Reference EllegårdEllegård’s, 1953, pioneering corpus-based quantitative study of the gradual appearance and disappearance of the auxiliary do in the interpersonal environments of the systems of mood and polarity in English over a period of roughly 250 years).

Comprehensive Descriptions

So the power of a holistic theory of language in context is very much in evidence in this book – as a resource for developing comprehensive descriptions of particular languages; and the methodology of shunting is part of the kind of relational theory that systemic functional linguists have been constructing since the 1960s (and in the decades before, since J. R. Firth’s system-structure theory was also a relational theory of language in context). The authors of this book show how this holistic theory empowers them to present descriptions of English, Spanish and Chinese – systematically and systemically, step by step; and these descriptions are on the path towards comprehensive descriptions.

In Chapter 6, the authors comment on the state of their descriptions: “For those of you familiar with English, Spanish or Chinese, we reiterate as a reminder here that the descriptions introduced above have been provided for pedagogic purposes. They are both preliminary and provisional. SFL grammarians working in the Sydney register of SFL (e.g. Reference Matthiessen, Yan and WebsterHalliday and Matthiessen, 2014) have developed more comprehensive descriptions of English than we have offered here.” They then provide some additional references.

This raises the question of what would constitute a comprehensive description of a particular language. We can, of course, review book-length systemic functional descriptions that have been published (including PhD theses). None of them have been designed as reference grammars; rather, the typical measure of comprehensiveness can be related to the degree to which they can support analysis of texts from a wide range of registers,Footnote ³ and in many cases, this has involved going ‘into the field’ to record and then transcribe a significant range of spoken texts in their community settings (see Reference RoseRose, 2001; Reference AkerejolaAkerejola, 2005; Reference KumarKumar, 2009; Reference MwinlaaruMwinlaaru, 2017; see also Reference Matthiessen and TeruyaMatthiessen and Teruya, forthcoming).

More generally, we can reason about comprehensiveness in terms of the semiotic dimensions of systemic functional theory; thus a comprehensive description should be geared towards:

• coverage of all metafunctions;

• coverage of all ranks;

• coverage of both axes (paradigmatic and syntagmatic patterns);

• coverage of a number of steps in paradigmatic delicacy, the elaboration in delicacy being an ongoing project;

• coverage of a number of ‘phases’ along the cline of instantiation from instances towards the total wording potential of a language, the expansion in coverage being a matter of extending the range of registers (functional varieties) accounted for.

Naturally, since the approach to lexicogrammar is ecological in SFL, coverage can also be measured ‘from above’ (semantics, and by another step context) and from below (phonology, graphology or sign): the description of the lexicogrammar of a language must be accountable to the neighbouring strata.

The first two dimensions from the list above, metafunction and rank, have been used by Halliday in designing function-rank matrices, as in Reference HallidayHalliday (1970). In such matrices, the metafunctions are set out horizontally as column headings, and the ranks vertically as row headings, with primary classes being distinguished for each rank. The cells defined by metafunctions and ranks (and primary classes) constitute the locations of all the lexicogrammatical systems of a given language. Such matrices have now been proposed for a number of different languages, and one can compare these and experiment with generalised ones (see Caffarel, Martin and Reference Matthiessen, Caffarel, Martin and MatthiessenMatthiessen, 2004). For example, if we compare the matrices of the lexicogrammatical systems of Chinese and English, we can see quickly that they differ in terms of grammaticalised process time – aspect and tense, respectively (see Reference MatthiessenMatthiessen, 2015b).

Characterology

Based on a comprehensive description of a given language, we can also begin to develop a profile of the language, identifying properties that seem to go together – not in causal relationships but rather in relationships of semiotic correlation. This is along the lines of Mathesius’ notion of the characterology of a language, an important contribution within the Prague School; and such correlations may of course be picked up in the formulation of implicational universals of the kind Reference Greenberg and GreenbergGreenberg (1966) proposed.

