Dyslexia: decoding and sequencing difficulties

According to phonological theory, dyslexia, which is a specific impairment in learning to read and write, is characterised by a deficit in the ability to establish correspondence between graphemes (letters) and phonemes (sounds), due to the fact that the phonological representations of these words are insufficiently fine or detailed. ULg researchers have shown the existence of another deficient process: people with dyslexia show difficulties retaining ordered sequences of words in short-term memory. The researchers thus showed that the ability of children in the last year of nursery school to retain short-term verbal information in order was the strongest predictor of their ability to read one year later. The Liège researchers believe that this experimental data further strengthens the value of using the 'phonic method' of learning to read, based on the sequential decoding of letters making up words. Furthermore, by highlighting the importance of sequential processing, they open up one complementary avenue for rehabilitation as a means of treating children with dyslexia.

Language first developed some two million years ago. It is believed that this would have been some kind of sign language. Over the course of genetic evolution, the spoken word emerged, some 150,000 to 200,000 years ago. Written language did not appear until even more recently, with writing dating back only 4,000 years. This is such a recent development that the brain has not had the time to evolve and create specific 'modules' dedicated to the cognitive operations underlying reading and writing. So, as Stanislas Dahaene (1) writes in his publication The Neurons of Reading, the brain has to practice a form of 'recycling', turning to pre-existing cerebral regions used for other functions, the versatility of which the brain uses to redirect them towards the identification of written signs and to connect them with the spoken word.

In this way, reading circuits involving the visual areas and spoken language areas of the brain are created. In 2010, the journal Science published the results of a study conducted by an international team led by Stanislas Dehaene (1), from Inserm (France) and including Régine Kolinsky and José Morais, from the Université Libre de Bruxelles.  This work addressed the cerebral activations in adults with varying degrees of literacy, when confronted with a battery of stimuli including written and spoken phrases, words and pseudo-words, i.e. words with no meaning such as 'tincheluche' or 'mupfteux', objects, faces and checker boards. Based on measurements conducted using functional magnetic resonance imaging (fMRI), their research took them to Brazil and Portugal, countries where education was not yet widespread as recently as a few decades ago. Sixty-three adult volunteers took part: 10 illiterate adults, 22 who became literate as adults, and 31 who were literate in childhood.

What did this study show? By comparing reading scores and corresponding cerebral activity, the researchers showed that the impact of literacy on brain function is much more extensive than previous studies had led to believe, affecting areas of the brain used both for vision and for spoken language. Learning to read thus increased responses from the visual areas of the cortex, not only in the region specialising in the recognition of letters - the brain's 'letterbox' - but also in the primary visual area, where all visual information ends up. Similarly, reading increased responses to spoken language in the auditory cortex involved in coding phonemes. 'These results may correspond to the fact that illiterate people are unable to conduct language games such as deleting the first sound of a word' indicate the study authors. In other words, they can't extract the sound sequence 'aris' from 'Paris' for example.

No critical period

However, the study conducted under Stanislas Dahaene showed that the ability to read led to an extension of the areas of language and bidirectional communication between spoken and written language networks. Comparison between good readers and illiterate adults is illustrative of this: in the former, the vision of a written phrase activates the whole area of spoken language, but in addition, hearing a word re-activates its orthographic code in the visual areas. In illiterate adults, on the other hand, language processing is limited to the auditory mechanism alone.

Another interesting result is that the visual area of the left hemisphere which readers use to decode written words is activated in illiterate people when recognising objects and faces. Readers also do this, but the intensity with which the area is activated decreases slightly as their reading ability increases. Thus, activations associated with the recognition of objects and faces are observed as partially moving to the right hemisphere. To summarise, readers' visual corteces reorganise themselves to respond to the new activity of reading, which seems to create competition with the more ancient activities of identifying objects and faces. Does this damage the ability to recognise these two entities? In particular, are people who know how to read less physiognomist than those who don't? We don't currently know.

One questions is particularly pressing: to what extent do the results obtained differ according to whether the ability to read is acquired during childhood or later, through adult literacy classes?  According to the authors of the article published in Science, there is no critical period for learning to read, and the effect of learning upon cortical functioning is almost identical for those who were educated in childhood and those who learned to read later in life. The authors conclude from this, happily, that the circuits involved in reading remain flexible throughout life.

A decoding exercise

Reading is certainly a complex cognitive operation, during which information is transferred from the visual module, enabling the visual perception of words, to the module dedicated to the spoken language. Thus, as Martine Poncelet, researcher in the Psychology, Cognition and Behaviour Department of the University of Liège (ULg) points out, reading may be conducted either through a process of converting graphemes into phonemes, called the 'decoding procedure', or through an overarching process called the 'direct access  procedure'.

