Welcome to EFL ONLINE in association with
Diploma in Teaching English Online |
|
Testing Taken from the LTTC
Diploma in TESOL Module Three Before considering what to test, or how, we should be sure
why we are testing. These are the
usual reasons:
to find out if students have learned what we have tried to teach
(a test of learning or progress)
to find out what students do or do not know, or can or cannot do
(diagnostic testing)
to find out if students have achieved a certain level of knowledge or
ability
(a test of attainment) In the first, the content of the test will be dictated by the
work that has been covered. The
second type of test will dictate some of the work to be done in the future.
The third type often dictates the syllabus and therefore the work done up
to the time of the test. We can perhaps distinguish between examinations, tests and
assessment in this way.
Assessment: The
ongoing evaluation by the teacher to check whether the
students
have learned, or whether they know something.
Test:
A more formal evaluation by the teacher at particular times,
for any purpose.
Examination: A
formal evaluation of attainment monitored by someone
other than the teacher - the English department, the school
or college, or by a nationwide or international board. Of our three reasons for testing, the first may be largely by
assessment, but also by more formal testing; the second by testing, but
sometimes by informal assessment or formal examination; the third normally by
examination. Assessment may very
often be oral, by question and answer (see 2.4).
Examinations are often largely or entirely written. What we test will depend on our reason for testing.
A test of attainment may wish to test all aspects of language - the four
skills, pronunciation, spelling, grammar, vocabulary, appropriateness, style -
or may concentrate on one or more. A
test of learning progress may be quite restricted in its scope.
A diagnostic test usually picks out known or suspected problem areas and
tests those very specifically. Consideration of what we test will affect not just the
setting of the items, but also the marking.
In a test of writing skill, how much weight should be given to
correctness of spelling? In a
written test of grammar, should we take note of spelling at all?
In an answer to a comprehension question do we penalize inaccurate
grammar? A student may understand a
reading passage and all the questions. Is
it just for a student to get zero out of ten because of an inability to form the
past tense correctly? As a marker
of tests, the decision is yours, but be sure to understand exactly what you are
testing. The specific items in a test will depend on the work covered,
or on what we expect the students to know and be able to do.
Some examples follow, but first we shall look at different types of test,
that is, how to test. Dictation (see 2.6.3) and cloze tests (p. 44) are global
in that they test a number of skills and abilities. Picture compositions, skeleton stories and essays are often
used to test writing skills; talks and conversations for speaking skills.
Some of these are marked subjectively according to a global impression.
It is possible to give more specific marks for, say,
use of correct prepositions
word order
spelling
correct tense formation
clause use
vocabulary
etc. Cloze tests however can be marked completely objectively (see
below). Dictation is entirely
objective in that the only correct response is the passage as read; the
subjective element is in the marking. Specific items ate better tested by discrete item
tests. It should be kept in mind
though that this is precisely what they test; they should not be interpreted as
an indication of general language ability, or communicative competence.
They are best restricted to a diagnostic role.
Discrete item tests are often of the gap filling or multiple choice type,
since the sentence gives them some minimal context.
Miriam _______ too much ice cream.
a) eat
b) ate
c) eaten
d) eating
It was very _______ of Henry to leave the gate open so that the cows got
out.
a) disresponsible b)
unresponsible
c) non-responsible d)
irresponsible Though these are just as easy to mark:
Give the past tense of eat.
What is the opposite of responsible? they are completely uncontextualized.
Better tests relate a set of questions to a single situation - a picture,
picture sequence, or a text. Many tests are really just a series of questions, and we have
discussed various types of questioning in 2.4. Some questions (or tests) are subjective and some are objective,
but this distinction does not apply to the setting of tests - all are subjective
in the setting. The test is
objective if all testers or markers agree not only on the answer, but on the
marking. It is often thought that multiple choice questions are
objective and WH questions are open ended and therefore subjective, since the
teacher has to evaluate the answer. This
is not strictly so.
Who was Hansel's sister? Probably all markers would agree that Gretel was the
right answer (assuming the question refers to the fairy tale), and the marking
could be standardized. It is
possible some teachers might insist on a full sentence for an answer; some might
disagree on how to mark Gretle or gereatle.
Giving a number of choices would certainly overcome this problem. Some multiple choice questions, especially of the 'Choose
the best answer' type, can be rather subjective. So can those where the setter's assumptions are open to
question.
We _______ return at 9.30.
a) shall
b) will
c) do
d) are Here the setter presumably disallows b), but many teachers
marking might allow it. A clear example of different assumptions is the 'odd man out'
question (see 3.4 for the use of this in teaching - valid because the
assumptions are always called for).
Give the odd one out among the following:
a) a giraffe
b) a zebra
c) a dog
d) a cat
e) a newspaper The answer is obviously a) since a giraffe is the only one
which cannot be black and white! We will mention three principles which apply to discrete item
tests, and in a general way to all tests. 1
The verbal context should not be more difficult than the item tested.
Miriam _______ so much ice cream that she complained of stomach
ache the following morning.
The sentence here (so .... that .... ) is rather complex for
testing a simple past tense form and
may confuse students who may in fact know the past tense of
eat. It depends of
course on what is considered difficult. It
may be worth testing advanced students in their use of articles, even though
they would have
been introduced to them at a very elementary stage. 2
Do not mix up what you are testing (see 2.4.2 on Multiple Choice); test
form or usage, not a bit of each.