Reference HallidayHalliday (2014) proposes seventeen features of Mandarin lexicogrammar and phonology as a basis of a characterology of the language; and he shows how sets of features correlate so that if one visualises these correlations (as I have done), they form a kind of connected graph. Such characterologies based on comprehensive descriptions can be very helpful in the further development of existing descriptions and also in the development of new descriptions of other languages.

They are a central part of systemic functional comparison and typology, and can, as just noted, support descriptive work. For example, as hinted at above, we can contrast the construal in Chinese of the actualisation of the process with the model in Spanish and English, and relate the Chinese construal not only to the system of aspect (as opposed to tense), but also to the system of phase. In Chinese, the actualisation of a process unfolding in time does not imply its completion – so completive phase plays an important role; in English and Spanish, the actualisation of a process unfolding in time does imply its completion (see further Reference MatthiessenHalliday and Matthiessen, 1999, pp. 306–8).

Choice of Languages

The orientation of this book is, of course, pedagogic – an invitation to explore the languages under description; it is part of a contribution to the enormously challenging task of instructing researchers in how to describe a particular language.Footnote ⁴ The languages the authors have chosen are three ‘major’ languages (however we define ‘major’ languages; but they are all included in Comrie’s edited series of sketches of the major languages of the world and come with long traditions of extensive linguistic descriptions couched in various theoretical frameworksFootnote ⁵), and one of them is the first – and so far only – global language, namely English (see Reference 445HallidayHalliday, 2003). They are all standard languages, so they embody – are aggregates of – the very wide range of registers (in Halliday’s sense of functional varieties) characteristic of standard languages, including those that have emerged as part of the evolution of institutions of administration, government, the law, education and science & technology. At the same time, they may have gradually shed registers characteristic of languages of small close-knit communities sustained by hunting and gathering or small-scale farming (e.g. Reference WiessnerWiessner, 2014Footnote ⁶; see Reference RoseRose, 2001, Reference Rose2005; Reference Halliday and WebsterHalliday, 2010/2013). And while personal and communal multilinguality is likely to have been the norm in human speech fellowships for most of our history (as emphasised by both Halliday and Evans, e.g. Reference EvansEvans, 2010), the users of standard languages are often effectively monolingual (but as standard languages get ‘exported’ around the world, they may be influenced by the mother tongues of substrate languages overrun by colonisers, as in the case of, e.g., Spanish and Mapuche in Chile or English and Celtic languages).

Regarding the choice of languages, the authors write in the final chapter: “In this book we have considered only three languages, which we selected because they are arguably three major world languages (and so address a worldwide readership) and because they reflect our personal expertise.” These two considerations are very good reasons – although I would have increased the number ‘three’ in “three major world” languages to ensure that a few other languages had been included, certainly Arabic (and Hindi is another good candidate, though of the same language family as English and Spanish); but it is up to those of us who are in a position to add to the descriptive pool presented in this book to expand the range of pedagogic models of how to reason in the development of descriptions of particular languages – including relatively minor ones that have been illuminated through systemic functional descriptions in the work by David Rose (Western Desert), Ernest Akerejola (Oko) and Isaac Mwinlaaru (Dagaare).

The expansion of the pool of pedagogically presented developments of language descriptions is an important task to undertake in order to support researchers wanting to develop systemic functional descriptions of particular lexicogrammatical systems – especially since, when viewed against the background of the rich or even exuberant lexicogrammatical variation around the languages of the world, English, Spanish and Chinese are not dramatically different; they are all modern standard languages (serving as the semiotic vehicles of a number of natio- states) and they are all (originally) Eurasian languages – which is significant even if one doesn’t subscribe to Joseph Greenberg’s account of macro-families.

