Reflections on Multilingual Grammatical Description: An Interview with J. R. Martin on the Occasion of His Seventieth Birthday Interview

In this interview, after briefly summarising his contributions to Systemic Functional Linguistics (SFL) up to this point in his career, Martin addresses key issues in relation to multilingual grammatical description and functional language typology. He reviews some prominent features of the grammatical descriptive work he is engaged in, elaborates on the notion of defeasible language typology, discusses key hierarchies and complementarities in SFL, comments on recent developments in multilingual grammatical description, and anticipates applications informed by robust functional accounts of diverse languages. J. R. Martin is Professor of Linguistics at the University of Sydney, where he is also Deputy Director of the LCT Centre for Knowledge-Building. His research interests include systemic theory, functional grammar, discourse semantics, register, genre, multimodality and critical discourse analysis, focusing on English, Tagalog, Korean and Spanish – with special reference to the transdisciplinary fields of educational linguistics, forensic linguistics and social semiotics. Recent publications include a book on teaching academic discourse on-line ( Genre Pedagogy in Higher Education , Palgrave Macmillan 2016), with Shoshana Dreyfus, Sally Humphrey and Ahmar Mahboob; a book on Youth Justice Conferencing ( Discourse and Diversionary Justice , Palgrave Macmillan 2018), with Michele Zappavigna; and a special issue of Functions of Language 2018 focusing on interpersonal grammar. Eight volumes of his collected papers (edited by Wang Zhenhua, Shanghai Jiao Tong University Press) have been published in China (2010, 2012). Professor Martin was elected a fellow of the Australian Academy of the Humanities in 1998, and was Head of its Linguistics Section from 2010-2012; he was awarded a Centenary Medal for his services to Linguistics and Philology in 2003. In April 2014 Shanghai Jiao Tong University opened its Martin Centre for Appliable Linguistics, appointing Professor Martin as Director.

theoretical and descriptive advances were driven by practical questions. A historical decade by decade approach is useful here, to highlight the main starting points for work that was carried on episodically and prosodically thereafter.
The main focus of my work in the 70s was cohesion, which I explored in relation to schizophrenic discourse (as noted above) and primary school children's story telling (my PhD research; Martin, 1983a). I was mainly focused on reference and conjunction at the time. This work developed through the 80s as the basis for our education initiatives. Its articulation as a stratum of discourse semantics realised through lexicogrammar is best known from Martin (1992) and its 'popularisation' as Martin & Rose (2003/2007. This involved reinterpreting cohesion as discourse system and structure (ultimately comprising ideation, connexion, negotiation, appraisal, identification and periodicity systems realised through co-variate structures).
The 80s saw the development of work on genre, as a level of context analysis extending SFL's traditional concern with register (i.e. field, tenor and mode in our stratified model of context). The model was inspired by analyses of types of writing in school, service encounters, doctor/patient interviews, dinner table conversations and interviews with dog breeders. Genres were conceived as recurrent configurations of meaning, realised through specific combinations of registers variables unfolding stage by stage in text. Conceived in these terms genre was used to design curriculum and pedagogy across sectors and disciplines in education and later on to analyse Youth Justice Conferencing genres and macro-genres. Martin & Rose (2008) is the canonical introduction to this perspective on context.
The best known initiative from the 90s was the emergence of the appraisal dimension of discourse semantics. Our work on story genres, critical responses to literature and art, history discourse and media texts led to the development of a model of evaluation, including attitude (the kind of feeling), graduation (its strength and precision) and engagement (sourcing the opinions in play). Martin & White's (2005) introduction to this framework continues to be my most cited publication, apparently filling a widely perceived gap in the study of interpersonal meaning across registers and genres, in many languages and cultures.
By the 00s, my PhD students and colleagues managed to drag me out of language per se and into the world of multimodality inspired by Kress & van Leeuwen's Reading Images (1990, 1996. I worked with Len Unsworth and Clare Painter on extending Reading Images' analysis of single images to deal with successive images in children's picture books (Painter et al., 2013), including preliminary work on modelling inter-modality. More recently I have been involved in a project developing a model of paralanguage, to be published as Ngo et al. (in production).
During the teens, another group of PhD students and colleagues lured me back to grammar, giving me a chance to extend the work I'd been doing on Tagalog throughout my career (Martin, 2004a) and develop an interest in Spanish and Korean. For my part this was largely motivated by the growing interest around the world in SFL, as researchers were attracted to our work on literacy, genre, appraisal and multimodality in particular. I felt that research in any of these areas needed to be grounded in a robust account of the lexicogrammar of the languages at stake, and that grammars had to be developed that described those languages in their own terms (and not simply as reflections of well-known SFL descriptions of English such as Halliday (1985) and subsequent editions). The results of this work are just now appearing (Martin et al., 2020, in production;Martin & Quiroz, 2020 and our grammar of Korean is well underway (Kim et al., in preparation).
