2 Models and fictions

How are we to name the activity we are engaged in when we

attempt to work out a principled approach to the solution of
practical problems in the acquisition and use of language? This
is not a trivial question. One does not have to embrace extreme
Whorfian doctrine to recognize that how a thing is called can
have a critical effect on how it is conceived. And this can have
important consequences on how we conduct our enquiries and
how they appear to potential sources of research funds. Our
activity has commonly been called applied linguistics. A meta-
thetic alternative has been suggested: linguistics applied. What
do we understand by these two terms and which of them pro-
vides the most appropriate designation for what we do?
I assume that, semantically, the difference between the two
expressions is that in the case of applied linguistics we have a
type classification, what Bolinger refers to as a ‘characterization’
(Bolinger 1952) whereas with linguistics applied we do not. We
might compare Lost paradise with Paradise lost. On the one
hand a type of paradise, a palm-fringed beach perhaps,
thronged with lotus eaters; on the other hand, the fall of man,
the unique loss of Eden. Thus applied linguistics can be under-
stood as a kind of linguistics, like historical linguistics or folk
linguistics. This presumably allows its practitioners to define an
independent perspective on the general phenomena of language
and to establish principles of enquiry without necessary refer-
ence to those which inform linguistics tout court—and tel quel.
With linguistics applied we do not have this option. Whatever we
do with linguistics, however we apply it, the informing prin-
ciples which define this area of enquiry, already pre-established,
must remain intact. Any other principles we invoke must be aux-
iliary operating principles and have to do not with theory as
such but with the technology of application.
 Models and fictions 23

It seems to me, then, that with linguistics applied the theory

of language and the models of description deriving from it must
be those of linguistics. As an activity, therefore, it is essentially
conformist. Applied linguistics, on the other hand, can develop
its own non-conformist theory, its own relevant models of
description. Both lines of approach have their dangers. The ten-
dency of linguistics applied will be to dance attendance to what-
ever tune is currently in theoretical fashion. The tendency of
applied linguistics will be to dance around in circles with no
tune at all. For linguistics applied, therefore, the question of cen-
tral concern is: how far can existing models of description in lin-
guistics be used to resolve the practical problems of language
use we are concerned with? For applied linguistics, the central
question is: how can relevant models of language description be
devised, and what are the factors which will determine their
effectiveness?
This first question takes us at once to the theoretical status of
linguistic models. Linguistics claims to be a science, like physics.
It must, therefore, conform to general principles of scientific
enquiry, and in particular it must avoid intuitive contamination
in the interpretation of its findings. And it is here that we come
against a major problem. Linguists have a way of claiming valid-
ity for solutions which seem peculiar and make no appeal what-
ever to ordinary intuition on the grounds that language, like
God, moves in a mysterious way and outside the range of the
common man’s awareness. On the other hand, this does not
inhibit them from commending solutions because they seem
‘intuitively correct’ and disapproving of those which seem to be
‘counter-intuitive’—these expressions are of high frequency in
the literature. So it is that the linguist finds a way of both having
his cake and eating it too. For although intuition may serve as a
source of hypotheses, it cannot scientifically be used to test them
or to provide them with a mark of approval. Geoffrey Sampson,
for one, is very insistent on this point. In his discussion of the
evidence for linguistic theories he says:
We do not need to use intuition in justifying our grammars,
and, as scientists, we must not use intuition in this way.
(Sampson 1975: 70)
Sampson argues that linguistics, like any other science, is, and
must be, empirical. But it seems that in order to meet this
requirement, linguistics has to restrict its attention to a fairly
24 Theory and practice

narrow strip of language phenomenon: most of what is of

human interest about language behaviour is left out of account.
This, of course, is Hockett’s objection to Chomskyan grammar:
it leaves out of account, he says ‘just those properties of real lan-
guage that are most important’ (Hockett 1968: 10). This is rem-
iniscent of E.M. Forster’s remark about Henry James: ‘Most of
human life has to disappear before he can do us a novel’ (Forster
1927: 147). This reference to literature is not gratuitous and I
shall have more to say about fiction presently.
Sampson argues that linguistics is a science, like physics.
Other people are not so sure. Here is Peter Matthews:
Is there a true analogy . . . between a ‘science’ like linguistics
and a ‘science’ like physics? The problem is not that linguis-
tics is at this stage ‘less advanced’: will the descriptive linguist
ever be able to propose an ordinary mathematical theory for
a language, will he ever be able to falsify a ‘theory’ by one
observation, will he ever be able to construct an absolute test
of his predictions? Surely we delude ourselves if we imagine
that linguistics will be ‘like physics’ sometime in the future.
. . . (The problem) is rather that the extent of our data is in
principle not precise. Languages change, and language inter-
acts continuously with other forms of social behaviour. . . .
We cannot even be expected to account for ‘all of the data’,
simply because we do not know what ‘all of the data’ means!
(Matthews 1972: 77 quoted in Bolinger 1975: 551)
We must surely accept the possibility that human knowledge
and behaviour are just not the same kind of phenomena as those
which are independent of human agency, that they are intrin-
sically participant, first and second person phenomena which can-
not be detached to be third person data of an objective kind and
are not therefore susceptible to scientific enquiry. But if this is so,
what then is the status of statements in linguistics?
This is not the place, and I am not the person, to enter into
the complexities of epistemology. But I would like to make an
informal observation or two which might have a bearing on the
question. Central to the issue is the matter of the privileged
position of the enquirer. In the physical sciences, the enquirer is
trained in detached analysis, in putting the phenomena he
wishes to investigate at a remove from the immediacy of percep-
tion and intuition. In this way he aims at correcting the distor-
tions of the human factor. What this involves, in effect, is the
 Models and fictions 25

