+44 (0) 7738 666 511 +44 (0) 1243 552 390 Info@superreading.com

Dr Ross Cooper, Assistant Director, LLU+, London Southbank University
(
r.cooper@LSBU.ac.uk)


We piloted Ron Cole’s SuperReading © course at LLU+, London Southbank University, with 15 adult dyslexic readers from January 2008. Despite my initial scepticism, both speed of reading and comprehension increased for all participants. In effect, reading effectiveness (including my own) doubled over the six sessions of the course (spread over 10 weeks). This was a pilot study to establish whether the claimed effect of the intervention appeared to work. The effect was dramatic and has been duplicated since with further readers. The pilot research is to be published in the peer reviewed Journal of Inclusive Learning in FE & HE (Sept 2009a). Since the pilot research, we have sought funds to finance further research involving control groups. Without this, it is not possible to establish with any degree of certainty, what is having the effect. Nevertheless, despite the initial relatively small group size, the effect was so large that the statistical significance of the effect was extremely high (p<0.001) and increasing with each further course as the sample size increases (currently less than 1 chance in a million that it could occur simply by chance).

Perhaps the most interesting feature of the effect is that those with the most difficulty reading nonwords make the most dramatic gains. This is despite the intervention involving no phonics. It involves metacognitive reading skills and an eye exercise designed to improve the visual processing of groups of words at speed. This effect cannot be explained away by ‘regression to the mean’, meaning that those with the weakest reading skills are making the most apparent gains as they move towards average reading skills, for two reasons. The first is that those with the most difficulty reading nonwords did not have the poorest reading skills in the dyslexic group. (We usually find that adult dyslexic readers who struggle with phonic attack develop very good reading for meaning strategies to compensate for difficulties with unfamiliar words allied with visual recognition of the meaning of words). Secondly, their reading skills generally developed beyond the mean, as we shall see below.

Perhaps the most obvious possible explanation for the effect is an inaccuracy in the measurement of it. However, unlike some recent research into phonics interventions with adults (Burton et al, 2009) which used the same tests before and after the intervention, we used different texts. The set of texts and questions the participant took before, during and after the intervention were randomised, so that differences in the difficulty or familiarity of a specific text or questions could not be an explanation for any differences across the group. In addition, standardised scores on TOWRE sight word tests increased by a mean of 7 standardised points (Cooper, 2009a), indicating that visual recognition of single words at speed improved (even though the intention was to improve the processing of multiple words, rather than single words).

We began to be left with very few possible explanations for the effect other than the intervention. We decided to see if the effect was duplicated when Ron Cole was not teaching the course, in case it was a consequence of his charisma. He trained a number of us to teach it, which we have begun to do. We also felt that this would increase our capacity if we could find the funds to finance larger scale research. This is where I feel that the results became even more interesting.

After the intervention, the mean reading speeds and comprehension of the dyslexic readers exceed the mean reading speeds and comprehension of the non-dyslexic readers (prior to the intervention). This is compelling evidence of a dramatic effect.


Measuring Reading Effectiveness

Two measures of both reading speed and comprehension are taken at each point in time. Participants are given a text of four hundred words to read in order to answer unseen questions. The reading speed is measured. The text is then removed and 10 questions provided. These questions are not multiple choice (which can generate false positives) and are designed to reduce the effect of prior knowledge by making sure that they focus on details in the text that are unlikely to be known prior to reading it. This means that the focus is on detail (and recall of the detail read) rather than general knowledge and inference. (This attention to detail and recall is usually particularly difficult for dyslexic readers and is likely, therefore, to underestimate the impact of the intervention on real reading where details can simply be looked up). Once completed, the questions and the first set of attempted answers are removed and the same text provided with the invitation to read it again to attempt to get 100% of the answers correct. This review reading speed is measured. Once answers to the questions are again attempted, comprehension for each set of answers is scored as a percentage.