As far as I can see into the future, my main preoccupation in the 20s will be work on individuation, including the allocation of discursive resources across communities and their uptake by users to affiliate in like-minded belongings. This will inevitably draw me back into work on discourse again, circling back to where my interest in language and linguistics began.
To more directly answer your question then, I would characterise my major contributions as having to do with the stratification of language's content plane as discourse semantics and lexicogrammar, the stratification of language's context plane as genre and register, the appraisal model of evaluation, axial relations and argumentation in language description, defeasible language typology, intra-and intermodal re/instantiation, and a tri-nocular perspective on hierarchies in SFL (realisation, instantiation and individuation). For more detailed accounts of the themes introduced above, from my perspective, see Martin (2014Martin ( , 2016. For a more objective appreciation see Doran (2020).
It is my impression that you are more widely acknowledged for your contributions in discourse semantics, genre and register, and appraisal systems, especially here in China, and many people may find it surprising that, as you mentioned, you went back to grammar over the past decade or so. What are the most prominent features of the grammatical descriptive work you are now engaged in?
That's interesting, isn't it? You have to go a long way down my Google Scholar citation list before you get to any of my descriptive work as a grammarian! That's probably because a lot of my descriptive work has been on Tagalog (Martin, 1981(Martin, , 1983b(Martin, , 1988a(Martin, , 1990(Martin, , 1995(Martin, , 1996a(Martin, , 2004aMartin & Cruz, 2018. I developed an interest in the language during one of my field methods courses during my MA at the University of Toronto and followed up with intensive language training and courses on Philippines languages and the Austronesian language family in general at the University of Hawaii's LSA Linguistic Institute in 1977. Then during the 80s I spent three sabbaticals in Manila, working on the language. Once I had children it wasn't possible to spend extended periods of time there, but I kept working on Tagalog grammar as occasions arose. It seems that this work is of interest to only some of the relatively small part of our community that works on describing languages other than English. Shanghai Jiao Tong University Press in fact refused to publish a volume of my Tagalog papers as part of my Collected Papers series (Martin, 2010(Martin, , 2012, reasoning I presume that there would be no interest in China in such a collection. I think that's an unfortunate attitude and certainly doesn't help encourage functional grammar work on languages in China, including minority languages, and languages in surrounding regions (such as the Philippines where Tagalog is spoken). Ironically my work on Tagalog is the part of my work of which I am most proud. As Halliday (1985, p. xxxiv) comments, 'Twentieth-century linguistics has produced an abundance of new theories, but it has tended to wrap old descriptions up inside them; what is needed now are new descriptions.' I think I've managed to produce a range of new descriptions of Tagalog, across ranks and metafunctions. 1 That's been very satisfying for me.
As for my 'grammatics' publications, by which I mean work that reflects on doing grammar, they are not I guess everyone's cup of tea (e.g. Martin, 1984Martin, , 1987Martin, , 1988bMartin, , 1996bMartin, , 1996cMartin, , 2004bMartin, , 2008. In more practical terms, I do think people have found the workbook for Halliday (1985 and subsequent editions) I got together with Christian Matthiessen and Clare Painter (Martin et al., 1997 useful. And as you know I've put a couple of practical grammar writing handbooks together (translated into Chinese by you and Zhu Yongsheng) -one about system networks  and another about language description (Martin et al., in press). So I guess people would mainly see me as someone who supports grammar analysis, without being a grammarian myself per se.
Of course my generation of linguists were by and large trained as grammarians in our coursework, and I have always worked in a linguistics department -teaching functional grammar in our undergraduate and graduate programs. There I've had the opportunity to supervise postgraduate work bearing on the grammar of several languages -Mandarin Chinese, Pitjantjatjara, Cantonese, Persian, Bahasa Indonesia, Spanish, Tagalog, Mongolian and Gija; and I've organised several intensive workshops to help foster these initiatives (in part reflected in functional language typology book I edited with Alice Caffarel and Christian Matthiessen (Caffarel et al., 2004)).
