Systemic functional linguistics is an approach to linguistics that considers language as a particular kind of system, a social semiotics system. It was developed by Michael Halliday, who took the notion of system from his teacher, J R Firth. Whereas Firth considered systems to refer to possibilities subordinated to structure, Halliday in a certain sense 'liberated' the dimension of choice from structure and made it the central organising dimension of this theory. In other words, whereas many approaches to linguistic description place structure and the syntagmatic axis in the foreground, Hallidayan systemic-functional theory adopts the paradigmatic axis as its point of departure. The term 'systemic' accordingly foregrounds Saussure's 'paradigmatic axis' in understanding how language works. For Halliday a central theoretical principle is then that any act of communication involves choices. The choices available in any language variety are mapped using the representation tool of the 'system network'. Systemic-functional linguistics is also 'functional' because it considers language to have evolved under the pressure of the particular functions that the language system has to serve. Functions are therefore taken to have left their mark on the structure and organisation of language at all levels. The organisation of the functional framework around systems, i.e., choices, is a significant difference to other 'functional' approaches, such as, for example, Dik's functional grammar (FG or as now often termed, functional discourse grammar) or lexical functional grammar Thus it is always important to use the full designation: systemic-functional linguistics rather than just functional grammar or functional linguistics.
For Halliday, all languages involve three very generalized functions: construing experience (meanings about the world), enacting social relations (meanings concerned with interpersonal relations) and the weaving together of these functions to create text. Because these functions are considered to come into being simultaneously [viz. one cannot mean about the world without having either a real or virtual audience], language must also be able to bring these meanings together: this is the role of structural organisation, be that grammatical, semantic or contextual. These three generalized functions are termed 'metafunctions'[1].
Contents |
The point of departure for Halliday's work in linguistics has been the simple question 'how does language work?'. Across his career he has probed the nature of language as a social semiotic system, that is, as a resource for meaning across the many and constantly changing contexts of human interaction. In 2003, he published a paper in which he set out the accumulated principles of his theory, which arose as he engaged with many different language-related problems. These principles, he wrote, 'emerged as the by-product of those engagements as I struggled with particular problems'[2]:1 as various as literary analysis and machine translation. Halliday has tried, then, to develop a linguistic theory and description that is appliable to any context of human language. His theory and descriptions are based on these principles, on the basis that they are required to explain the particularly complexity of human language. These principles are:
The label "Systemic" is related to the System Networks used in the description of human languages. System networks capture the dimension of choice at each stratum of the linguistic system to which they are applied. The system networks of the lexicogrammar make up systemic-functional grammar. A system is a theoretical tool to describe the sets of options available in a language variety; it represents abstract choice and does not correspond to a notion of actual choice or make psychological claims. Formally system networks correspond to type lattices in formal lattice theory, although they are occasionally mistakenly mistaken for flowcharts or directed decision trees. Such directionality is always only a property of particular implementations of the general notion and may be made for performance reasons in, for example, computational modelling. System networks commonly employ multiple inheritance and 'simultaneous' systems, or choices, which therefore combine to generate very large descriptive spaces.