Multiple Matching

Multiple matching (MM) tasks are really versatile in their format: matching pictures to the words/phrases, matching words to the definitions, matching sentence beginnings to the sentence endings, matching headings to the paragraphs, etc. etc.

The items for which test-takers need to find a match are called ‘premises’. The matching options are called ‘responses’.

MM tasks are flexible in terms of constructs that can be measured. In language testing, they are often used to test listening and reading comprehension, and this format is particularly suitable for assessing global understanding – the gist of an oral or written text. MM tasks are also widely used to test grammar (e.g., matching sentence beginnings to the endings) and vocabulary (e.g., matching words to their definitions).

MM tasks are somewhat easier to construct compared to multiple-choice tasks. However, producing a good MM task also takes a degree of item-writing skill and the knowledge of pitfalls to watch out for.

Recommendations for writing MM tasks

  • Have more alternatives than the task requires. It is usually recommended to have two distractor responses which do not match with any of the premises (for a six-item task).

There are 9 premises and 9 responses: once 8 pairs have been matched, the last one can be done by default. Moreover, it is easy to find many matching pairs through word spotting. For example, the word ‘name’ is both in the question (“What is your name?”) and only one answer (“My name is Jane”).

  • Make sure there is only one correct matching response for each premise.
  • Make sure each premise has a matching response.

On the plus side, this task has four premises but five responses, which means there is one distractor. All words also come from the same lexical set (people, professions). However, Premise 1 does not have a matching response and seems to be written in jest

  • Avoid having too many items in a MM task. More than six items might be too many. This is because, to answer the task, test-takers have to keep all the matching options in mind so the test-takers with larger working memory capacity will have an unfair advantage.
  • Place the premises and responses in two columns alongside each other – this will allow test-takers to easily see all premises and responses. The responses should be placed on the right-hand side of the premises. Avoid placing the responses under the premises. The whole task should fit one page.

This collocation matching task has a strange layout – the gap is on the wrong side of the collocation. Moreover, there are multiple correct responses, e.g., “clean my face / the house / the table / the floor” are all possible.  The task also has the same number of premises and responses.

  • All responses should be homogeneous. E.g., for the task that requires matching sentence beginnings to the sentence endings, all endings should start with the same part of speech.  For vocabulary matching tasks, the responses should come from the same lexical set.

In this task, responses seem to be a random collection of words. For example, “turtle” is the only animal, “nurse” is the only word for a person. A test-taker could match words to their definitions by having a vague idea of the semantic group the words belong to and without knowing the words’ precise meaning. Moreover, there are 7 premises and 7 responses, so the last pair can be matched by default.

  • Make sure both premises and responses are concise – they should not be so long as to depend on a test-taker’s working memory capacity for the correct answers.

Further reading

Miller, M. D., Linn, R. L., & Gronlund, N. E. (2009). Measurement and assessment in teaching (10th ed.). Pearson Education Ltd. – pp.186-191 of this book contain some useful recommendations for constructing matching tasks, as well as a checklist for the task review.