Reading Effectiveness (RE) is measured for both the ‘first read’ and the ‘review read’ by multiplying the words per minute with the percentage of correct answers. So, for example, if the reader read at 200 wpm, and got 50% of the correct answers, the RE score is 200x50/100=100. If they read at 100wpm and got 80% of the correct answers, the RE score is 100x80/100=80. In this way, RE scores the effectiveness of reading for comprehension within a timescale. RE scores would increase if just speed or comprehension increased. However, both speed and comprehension increased for all the participants.

It was suggested by one of the participants that perhaps practicing the comprehension tests itself leads to the improvement. In the absence of a control group, we found 11 volunteers to take the comprehension tests as many times as the course participants in as close to the same circumstances as we could provide. Overall, neither their reading speeds, nor comprehension improved at all.


Making comparisons

We now have a larger body of participants and can begin to make comparisons based on estimated standard deviations and mean scores. While this is no substitute for large scale independent research including control groups, it can provide very interesting perspectives on the comparative improvement of dyslexic, compared to non-dyslexic, readers. The overall group of non-dyslexic readers is still very small for estimating standard deviation (n=27). However, most of the additional non-dyslexic participants are teachers or teacher trainers. Consequently, it is likely that their initial mean scores are high compared to a random group of readers. I therefore intend to use these for illustrative comparative purposes knowing that this likely sets the bar high, and that the standard deviation measures are estimates.

One standard deviation below the mean is currently used as the cut-off point on standardised tests below which a learner can achieve access arrangements for exams. This therefore gives us a rough guide against a cut-off point familiar to dyslexia support teachers. More importantly, it gives us a comparative scale to judge the improvement in reading skills of the dyslexic group compared to the non-dyslexic group.

For each measurement I shall provide the mean score of the dyslexic participants (n=20) before the intervention, compared with the mean score of the non-dyslexic participants (n=27) before any intervention. I shall then give the mean score of the same dyslexic participants after the intervention. In addition to this I shall provide the percentage of participants who scored above the mean as well as the percentage that scored one estimated standard deviation above and below the mean. If we were making comparisons with a large random sample, we would expect the percentage above and below the mean to be 50%, and those one standard above or below one standard deviation from the mean to be 16%. With a small group such as this, the distribution of scores is likely to vary. This therefore gives us a more realistic comparison between the dyslexic and nondyslexic samples, recognising the limitations of the data. As I am comparing the dyslexic sample against the non-dyslexic sample, the non-dyslexic mean and estimated standard deviation is taken as the comparative yardstick.


Overall comparisons

Before looking at detail at each measure of speed and comprehension, I shall present an overview of the impact on participants through looking at ‘combined RE’ scores. This is a convenient way of comparing the total time taken from both reading and reviewing for the final comprehension. It is measured by adding the time taken for both sets of reading, divided by the number of words (giving a ‘total wpm’ score) multiplied by the percentage comprehension achieved after the review. This allows us to measure the combined effect of both sets of reading. It measures total comprehension scores against total time taken, thus providing the best overall comparative measure.

 Mean

combined RE % above the mean % below 1 st. dev below mean % above 1 st. dev above mean

Pre-course dyslexics 59 15% 40% 5%

Pre-course non- dyslexics 83 52% 3% 11%

Post-course dyslexics 114 55% 5% 35%

The estimated standard deviation is 38. The improvement made by the dyslexic group is 55. Here we can see that the dyslexic group have progressed from being significantly weaker in their reading effectiveness than the non-dyslexic group with only 15% above the mean and 40% below one standard deviation below the mean, to being superior to the non-dyslexic group. Now 55% are above the mean and 35% one standard deviation above the mean.

The strategies taught on the SuperReading course of previewing, questioning text, using eye-hopping and ‘pattern reading’, and identifying what has been understood and reviewing the material all have an accumulative effect. They work together to produce an overall effect. Nevertheless, at every measure, significant improvements in speed and comprehension are made, although this is more pronounced in some measures than others.