For the past few years I've been working closely with Mira Kim and Gi-Hyun Shin developing a grammar of Korean (Kim et al., in preparation). The challenge of building a functional grammar across ranks and metafunctions has been a fascinating one for me, and I think we have been able to make some important contributions in terms of pushing the description beyond clause rank system and structure through the group/phrase and word rank systems and structures through which they are realised. I've also had the opportunity to work with Beatriz Quiroz on Spanish, and through that work make a small contribution to translation studies (Martin & Quiroz, 2021) and to language typology (Martin & Quiroz, 2020).
People who are familiar with my work on appraisal will probably not be surprised that one thing I have especially tried to push along is grammatical work on interpersonal meaning. This is reflected in a special issue of Functions of Language I edited (Martin, 2018) and a collection of papers edited by myself, Beatriz Quiroz and Giacomo Figueredo for Cambridge University Press (Martin et al., in production). For most linguists, as you know, this is a relatively neglected dimension of grammar and it's important to keep encouraging grammar work in this domain.
I have also been concerned recently about the lack of work from an SFL perspective on nominal group system and structure across languages. Yaegan Doran, Dongbing Zhang and I are currently editing three special issues of the journal Word to try and open up this dimension of SFL grammatical description.

Can you elaborate a bit on what is 'defeasible' language typology?
This notion is inspired by work by Matthiessen (2018) on multilingual language description and elaborated in Martin & Quiroz (2020. Matthiessen makes the critical point that diversification across languages can be usefully explored by asking how far it extends 'upwards' -along various dimensions of SFL theory. And he proposes a general fractal principle to guide this exploration -a cline from more similar to more different: In the comparison and representation of two or more languages, we can postulate a general fractal principle in the form of a cline from 'most similar' to 'most different'. The principle is fractal in the sense that it is manifested along a range of semiotic dimensions in the overall 'architecture' of language… . The dimensions… are the hierarchy of stratification, the hierarchy of rank, the hierarchy of axis and the cline of delicacy. (Matthiessen, 2014, p. 43) This principle is outlined in Figure 1 below, taken from Matthiessen's draft paper (but unfortunately omitted from its final publication).
The general rule of thumb we are looking at here is that when things look different, we need to move 'up' and look again (Matthiessen, 2014, p. 42). From the point of view of axis, different structures may be realising comparable systems; from the point of view of constituency, structures at different ranks may be realising comparable systems; and from the point of view of stratification, an array of lexicogrammatical resources may perform comparable discourse semantic functions. In SFL we can in principle push these two steps further on, asking how different discourse semantic systems realise comparable field, mode and tenor variables, and how in turn different combinations of register variables cooperate to realise comparable genre systems. As far as functional language typology is concerned, what is important is to be explicit about the perspective we are comparing languages from -which stratum, which rank, which axis? We also need to keep in mind that any generalisations we make will have to be treated as defeasible, since what we abduce as the same or different from one vantage point can look rather different from another. The metaphor of shunting has long been iconised in SFL, deriving as it does from Halliday's (1961) seminal articulation of the model. Halliday used the metaphor in relation to the process of describing a single language. But we have to rearticulate the metaphor for cross-linguistic work -encouraging linguists to compare and contrast languages from multiple perspectives -and not privilege categories over relations (class over function in SFL terms), syntagmatic relations over paradigmatic ones (structure over system in SFL terms), morphemes over clauses (lower ranks over higher ones in SFL terms), syntax over semantics (lexicogrammar over discourse semantics in SFL terms), semantics over pragmatics (ideational meaning over interpersonal and textual meaning in SFL terms) or language over context (text over field, tenor, mode and genre in SFL terms). All of these complementarities need to be respected as such (cf. Halliday, 2008), with our gaze shifting systematically upwards and downwards to recontextualise what we find.
Does it mean that, of those complementarities, SFL grammatical description privileges things the other way round, i.e. function over class, system over structure, higher ranks over lower ones, discourse semantics over lexicogrammar and context over text?
Yes, absolutely. Having declared that, we need to always keep in mind that we are dealing with complementarities -you can't have one perspective without the other! Most of the privileged part of those complementarities are represented in diagrams as the 'up' side (e.g. system, higher rank etc. as shown in Figure 1

above). But aren't 'ideational', 'interpersonal' and 'textual' designated as three parallel components of meaning? How might we understand the privileging of one kind of meaning over another?
