Knowledge of Language and Ability for Use
H. G. WIDDOWSON
University of London
This paper seeks to clarify the notion of competence in language and to draw on
its relevance to language teaching practices. It is suggested that Hymes's
account of communicative competence as incorporating language beyond
grammar and ability as well as knowledge raises problematic issues concerning
the analysability and accessibility of knowledge and the scope and application
of linguistic rules. A consideration of these issues suggests the possibility that
competence for use may involve not so much the generation of expressions by
direct reference to rules as the adjustment of pre-assembled and memorized
patterns. The ability to use language, therefore, may have to do with access
which is relatively independent of the analytic knowledge of grammar as
defined in Chomsky's original concept of competence.
The influence of ideas, whether for good or ill, does not depend upon their being
fully understood in their own terms. Usually, indeed, it depends upon them
being recast in different terms to suit other conditions of relevance. This should
not be a matter of surprise or regret. The more influential an idea, the less
dependent it is, obviously enough, on the particular context of its conception.
All interpretation, after all, is a matter of reformulating ideas so that they key in
with one's own frame of reference. And so it is with the idea of communicative
competence. It has been adapted, interpreted, and exploited, keyed in with the
concerns of applied linguistics and language teaching pedagogy. It has inspired a
good deal of innovative activity in the field. Some of this comes from partial
understanding of the concept, from a distortion indeed of the original. This does
not matter. Many a new dish has emerged from a mistake with a recipe. We do
not condemn this; we call it being creative.
But as with other influential ideas, it is sometimes enlightening to renew
acquaintance with the primary textual source and scrutinize it for meanings
which we might have missed or misinterpreted and which we might find of
relevance to our current concerns. I believe that there are matters arising from
Hymes's original formulation of the notion of communicative competence that
have a bearing on a number of issues which have emerged over recent years in
research in second language acquisition and in the design of language
programmes. It is the purpose of this paper to explore these matters.
As I was in the process of composing this paper, I discovered, rather to my
chagrin, that its purpose had in part been anticipated by an article by David
Taylor in the most recent issue of Applied Linguistics (Taylor 1988) which
examines the concept of competence with impressive perception. Other
Applied Linguistics, Vol. 10, No. 2 © Oxford University Press 1989
H. G. WIDDOWSON 129
scholars too have had things to say on the issues I raise here, as will be clear from
my references. So this paper will be full of other men's flo\.~ s (and other
women's too, I should add). 'Only the string that binds them is my own.'
We can begin with the authorized version that every student is schooled in.
Hymes proposed his concept of communicative competence in reaction to
Chomsky, and it is customary to present it as an improvement in that it covers
aspects of language other than the narrowly grammatical. It accounts for the fact
that knowing how to use a language involves more than knowing how to
compose correct sentences. 'There are rules of use... ' etc.: a familiar quotation.
But it needs to be recognized that the two concepts are (as Feyerabend would
note) not really commensurate. Chomsky and Hymes are playing in different
kinds of game. It is true that there are rules of use, etc., but then Chomsky does
not deny it. It is simply that he is not interested in language use. Indeed he is not
really interested in language as such at all. He is interested in grammar. As far as
he is concerned, as he himself says 'language is a derivative and perhaps not a
very interesting concept' (Chomsky 1980: 90). Other linguists, inspired by
Chomsky's example, have, of course, taken the same line. In his inaugural
lecture at University College London, for example, Neil Smith tells us:
'Linguistics is not about language, or languages, at least that is not its main focus,
it is about grammars' (Smith 1983: 4).
The point then is that for Hymes linguistics is about language and for
Chomsky it is not. So Chomsky's notion of competence has nothing whatever to
do with the actualization of language behaviour, communicative or otherwise.
So really it does not fit into Hymes's scheme at all. For Chomsky, competence is
the knowledge of something much more abstract than language: it is a
knowledge of systems of rules, of parameters or principles, configurations in the
mind for which language simply serves as evidence. How these mental
abstractions are realized in actual instances of language use, or how they are
related to other aspects of language or to other kinds of abstract knowledge are
matters entirely beyond the scope of his enquiry. He invokes the notion of
modularity, and leaves it at that.