English and Spanish are, in many ways, typical of languages of western Eurasia (allowing for changes since they began to spread around the world half a millennium ago as part of European colonisation) and Chinese is a fairly typical language of (South-)East Asia (with the exception of, e.g., Korean, Ainu and Japanese). This is how Reference HallidayHalliday (2014) characterises Mandarin Chinese within the area of East Asian languages:

Thus, if we consider the grammatical model of process time in English, Spanish and Chinese, we can note that English and Spanish construe process time grammatically in terms its location relative to the now of speaking, the basic choice being past vs present vs future – in terms of logical tense systems potentially involving intermediate reference times between the now of speaking and the time of the process itself (serial tense construing serial time; cf. Chapter 3, Section 3.2.1 on English and Section 3.3.1 on Spanish), and we find similar tendencies in other languages in western Eurasia. In contrast, Chinese construes process time grammatically in terms of the boundedness of the unfolding of the process in time, the basic distinction being ‘perfective’ (actualisation of process bounded in time) vs ‘imperfective’ (actualisation of process unbounded in time) – in terms of experiential aspect systems; and in this respect Chinese is like a number of other languages of South-East Asia, including ones from different genetic families such as Tagalog, Vietnamese and Thai.

These two models of process time – the tense model and the aspect model – would appear to be incompatible alternatives, but it turns out that they are complementary perspectives on process time (see Reference HallidayHalliday, 2008, on the fundamental importance of recognising complementarities both in languages and in linguistics), and there are languages between the outer poles of Eurasia with mixed tense-aspect systems, including members of Slavic languages, Indo-Aryan languages and (in my view) Classical and Modern Standard Arabic. (And when we move to other parts of the world of languages, we can expect to find yet other models, including tense systems of an experiential rather than logical nature where time is not serialised but rather taxonomised and the different system of Hopi, as described by Whorf.) Other such systemic complementarities we can expect in different ‘mixtures’ include the models for construing participation in the process (the complementarity of the transitive and ergative models) and ‘protocols’ for assessing information (the complementarity of modality: modalisation and evidentiality). Such complementarities are important to look for and identify, since systems are often labelled ‘from below’, where the complementarities may not be clearly discernible.

When we move to other lexicogrammatical systems, English, Spanish and Chinese will group themselves in other different ways. For example, all three languages have highly grammaticalised pronominal systems – in contrast with languages such as Thai (see Reference Iwasaki and IngkaphiromIwasaki and Ingkaphirom, 2005), which tend towards lexical or lexicogrammatical rather than grammatical systems. Naturally, they differ in particular details, as do pronominal systems of different varieties of Spanish (even within particular regions as in Colombia); but on a ‘global’ scale they are fairly similar.

The general point is, of course, that the lexicogrammatical resources of any particular language embody many lexicogrammatical systems, and each system can be taken as a reference point when we compare, contrast and typologise languages. This is indeed why Halliday has always pointed out that language typology should be focused on systems within languages rather than on whole languages (e.g. Reference Halliday, Halliday and McIntoshHalliday, 1959/60), and I think the consensus around this view has grown considerably over the decades of empirical language typology. In other words, we have to recognise that the nineteenth-century attempts to typologise languages based on a single ‘feature’ of word grammar is quite unhelpful (see already Reference SapirSapir’s, 1921, contribution to the separation of different aspects of the standard nineteenth-century typologies).

Paradigmatic Orientation; Metafunctions

Here, as in the present book, the notion of system is foregrounded – i.e. we give priority to the paradigmatic axis over the syntagmatic axis (see Reference HallidayHalliday, 1966/2002; and see Reference Matthiessen and WebsterMatthiessen, 2015a). This has been important in systemic functional language description, comparison and typology. The primacy of systemic organisation makes it possible to bring out patterns of similarity and difference when we move around the languages of the world that might otherwise not have been highlighted.

At the same time, this means that syntagmatic patterns are shown to serve as realisations of terms in systems, and they are thus always ‘contextualised’ systemically (paradigmatically). One consequence of this is that terms in one and the same system may be realised syntagmatically at different ranks (or even by either grammatical items or phonological features); and it becomes possible to discern the continuity in comparison of different languages (or different historical phases of one language or language family, as in the well-known case of Romance languages) between syntagmatic patterns that might at first seem very different – e.g. the realisation of participants (transitivity) as nominal groups (or adpositional phrases) serving as elements in the clause (whether pronominal or non-pronominal), as pronominal clitics partially ‘integrated’ in the structures of lower-ranking units such as verbal groups or as pronominal affixes serving within the structure of units at word rank such as verbs. So in this area it is possible to see a cline from, say, English and German to French, Spanish and Portuguese to Mapuche, Panare and Inuit. (And while it is significant that, e.g., personal references may be realised pronominally as pronominal nominal group serving as elements of the clause, or as pronominal ‘clitics’ that are in a sense halfway down the rank scale from clause to group function, or as pronominal affixes serving at word rank, one should not exaggerate such differences. Indeed, they often represent different phases in the evolution of a particular language or family of languages; and there is often uncertainty among linguists reflected, e.g. in the orthographies they propose, as in the case of a number of languages spoken in West Africa where pronominal Subjects are written as separate orthographic words in some languages and as prefixes in other languages even where there are no linguistic differences.)

