Explorations in Applied Linguistics 2
Explorations in Applied Linguistics 2
Explorations in
Applied Linguistics 2
H. G. Widdowson
Introduction 1
Bibliography 271
Acknowledgements
The papers in this book are further enquiries into issues raised
in its predecessor of the same title, and in particular they seek to
discharge the responsibility of applied linguistics, as stated in
the earlier book (and restated in the first Section of this one), to
devise a coherent model of linguistic description which will be
relevant to language teaching.
As before, the papers here are in some respects a varied assort-
ment. They were prepared at different times and so mark differ-
ent stages in the conceptualization of the matters concerned.
Some focus attention on descriptive issues, others attend more
closely to pedagogic ones. Again, some papers were written
within an academic framework and key their arguments in with
the publications of other people, acknowledged by citation and
reference in the approved scholarly fashion; other papers, more
popular in intention and designed to be performed, dispense
with these accoutrements of scholarship altogether.
But beneath the appearance of variety of presentation, there
is a pattern of recurring ideas which are taken up from the pre-
vious book and bear upon the common applied linguistic pur-
pose I have referred to. They are all concerned with the
formulation of a relevant model of language which will serve as
a source of reference for a principled approach to language
teaching. And they all subscribe to the assumption that if lan-
guage teaching is to develop the ability to use language as a
resource for communication in the natural manner, then what is
required is a model of language use rather than one based on lin-
guistic analysis of the kind that has conventionally informed the
practice of teaching.
The formulation of such a model has to be expressed in terms
which allow it to be assessed as description, but it also has to be
shown to be relevant to pedagogy. Hence the varying emphasis
2 Introduction
The papers in this first section provide the conceptual setting for
those which follow. They are all concerned with the principles
which relate theory to practice and which, in my view, define
applied linguistics as an area of enquiry bearing on the tech-
niques of language teaching.
Paper 1 is a variation of a paper published in the first volume
of Explorations in Applied Linguistics under the title ‘The par-
tiality and relevance of linguistic descriptions’. Indeed, certain
passages from the earlier paper are repeated here, placed now in
a different pattern of argument but contributing to the same
conclusion: that in applied linguistics our task is to look for
models of language description which relate to the experience of
the learner as user. This paper, then, indicates quite explicitly a
continuity of thinking from the previous book, and sets the
scene for this present one.
Paper 2 takes up the same theme but associates it with more
general epistemological issues. It seeks to show the culturally
relative nature of all systematic enquiry and the implications of
this for the status of statements which claim to give a scientific
account of human behaviour. This paper bears on the whole
question of whether we can, or should, expect any such state-
ments to provide explanations rather than partial representa-
tions of reality which can facilitate our understanding (cf.
Ochsner 1979).
Both of these papers talk about the difference between ana-
lyst-oriented models of language system and participant-
oriented models of language use, claiming primary relevance for
the latter with regard to the practices of language teaching and
learning. Both are somewhat circumspect about theory. The
third paper in this section, on the other hand, proclaims its
importance.
6 Theory and practice
The term applied linguistics suggests that its concern is with the
use of findings from theoretical studies of language for the solu-
tion of problems of one sort or another arising in a different
domain. The close association of applied linguistics with lan-
guage teaching (to the extent, in some quarters, of virtual syn-
onymy) is based on the belief that such findings must necessarily
be relevant to the practical teaching of languages. It would seem
to be self-evident that since linguists study language and teach-
ers teach it, there must be a relevant connection if only one
could find it. It seems perverse to question such a belief. Never-
theless, that is what I propose to do. The relevance of linguistics
cannot, I think, be taken for granted because it is not obvious
that the way linguists conceive of language is the most appro-
priate for teaching purposes. I want to suggest that the main
business of applied linguistics should be the establishing of
appropriate concepts or models of language in the pedagogic
domain without prejudging the issue by supposing that a rele-
vant model of language must inevitably derive from a formal
model of linguistic description in a technical sense.
The following is a representative statement of the belief that I
wish to question:
since they just have not been encoded as such in the language
system, and because language behaviour has to be imprecise if it
is to function effectively as communicative interaction. I shall
return to these points presently.
