One of the central tasks in learning to read is the accurate detection of the complex pattern of associations between graphemes and the phonemes (letters, groups of letters, and sounds).

In English, the relationships between graphemes and phonemes is particularly complex, for example, consider the role of;

the letter ‘o’ in on, once, only, woman, women, worry;
or the ee sound in leap, people, here, weird, chief, police, me, ski, key;
or the oo sound in rude, shrewd, truth, group, move, fruit, tomb, through, blue, shoe.

(Visit www.englishspellingproblems.co.uk or read the book by Marsha Bell, 'English Spelling problems', for a thorough analysis of the problem.)

In general, our neural networks are amazingly good at detecting closely coupled associations in time and space – IF - the relevant neural networks receive consistent, high quality inputs, but our performance falls of rapidly if the sensory inputs are inconsistent, or worse still, internally contradictory.

It is an unfortunate aspect of the English language that many of the words we encounter in early childhood are atypical - unlike the rest of the language. So inconsistency is built into the language from day one. The destabilising effects of this inconsistency can be reduced if care is taken over the sequence of exposure, and if detailed explanations are given that help the developing child distinguish between different contexts (Germanic, French, Latin, Scandinavian, Greek, etc.) and between regular and irregular grapheme/phoneme associations. Over-simplification is counter productive because it lends adult approval to language rules and categories which the child can clearly see are irresolvabley internally inconsistent.

Artificial neural networks can be programmed to recover from exposure to early errors of this sort. A hundred good example can rectify damage done to a neural network's model of reality by early exposure to atypical examples – but human neural networks are not so good at this. For us, early exposure to bad (atypical) examples can cause damage from which it is very hard to recover – particularly if strong negative emotions come to be associated with particular stimuli.

I want to draw your attention to three issues that may be contributing to the current difficulties many children seem to be having learning to read in England.

1 disrupted visual inputs

2 disrupted aural inputs

3 a wide variety of accents, grammatical structures, meanings.

The human brain is not evolved for reading. Reading works by riding on the back of mental abilities that evolved for other purposes. The visual system was not evolved for reading letters. The speed with which humans can see/read letters is subject to a same sort of normal distribution curve as any other human ability. Some people have very fast visual systems, some relatively slow, and most will be somewhere in the middle.

Take the Visual Persistence test at www.gts-training.co.uk to see how fast your visual system is. Invite other people you know to take the test and see how you compare.

Early results show that some people can distinguish letters quite comfortable at 70 ms and less whilst others are struggling at 300 ms. Most people find that they experience optical illusions if they try to read faster than their visual system can comfortable cope with. The mechanism underlying these optical illusions (visual persistence - masking) is analogous to a buffer overflow and can result in some letters getting lost entirely, and strings of letters being perceived in a different order from the sequence that they actually appear on the page. Take the test and try this for yourself.

So for people with a long visual persistence the visual input to the grapheme phoneme detection process is disrupted at normal reading speeds – letters missing – letter changing their sequence – letters changing their dominance. It is easy to see how this could seriously interfere with the detection of grapheme phoneme relationships.

Similar problem seem to be occurring in the aural system as well. Recent research has discovered that we don't all hear the sound patters in speech in the same way. We don't yet have an on-line system that enables you to try this for yourselves but you can read research at.

Richardson, U., Thomson, J., Scott, S.K., & Goswami, U. (2004). Auditory Processing Skills and Phonological Representation in Dyslexic Children. Dyslexia, 10, 215-233.

http://www.educ.cam.ac.uk/people/staff/goswami/

It turns out that people who are considered to be insensitive to phoneme patterns in speech have a different perception of the beats and rhythms in human speech from people who are considered to be good at detecting phonemes in human speech. (This insensitivity to phoneme patterns is considered to be a good predictor of reading ability.)

So - humans naturally vary in their visual ability to process text and in their aural ability to detect some particular, subtle, patterns in speech. These differences may interact to produce a range of difficulties learning to read and a wide range of natural reading abilities in terms of speed, endurance and accuracy.

Thirty years ago a child growing up in the UK would have been exposed to just two dominant accents (phoneme/grapheme and grammatical systems). A local accent and the received pronunciation from the BBC and the local Vicar. Today's children are exposed to a huge variety of different accents and NO dominant system. So how on earth are they supposed to build up an understanding of grapheme/phoneme relationships?

Some people happen to be very fast and comfortable readers. Many are not. Some really struggle. Environmental factors may be making the problem worse (multiple accents – over relaxed and inconsistent trendy writing styles in the popular media, lack of practise, inappropriate over-simplistic teaching systems, etc.)

I am one of the people who struggle with the visual element of reading. My visual persistence is around 280 ms, which is useless for reading but very good for motor bike racing and detecting subtle patterns in a changing environment. My difficulty with text had the consequence that almost ever school lesson and university lecture was ruined, for me, by the teacher presenting the crucial information and concluding ideas in the form of a swirling mess of text.

I now work with dyslexic prisoners (did you know that our prisons are full of people excluded from society because of their problems with text) and excluded school age children. The introduction of the literacy hour (UK) and the general obsession with integrating text into every lesson has had a spectacularly counter productive effect for many kids. When I was at school I only had to skive-off on Tuesdays and Thursdays to avoid the spelling tests and reading-around the class. If I was at school now – I would be a persistent truant.

So. We should accept that there is a wide range of reading abilities in the general population, and that reading ability is not necessarily an indicator of other abilities and talents. Indeed it may even be an indicator of a lack of some specific and important abilities and talents. Because the current education system (run by a self-selecting group of particularly fast readers) considers text to be a particularly important measure of something or other, people with reading difficulties unfortunately get a very bad education and thus emerge with few qualifications, but this should not automatically be mistaken for a lack of intelligence, ability or talent.

So – please take part in our Visual Persistence Survey (above) as this will give us the data we need to get something done about the situation.

Thank you.

John Evans