Many contributions outside SFL to language typology are, of course, focused on syntagmatic patterns. One reason for this is theoretical in the sense that researchers outside SFL work with syntagmatically oriented conceptions of language. Another reason – which is directly related – is that when we examine the literature, it is much easier to find observations about syntagmatic patterns; paradigmatic patterns tend not to be in focus outside phonology and morphology. One obvious example is the extensive work on ‘word order typology’ based on observations about the sequence of elements in different grammatical units, perhaps primarily the clause, the nominal group and the adpositional phrase, starting with Reference Greenberg and GreenbergGreenberg (1966) and continued (and corrected) by Matthew Dryer and many other scholars. Some early generalisations have been confirmed (like the ‘resonance’ between V • O and sequence in adpositional phrases) but others have now been rejected as it became possible to consult larger more representative samples of languages – like Greenberg’s early correlation between the sequence of elements in the clause and of elements in the nominal group (adjective and noun) – predictably, from a systemic functional point of view since clauses and phrases are both exocentric constructions but groups are endocentric rather than exocentric.

In explorations of ‘word order’, metafunctional considerations are crucial: the myth of ‘fixed word order languages’ vs ‘free word order languages’ has turned out to be remarkably resilient despite the increasing evidence that the fundamental issue is not whether the ‘word order’ is ‘fixed’ or ‘free’ but rather how the sequence of elements is deployed as a mode of realisation in any given particular language by the different metafunctions (see Reference Matthiessen, Caffarel, Martin and MatthiessenMatthiessen, 2004). Halliday wrote a paper designed to demolish the myth, “It’s a fixed word order language is English” (Reference Halliday and Webster1985/2005); any linguist who reads it could be expected to question the distinction between ‘free’ and ‘fixed’.

Most commonly, it’s the textual metafunction that’s ignored when linguists make claims about ‘free word order’, which is a consequence of relying on (elicited or constructed) clauses deprived of their discursive environment (i.e. a non-ecological approach to the investigation of grammar). This is one of many areas where the present volume makes an important metafunctional contribution. For example in Chapter 6, which is concerned with the textual metafunction, the authors write:

(I would add to the insightful discussion in this chapter that ‘Themes’ appearing after the Predicator of a clause might be interpreted as thematic Reprises or Afterthoughts.)

This is one of many areas where it’s crucially important to distinguish between the general theory of language and specific descriptions of particular languages – a distinction that can be found already in Firth’s work (theoretical and descriptive terms are distinguished clearly in Reference Matthiessen, Teruya and LamMatthiessen, Teruya and Lam, 2010). Various frameworks used in the description of particular languages and in linguistic comparison and typology such as LFG, FG and RRG would include as part of the ‘theory’ notions like Subject or Pivot, Object, Predicate, Topic, Undergoer, Patient and many other categories that would be posited in SFL in the course of the description of particular languages based on empirical evidence derived trinocularly – as in this volume for English, Spanish and Chinese. The fact that such categories are not part of systemic functional theory is not a bug but rather a feature – one informed by extensive engagement in the Firth–Halliday tradition with a very wide range of languages. Linguists are encouraged to do the hard work of showing that the descriptive categories they posit in their interpretation of any particular languages can actually be motivated empirically based on weighty evidence coming from the examination of registerially varied corpora of texts in context.