For Chomsky, then, competence is grammatical knowledge as a deep-seated
mental state below the level of language. It is not an ability to do anything. It is
not even the ability to compose or comprehend sentences, for knowledge may
exist without its being accessible and, as Chomsky insists, actual behaviour is
only one kind of evidence, and not a criterion for the existence of knowledge
(Chomsky 1980: 54). For Hymes, on the other hand, competence is the ability
to do something: to use language. For him, grammatical knowledge is a resource,
not an abstract cognitive configuration existing in its own right as a mental
structure. How such knowledge gets realized as use is therefore a central issue,
and it is necessarily a component of communicative competence.
Hymes's communicative competence, then, is defined as 'the capabilities of a
person' and, as he says, 'it is dependent upon both [tacit] knowledge and [ability
for] use' (Hymes 1972: 282). What Chomsky takes such care to eliminate from
his definition, Hymes reinstates. Having identified his four parameters of
130 KNOWLEDGE OF LANGUAGE
communicative competence (possibility, feasibility, appropriateness, and
attestedness in actual performance), Hymes observes:
Knowledge also is to be understood as subtending all four parameters of communica-
tion just noted. There is knowledge of each. Ability for use also may relate to all four
parameters. Certainly, it may be the case that individuals differ with regard to ability to
use knowledge of each: to interpret, differentiate, etc. The specification of ability for
use as part of competence allows for the role of non-cognitive factors, such as
motivations, as partly determining competence. (Hymes 1972: 282-3)
There are, then, two aspects to competence in respect to each of the four
parameters: knowledge on the one hand, and ability on the other. Hymes thus
extends the Chomsky concept of competence in two ways. He includes
knowledge of aspects of language other than grammar—of what is feasible,
appropriate, actually performed. And he includes ability for use. There are
therefore eight elements to the model where Chomsky only has two. So, in
principle, we ought to be able to say what constitutes a knowledge of feasibility,
for example, as distinct from an ability to be feasible; what constitutes
knowledge of appropriateness conventions, as distinct from an ability to be
appropriate, and so on. I have no idea how one would set about doing this, and
Hymes gives us little indication. The problem, as always with models which
simply provide a list of features (if indeed these can be called models at all), is
that it is difficult to know what the relations between the features might be. I will
not attempt a systematic exploration of all the relational possibilities in this
paper but shall concentrate on one or two, and in particular on the relationship
between knowledge and ability. This alone will give us plenty to think about. For
the distinction, and the problem of the relationship between the two concepts,
appear in various guises in work in applied linguistics and language pedagogy
over recent years.
Consider first the question: is it possible for somebody to have knowledge in
respect of what Hymes calls possibility, that is to say of the formal properties of
a language, and fail to have ability in respect of appropriateness? We might
expect a resounding 'yes' to come from the proponents of the communicative
approach to language teaching. For it would seem, on the face of it, that it was
because so many learners seemed to be in precisely this state as a result of
structure-based teaching that the communicative approach was proposed in the
first place: learners full of knowledge that they apparently did not have the
ability to put to effective, that is to say, appropriate use? And it would appear
that Chomsky too envisages just such a state of affairs, or (to be more exact),
state of mind: 'I assume' [he says] 'that it is possible in principle for a person to
have full grammatical competence and no pragmatic competence, hence no
ability to use a language appropriately, though its syntax and semantics are
intact' (Chomsky 1980: 59). But there is a confusion here. We should notice, to
begin with, that Chomsky is inconsistent in his use of terms. For by grammatical
competence he means a kind of knowledge, but it is evident that by pragmatic
competence he means a kind of ability. He does not, as Hymes does, conceive of
a knowledge of appropriateness, nor indeed for that matter of a grammatical
H.G.WTODOWSON 131
ability. So we might now raise the question: is it possible, in principle, to have
grammatical ability without pragmatic ability; that is to say, can we conceive of
people who can, in Hymes's terms, process language along the possibility
parameter without regard to appropriate use? The answer here would again
seem to be 'yes', and now we return to the supposed shortcomings of the pre-
communicarive paradigm in language teaching. For on reflection it is really this
state of affairs which the communicative approach sought to rectify. It was not
that learners who suffered under the structuralist regime were full of knowledge
which had no behavioural output. There was plenty of output, but it was not of a
pragmatically approved kind. The classrooms were full of activity for the
achievement of grammatical ability. The trouble was that it was often carried out
in dissociation from any consideration of appropriateness. In other words, to
use my own terminology, what was promoted was the ability to manifest
language as usage rather than to realize it as use.
