How does Lucid Exact measure reading comprehension speed?

Lucid Research Ltd
3 Spencer Street
BEVERLEY
East Yorkshire
HU17 9EG
United Kingdom
T: +44 (0)1482 882121
F: +44 (0)1482 882911
[email protected]
How does Lucid Exact measure reading comprehension speed?
Reading comprehension is a complex skill that depends on many cognitive sub-skills, including word recognition,
vocabulary knowledge, inferential thinking and working memory, to name but a few. During the process of reading
the nature of the task and the complexity of the material to be read interact with limitations imposed by the varying
degrees of competence with which the component sub-skills can be executed by the reader. The outcome affects
both the accuracy and speed of how we can read with understanding.
Each person does not necessarily read at the same speed all the time: speed depends on a number of factors,
including the difficulty of text content for that individual, complexity of the grammar, familiarity of the vocabulary,
physical/environmental conditions such a lighting, font size and distractions, as well as personal factors such as
purpose, motivation, interest and tiredness. One may liken it to driving a car: the speed at which we drive is
dependent on factors such as road and weather conditions, amount of traffic, speed restrictions, purpose of the
journey and whether we are late. These factors can vary continuously throughout a journey, allowing us to speed up
or slow down as required, just as we do when reading.
Given these complex factors it can be appreciated that assessing reading comprehension ability is far from
straightforward. In order to explain how Lucid Exact measures reading comprehension speed, it is first worthwhile
considering the disadvantages of other methods (see Table 1).
Table 1. Disadvantages of various method of measuring reading speed.
Method
1. Measure the time taken to read a
list of words aloud and convert into
words per minute (wpm).
2. Measure the time taken to read a
passage of connected text aloud
and convert to wpm.
3. Measure the time taken to read a
passage of connected text silently
and convert to wpm.
4. Measure the time taken to read a
passage of connected text silently
and convert to wpm, and then ask
them to answer comprehension
questions.
FAQ_le15
Problems / Disadvantages
Not a true measure of reading speed. What is actually being
measured is a combination of decoding speed and
pronunciation speed.
What if the person does not recognise a word or cannot
pronounce a word correctly?
Same as above.
Not the normal way in which people read text.
Cannot be sure the person has read every word and not
skipped over some.
Cannot be certain how much of the text has been
understood.
Difficulty of factoring comprehension accuracy into the
calculation in such a way that the measure of reading speed
does not become actually a measure of reading
comprehension accuracy.
What happens if the person gets a lot of the comprehension
questions wrong?
Nov 2012
It can be appreciated that there are many factors which make the task of measurement of reading comprehension
speed tricky. When reading aloud (which in itself is problematic because it is not the usual way in which people read)
pronunciation speed becomes a major influence; not only can some people speak faster than others, but longer
words require more time to pronounce, so the person’s reading speed will appear slower when tackling longer
words. When a person is reading silently, however, we cannot be sure they are reading every word and not skipping
some, and unless we check we do not know whether the person has understood what they have read. With any
reasonably complex texts and over a certain basic speed, there is a broad trade-off between reading speed and
comprehension: the faster you read the less you are likely to understand.
To illustrate the consequences of this fact for the process of assessment, say a person reads a passage of 240 words
silently in two minutes and then answers 10 questions about it, and they get six questions correct. How shall we
most accurately calculate their reading speed? We could simply divide 240 by 2 to get 120 wpm, but since they only
got six questions right how we cannot be sure whether they actually read the whole passage properly.
Unfortunately, if we try to deal with that by taking 60% of 120, i.e. 84 wpm as their ‘true’ rate of reading on the basis
that they only understood 60% of the passage, we are left with a measure that is much a function of reading
comprehension accuracy as it is of reading speed. Conversely, our ‘direct’ measure of reading comprehension
accuracy (i.e. 60%) is also as much a function of reading speed because we do not know whether the person would
have obtained a better score if they had taken more time to read the passage. Although measures of reading
comprehension accuracy and reading comprehension speed can never be completely independent of one another, it
would be downright misleading to have two measures (accuracy and speed) that are purported to be different and
distinct but which actually turn out to be measuring more-or-less the same thing.
Most conventional tests of reading comprehension have a fixed time limit. Timed tests of reading comprehension
have been shown to have particular value over untimed tests when assessing student and adults, because untimed
tests fail to distinguish adequately between better readers and poorer readers, whereas the former do not have this
limitation (Lesaux, Pearson & Siegel, 2006). However, these tests inevitably yield a score that is a product of both
reading comprehension accuracy and reading comprehension speed, and problems of interpretation of results can
arise. For example, if person A scores, 60% having taken all the time available on a test of reading comprehension,
and person B also scores 60% having taken only 80% of the time available on the same test, it suggests that B is a
faster reader than A, but that will not be apparent from their score, which assumes that everyone uses the same
amount of time.
How Lucid Exact overcomes these various limitations
Lucid Exact overcomes the limitations of other methods by using a complex algorithm that factors into the
calculations not only time taken but also the number of questions attempted and the proportion of those questions
correct, which are proxy measures of the amount of the text that has been processed. The algorithm ensures a
reasonable degree of separation between the two key measures of reading comprehension accuracy and reading
comprehension speed. The success of this may be judged from the results of the validation of Exact, in which, among
other things, Exact was compared with the Edinburgh Reading Test (a timed test that is widely used for exam access
assessments). The findings were that Exact Reading Comprehension Accuracy score and Exact Reading
Comprehension Speed score both correlated very highly with Edinburgh Reading Test score (showing that Exact is as
good as the Edinburgh Reading Test for assessing reading comprehension) but also that the correlations with
Edinburgh Reading Test were both higher than the correlation between the two scores derived from Exact Reading
Comprehension. This shows that Lucid Exact achieves a satisfactory separation between the two key measures of
reading comprehension accuracy and reading comprehension speed
Reference
Lesaux, N.K., Pearson, M.F. & Siegel, L.S. (2006) The effects of timed and untimed testing conditions on the reading
comprehension performance of adults with reading disabilities. Reading and Writing, 19, 21-48.
FAQ_le15
Nov 2012