The place of systemic functional linguistics as a linguistic theory in the twenty-first century

Abstract

The scaffold for comparison that I will employ will be defined by treating the various domains within which ‘language’ manifests itself as ‘starting points’ for systematic investigation. Such manifestations of language are to be understood in a very broad sense, taking ‘language’ as an observable phenomenon already somewhat for granted. In any situation of assumed language use, the systematic study of that phenomenon can adopt one of four points of departure. Language can be studied in terms of the properties of performances and products (‘in texts’); in terms of the coordination and interaction of collections of speakers, from small groups to entire societies (‘in groups’); in terms of the operations and mechanisms that might be responsible for the capabilities and actions of the individual speakers observed (‘in heads’); and in terms of properties of those situations, generally social, in which language is occurring (‘in contexts’). These starting points are characterised graphically by the four vertices of the pyramid shown in Figure 2.1. The labels selected for these vertices deliberately avoid established theoretical constructs proposed in the linguistics literature so as also to avoid the accompanying theoretical ‘baggage’ that often makes morefoundational comparisons across approaches difficult. There are also naturally differences in how precisely such starting points are handled – that is, the methods drawn upon by particular approaches when drilling deeper into their selected vertices may differ. For example, within an ‘in texts’ perspective, accounts may, on the one hand, employ increasingly formal techniques for building models that ‘generate’ the patterns found in the adopted data, as pursued in (but not only in) generative models; on the other hand, they may rely more on statistical methods for producing distributional descriptions of that data. In contrast, cognitive approaches, most readily aligned with the ‘in heads’ perspective, commonly distinguish themselves from generative or ‘formalist’ approaches on the basis of very different foundational assumptions. Whereas for cognitive approaches observational data with respect to actual language behaviour plays a central role, for generative approaches introspection and judgement data generally play the primary role. This marks a long-running debate summarised well by Butler and Gonzalvez-Garcia (2014: 6-17), drawing on positions articulated in Newmeyer (1998), Croft (1999), Haspelmath (2000) and others, and to which we will return later in the chapter. The labels in the scaffold thus mark out common ‘fault lines’ drawn between accounts that we can use to characterise those accounts more abstractly.