The theory of language is designed to empower linguists to imagine a wide range of possible languages, guiding them in their approach to particular languages by providing them with a general conception of human language as it must have evolved together with AMHs (Anatomically Modern Humans) – Homo Sapiens Sapiens – on the order of 150,000 to 200,000 years ago, and as it emerges when young children make the transition from the protolanguages they have constructed in interaction with their immediate caregivers starting around the age of 5 to 8 months to the mother tongue(s) spoken around them somewhere halfway into their second year of life (see Halliday, in Reference WebsterWebster, 2004).

Parallels and Descriptive Generalisations

Empowered by the general theory of language provided by SFL, researchers can develop descriptions of particular languages, reasoning in the way illustrated in this book as the authors develop the descriptions of English, Spanish and Chinese – always deploying Halliday’s trinocular vision. In addition to such theoretical guidance, one might also draw on descriptive generalisations of the kind put forward in Reference Matthiessen, Caffarel, Martin and MatthiessenMatthiessen (2004), Reference Teruya, Akerejola, Andersen, Caffarel, Lavid, Matthiessen, Petersen, Patpong, Smedegaard, Hasan, Matthiessen and WebsterTeruya et al. (2007), Reference Arús-Hita, Teruya, Bardi, Kumar and MwinlaaruArús-Hita et al. (2018) and, with an areal focus on West Africa, in Reference Mwinlaaru, Matthiessen, Akerejola, Agwuele and BodomoMwinlaaru, Matthiessen and Akerejola (2018).

If one is interested in contributing descriptions of Spanish or languages like Spanish, it makes sense to consult the book-length systemic functional description of Spanish by Reference Lavid, Arús and Zamorano-MansillaLavid, Arús and Zamorano-Mansilla (2010) and in addition other accounts of specific areas of Spanish lexicogrammar by Estela Moyano, Anne McCabe and other SFL scholars mentioned in this book. Also relevant is the work on French by Alice Caffarel, including her Reference Caffarel2006 book and her engagement with the challenge of interpreting tense in Romance languages in Reference CaffarelCaffarel (1992). See in particular Chapter 3, Section 3.4 in her book – on the role of a corpus in the investigation of verbal groups; see Reference BenvenisteBenveniste (1966). Systemic functional descriptions of areas of Portuguese would also be relevant, although there is as yet no book-length overview. Taking account of these different Romance languages and the systemic functional descriptions developed so far would shed light on the verbal group in relation to the clause – and the rank-related changes since Latin. These illustrate a common phenomenon around the languages of the world – namely that of cycles along the rank scale, with realisational items drifting down (as recognised in studies of grammaticalisation) and gradually being replaced as new items ‘drift’ at higher ranks along the cline of delicacy from lexis to grammar.

Similarly, if one is interested in contributing descriptions of Chinese or languages like Chinese, one can now consult a number of book-length systemic functional accounts in English (not to mention the ever-growing systemic functional literature in Chinese), including Reference LiLi (2007) on Mandarin and Reference TamTam (2004) on Cantonese, and accounts of particular systems of Chinese, including McDonald (Reference McDonald1994, Reference McDonald1998), Reference Fang, McDonald, Cheng, Hasan and FriesFang, McDonald and Cheng (1995), Reference 443FungFung (2018). But systemic functional work on Vietnamese (e.g. Reference Thai, Caffarel, Martin and MatthiessenThai, 2004) is also relevant, as is work on Thai (e.g. Reference PatpongPatpong, 2005) – for example, in the interpretation of interpersonal particles serving as realisational resources in systems of mood and modal assessment (including modality and evidentiality) at the end of the clause, the construal of aspect, and the location of so-called serial verb constructions metafunctionally in relation to experiential transitivity and logical complexing.

But, of course, helpful linguistic descriptions are not limited to languages that are genetically related to the language under description or spoken in the same linguistic area. There may be particular systems or syntagmatic patterns that are illuminated in the descriptions of languages which are at some or several removes from one another historically or geographically. This is brought out at various points in this book; for example, in Chapter 6, the authors note: “It is simplistic to suggest that the reason Spanish can realise co-referential nominal groups before or after the Predicator is that it has person and number agreement. Comparable flexibility is found in many languages without such agreement, as we will illustrate in the Chinese section of this chapter.”

