Towards a Metagrammar for Applying Systemic-Functional Linguistics to Literature

By Maria C. Casas

The problem

The problem that motivates this paper arose during a grammatical analysis of a short story called "Direction of the Road", by Ursula LeGuin. The story is a fantasy, a monologue in which a non-human creature laments a gradual change in its relationship to humans. It opens like this:

They did not use to be so demanding. They never hurried us into anything more than a gallop, and that was rare; most of the time it was just a jigjog foot-pace. And when one of them was on his own feet, it was a real pleasure to approach him. There was time to accomplish the entire act with style. There he'd be, working his legs and arms the way they do, usually looking at the road, but often aside at the fields, or straight at me: and I'd approach him steadily but quite slowly, growing larger all the time, synchronizing the rate of approach and the rate of growth perfectly, so that at the very moment that I'd finished enlarging from a tiny speck to my full size -- sixty feet in those days -- I was abreast of him and hung above him, loomed, towered, overshadowed him. Yet he would show no fear. Not even the children were afraid of me, though often they kept their eyes on me as I passed by and started to diminish. (LeGuin 1976:244)

In the good old days, this creature had the leisure to grow and shrink slowly, in synch with the pace of the walking creatures of the earth; but, as the story progresses, the speaker reveals that industrialization has brought humans in cars, which constrain the speaker to grow and shrink at the breathless pace of the automobile. Gradually, the reader realizes that the narrative voice is that of a tree; and that all the laws of perspective that humanity has conceptualized and encoded in human terms are really, according to the tree, maintained by the care of trees in keeping up illusions for the human "loose creatures" that they pass by.

A linguistic analysis of this text raised a number of interesting questions about stylistics, alterity, and semiotics. The particular type of grammar used to make the linguistic analysis was Systemic-Functional grammar, described in M.A.K. Halliday's Introduction to Functional Grammar (2nd ed.). It was chosen because it is one of the most "semiotic" of formal grammars: it relies on structures created through the interplay of meanings in order to formalize an analysis of text. As Halliday writes in the "Introduction" to IFG, the thrust of the grammar is to describe verbal meaning in terms of function:

A language is interpreted as a system of meanings, accompanied by forms through which the meanings can be realized... This puts the forms of a language in a different perspective: as means to an end, rather than as an end in themselves. (xvi)

The relation between the meaning and the wording is not, however, an arbitrary one; the form of the grammar relates naturally to the meanings that are being encoded. A functional grammar is designed to bring this out; it is a study of wording, but one that interprets the wording by reference to what it means. (xvii)

When we use such a framework, as a way of getting from the meaning to the wording, we make the assumption that there are typical ways of saying things... (321)

As a concrete example, the functional grammar will label a noun as an Actor, or as a Senser. These categorizations interpret the forms of the language (nouns) as meanings, rather than forms in themselves. A Systemic-Functional grammar as presented in IFG, is essentially a semantic grammar. Its functionality inheres in the meanings that the system presents. The question central to this paper is: When a semantic grammar is based on the assumption that "there are typical ways of saying things", how can that grammar describe untypical, fictional ways of saying things? Of seeing things? Of being?

According to the grammar's transitivity system, there are six general ways of describing actions in English: as material processes (happening, creating, doing), behavioural processes (behaving), mental processes (seeing, feeling, thinking), verbal processes (saying), relational processes (symbolizing, having identity, having attribute), and existential processes (existing). Each of these processes includes as components both the participants that typically take part in the process and the circumstances (typically) associated with the process. A relational process, for example, is contained in the clause, "The cat is on the mat by the front door."

LeGuin's "tree" text, however, challenges this transitivity system. For, if a tree can grow and shrink at will; can walk and canter and gallop in place; then human conceptions of transitivity are called into question. Must reality, from the point of view of a tree, always involve agents that are Actors, Sensers, Attributes? These, together with their "typical" circumstances, are also called into question.

The text

To take an example from the text quoted, the phrase "I'd approach him..., growing larger all the time" captures the double perspective at the heart of the passage. The phrase "growing larger" is an event, so it is analysed as a transitive process. Normally, "to grow larger" is a relational process attributing to the subject the quality of more size. That is, "to grow" is meant in the sense of " to become". This would be the ordinary human reading for a description of the event of passing a landmark on the road. As a synonym for "becoming", this sense of "growing" is understood to be non-physical. The object does not physically grow, but only becomes larger in the eye of the beholder.

