Sequencing as an item type

Abstract

A text’s coherence clearly depends upon the way ideas are related within that text, both in terms of their logical relations, as well as the cohesive devices that show, or create, the links between ideas, across paragraphs and sentences. Thus, it would appear that part of the ability of a competent reader is to recognize the appropriate order of ideas in text, to identify cohesion and coherence in text in order to relate the ideas to each other, and to understand authorial intention with respect to the sequence of ideas. It follows from this that a potentially useful test method that might tap such abilities is to require candidates to inspect text in which the elements are out of order, and to reconstruct the original order. This, it might be supposed, would require candidates to detect the relationship among ideas, to identify cohesive devices and their interrelationships.

Such test methods are, indeed, increasingly common in so-called reading tests. However, we know of no reports of research into, or even descriptions of the use of, this promising task type. In this article we report on potential problems in scoring responses to sequencing tests, the development of a computer program to overcome these difficulties, and an exploration of the value of various scoring procedures.