Similarly, in exploring the system of phase in Chinese, I find it helpful not only to consider so-called ‘serial verb constructions’ common in languages spoken in South-East Asia but also to compare them with such constructions in languages spoken in West Africa such as Akan, Dagaare, Yoruba and Oko. Likewise, at some point after I had begun to try to learn Spanish and (Modern Standard) Arabic (MSA) in the 1970s and was moving into linguistics, it struck me how helpful it was to use (descriptions of) these two languages to illuminate one another. Thus when systemic functional linguists have discussed and debated the demarcation of Theme in Spanish, I have realised that comparison with MSA would be quite helpful (and not only with MSA, obviously). For instance, when one approaches the clause grammar of MSA registerially, it becomes clear that the classification of the language as ‘VSO’ includes a typical pattern in narrative texts (often simply VO or V), but under other registerial conditions, SV or SVO is motivated, as in taxonomic reports. The value of register-sensitive description, comparison and typology has been brought out in a number of SFL studies, including Reference TeichTeich (1999), Reference Lavid and VentolaLavid (2000) and Reference RoseRose (2005); see also Reference Matthiessen, Arús-Hita and TeruyaMatthiessen, Arús-Hita and Teruya (2021).

Comparison and Typology: Etic Pools

Naturally, one should not limit oneself to languages one happens to be familiar with (as a systemic functional linguist), but the exploration needs to become more systematic. Here the long-term project of developing systemic functional comparison and typology is directly relevant. In Reference Teruya, Matthiessen and WebsterTeruya and Matthiessen (2015), we provide a brief overview of systemic functional descriptions of various languages, and Reference Mwinlaaru and XuanMwinlaaru and Xuan (2016) report on a systematic review undertaken by them of systemic functional typological work, with Reference Kashyap, Thompson, Bowcher, Fontaine and SchöntalKashyap (2019) adding another overview (see also Reference Matthiessen, Wang, Ma and MwinlaaruMatthiessen et al., 2022, chapter 7); and in Reference Matthiessen, Caffarel, Martin and MatthiessenMatthiessen (2004), I try to provide links to relevant work in other traditions – though a decade and a half later, this now needs updating (a task I’m working on: a multilingual version of IFG).

Complementing systemic functional theory of language as an empowering resource, typologically oriented generalisations provide further help – minimally, to put it in Tagmemic Linguistic terms, as an etic pool.Footnote ⁷ In working on different languages and in ‘supervising’ research by doctoral students, I have always thought that we need to be nudged to expand the horizon of our imagination – to think more creatively in an informed way about the various ways in which language may evolve in different eco-social environments (I think this was one of the points Whorf tried to get across to us). So we need holistic theories of language and we need to accumulate and process and profile comprehensive descriptions of particular languages. (Developing etic pools based on descriptions of the expression planes of languages is, naturally, much easier than doing the same for their content planes; but the contributions by two British phoneticians who had been influenced by Abercrombie in Edinburgh and moved to the USA – J. C. Catford and Peter Ladefoged – still stand out as remarkable achievements; they give an indication of what systemic functional phonetics might be (cf. Reference Matthiessen, Teruya, Canzhong and SladeMatthiessen, (2021)) and can serve as good models for thinking with also when we turn our attention to the content plane and try to conceptualise and develop etic pools within that planeFootnote ⁸.)

Insiders and Outsiders

Rich holistic theory supported by descriptive aggregation and generalisation can help us as we develop descriptions of particular languages whether we have mastered these languages as insiders, members of the relevant speech fellowship, or approach them as outsiders. The descriptions of English, Spanish and Chinese presented in this book have, of course, been undertaken by ‘native speakers’ (I have added the quotes since the notion of native speaker has been problematised in the last couple of decades – importantly by leaders in applied linguistics, including Alan Davies and Lourdes Ortega), and they are able to enrich their accounts, not only by analysing and interpreting texts in contexts, but also by probing paradigms without the intermediate step of working through elicitation with language consultants.