acquisition of a secondary culture of a non-participant kind

defined by the philosophy of science and apart from the popular
beliefs and values of participant primary culture. From the van-
tage point of this secondary culture, the scientist can reveal a
different, third person reality, which will often run counter to
that which is popularly accepted. He can, furthermore, demon-
strate its truth, though until the respectability of such a second-
ary culture was established and its practical consequences made
clear this truth was commonly condemned as heresy.
But now how does all this apply to the linguist as scientist?
Can he detach himself and take up a privileged observer’s pos-
ition in the same way? I do not think so. The linguist is of neces-
sity a participant: like other social scientists he enacts the data
he analyses so that there are no separable facts out there, as it
were, open to objective validation. We get a model of behaviour
(that of the linguist) used to analyse another model of behaviour
(that of the common man), one set of cultural assumptions,
which are secondary and superposed, used to interpret another
set of cultural assumptions, which are primary and acquired
through ordinary socialization.
So it is that the linguist is not a representative speaker/listener.
He comes from a different culture and so is able to analyse in
detachment. The data he produces are devised to demonstrate
the general underlying rules of language he has inferred. The
sentences that linguists present have the same sort of status as
the components of kinship analysis presented by anthropolo-
gists: devices for the design of ideal systems. But what cognitive
reality do they have for people who use language and enact kin-
ship relations in the ordinary business of social life? What is
their empirical status?
As soon as a language user is presented with a sentence he
quite naturally converts it into an utterance. His culture prevents
him doing otherwise: he will associate the isolated linguistic
expression with a likely context of occurrence and interpret it
accordingly. If you train him to do otherwise, he loses his inno-
cence and at the same time his usefulness as an informant. He
becomes tainted by foreign influence.
The linguistic analyst, in conformity with the secondary
cultural principles he embraces, represents language as a system
in a steady state, with each term in its place in peaceful co-
existence with others—a picture of classical harmony and well-
formed order. An artefact, a cultural construct. For really there
26 Theory and practice

is conflict, an intrinsic instability—with rules moving into dom-

inance or decline. And every use of language is a resolution of
this conflict, the ex tempore exploitation of the meaning
resources available for expressing the variety and contradictions
of human experience; in short, as T. S. Eliot puts it:
a raid on the inarticulate
with shabby equipment always deteriorating.
(from East Coker)
The linguist presents his view of the world, his model of real-
ity. Shift the perspective and you get a different view, and a dif-
ferent reality. The linguist’s model has no privileged patent on
the truth, it is open to investigation by another, that model by
yet another, and so on. Now, of course, it can be very interesting
and enlightening to see how one set of cultural values interprets
another and the exercise can yield all kinds of insights, but there
is no way, as far as I can see, of establishing which is true and
which is not.
What I am saying is that the linguist, in common with other
social scientists, simply fashions his data to fit a particular set of
cultural assumptions. Further, he will be inclined to interpret
behaviour according to his own predispositions. Liam Hudson
has this to say about psychologists and his remarks apply,
mutatis mutandis, to linguists as well:
The truth is that, to a remarkable extent, psychologists do
research in their own image. One notices this daily. Psycholo-
gists with high I.Q.s do research that reflects well on those
with high I.Q.s. Neurotic introverts show that neurotic intro-
verts do well at school and university. Convergers do research
that bodes well for convergers; divergers, for divergers. . . .
Just as novelists draw on their experience, so too do psych-
ologists. We would both be cut off, otherwise, from the springs
of our intellectual vitality. (Hudson 1972: 129)
We return to fiction. It seems to me that models of human
behaviour in the social sciences are comparable in status and
function to the representations of human behaviour in novels or
plays or any other art form. Both depend upon idealization pro-
cedures which in effect yield prototypes of a kind which we can
set into correspondence with actual and non-idealized reality.
There is not, and cannot be, any direct empirical link between
either of them and the external world. Descriptive models and
 Models and fictions 27

