Integrational Linguistics and Systemic Functional Linguistics: Prospects of Dialogue?

This is not intended to be an analysis of the arguments presented, or even a comprehensive presentation of the arguments presented. It is a personal view of the proceedings.

The day was organised into a series of half-hour papers, each followed by 15 minutes of audience interrogation. At the end of the day there was an hour of open discussion. The meeting was ably chaired by Charles Owen of Birmingham University. The interrogation and discussion ranged from the inquisitive to the inquisitional, and it was clear that no conclusion to the dialogue was likely in the seminar. However, the seminar provided clear and certain starting points from which the dialogue could proceed.

"There is no such thing as data" (Roy Harris)

"I do not see linguistics as a science" (Roy Harris)

"Where is the data?" (Nigel Love)

"What is the IT we have to describe?" (Nigel Love)

"Rules have nothing to do with resources" (Theo Van Leeuwen)

"To have construal, language must have its own role" (Geoff Thompson)

"Meaning is shaped wholly by context, context is shaped partially by participants" (Peter White)

«» What is a system? Both sides were arguing for and against this concept, but it seemed that several definitions of the concept were being used by different sides: model, function, tool, structure and rules repository were just some of the definitions that seemed to be in use.

«» Is the SOS an emergent from the sum of other systems, or are the other systems emergent from the SOS? It seemed to be used both to justify the search for systems and as a product of the search (not an unreasonable philosophy, but it can lead to grail-searches).

«» If there is no data, is there language at all? What is language other than the transfer of sounds/squiggles (phonological/orthographic data) from sender to receiver with the intention (pragmatic data) of producing consonance of meaning (semantic data)?

«» Systems can be emergent from perceived rules or from actual rules. To use an example from Hamer and Copeland (Living With Our Genes, Macmillan, 1998, ISBN 0 333 76017 4): Ohioans uses knives and forks, Tokyoans use chopsticks; there are genetic differences between the two populations, but to look for a "cutlery gene" is futile. Cutlery use is a description of perceived rules, not a system of actual rules. How can we differentiate perceived from actual?

The problem of defining the word system is not limited to linguistics, and has been addressed in the discipline of computing. Here, the term has achieved a relatively stable definition, at least in reference to the production of software. The system, in computing, is both a logical object and a dynamic effect. On the logical side it is an identifiable and self-contained entity, although not a physical one. It can be subdivided into modules, which are identifiable entities contained within the single system; and components, which are the basic units from which a system is developed, and which compose the system. On the dynamic side the system enables a task to be carried out. It is composed of processes, which subdivide the tasks; and functions, which are designed to produce limited and highly predictable effects (most definitions say that functions have a single output, but this is really a mathematical and not a computing definition).

The module/process level in these models is deliberately vague: the system is the total desired effect (and the objects necessary to achieve it), and the components/functions are the tools available to build the system. In between is the building itself, which can be as simple or complicated as the builder wants. This six-part definition does not satisfy the direct question "what is a system?", but it gives a useful way of analysing the approach to and structure of a task in terms of computer development. To deal with the duality of system the word structure is sometimes used for the logical meaning, leaving system as the dynamic meaning only.

If we remove the computer references from the definition we are left with a model which can be summarised as follows:

The problem with this model is that it describes two things. These things are qualitatively different but they appear to be expressed in the same way. This expression, however, is an emergent feature of the nature of the things, and is not the nature itself. The two things are:

the structure that is describable as a set of components;
and
the set of components that is describable as a structure.

To give examples, the human brain is definitely a structure with an identifiable system (the mind). However, it is possible to divide this structure into modules, some controlling other modules, which process thought to produce the system of mind. At bottom are a series of brain components, some with fixed function (like the amygdala) and some with more varied function (like the frontal lobes). The structure is reasonably non-controversial, but the components are less so, with location and function both being indeterminate in some cases.

Human society, on the other hand, is not a structure with an identifiable system. The components (human beings) are discrete and identifiable, but their individual social functions are not knowable in a given situation (module). Yet their average contribution to a social process is often predictable, which means that the system of human society is amenable to planning. This gives the appearance of an identifiable structure, but this structure is an emergent property of the components. It is, therefore, constantly vulnerable to unexpected component effects.

While it would appear that the module level is an unnecessary fudge between the whole and the sum of its parts, it cannot be left out. A structure cannot be described as the sum of its components, because those components have different weight and purpose in the structure. It is the module level that identifies these weights and purposes.

The problem for language would appear to be that it is both a top-down and a bottom-up structure at the same time. In a single mind the structure is very likely top-down: everyone knows the language they speak, and they are able to operate it both to manipulate ideas in their own heads and to produce code-streams to transmit ideas to other heads; however, each individual would be hard-pressed to analyse their language structure into a series of rules (modules), although they may well have an idea of the resources (components) they use in language. At the same time, we are aware that the language resources of individuals produce messages which reliably transmit ideas from one person to another; we know that there must be a structure to this, but as yet we have no agreement whether the structure will explain cognitive consonance or determine it.

This was the first professional linguistics seminar I had ever attended, and I found that it was full of names from my bookshelf. Roy Harris, Theo Van Leeuwen and Geoff Thompson are all quoted in my MA, and it was the discovery of Hallidayan linguistics (via Geoff's book) that got me started on the subject in the first place.

I also found out that THE Michael Toolan is the Michael Toolan I went to school with, over 30 years ago. It was strange to compare two very different lives that had started out in the same place, separated, and then collided again.

I was a little frustrated by the entrenched positions, but I gained a very clear understanding of why the positions are held, and what the territory is between the two positions. However, I continue in my prejudice that there is a middle way between the two positions, definable if not discoverable. From the Integrationist point of view the middle way will be an ad hoc compromise; from the SF point of view it will be defining structure. What it is doesn't really matter, as long as it is defined or found.