What then, about the reverse situation? Can we have pragmatic ability, an
ability to use language appropriately, without grammatical knowledge or the
ability to compose or decompose sentences with reference to it? There is
evidence that excessive zeal for communicative language teaching can lead to
just such a state of affairs.
Learners can acquire a repertoire of phrases for deployment in a range of
correlating contexts but fail to analyse them fully into constituents as linguistic
structures. Grammatical competence is not necessarily inferred from use, and
so use is itself constricted. For, as Bialystok observes, 'The assumption is that if
knowledge is analysed, then certain uses can be made of that knowledge which
cannot be made of knowledge that is unanalysed' (Bialystok 1982:183). So the
learner might acquire a repertoire of phrases with contextual associations,
chunks, so to speak, which have not been analysed into grammatical knowledge.
One might indeed suggest that the very focusing on the relationship between
linguistic expressions and external context inhibits the process of establishing
internal relationships of linguistic expressions with each other.
We are, of course, talking of tendencies. Some grammar is likely to be contin-
gently analysed out of communicative activity just as there will be some
pragmatic side-effects from a concentration on analysis by focusing on form. It
is a matter of degree. Nevertheless, we seem to have something of a paradox
here. If we assume with Bialystok that the analysis of language into grammatical
elements allows for the flexible adaptability of the language to a range of
different uses, then it would be reasonable to conclude that an approach to
teaching which induces such analysis, the structural approach indeed, is more
communicative, in its potential effect at least, than an approach which deflects
attention away from analysis and which links holistic expressions with a fixed
range of contexts. In other words, if the ability to use linguistic knowledge
effectively for communication depends on the extent to which that knowledge is
analysed, then the structural approach would seem to provide the basis for the
development of communicative competence in a way which the communicative
approach, at least in some of its manifestations, does not.
132 KNOWLEDGE OF LANGUAGE
But of course, as Chomsky makes clear, knowledge, analysed or not, needs to
be accessed. It may be the case that the more analysed language is, the more
potentially generalizable it is across a range of uses, but the potential has to be
realized. Bialystok recognizes this. She refers to the factor of automaticity, 'the
relative access the learner has to the knowledge, irrespective of its degree of
analysis' (Bialystok 1982:183). In a subsequent paper, written with Sharwood-
Smith, language ability is said to involve 'knowledge systems on the one hand,
and control of these systems on the other' (Bialystok and Sharwood-Smith
1985: 106).
So, to summarize the story so far, and to get the terms straight, we seem to
have the following positions. For Chomsky, competence is knowledge. For
Hymes, competence is knowledge and ability. For Bialystok and Sharwood-
Smith ability is knowledge and control. Since it is clear that Bialystok and
Sharwood-Smith are restating the Hymes distinction in different terms, let us
regularize the terminology by saying that competence has two components:
knowledge and ability, and that these in principle relate to all four of Hymes's
parameters (possibility, feasibility, appropriateness, performance) which in turn
can be reformulated as grammatical competence (the parameter of possibility)
on the one hand, and pragmatic competence (all the other parameters) on the
other. Knowledge can be characterized in terms of degrees of analysability,
ability can be characterized in terms of degrees of accessibility. We might now
hazard the generality that the structural approach accounts for one aspect of
competence by concentrating on analysis but does so at the expense of access,
whereas the communicative approach concentrates on access to the relative
neglect of analysis. These concepts then enable us to make a general character-
ization of different pedagogic practices. So far so good. But it would be as well to
look at these concepts a little more closely.