But the phrase "growing larger" is also redundant, since the word "growing" has the meaning "to become larger". Here is where the peculiar tension of the double perspective comes in: the tree means to say that he is willing himself to grow, quickly and physically, in order to increase its actual size in seconds; whereas human transitivity rejects that possibility and allows "grow larger" only a relational function: something becomes larger.

Furthermore, an attempt to force the semantic grammar to encode the tree's meaning quickly runs the analysis into limitations. Because the phrase also describes the second step in a sequence of material actions: "There he'd be, working ..., ...looking ...; and I'd approach..., growing..., synchronizing...", the clause can also be glossed as "and I'd approach, while at the same time increasing my size volitionally". This meaning includes a participant capable of volition. In a human, Systemic-Functional transitivity, the only processes that can include participants capable of volition are material processes. The grammatical limitation is that the element "larger" becomes unplaceable in the material analysis because it is not a goal, the only possible participant in a material process, according to the grammar. The presence of the adjective "larger" constrains the phrase as relational.

A further problem with the semantic-grammatical analysis is that it does not necessarily encode the meaning we know to be in the text. "I grow tomatoes" is a volitional process, but it means "I nurture tomatoes", rather than "I cause tomatoes to become larger instantly". On the other hand, "I grow" means that the process is not willed, because the participant " I" is the medium and a medium of growth cannot, in our senses of transitivity, be an agent of growth. In order to encode the reality of a conscious entity willing an organic enlargement, and not simply undergoing it, we need both agency (as in growing tomatoes) and a reflexive participant, as in "I grow myself". But "I grow myself" is semantically incorrect.

Grammars as codes

Umberto Eco's views on codes in A Theory of Semiotics offer several ways to evaluate the Systemic-Functional grammar as code, to sort out the semiotic limitations of a semantic grammar, and perhaps to place the tree's sense of "grow larger" in a more satisfying framework.

As Eco points out, many linguists since the nineteen-fifties have used the term "code" metaphorically. Halliday himself inserted in his "Preface" to the second edition of Introduction to Functional Grammar a clarification of his use of the term "code": "I have (perhaps unwisely) used the metaphor of coding at various places, because it is a convenient way of making certain points clear." (xii) Among the passages he refers to are the following:

...both the general kinds of grammatical pattern that have evolved in language, and the specific manifestations of each kind, bear a natural relation to the meanings they have evolved to express...
The adult language has built up semantic structures which enable us to "think about" our experience -- that is, to interpret it constructively -- because they are plausible; they make sense and we can act on them. ...lexicogrammatical structures are likewise plausible: hence we have verbs and nouns, to match the analysis of experience into processes and participants. This is how children are able to construe a grammar: because they can make a link between the categories of the grammar and the reality that is around them and inside their heads. They can see the sense that lies behind the code. (1994:xviii)

What is the relation between the code and the culture which creates it, and which it transmits to the next generation?...Only the grammatical system as a whole represents the semantic code of a language...Just as each text has its environment, the 'context of situation' in Malinowski's terms, so the overall language system has its environment, Malinowski's 'context of culture'. The context of culture determines the nature of the code. As a language is manifested through its texts, a culture is manifested through its situations; so by attending to text-in-situation a child construes the code, and by using the code to interpret text he construes the culture. Thus for the individual, the code engenders the culture; and this gives a powerful inertia for the transmission process. (1994:xxxi)

This gives a powerful inertia for the transmission of posited "natural" meanings; but what happens when meanings are part of an alternate world? A fantasist such as LeGuin can present us with another world; but there are also many situations of linguistic contact and cultural conflict that demand a larger framework than those of a semantic grammar based on "natural" ways of saying things.

In his "Preface", Halliday's caveat continues: "Language is not a code... there are no meanings waiting around to be encoded; the meaning is created in language." (xii) However, as he subsequently made clear, he conceives of it as a dialectical process, in which the meanings that are created in language are determined by contexts of situation.

