Málfríður - 15.05.1989, Page 4
Dave Allan:
Should English language tests
be more ‘communicative’?
Dave Allan studied languages and
linguistics at the Universities of
Cambridge and York, before work-
ing first as a state school teacher of
French and German, then of Eng-
lish. He is now Principal and teacher
trainer at the Bell School of Lan-
guages, Bowthorpe Hall in Norwich.
His other professional positions in-
clude: Chief Examiner for the Cam-
bridge/RSA Tests in the Communi-
cative Use of English; test writer and
author for O.U.P. (the Oxford Place-
ment Tests); member of the British
Council Recognition Advisory Com-
mittee; specialist lecturer overseas
for the British Council; lecturer/tu-
tor at the University of East Anglia.
Dave Allan has travelled widely as
a teacher and teacher trainer and has
worked with Icelandic teachers both
on the ‘Norwich’ course in England
and, in 1988, in Iceland.
In an ideal world, the processes of
learning, teaching and testing a lan-
guage should be interdependent and
mutually supportive. Our evalua-
tion procedures, and the tests and
examinations which are part of
them, should reflect our current be-
liefs as to how people learn a lan-
guage most effectively and what it is
to be ‘competent’, ‘fluent’, ‘profi-
cient’ or ‘near-native speaker’ in a
foreign language. To put it another
way, if we are concerned with the
teaching of English as a communi-
cative medium, and not just as an-
other classroom discipline, then we
must ensure that we test the lear-
ner’s communicative performance;
the testing must reflect the teaching
given or foreshadow the teaching in-
tended; the tests must be tests not
just of knowledge (as one can have
knowledge of ancient, dead lan-
guages) but of knowledge and skills;
communicative syllabus design and
communicative methodology must
be matched by what has been called
‘communicative testing’.
The problem about the idealized
model, though, is that while almost
all languagé use in the real world has
a communicative purpose, language
tests are usually intended for other
purposes and are designed to be pre-
dictive or representative or diagnos-
tic or discriminatory, or perhaps all
of these. In short, we often test for
reasons very different from those
which motivate us to learn and use a
language. These reasons for testing
are many and diverse, some of them
having very strong implications for
the test content and test types that
will be appropriate. Whatever the
‘approach’ to the teaching (e.g.
‘grammar-translation’ v. ‘audio-lin-
gual’ v. ‘communicative’) there is a
world of difference between the test
design and procedures that are ap-
propriate for a placement test to be
taken by thousands and those that
would be suitable for an end-of-unit,
in-class test for a group of 20-30. The
design, administering, scoring and
evaluating of language tests will al-
ways have to be a compromise be-
tween the ideals of test construction
and the constraints imposed by the
specific context or type of context
for which the test is being produced.
Factors such as testee availability,
test economy, administrability and
markability all have to be taken into
account, though they should never
be allowed to outweigh validity and
reliability in tests that have any real
significance for the learner’s future.
A teacher giving frequent ‘snapshot’
class tests and maintaining a well-
informed professional overview of
progress can afford not to worry too
much about formal procedures to
ensure the validity and reliability of
every single test, but anybody con-
cerned with the design of language
tests should have a clear under-
standing of why validity and reliabil-
ity matter.
There are many different kinds of
validity, including ‘predictive’, ‘con-
tent’, ‘construct’, ‘concurrent’ and
‘face’ validity. These are discussed
in detail in the literature on testing
(see selective bibliography below),
but two of them have particular sig-
nificance for the design of tests ap-
propriate to ‘The Communicative
Approach’, viz:
(i) Construct validity: a test is val-
id only if based on a valid construct
of language learning. Thus tests of
translation were/are appropriate in
the context of a grammar-transla-
tion approach to language teaching.
Equally, if you believe that a lan-
guage should be taught as a means of
communication, your testing as well
as your teaching should reflect that
view.
(ii) Content validity: The content
of a test in terms of the balance of
focus between knowledge and skills,
and of the weighting of different
items in the test, should reflect the
teacher’s/tester’s view of the relative
importance of the different aspects
of language taught or to be taught.
Any test of ‘language as communi-
cataion’ should thus give significant
focus to communication skills. To
give an extreme example, an Eng-
lish test without an oral component
could not claim to be directly repre-
sentative of a learner’s overall profi-
ciency in English, though many tests
make use of statistical correlation to
test a wider range of skills indirectly
4