Consider first the concept of analysability. One can discuss this in reference
to the interlanguage of the learner and talk about the extent to which the
language data have been reduced to rule as required, thinking of competence
essentially, as Chomsky and Hymes both do, as the knowledge of abstract
systems. But the question arises as to how far knowledge of language is
systematic and circumscribed by rule in that way. Grammar of course is, by
definition. But there is a great deal that the native speaker knows of his language
which takes the form less of analysed grammatical rules than adaptable lexical
chunks. Pawley and Syder refer to these as 'lexicalised sentence stems' and
suggest that 'the stock [of them] known to the ordinary mature speaker of
English amounts to hundreds of thousands'(Pawley and Syder 1983:192). This
stock is, then, a not inconsiderable component of competence. It cannot just be
dismissed as somehow incidental. Apart from anything else, there is evidence
that these holophrastic expressions figure prominently in first language
development (Peters 1983) and indeed might be regarded as the primary units
of child acquisition, with rules emerging as a function of gradual and differential
focusing, as the child detaches the language from its direct contextual connec-
tions and generalizes features of context into grammar.
H. G. W1DDOWSON 133
These lexicalized sentence stems are of course subject to differing degrees of
syntactic modification. At one end of the spectrum, we have fixed phrases that
cannot be dismantled, compound lexical items in suspended syntactic anima-
tion. Knowledge of these, like so much lexical knowledge, is a matter of memory.
At the other end of the spectrum we have collocational clusters which can be
freely adjusted as sentence constituents. So we have a scale of variability in the
legitimate application of generative rules. And competence must be a matter of
knowing how the scale is to be applied; when analysis is called for and when it is
not. Pawley and Syder, who discuss this issue in some detail, point out that
'native speakers do not exercise the creative potential of syntactic rules to
anything like their full extent, and . . . indeed if they did so they would not be
accepted as exhibiting nativelike control of the language' (Pawley and Syder
1983: 193).
But notice the control that is referred to here is not the same as that which
Bialystok and Sharwood-Smith talk about. It is not a matter of inadequate
access to knowledge in the exercise of the ability to use language, but inadequate
knowledge of the internal operation of language itself. The over-application of
grammatical rules in English, for example to yield goed instead of went, or He
explained me the problem instead of He explained the problem to me is
recognized as being in need of remedy. But just the same principle applies with
expressions like, for example, Before you leap, look, All that has ended well is
well, By hook or crook, and so on. The difference of course is that these latter
examples are grammatical, the former ones are not. But they are still not English
in respect of native speaker norms. And anybody producing these syntactic
variants of fixed idiomatic phrases would nevertheless be adjudged incompe-
tent in the language.
An ignorance of the limits of analysability, of the variable application of
grammatical rules, constitutes incompetence. One might suggest this aspect of
incompetence is covered by the Hymes parameter of performance: expressions
that come from an over-application of rules are grammatical but uncommon. 'A
sentence', says Hymes, 'may be grammatical, awkward, tactful and rare' (Hymes
1972:281-2). But what I am saying is that such expressions are not just rare, by
accident as it were, but are illegitimate, rule-violations, not possible, and the
native speaker knows this full well. Such sentences as manifestations of
knowledge, as instances of usage (to use a term of my own), are grammatical and
yet linguistically ill-formed. Linguistically ill-formed, notice. This is not a
judgement about acceptability. I am talking about linguistic well-formedness
and I am claiming that this notion applies to the variably analysed formal
properties of a language which are not just syntactic but lexical as well, and
indeterminate between the two, properties which are holistically remembered
as well as those which are analytically reduced to rule.
I have been talking about variability in rule application and of course Labov
comes to mind. He had demonstrated how internal linguistic environments of a
morphological kind condition the appearance of particular phonological
variants (Labov 1972). All I am saying is that similarly there are