However, methodologically, the approaches to language description by ‘insiders’ and ‘outsiders’ are productively complementary – not mutually exclusive; in fact, for any given language we ideally need both – a methodological point which I hope will encourage people from different language backgrounds to make contributions not only to their mother tongues but also to other languages. The advantages of being an ‘insider’ are quite obvious; but there are also advantages in being an ‘outsider,Footnote ⁹ two central ones being the experience of reflecting on the language as a learner (if one gets to the point of learning the language in one’s descriptive project) and the comparative point of view.Footnote ¹⁰

In my own work, I have always been an ‘outsider’ – I have hardly done any work on my own mother tongue, Swedish, partly because I was always keen to transcend my own environment, partly because since Swedish is hardly a ‘major’ language (having approximately as many speakers as, say, Akan), I knew that international interest would be very limited. Even when I have worked on English, which has now become my primary language, this has always been against the background of the experience of studying and learning the language, moving in from the outside. So I am keenly aware of the value of being an outsider – also in the case of languages I have never been a learner of but have dabbled with, like Japanese, Kannada and Marathi.

When I worked on Akan as a linguist – initially over three decades ago but now again together with Isaac Mwinlaaru, I would have benefitted enormously from the present volume. Exploring the language in the mid-1980s with a wonderful language consultant, I tried my best to work systemic-functionally; but I had to develop the kind of trinocular reasoning modelled throughout this book on my own (just as Martin had to do when he worked on Tagalog). By taking another step – maybe in this case a knight’s move, I am sure that we can draw on the trinocular reasoning in this book even when we work as ‘outsiders’ with language consultants – or, even better, as members of teams of insiders and outsiders.

Descriptive Methodology

As I recall my early work on Akan, I would like to return once more to descriptive methodology as a way of rounding off my afterword. In SFL, methodology is derived from the theory in the following sense: the systemic functional theory of language in context is a relational one rather than a modular one; the relations that are posited have as their domains the intersecting semiotic dimensions that constitute the ‘architecture’ of language in context according to SFL (see Reference MatthiessenMatthiessen, forthcoming, including comments on problems with the term ‘architecture’). The methodology is essentially one of shunting along these dimensions, exploring the relations trinocularly. This methodology has been foregrounded very effectively in this book; the authors show the power of reasoning ‘from above’, ‘from below’ and ‘from roundabout’ in terms of the hierarchy of stratification (but also in terms of rank, axis and – working sideways, as it were – metafunction) as they develop their descriptions.

The method of shunting applies equally to the cline of instantiation. Because of the importance of authentic natural text in context in empirical, evidence-based descriptive linguistics, the authors show how to move in ‘from below’ along the cline of instantiation, and how to generalise instantial patterns inductively by positing systems (while at the same time being informed by stratal considerations ‘from above’ – guidance coming from semantics and context). And it is possible to imagine how this process could be partly (but only partly!) automated by using the tools and techniques (including machine learning) of NLP (Natural Language Processing), which tends to be (far) ahead of corpus linguistics in terms of computational power and sophistication (inside SFL in the identification of register profiles of different disciplines, Reference Teich, Degaetano-Ortlieb, Fankhauser, Kermes and Lapshinova-KoltunskiTeich et al., (2016); outside SFL, see Reference Lee, Brentari and LeeLee (2018) on unsupervised machine learning in the development of descriptions, with reference to the Linguistica morphological analyser developed by Goldsmith and his team, e.g. Reference Lee and GoldsmithLee and Goldsmith (2016)).

Whether the process of moving in ‘from below’ in terms of the cline of instantiation is manual or partially automated, the selection of text types or registers is methodologically very important since different registers may put different lexicogrammatical sub-systems at risk, as illustrated in this book by the selections of texts from different registers (stories, interviews, procedures, service encounters, court trials and conferences: see Table 1.2) – as outlined in Chapter 1, Section 1.2.1: see Table 1.1 for the use of the registerially different texts in the exploration of the different areas of the grammars of the three languages.

