Málfríður - 15.05.1989, Blaðsíða 6

WHAT TO TEST (= ENGLISH)
KNOWLEDGE
SKILLS (INTEGRATED
COMMUNICATION SKILLS)
FORM RULES OF DISCOURSE LISTENING/
SPEAKING/
READING/
WRITING
GRA 4AR
LEXIS
STRUCTURE
PHONOLOGY
e.g. ENABLING SKILLS/
MICRO SKILLS
LISTENING
FOR GIST
PHONEME
DISCRIMINATION , , TCTCKITK,r
LISIENING
FOR SPECIFIC
INTONATION INFORMATION
DISCRIMINATION ETC. ETC.
4.4. Sometimes it is appropriate
to test the ‘WHAT’ by means of
'pragmatic’ test-types, such as dicta-
tion or cloze (see 5 below). In such
cases the 'WHAT’ is an overall no-
tion of General Language Proficien-
cy, a crude but often useful label.
5. HOW
5.1. The ‘HOW’ is the means by
which a scorable sample of perform-
ance is provided. Some of the most
valid test types in terms of content
validity are the hardest to score re-
liably. Some of the easiest test-types
to score (e.g. true/false, multiple
choice), while having significant ad-
vantages in test situations requiring
the testing of large numbers, have
problems in terms of validity and are
beset with statistical dangers.
5.2. The list in 5.3. below repre-
sents a shopping list of the test-types
most commonly used world-wide.
They should be classified and cross-
referenced by the test-designer in
terms of a number of selection crit-
eria, though the relative weight of
the criteria will change according to
the reasons for testing, the ‘WHO’
and the ‘WHY’. Important selection
criteria would be:
(i) Scoring procedures: subjective
v. objective
(ii) Markability: are the tests easy
to mark? will the scores be meaning-
ful?
(iii) Efficiency: how many mean-
ingful test items can be covered per
hour?
(iv) Producability: are tests of this
type easy/difficult to construct?
(v) Co-operability: could a group
of colleagues use this type effective-
ly together?
5.3. The most popular test-types/
formats are:
Cloze
Completion
Composition
Correcting (e.g. errors in text)
Dictation
Essay
Gap-filling
Interrogation
Interview
Matching
Note-taking
Precis
Reading aloud
Reproduction
Resumé
Re-ordering
Report-writing
Role-play
Simulation
Summary
Transcription
Canon
Rétti tíminn til
reiknivélakaupa.
Mikiö úrval.
Hagstætt verö.
Suðurlandsbraut 12.
S: 685277 — 685275
6