fictional representations create idealized norms of human

behaviour which we cannot accept as a plausible pattern against
which actuality can be compared. Their function is not to be
correct but convincing, to serve as a means towards a more per-
ceptive awareness of what we do and who we are. Hamlet and
Heathcliff do not carry conviction as characters because they
have empirical counterparts among our acquaintances. If they
did, they would diminish into insignificance. So it is with models
of linguistic description. Their value, as a kind of art form,
depends on idealization and a detachment from empirical
validation.
Linguistic theories, I suggest then, are fictions. This is a
perfectly respectable thing to be. As Harré points out, the
fictionalist view of theories has a long history. He describes it
like this:
As an aid to thinking a plausible theory may be a very power-
ful tool, and by using known laws of nature in describing
the behaviour and nature of the entities with which it deals it
gains plausibility. But the entities themselves have no more
reality than the characters of fiction, and the terms which are
used to describe and particularly to refer to them are like the
names and addresses of characters in novels. (Harré 1972: 81)
I find the notion of a theory as an aid to thinking an appealing
one. You do not apply a theory which is a fiction, any more than
you apply a novel. What you do is to use it to develop your
awareness of what a useful model of behaviour might be for
your particular purposes. In this view, a linguistic model is of
value only to the extent that it can help us towards a definition
of what an applied linguistic model should be.
So we come to the second question I posed at the beginning.
What would a relevant model of language in applied linguistics
look like: what assumptions, beliefs, values would it need to sat-
isfy? What would be its cultural character, if you like? It would,
I think, have to be congruent with the knowledge and attitudes
of language users, and be developed, therefore, in accordance
with membership categories of the kind that ethnomethodolo-
gists have made familiar. It would attempt to record the
language user’s intuitive, imaginative as well as rational aware-
ness of the nature of language and would accord as much
respect to the operations of the right as to those of the left
cerebral hemisphere.
28 Theory and practice

It naturally follows that applied linguistic models of language

would vary according to the kind of language user concerned.
They would be consumer based. Halliday (1964) (severely sav-
aged in certain purist quarters) with its argument in favour of
relating adequacy to specific relevance indicates an appropriate
rationale for the applied linguist. Again, Halliday (1969), con-
cerned as it is with relevant models of language points us, I
think, in the direction we might go. Thus the relevant model of
language for teachers of literature will take into account the cul-
tural perspectives of literary criticism and aesthetics. The rele-
vant model for students of physics doing a course in ESP will
take into account the philosophy and methodology of science.
The relevant model for learners in a particular cultural context
will take into account the traditions, values, beliefs, and cus-
tomary practices of that culture and the way language is used to
express them. In all such cases, the starting point is the need and
purpose that the model of language must provide for. You do
not start with a model as given and then cast about for ways in
which it might come in handy. You start instead with a charac-
terization of the learner and his circumstances: his behaviour
and beliefs, his habitual patterns of thinking, his conventional
attitudes to language, the extent and influence of literacy and
other social patterns of cultural transmission and control, and
so on. All these, and more, are factors which might affect the
model of languages to be presented. The relevance of the
learner’s first language has long been recognized and interesting
work has been done recently on the matter of language distance
(see Corder 1979, Kellerman 1979). But questions of social and
psychological distance, as discussed, for example, by Schumann
(1978), are of equal importance in the design of effective applied
linguistic models. We are concerned with language as the right-
ful property of language users, not as the special preserve of the
linguist.
The issue of directionality is critical here, as it is for sociolin-
guistic studies (see Cooper 1980). Linguistics applied works in
one direction and yields descriptions which are projections of
linguistic theory which exploit the data of actual language as
illustration. Applied linguistics, as I have characterized it here,
works in the opposite way and yields descriptions which are pro-
jections of actual language which exploit linguistic theory as
illumination. My own belief is that it is only by preferring
applied linguistics to linguistics applied that we shall achieve
 Models and fictions 29

something which is relevant and accountable in terms of useful-

ness and avoid the kind of ethnocentrism and cultural impos-
ition that has marked so much of language study and teaching in
the past. There are signs of this preference emerging in recent
work. But we have a long way to go and there will be many an
alluring linguistic Will-o’-the-wisp to lead us astray from the
path of our intentions.

Note
Paper presented at the BAAL Annual Meeting, Manchester Poly-
technic, September 1979, and revised for publication in Applied
Linguistics Vol. 1, No. 2, Summer 1981.