Cloze tests
Cloze tests are texts of a few sentences in length from which words have been omitted at regular intervals (eg every 5th, 8th or 10th word). Originally used to assess first language reading, they are also considered to be an effective way to assess overall proficiency in the written language of FL learners.
Typically a text will begin with a complete sentence before the blanks start to appear in the second sentence. For example:
Activity 24
Fill in the gaps in the following text. Are all the answers obvious? Are
there any you might have problems drawing up a mark scheme for?
| Many able students are not very proficient at L1>FL translation. They discover that, although they have _____ understanding of the target language text, _____ simply cannot find the appropriate English _____ which to express the foreign language. _____ reasons for this can vary greatly, _____ it may simply be due to _____ knowledge of L1. It seems to _____ particularly true of able learners who _____ learnt the FL naturalistically, with minimal _____ to their mother tongue. In such _____, the FL is processed in ways _____ akin to the processing of the mother tongue. |
A full version with feedback can be found by clicking 'Commentary'.
Marking of cloze tests can be based on the principle of there only being one acceptable answer, ie the original deleted word, or else any alternative might be accepted that is grammatically correct, fits the context and does not render the text incoherent. The first approach is the quicker and more reliable method: the marker has a simple right or wrong decision to make. The second slows marking down considerably but rewards candidates' creativity, and if more than one marker is involved, it may detract from the test's reliability since one person may deem something acceptable that another would not.
A critical issue in writing cloze tests is to get the level of the text right: it must not be too difficult, as blanked words will make it even more inaccessible. Factors to consider include:
|
The blanks in the text in Activity 24 were chosen on the basis of 'fixed ratio deletion', ie every 7th word. The alternative approach is 'rational deletion' in which something specific, such as past-tense verbs or conjunctions, is deleted so as to test students on just that item. Generally speaking, fixed ratio deletion makes for a harder test.
Reservations concerning cloze tests include doubts about their reliability and validity. Fixed ratio deletion frequently results in isolated items that are impossible for the candidate to predict. Furthermore, it is not a consistent procedure. As Weir states: 'If one changes the text, changes the deletion rate, starts at a different place or alters the scoring procedure, one gets a different test in terms of reliability and validity coefficients and overall test difficulty' (1990: 47). This points to the need for rational deletion, where deleted items are consciously selected and maybe trialled to assess their difficulty. However, even this does not address the major criticism of cloze tests, namely that they are based solely on individual words, and that gaps can be filled by reference to the immediate context; consequently, reading skills above the level of the word, phrase or sentence are usually not assessed.
 |
|
|
 |
 |
|
|
|
|
 |
 |
|