But what about other methods widely used in language description? If we add typological generalisations based on many descriptions of particular languages to our theoretical resources, we can operate with certain expectations. For example, we might tentatively posit a specific system as we develop a description of a particular language, and then look for examples (what I called ‘paradigm probing’ above) – possibly using elicitation in working with a language consultant. This would appear to be a deductive method; but it should really be abductive since we will need to adjust the systemic description as we go along, testing the description as it is being developed against new samples of texts from different registers. This approach is of course a staple in traditional linguistic fieldwork, but we can enhance it by taking semantic and contextual considerations into account. Thus in developing a description of any given language, we actually need to keep shunting up and down the cline of instantiation. Michael Halliday has observed somewhere that people don’t speak in paradigms, so we need to posit potential paradigms as we describe a particular language and then probe them.

As long as we shunt along the cline of instantiation, the hierarchy of stratification and the other semiotic dimensions presented in this book, we remain – as it were – within our own projection of the emergent description we are working on. But there may be other descriptions that can serve as secondary sources – as there will be in the case of languages such as English, Chinese and Spanish with a long descriptive history; and we should make good use of them while obviously assessing their quality at the same time (for example, are they based on naturally occurring texts in context or not?).

Making use of secondary sources is a form of meta-translation – reinterpreting the descriptions they present in systemic functional terms. This requires a great deal of training and practice, and the ‘harvesting’ of secondary sources would require another book (which would include guidance in the use of typological accounts).

The degree to which we have access to secondary sources will vary considerably as we move around the languages of the world. In the case of ‘major languages’, there will very likely be more sources than is humanly possible to consult (as in the case of Chinese, English and Spanish), and since they will be of variable relevance and quality, one has to develop methods of informed selection. In the case of ‘non-major languages’, the availability of secondary sources will vary considerably – say from descriptions with reasonable coverage via brief field notes to nothing at all. But it is vitally important to do a stock-take at the outset of any descriptive project, preferably using a tentative function-rank matrix to develop a sense of the nature and degree of previous descriptive coverage.

In my own descriptive projects, I have ranged from relying largely on primary sources to relying largely on the reinterpretation of secondary sources. Thus when I developed my systemic functional descriptive sketch of the lexicogrammar and phonology of Akan a bit over three decades ago, I had to rely mainly on primary sources – texts and elicited examples provided by a great language consultant. The most comprehensive secondary sources at the time were the grammar and dictionary by Christaller from the 1870sFootnote ¹¹ (though some relevant work had been done in generative phonology).

At the opposite end of this scale from primary to secondary sources, I was given the task in the last couple of years by my friend and colleague Professor Bhimrao Bhosale to sketch a systemic functional description of Marathi as input to a series of lectures we gave in May 2018 at Dr Babasaheb Ambedkar Marathwada University in Aurangabad. As I was preparing material for the lecture series, all I had were secondary sources, so I set out to interpret them (including the examples provided) in systemic functional terms, drawing on similar ‘exercises’ I have undertaken since the late 1970s (originally with MSA).

This is a perfectly reasonable way of preparing for the development of the description of a particular language – as long as one recognises it for what it is: a preliminary guide – perhaps akin to a transfer description (as discussed by Halliday and by Elke Teich). It is a stepping stone – but like other forms of scaffolding and tacking, it needs to be set aside once the description gets airborne (actually, I’m just testing the tolerance for mixing figures of speech in close clausal proximity).

The Need for Descriptive Guidance

On one of our many hikes in the Santa Monica Mountains, Halliday and I discussed the need to produce a paper on how to develop systemic functional descriptions of lexicogrammatical systems. I was well enough into the development of the systemic functional computational grammar known as the ‘Nigel grammar’ and may have started my work on the systemic functional description of Akan, so I was experienced in developing systemic functional descriptions. We thought we might use as an example the development of a systemic description of bound clauses in English since this was one area that still needed some basic work, and we added the task to our agenda.

That was over thirty years ago, and the task still remains on the agenda – except now we have this book, the third step in the series of steps in the development of systemic functional theory and systemic functional descriptions. So many thanks to Martin, Quiroz, and Wang – and congratulations on your achievement!