It's useful to bring Matthiessen in again here. Summing up his discussion of choice in translation, he poses a number of research questions which need addressing in future research informed by SFL (Matthiessen, 2014, p. 322): -what degree of systemic separation do choices produce in translation -minimal "equivalence" or maximal "shift"; and what are common shifts in terms of delicacy -more or less constant delicacy, increase in delicacy, decrease in delicacy? -to what extent are choices made within the same metafunction and to what extent do they entail a shift from one metafunction to another; and what are common shifts in metafunctional modes of meaning -are there favoured directions of shift such as the experientialisation of meaning (as in grammatical metaphor)? -how far up do we have to ascend in terms of rank and in terms of stratification to locate the systems where meanings are located through choices in the target language? We discussed his third point above. You are now bringing his second into our discussion. The key point here is that in SFL metafunctions are not a hierarchy -as you say, they are simultaneous bundles of features that each make a contribution to any instantiation of a linguistic system. In this respect SFL contrasts with other models of language, which tend to stack different kinds of meaning up (in a phonology, morphology, syntax, semantics and pragmatics hierarchy which in effect privileges ideational meaning and blurs the boundary between interpersonal and textual perspectives). So it would be a mistake in SFL to privilege one or another metafunction in language typology research.
That said, for a particular applied purpose, we might want to privilege one metafunction -at least as a starting point. In translation studies for example a translation would probably not be considered a translation unless there was some degree of ideational convergence; so in a sense translation studies have tended to privilege ideational meaning over interpersonal and textual considerations. There's a price to be paid for this, and more recently a lot of effort has gone into bringing textual and interpersonal meaning into the picture (cf. Munday, 2012;Kim et al., 2021).
And moving beyond grammar, into the realm of text analysis, I can perhaps be accused of sometimes privileging textual meaning (e.g. Martin, 2009) when I suggest that analyses of ideational and interpersonal meaning need to keep firmly in play the hierarchy of periodicity that the textual metafunction contributes to the organisation of discourse.
As ever it all depends on the purpose of one's analysis whether we privilege one metafunction over another. But this is a practical, not a theoretical concern. Theoretically speaking, SFL does not privilege one kind of meaning over another.

Halliday (1985, p. xxxi) explicitly warns us against 'foist[ing] the English code on other'
languages, and functional language typology gives prominence to a number of general descriptive principles (e.g. axial argumentation, approaching grammar from above, the trinocular perspective), which are designed to be applicable to description of all languages. I think the general theoretical model and particular languages also constitute a sort of complementarity -on the one hand, we use SFL theory to account for the resources through which a language makes meaning; on the other hand, we conduct research on various languages to attest to the appliability of the theory, or perhaps adjust the theory based on specific descriptions. Over the past decade or so, what kind of contribution have the multilingual grammatical descriptions made to the overall theoretical framework of SFL proposed by Halliday? Have they brought about any modification to the theory? And what contributions do you expect grammatical descriptive work to make to SFL theory in the next decade and beyond?
You've saved your most challenging questions for last, Pin! By way of answering let me first be clear about how I understand the term 'theory' in relation to SFL. I take SFL theory as comprising a number of basic understandings which give rise to descriptive motifs and generalisations gleaned from the description of particular languages. At the heart of these theoretical understandings is axis (i.e. system/ structure relations), from which we can derive rank, metafunction and stratification  according to the way systems bundle together in relation to units of different size (rank), kinds of meaning (metafunction) and levels of abstraction (stratification). To this we need to add our three hierarchies (Martin, 2010): realisation (system and structure -i.e. axial relations across strata, ranks and metafunctions), instantiation (system and text) and individuation (system and allocation/affiliation).
Based on these categories of our theory we construe our descriptions of particular languages, constantly adding to our reservoir of descriptive motifs and generalisations as we go. This is where our various typological complementarities come into play -transitive and/or ergative transitivity systems, modality and/or evidentiality assessment systems, tense and/or aspect temporal systems, recursive and/ or non-recursive agency systems and so on (Matthiessen, 2004). I feel it is here that work on particular languages has made and continues to make significant contributions to SFL (e.g. Martin, 2018;Martin et al., 2020, in production;Mwinlaaru et al., 2018); Mwinlaaru & Xuan (2016), Arús-Hita et al. (2018) and Xuan & Chen (2020) provide a useful survey of this work.