This is imperative if the test is diagnostic.
It was very ______ of Henry to leave the gate open so that the cows
got out.
a) irresponsible
b) irrepressible
c) irresistable
d) unreliable
[usage]
a) irresponsible
b) unresponsible
c) irresistable
d) unresistable
[mixing form and usage]
Henry left the gate open and the cows got out.
Henry was a) thinking
b) unthinking
c) thoughtful
d) thoughtless
Though most might agree that d) is the best answer, a case could be made
for b). Moreover the student might
be justified in answering a); Henry was lost in thought and that is why he left
the gate open. We said that discrete item tests are not reliable indicators
of the ability to communicate. There
is now a move in the field of testing, as there has been for some time in
teaching, towards appropriateness and communicative use of English.
This may be why global tests such as dictation have come back into favour.
Communicative tests tend to revolve around the situational factors
mentioned in 1.3. Often more than
one answer is allowed, and they can be somewhat objective.
Where might you hear the following?
Er, excuse me, is this seat taken?
a) in a restaurant b)
in a car c)
in a train
d) in a police station e)
in a furniture shop
[setting]
Who do you think might be speaking to Charles?
Late again, Charles? I'm
afraid this just won't do, you know.
a) his daughter
b) his boss
c) a shopkeeper
d) a priest
e) a bus conductor
[role/status]
Which of the following refer to a court case?
a) Brown was sentenced to
fifteen years.
b) Brown was away for
fifteen years.
c) This sentence is rather
hard to read.
d) The jury were out for ten
hours.
e) Don't judge a situation
before you know all the facts.
[topic] There are also activity, or performance tests, where students
are given a meaningful task to perform. It may be as one participant in a role play, or the student
may be given a problem to solve. These
are difficult to create and difficult to administer to a large group of
students. They are subjective but
do give a real indication of the ability of the students to use language
effectively. Unseen full dictations are extremely difficult for learners
up to intermediate level, but there are variations which can be more suited to
earlier learners. The students can
write down only the first sentence in each paragraph, or a sentence you
indicate; or the first word in each sentence.
You can ask them to prepare the passage beforehand.
You can give them a skeleton, either as a photocopy or on the board,
which they have to flesh in. You
can give them the entire passage with multiple choices at various points:
The three bears / pairs /
hares knocked
on / in / of the door of the
house.
They listened carefully but they couldn't hear
a / an / any reply. . . . . Finally, cloze tests are a particular form of
gap-filling test which has become popular because it is extremely simple to set,
and in its objective form, easy to mark. In
a cloze test every nth word of a reading passage is omitted.
In a difficult test it might be every fourth word; in an easier test,
every tenth word. But whichever
number is chosen for n, it must be strictly adhered to - you cannot change from
sixth to seventh within one text just because at one point the seventh word
seems to test something more relevant. However,
there are gap-filling tests with gaps at varying intervals, which go by the name
of cloze test. There are two views about marking cloze tests; it can be done
objectively or subjectively. The
objective view is to say that only the original word can be accepted as correct.
Though easy to mark, it may seem a little unfair.
Those who prefer subjective marking accept synonyms or indeed any word
that fits the structure and makes sense. Try
this:
Just as Heracles was ___1___ out in the morning, ___2___ heard a cow
lowing ___3___ in the hillside: for ___4___ had very foolishly taken
___5___ of the
cows who ___6___ a calf, and left ___7___ with the herd.
___8___ at once counted the ___9___, found that some were ___10___, and
set out to ___11___ with the thief.
Before ___12___ he found the door
___13___ the cavern, but Cacus, ___14___ that he was discovered, ___15___
the chains and ___16___ which should have raised ___17___, and even
Heracles could ___18___ lift the enormous block ___19___ stone and slide
it ___20___
its grooves. Did you decide 1 was starting, setting, or perhaps stepping?
Cows or cattle for 9? Proponents
of objective marking say that statistically it does not matter.
Scores will be lower but the rank order will be much the same.
Opponents maintain that is all right if all you want is a rough grading.
If the test is to be used diagnostically we should allow synonyms or any
word that makes sense. It then
becomes much more time consuming to mark. Advocates of cloze testing claim many of the advantages
attributed to dictation: it tests vocabulary, collocation, command of structure,
spelling, and the ability to grasp the total drift as well as specific meaning.
It is, incidentally, a useful exercise in preparing students for
listening to lectures - a speaker's words are often lost when someone coughs or
a car hoots outside. Though the
medium is different, it is an instance where it might be wise to practise the
reading before the listening.
[The answers to the cloze passage are: 1 setting, 2 he, 3 somewhere, 4
Cacus, 5 one, 6 had, 7 the, 8 Heracles, 9
cattle, 10 missing, 11 deal, 12 long, 13 of, 14 realizing, 15 hastily, 16
balances (you were brilliant if you got that one), 17 it, 18 not, 19 of, 20 up.
Notice that one has to read with total understanding to get 4 as Cacus.
Applied correctly, a cloze passage should have a few sentences at the start without gaps.] Copyright
LTTC 1984
|