What can Eco's work contribute to the concept of "code"? In Chapter One of A Theory of Semiotics, Eco makes some useful distinctions around the term. An s-code, according to Eco, is the arrangement of a system, and as such, a system that can "subsist independently of any sort of significant or communicative purpose"(38). It is "a system (i) in which every value is established by positions and differences and (ii) which appears only when different phenomena are mutually compared with reference to the same system of relations." (38) A correlational code, on the other hand (or what Eco calls simply a "code"), is a rule coupling items from at least two different s-codes. A traditional grammar is a type of Eco's correlational code in that it is a structure meant to correlate an s-code of the expression plane (a syntactic structure) with an s-code of the content plane (a semantic structure). Although Halliday claims that traditional grammars "have verbs and nouns, to match the analysis of experience into processes and participants" (1994:xviii, quoted above), in fact these grammatical categories are, in the last analysis, constructed according to their distributive properties in the syntax alone (or the s-code of the expression plane).

From this perspective, a Systemic-Functional grammar is overcoded in the sense that the correlational code relates the expression plane to the content plane in a less than arbitrary fashion: the elements of the expression plane (Actor, Senser, etc.) relate to the elements of the content plane according to the (semantic) properties of the content plane. Thus, the syntax of Systemic-Functional grammar is arranged according to the "semantic structure" (or s-code) of the content plane. In turn, the s-code of the content plane reflects a set of situations, which, taken all together, manifest a culture. This seems to be Halliday's meaning of the word "code": a pattern whose configuration mirrors the configuration of a culture. In Hjelmslevian terms, the Systemic-Functional grammar is the Form of an expression plane called "language", and it is identical to the pattern perceptible in the Form of the content plane "culture".

How do we understand the tree's meaning if it is beyond the scope of our culture/semantic grammar? The answer seems to be in another of Eco's remarks about codes: "Lies are both authorized and elicited by the code's existence. The code does not stop us from understanding a proposition which is commonly believed to be false. It allows us to understand it and to understand that it is 'culturally' false." (1976:64) Eco's use of the term "code" here means correlational code; his remark, however, implies that no sort of code is a complete description of our knowledge.

Semantic theories

At this stage, it may seem wise to abandon a consideration of codes as an approach to semantic grammars and ask if the process at work in understanding a talking tree can better be explained by a theory of sign production and/or reader comprehension. But sign production/comprehension cannot be explained without an underlying code, just as a theory of codes cannot be constructed without reference to sign production. The original question remains, how can a functional grammar describe fictional worlds?

If we were to look for the site of the referent of the particular sign-function "growing larger" in the sense in which it is meant in this context, we are forced to redefine it in terms that are more detailed than those a Systemic-Functional grammar allows. These terms can be considered "cultural units":

Within the framework of a theory of codes it is unnecessary to resort to the notion of extension, nor to that of possible worlds; the codes, insofar as they are accepted by a society, set up a 'cultural' world which is neither actual nor possible in the ontological sense; its existence is linked to a cultural order, which is the way in which a society thinks, speaks and, while speaking, explains the 'purport' of its thought through other thoughts. Since it is through thinking and speaking that a society develops, expands or collapses even when dealing with 'impossible' worlds (i.e. aesthetic texts, ideological statements), a theory of codes is very much concerned with the format of such 'cultural' worlds, and faces the basic problem of how to touch contents....
The semiotic object of a semantics is the content, not the referent, and the content has to be defined as cultural unit... (Eco 1976:61-62)

The terms we are looking for in order to describe the contents of the speech of the talking tree are also cultural units in the sense in which the categories of the Systemic-Functional grammar are cultural units; but they fall outside, or beyond, the code established by the Systemic-Functional grammar. Furthermore, they can never be "touched" in any definitive way.

Throughout A Theory of Semiotics, Eco relies on C.S. Peirce's conception of semiosis as an unlimited chain of interpretants. All meanings are grasped through other meanings; and the chain of interpretants involved is long, leading to an "unlimited semiosis". It is actually this unlimited semiosis that introduces untidiness into the Systemic-Functional grammar. Semiosis is not like a Systemic-Functional grammar.

