This is the first in a series of posts describing specific terminology in ELT. Today we are focusing on assessment.
Backwash (Also known as washback)
Backwash is the effect that testing has on the teaching and learning that precedes the test. So if a test is heavily weighted on speaking, for example, then the probable backwash will be that you will teach a lot of speaking leading up to the test.
Discrete-point test and Integrative test
Discrete point tests are designed to test only one specific point. This type of test has been associated with multiple choice tests in which selection of a choice of grammar answers is available.
The opposite of the discrete-point test, the integrative tests lots of elements of language at the same time. An example of this maybe an interview, or
A placement test is generally done before the commencement of a language course. It is often done to ascertain a students ability, to place them into an appropriate class.
Validity is a particular criteria in which a test is judged and is categorized as follows.
Content validity is concerned with whether the correct things are being tested. For example, a grammar test needs to have examples to test control of grammatical features.
As the name suggests the construct validity concerns itself with how a question is put together, but its main focus is on how that construct tests what you want to test. For example, if we were to test a certain grammar point, we could test it using several different methods, and we should get similar results. This should be a fair certainty, unless a particular type of question was difficult to construct to test that certain grammar point. This example would lack construct validity.
Face validity concerns itself with how it is received. A test needs to have face value, if a candidate is to take it seriously.
There are two types of criterion related validity: concurrent and predictive. Concurrent is basically when you compare the results of the test with examples of an independent criteria. In theory, the students test scores should tally with that criteria.
Predictive is when you use the test to examine and predict future performance. An example of this would be a placement test, you are testing to see how well a student will perform in a particular class.
The reliability of a test is concerned with the quality of the scores we get. How much can we rely on the score?
This is the measure of how easy the test is to administer, how much it costs to produce and mark, and much time it takes to do also. Hence, why multiple choice tests are very popular – they are cheap to make, and can often be marked by computer. Its a fine balance between practicality, and reliability and validity.
A test designed with every nth word missing from a piece of text.
Criterion-referenced testing and Norm reference testing
The different referencing concerns itself with what we do with the test scores. Norm referencing is when we use the students score to compare them with a range of other students doing the test. I.e. The student is in the top 5% of their class.
Criterion-referencing is when we use the scores to compare them with criteria. We scale them according to what the student can do. I.e. Student shows a wide range of appropriate lexis.
Direct testing and Indirect testing
If we wanted to test a students letter writing ability for example, we could get the to write a letter. This an example of direct testing.
If we were to ask the students to analyze and discuss an example of a letter, it would be an example of indirect testing. Their scores would indicate how well they understood the features of letter writing.
Arthur Hughes (1989). Testing for Language Teachers. Cambridge: CUP
Scott Thornbury (2006). An A-Z of ELT. Oxford: Macmillan Education