First Reading Speed

 Mean wpm % above the mean % below 1 st. dev below mean % above 1 st. dev above mean

Pre-course dyslexics 158 20% 30% 5%

Pre-course non- dyslexics 192 33% 4% 11%

Post-course dyslexics 201 30% 15% 20%

The estimated standard deviation is 75 wpm. The mean increase for the dyslexic group is 43 wpm. Both groups show that most of the participants score below the mean, which means that a few readers are much faster than the others, although the percentage above one standard deviation above the mean remains small. Following the intervention, the dyslexic group mean score exceeds the mean reading speed of the non-dyslexic group and the percentage who are one standard deviation above the mean is also superior. Overall, the dyslexic group have progressed from being significantly below the non-dyslexic group in speed to becoming very similar (with a broader spread of results and higher mean).

First Reading comprehension

 Mean

comprehension % above the mean % below 1 st. dev below mean % above 1 st. dev above mean

Pre-course dyslexics 45% 25% 40% 0%

Pre-course non- dyslexics 60% 52% 8% 19%

Post-course dyslexics 57% 55% 30% 15%

The estimated standard deviation is 17.5% The mean improvement of the dyslexic group is 12%. Again, the dyslexic group have improved from being significantly less able to answer comprehension questions at the first read to being quite similar to the non-dyslexic group. Although the mean percentage comprehension is slightly below that of the non-dyslexic readers, the percentage of individuals above the mean has become greater for the dyslexic group. In contrast to this, the percentage of those one standard deviation below the mean remains higher.

First Reading RE

 Mean RE % above the mean % below 1 st. dev below mean % above 1 st. dev above mean

Pre-course dyslexics 74 15% 40% 10%

Pre-course non- dyslexics 113 37% 8% 8%

Post-course dyslexics 118 40% 30% 20%

The estimated standard deviation is 45. The mean improvement of the dyslexic group is 44. Since reading quickly can reduce comprehension, while reading more slowly can, all things remaining equal, improve comprehension scores, these RE scores are more significant than just the speed or comprehension scores alone, since they measure reading comprehension against time. As we can see, the post intervention dyslexic mean score exceeds the non-dyslexic mean score, while the percentage above the mean and above one standard deviation above the mean are both superior to the non-dyslexic group scores. The only score which remains below the non-dyslexic score is the percentage one standard deviation below the mean. This is indicative of a minority of dyslexic readers who are still below the reading effectiveness of the non-dyslexic readers. Nevertheless, that 30% of the dyslexic participants all individually improved their RE scores by from 9 to 39 points, achieving a mean improvement of 19.

Review Reading Speed

 Mean wpm % above the mean % below 1 st. dev below mean % above 1 st. dev above mean

Pre-course dyslexics 164 10% 30% 5%

Pre-course non- dyslexics 261 22% 3% 19%

Post-course dyslexics 389 90% 10% 30%

The estimated standard deviation is 119 wpm. The mean improvement of the dyslexic group is 225 wpm (more than doubling their mean reading speed while also, as we can see later, improving their comprehension). Here we can see a large improvement over the non-dyslexic group. This includes a mean additional 128 wpm more than the non-dyslexic group. To put this into some perspective, reading for 10 minutes, the non-dyslexic group would have reviewed just under 7.5 pages, while the dyslexic group would have reviewed over 11 pages. 90% of the dyslexic participants are now above the mean and 30% above one standard deviation above the mean. The group have clearly learned how to review texts effectively (see below) at much greater speed than is normally the case for non-dyslexic readers.

Review Reading comprehension

 Mean

comprehension % above the mean % below 1 st. dev below mean % above 1 st. dev above mean

Pre-course dyslexics 73% 35% 55% 5%

Pre-course non- dyslexics 87% 59% 11% 31%

Post-course dyslexics 91% 80% 0% 40%

The estimated standard deviation is 13%. The mean improvement of the dyslexic group is 18%. Again, the dyslexic group have improved significantly beyond the non-dyslexic group scores. The mean score is 4% higher, but more significantly, 80% are now above the mean comprehension score. Previously, 55% were below one standard deviation below the mean, whereas now none are and 40% are above one standard deviation above the mean. They understand more in less time.