In the case of the talking tree whose will guides its rate of growth, we understand "tree", we understand that in some possible world trees can talk, and we also understand that perhaps they move quite fast in their rates of growth in that possible world. The problem at the heart of this investigation is not that there are multiple interpretants involved in this particular cultural unit ("to grow larger" in its cotext) , but the way in which they do not match their description as it is put together in the grammar: the semantic s-code of the unit, so to speak. As such, the Systemic-Functional grammar duplicates the problems of formal semantics, which offer impoverished and overly-rigid definitions of semantic items (such as tree: immobile, non-sentient, organic entity). Eco's solution to the drawbacks of such rigid semantics, which he calls "dictionaries", is to call for a more encyclopedia-like semantic theory: that is, definitions that include connotations, and contextual and circumstantial elements.

However, even encyclopedic semantic theories are limited by the need to code contexts and circumstances. A theory of such settings requires that external circumstances also be subject to convention. And this is the stage at which a System-Functional grammar remains. It conventionalizes circumstances and contexts. Halliday imagines that the configuration of a culture, its cultural s-code, can be mapped through its verbal manifestations; but in order to produce such a map, he must introduce a degree of overcoding in relation to the possibilities that exist for human semiosis.

Eco's answer is the model "Q". Unlike the dictionary model, it does not include pre-coded definitions. Instead, the entries are reduced to constellations of elementary cultural units. This is both a problem and a solution, because "every semantic unit used in order to analyze a sememe is in its turn a sememe to be analyzed" (1976:121) As a solution to the problem of the Systemic-Functional grammar's overcoding, it involves breaking down the configurational meanings posited by a system of processes such as "material", "behavioural", "relational", with their typical participants and circumstances. Instead, each process, participant and circumstance in the transitivity system is taken up on its own terms; and in the model Q, the result of a material process such as the "growth" of an active tree can be linked through more primitive semantic units to non-sentient beings such as trees -- in one of their possible incarnations.

Eco describes the s-code of such a semantic system thus:

We can imagine all the cultural units as an enormous number of marbles contained in a box; by shaking the box we can form different connections and affinities among the marbles. This box would constitute an informational source provided with high entropy, and it would constitute the abstract model of semantic association in a free state. According to his disposition, his previous knowledge, his own idiosyncrasies, each person when faced with the sign-vehicle /centaur/ could arrive at the unit <<atomic bomb>> or <<Mickey Mouse>>.
But we are looking for a semiotic model which justifies the conventional denotations and connotations attributed to a sigh-vehicle. And so we should think of magnetized marbles which establish a system of attraction and repulsion, so that some are drawn to one another and others are not. A similar magnetization would reduce the possibility of interrelation. This would constitute an s-code. (1976:124)

What this model gains in its power to explain creativity it sacrifices in its ability to explain syntactic structure. Eco is thus forced to conclude that "[t]he creation of a complete semantic structure must thus remain a mere regulative hypothesis...A semiotics of the code is an operational device in the service of a semiotics of sign production." (1976:128)

Conclusion

Although there remains the problem of how we understand what we understand when signs are produced, some sort of compromise such as the one Eco suggests is probably necessary. He adds to his conceptualization of the process, at a point somewhere between a code and a practice of sign production, extra-semiotic and "uncoded determinants of the interpretation". That is, there are determinants that allow understanding that are not part of a correlative code. These include abduction, under-coding, extra-coding, and overcoding, all operating at the same time. "Thus the criss-cross play of circumstances and abductive presuppositions, along with the interplay of various codes and subcodes, makes the message (or the text) appear as an empty form to which can be attributed various possible senses." (1976:139) Is there a solution to the problem of producing a Systemic-Functional grammatical description of the actions of a talking tree? Perhaps not. The ambition to relate the activity of the tree to other coded actions in its particular semantic universe must be set aside for now. It becomes evident that the nature and scope of the problem is not so much an incorrect semantic grammar as an currently underdeveloped conception of the nature and limitations of grammars as codes, and of codes as static phenomena.

References
Eco, Umberto. A Theory Of Semiotics. Bloomington, Illinois: Indiana University Press, 1976.
Halliday, M.A.K. An Introduction to Functional Grammar. 2nd ed. London: Longman, 1994.
LeGuin, Ursula. "Direction of the Road". The Wind's Twelve Quarters. Toronto: Bantam, 1976.

[ Return to Top ]