Word recognition

Word recognition is a speeded test of the ability to recognise individual real words out of context (see Section 1.1.2 for explanation of speeded tests). It depends on fluent and efficient reading accuracy. In each item the student is presented with six real words shown on a circular location space in the centre of the screen (a circular location space is used, rather than random locations across the screen, to minimise any possibility of the results being adversely affected by visual tracking, fine motor skills and/or eye-hand coordination difficulties). One of those words (the ‘target’) is spoken by the computer, and the task is to identify the target word by clicking on it using the computer mouse as quickly as possible. (It is essential that students doing this test on a laptop use a mouse rather than a touchpad, because use of the latter creates an unacceptable delay in response times.) A total of five seconds is allowed for a response, with an audible prompt being given after three seconds. The remaining five words (the ‘distractors’) were selected in order to maximise lexical and phonological similarity with the target word, e.g.:

Target word: century

Distractors: centre, sanctuary, centrally, scented, central

Each of the two forms of the word recognition test comprises a total of 60 items, with equal numbers of regular and irregular target words. Difficulty – in terms of the frequency of the target words in written English – has been balanced across the two forms. The test begins with two practice items.

Because the distribution of raw scores for this test tends to be negatively skewed – i.e. scores tend to bunch towards the upper end of the scale – the raw score has been transformed to give the normal (bell-shaped) distribution that is required for generating standard scores. This transformation takes into account the speed of response, and has minimal effect on the scores of students with a raw score below the mean, but a somewhat greater effect on scores above the mean. For students who have below average word recognition, over 70% of the variance in transformed score is predicted by raw score – i.e. for these students the standard score produced by this test is largely a function of their reading accuracy rather than speed. For students whose word recognition is highly skilled, however, differences are largely a function of speed rather than accuracy.

Results are provided for overall words (see Section 3.1.1), and for regular words and irregular words (see Section 3.1.4). In practice, students do not use a phonetic approach when recognising words at speed, and their scores are usually very similar for both phonetically regular and irregular words. The standard scores in this test correlate well with spelling ability, and a standard score below 85 is a useful pointer to literacy problems such as dyslexia.