Review Reading RE

 Mean RE % above the mean % below 1 st. dev below mean % above 1 st. dev above mean

Pre-course dyslexics 125 5% 45% 0%

Pre-course non- dyslexics 224 30% 8% 19%

Post-course dyslexics 358 65% 0% 45%

The estimated standard deviation is 124. The mean improvement of the dyslexic group is 233 (almost trebling their score). Here, the improved dyslexic scores are far superior to the non-dyslexic group. They have reversed the 45% that were one standard deviation below the mean, to having 45% one standard deviation above the mean. Previously 5% were above the mean, and now 65% are (compared to just 30% of the non-dyslexic group). This is only possible because of the accumulative effect of increasing both speed of reading and comprehension.


Conclusions

Although we are still seeking funds for larger scale research involving control groups, the evidence of an effect is compelling. Perhaps most tellingly, overall across all scores measured, 35% of the dyslexic group converted one or more score from one standard deviation below the mean prior to the intervention, to one standard deviation above the mean after the intervention; more than two estimated standard deviations.

To put all this data in context, if an average book is 120,000 words, prior to the intervention it was taking the dyslexic group almost 25 hours to gain a 73% comprehension/recall level. This compared with the non-dyslexic group taking slightly more than 18 hours for an 87% comprehension/recall level. After the course, the dyslexic group are taking 15 hours for a 91% comprehension/recall level. Controlled research could tell us more about what is causing it, but the evidence for the effect is now significant. Many participants on the course tell us that they wish they had done the course before entering university. Others tell us that text has stabilised so that they can now read music, or tables of numbers that they had great difficulty with before. Four of the dyslexic participants (20%) have read books for the first time in their lives. I personally save a day a week at work through more effective reading (and marking!) . However, it is the implications of the effect that are most far reaching. The Rose Review (2009) of dyslexia has based most of its demanded phonics intervention on the assumption that phonics intervention is a necessary prerequisite for improving literacy. Singleton (2009) argues that there is no research in peer reviewed journals to support ‘reading recovery’ methods with dyslexic learners and (perhaps forgetting that a lack of counter-evidence does not mean an argument is won) goes one unsupported step further, to assume that nothing but phonics would work on the basis of dyslexia being a ‘phonological deficit’. I have argued elsewhere (Cooper 2009b) that dyslexia is not a deficit, but a difference in the way we process information; that we are entitled to learn in ways that suit us, rather than be ‘remediated’ in the phonological deficit image. One of the most common characteristics of being dyslexic is a preference for and a facility in holistic processing of information, along with a difficulty with sequential processing of information. Basing the teaching of reading on phonemes is intrinsically sequential and bottom up rather than holistic and top down. In other words it starts with the meaningless mechanics, rather than with the purpose and meaning. This embryonic research is beginning to build a convincing evidence base that there is an alternative to phonics for dyslexic adults. I am very excited by it and we are now offering SuperReading to dyslexic students entering university. I have never known an intervention about reading (or indeed any other aspect of literacy) to have such a dramatic effect so quickly. That it can do so through group teaching with no differentiation of reading materials so cost effectively is, in my view, even more remarkable.


References

Burton, M et al (2009) Burton, Davey, Lewis, Ritchie & Brooks (2008) Improving Reading: Phonics and Fluency. NRDC.

Cooper, R (2009a), An evaluation of SuperReading with Dyslexic Adults, Journal of Inclusive Learning in FE & HE, September, 2009

Cooper, R (2009b) Dyslexia, in Pollak, D (Ed.) Neurodiversity in Higher Education; Positive responses to specific learning differences, Wiley-Blackwell