What I am suggesting is that the very exactitude required of
the analyst’s model makes it essentially inadequate as an ex-
planationof userbehaviour.Italsomakesempiricalvalidationvery
tricky, perhaps ultimately impossible. It is not a matter of per-
formance being an imperfect reflection of competence, of the
analyst’s model being but poorly represented in behaviour
because of a variety of rather tiresome incidental circumstances.
The point is that they are two distinct modes of reality. Compe-
tence in the Chomskyan sense is a set of abstract rules which are
manifested through analytic models of description made as
explicit as possible in the interests of scientific enquiry. Perfor-
mance is a realization of language resources for communicative
purposes. Some of these resources will, of course, correspond to
rules specified by the linguistic analyst, in which case the rules
can be said to have ‘psychological validity’. In other cases, how-
ever, the user will have concepts, will use communicative pro-
cedures which the analyst cannot capture, or the analyst will
have rules which he has to apply but which the user will need to
ignore. In brief, the analyst devises rules which represent an ideal
knowledge system whereas the user draws on the whole range of
communicative resources, in Halliday’s terms the whole mean-
ing potential, of his language and employs procedures for mak-
ing sense. These are alternative models of language: the analytic
representation of formal language on the one hand and commu-
nicative realization of natural language on the other. Obviously
they must correspond in some way; otherwise linguistics would
have no empirical content at all. But (to use Chomsky’s own
words) it is a relationship which ‘must be demonstrated, and
cannot be presumed’.1
I have claimed that there is a radical difference between the
analyst’s model of language and the user’s. It is a claim that I
must now try to substantiate. To do so I want first to return to
the conceptual definition of the noun. For analytical purposes it
is necessary, or so it would appear, to reject this definition in
favour of a formal one. A noun, the analyst tells us, is not really
the name of a person, place or thing, but a distributional class.
Yet the language user does, it seems, associate this syntactic class
with a semantic concept, so that reflection, suffering, pride, and
Applied linguistics 11
The sun is both human and inanimate and, as I point out else-
where (Widdowson 1974), it is crucial to an understanding of
the poem that this ambiguity remains unresolved.
The language user, then, is able to understand even when the
rules which purport to represent his competence are violated.
How can this be? Here was another troublesome question. A
grammar not only represented rules which would account for all
the sentences of the language but which accounted also for the
language user’s ability to understand all the sentences. Yet there
were sentences—semi-sentences—which violated the rules and
still were comprehensible. After considering this problem, Katz
comes to the conclusion:
The task a speaker performs when he understands a semi-
sentence involves, in addition to his use of grammatical
knowledge, the use of knowledge of another kind. (Katz
1964: 402)
This knowledge of another kind is what I have been referring to
as an awareness of the potential meaning resources of the
language and of the ways in which they can be realized. This
realization occurs in contexts of actual communication. It
makes no sense to ask how a language user manages to under-
stand a deviant sentence because a sentence is not a unit of use
but of analysis. It is a device for manifesting rules, not for real-
izing resources. This is why it is such a frustrating business try-
ing to get subjects in experiments to make judgements on
sentences: there will always be a natural tendency to shift from
analysis to use and to try to process the sentences as commu-
nicative expressions. This is the problem of empirical validation
which I mentioned earlier: how do you get your subjects to
assume the required analytic role, and if you do, what is the
point of your experiment, since you have lost them as normal
language users? I think it is likely that the sentence is, in the
words of T. S. Eliot:
. . . an abstraction
Remaining a perpetual possibility
Only in the world of speculation.
(from Burnt Norton)
A fair amount of poetry seems to be getting into this paper, and
it might be thought that the points I have made about user inter-
pretation and analyst description only apply to such rather
14 Theory and practice
Notes
First published in Kaplan 1980.
1 For a brief, but particularly lucid discussion of the principles
of formal analysis and the relationship between formal and
natural languages, see Lyons 1977, Appendix 1. The discus-
sion ends with the following comment:
The goal of theoretical linguistics can be described as that
of constructing a class of formal languages, all of whose
members share certain general properties and each of
whose members can be put into correspondence with
some actual or potential natural language. It is as yet
unclear whether this goal can be achieved. (Lyons 1977a:
169)