As far as SFL theory is concerned, I would say that to date, multilingual grammatical descriptions have not engendered modifications to SFL theory -SFL is still flexing its extravagant theoretical musculature we might say in relation to grammars of the particular languages it encounters. That said, work across languages has certainly refined our understanding of our hierarchies, our realisation hierarchy in particular (i.e. axis, rank, metafunction and stratification). As far as axis is concerned, this work has improved our appreciation of the cryptogrammatical reasoning that affords our distinctively penetrating analyses of grammar as a meaning making resource. As for rank, work across languages has improved our understanding of the distribution of the realisation of clause systems across clause, group/ phrase and word structures -both within and between languages -including the development of work on morphology (Matthiessen, 2015). With respect to metafunction, multilingual research has deepened our appreciation of the association of types of structure (particulate, prosodic and periodic) with specific metafunctions (ideational, interpersonal and textual respectively), including the co-option of logical particulate structure (i.e. hypotactic dependency) to specify the prosodic scope of modal assessment systems (e.g. Martin & Cruz, 2018;Martin et al., in press). Turning to stratification, approaching grammar from above and taking discourse semantics (contextualised by register and genre) as point of departure has certainly helped clarify how far lexicogrammatical descriptions need to be pushed by way of accounting for how meaning above and beyond the clause gets realised in clause complexes, clauses, groups/phrases, words and morphemes (e.g. Zhang, 2020a, b). To date the main contributions to instantiation derive from Figueredo's modelling of logogenesis in Brazilian Portuguese (e.g. Figueredo & Figueredo, 2019;Figueredo & Sant'Anna de Lima, in press). And finally, for individuation, we have Hao & Martin (in preparation) to look forward to -with its focus on history discourse (including its axiology) -across English, Spanish and Chinese languages and cultures.
What are my expectations for the future? I think we'll get better at cryptogrammatical reasoning, especially with respect to knowing what kind of reactances to look for when working on specific regions of ideational, interpersonal and textual meaning. We'll improve our understanding of how to formalise the distribution of meaning across ranks, including distinguishing more clearly among particles, clitics and affixes and regularising our terminology (on a language specific basis) for clause, group/phrase and word systems and function structures. We'll fine-tune our understanding of types of structure -distinguishing recursive from non-recursive orbital and serial particulate structures, more effectively modelling scope in relation to prosodic structure and finding better ways to represent waves of information flow at clause and group/phrase ranks (cf. Martin, 1996cMartin, , 2004bMartin, , 2008. One of the most exciting developments I'm looking forward to has to do with the ways emerging lexicogrammatical descriptions will rebound on our modelling of discourse semantics. We are already beginning to see this emerge through work on modal assessment in relation to appraisal (engagement in particular) and negotiation (especially move rank systems); cf. Martin & Cruz (in production), Zhang (2020a, b). More work on interfacing transitivity and logico-semantics with Hao's model of ideation and connexion should be equally productive. And our understanding of the nature of language as a probabilistic system will surely evolve -both with respect to Halliday's proposals for equiprobable and skew systems (Halliday & James, 1993;Halliday, 2005) and embryonic work on the re-weighting of these inherent probabilities by register and genre in Nesbitt & Plum (1988).
Ultimately of course what will really kick things around will be applications building on functional grammar descriptions which present the meaning potential of languages in their own terms -not in terms dictated by extant descriptions of well-described languages like English or Chinese. Putting multilingual SFL research to work in educational linguistics, clinical linguistics, forensic linguistics, ecolinguistics, translation studies and so on is what has led to the most significant theoretical and descriptive developments in the evolution of SFL (Martin, 2014(Martin, , 2016. I'm sure it will continue to do so in future as applications develop around the world. Recent interventions designing multilingual pedagogy show special promise (e.g. Crane, in press;Kartika-Ningsi, in press;Kartika-Ningsi & Rose, 2021;Ramírez, 2020).
I'll see some of this unfold, Pin; but you'll get to see much more. Do enjoy the panorama, as time flies by. Note 1. We can't really go into detail in an interview of this kind, but I would note in passing -a Whorfian perspective on syndromes of meaning (family, face and fate) in Tagalog grammar (Martin, 1988); an integrated account of interpersonal meaning (mood, modality, modal assessment) and its co-option of logical structure to establish prosodic domain (Martin, 1990); fractal marking of hypotactic dependency across ranks in relation to both recursive and nonrecursive systems (Martin, 1995); a model of transitivity relations based on a centrifugal/ centripetal opposition (Martin, 1996) rather than the usual transitive or ergative perspectives; recognition of the unmarked conflation of Theme and New at the end of a Tagalog clause (Martin, 1983(Martin, , 2004; generalisation of relational clause oppositions across process types in relation to modal responsibility (Martin, 2004a;Martin & Cruz, 2018); co-option of identifying relational clause structure for projection and elemental interrogation (Martin, 1990(Martin, , 2004aMartin & Cruz, 2018; and re-interpretation of model assessment clitics in relation to heteroglossic engagement (Martin, 1990(Martin, , 1993Martin & Cruz, in production).