Descriptive Grammar - Descriptive grammar: the systematic study and description of a language. Descriptive grammar refers to the structure of a language as it is actually used by speakers and writers. As with any scientific theory, the basic concepts (things like stratum and metafunction, realization, instantiation, and delicacy and rank) are not empirically verifiable; they form an abstract framework, or model, set up as a means of exploring and explaining.
(An abstract model of this kind is itself a semiotic system: see Section 4 below.)
Such a model is all the time being refined and elaborated in use.By contrast with these general theoretical categories, the descriptive categories are set up by reference to specific features of given languages, and hence in principle are verifiable. That is to say, they can be characterized in ways which make it possible to decide whether or not something is an instance, and whether or not some category is present in the system at all. Two kinds of considerations are typically involved.
(i) Most generalizations made in describing grammar are valid only with some degree of probability. They allow us to identify "exceptions", for which we then seek a further generalization; and so on.
For instance: we say that the English Moodtag consists of Tagfinite + Tagsubject, where the former repeats the Finite operator, and the latter repeats the Subject (each grammaticized in deictic form, as in Mary knew, didn't she?). Typically, the tag does repeat the Finite; but we also come across instances like She'll like fairy tales, does she? where Finite will is replaced by does. The meaning is 'I expect she likes fairy tales; will you confirm my expectation?'. This contrasts with She'll like fairy tales, will// won't she?, which would mean 'Do/ don't you share my expectation?'; and it is an instance of a more delicate subclass whereby a modal Finite in the clause is replaced by a temporal one in the tag (cf. It must have been crowded in there, was it?).
And where the tag system in general favours reversing the polarity (positive > negative, or negative > positive), in this subsystem the local probability is for the polarity to remain constant . though in both cases the other, marked option also occurs.
Such probabilistic generalizations can still be tested, by reference to quantitative patterns in the corpus. They are significant because they enable us to define the meaning of doing something by contrast with the meaning of not doing it, or of doing something else. Certainty then appears as the limiting case of probability . the point where no further meaning is created. Thus if the second part of the generalization above ("repeat the Subject") holds true in all instances, there is no more delicate choice point here at which more semantic space is opened up. We then go back to the corpus to check; we find But my husband heard it too, didn't you? and we have to decide whether that fits the pattern or not. This leads into the next principle to be considered.
(ii) Clearly the answer to whether you in the tag in the last example is a repetition of the Subject or not depends on how we are looking at it. If we are looking at it 'from below', at how it is realized in form, then it is not; the pronominal of my husband must be he. If we are looking at it 'from above', at the meaning which is being realized, then it is; you is functioning anaphorically and the two are co-referential. Probably the speaker turned towards her husband while moving into the Tag; but note that we do not need any information of this kind from outside the text .it is the text itself that construes the meaning and the context of situation for us. Because we know the MOOD system of English, including the principle of the Tag (based on the meaning of Subject as the modally responsible element in the proposition), we are able to interpret the instance by locating it in its place in the meaning potential.
And here we are adopting the third perspective, looking at it 'from around'.
As we saw in Section 1, it is a critical feature of systemic grammatics that the grammarian has trinocular vision, looking at any phenomenon from each of these three stratal perspectives. We may choose to privilege one or another; but all are taken into account, and since they will typically conflict the optimum description for any particular occasion will almost always be a compromise. Traditionally grammarians have begun by looking 'from below', because this is the most obvious way in: we ask questions like "What is the meaning of wa in a Japanese clause?" first identifying a form (this then becomes the grammatical 'fact'), and then asking what this form means. But in a functional grammatics such as systemics, relatively greater priority is accorded to the perspective 'from above', where the question is one such as "How does the Japanese clause construe the flow of information?". Interestingly, the perspective that seems to be most often ignored is that 'from around', where we construct the paradigmatic environment: the set of options that constitute the local grammatical potential. In this example, we would be setting up the network of systems that constitute the textual resources of the clause: on the analogy of English, the systems of THEME and of INFORMATION, and their realization through the structural elements of Theme + Rheme and Given + New.
The analogy with English may not hold, of course; and here again it is the trinocular perspective that is vital. There is no objective criterion for deciding how much alike two phenomena must be for us to call them by the same name: in the last resort questions such as "Is there a passive in that language?" or "How do we recognize the Subject in this language?" are questions about whether to transfer labels in comparing one language with another. There is nothing wrong with making predictions about one language on the basis of what is known about others; this is a normal way of proceeding. But one has to guard against foisting the categories along with the labels.
The systemic approach would be, rather, to ask a question such as "Is there a system which redistributes the participants into different textual statuses?" (as English Actor, Goal, etc. are redistributed into different patterns of thematic and information structure). If there is, we call it the VOICE system; and then if it sets up some kind of unmarked/ marked opposition, we may call the unmarked term "active" and the marked term "passive". Similarly, instead of "How do we recognize the Subject?" we might ask "Does the grammar incorporate an element having a specific function with regard to validating a proposition?" if so, we justifiably refer to this as the "Subject". All these steps depend on adopting the same 'trinocular' stratal perspective.