Young children who do not know how to read or write recognise certain symbols, which point them towards identifying product brands, for example, or certain simple words which they perceive around them using their visual characteristics. Thus, in French the word 'moto' (meaning 'motorbike') may, because of the two 'o's which appear in it, suggest the idea of two wheels. However, when children start to learn to read, they use the decoding procedure. They undertake a decoding exercise which enables them to establish a connection between the graphemes and corresponding phonemes. This same technique is used by illiterate adults when confronted with new words, some proper nouns or invented words used by psychologists for experimental purposes (pseudo-words). 'According to current thinking about learning written language, the ability to apply the strategy of converting graphemes into phonemes is a necessary condition for good reading', says Poncelet. The task of decoding is indispensable. '

Apart from examples where a word like 'moto' is identified because of its visual particularities, the direct access process only takes place at a subsequent stage. Clearly, it can only achieve be fully utilised by building upon knowledge previously acquired using the decoding process. When a word which has already been decoded is encountered repeatedly within a relatively short period of time, it becomes integrated in the reader's memory, in such a way that its recognition usually becomes instant when they see it. For experienced readers, reading is therefore based upon the direct access pro procedure; the decoding process is only, in theory, used to decipher unknown or forgotten words. 'A good reader is someone who automatically, quickly and faultlessly identifies written words and who can therefore focus on the meaning of what he or she is reading', stresses Poncelet.

Fluid, expert reading is built upon clearly identified cognitive abilities, in particular relating to processing the sound of speech, but also, as we will see, in processing the order of phonemes (sequentiality). Although this is a necessary condition, however, it is not sufficient. Because, although these capacities are deficient in dyslexic children, they are not lacking in those children who are labelled 'poor readers'. 'These children show no such fundamental difficulties.' explains Poncelet. 'Their representations of the sounds of speech are therefore normal. The problem lies elsewhere. Either they haven't benefited from good learning methods or the necessary support, or they suffer from concentration difficulties or lack of motivation. '

Phonological conscience

Poncelet goes on to specify that establishing the distinction, in adulthood, between dyslexics and poor readers is sometimes difficult. Why? Because a good experience of reading has an influence on the representation of sounds. It isn't unusual for people to say: 'When I hear a word, I see it in its written form' or 'When I hear a word I don't know, I try to imagine it written down'. While phonological representations influence spelling, spelling also influences phonological representations. 'Eventually, the level of understanding the sounds of language is not the same in someone who has learned to read and someone who has not, despite having the required prerequisites', says Poncelet. By adulthood, it is difficult to make a differential diagnosis and a battery of appropriate tests is required, all pointing to the question of evaluating the degree of deficiency in decoding graphemes. The problem is even more taxing when the forest can't be seen for the trees: a dyslexic who has reached university level, having benefited from appropriate support during childhood, often reads better than a poor reader.

There are several theories of dyslexia, but the most solid by far is that of phonological theory. Focussing respectively upon hearing, sight and motricity, other approaches have received less empirical support.

According to phonological theory, the deficiency in the decoding process, i.e. the ability to establish a correspondence between graphemes (letters) and phonemes (sounds) which characterises dyslexia, is directly due to the fact that these representations of the sounds of speech (or phonological representations) are insufficiently fine or detailed. It follows that the person also suffers from poor 'phonological awareness: he or she is unable, or manages much more slowly than a normal child, to divide words up into their different parts, segment them into syllables and, further, into phonemes. These children find it difficult to see that a word such as 'cahier' (French for 'exercise book') consists of two syllables (ca-hier), and particularly that it consists of four phonemes (k-a-i-é).

A deficiency in short-term phonological memory, the system responsible for temporarily storing verbal information, has also frequently been highlighted in dyslexics. The proof is that they appear to experience much greater difficulty than other children in repeating pseudo-words, words with a fairly high number of syllables, or sequences of numbers of a certain length. 'The results obtained from this type of task prior to entry into primary school are predictive of the child's reading level a few years later', notes Poncelet. Systematic application of such tasks would facilitate testing for potential dyslexia.'

The right order

In an article published in November 2010 in the Journal of Cognitive Neuroscience, Steve Majerus, an F.R.S.-FNRS senior scientist working in the Psychology, Cognition and Behaviour Department of the ULg, shows that, in contrast to widely received wisdom, short-term verbal memory is not a specific entity with its own existence (read the article Short term memory revisited). Rather, it can be compared to a verbal information retention function which emanates from the interaction of three more general systems: the language system (in particular long-term verbal memory), attentional control and the system of 'serial order' processing (sequentiality).

Recently, Trecy Martinez Perez, Steve Majerus, Aline Mahot and Martine Poncelet applied this three-pronged strategy to understanding short-term memory impairment in dyslexics. Their work was the subject of two articles, one published in the journal Dyslexia, entitled Evidence for a Specific Impairment of Serial Order Short-term Memory in Dyslexic Children (2), and another published in the journal Research in Developmental Disabilities, entitled Impaired Short-Term Memory for order in adults with dyslexia (3). 'Short-term verbal memory impairments may have several origins', says Steve Majerus. It has been established that one of these, in dyslexics, is the difficulty of creating fine, individualised, well-segmented phonological representations. However, while learning to read requires good phonological awareness, it also requires the ability to connect phonemes in the right order. Because the number of sounds is limited in each language, this order constitutes one of the key factors enabling the distinction between words of the same length and consisting of the same phonemic units to be made.'

The question is therefore as follows: do dyslexics struggle when they have to store the sequence of words in their memory, i.e. the order of phonemes? The answer is yes. In the course of their work, the ULg neuropsychologists have shown not only that retaining 'sequentially ordered' information (the sequence of phonemes) was impaired in dyslexic children and adults, in the same way as phonological knowledge, but, furthermore, this deficiency was generally associated with deeper dyslexic difficulties. 'Moreover, we discovered that in children in the last year of nursery school, their ability to memorise the order of verbal information was the biggest predictor of their reading level one year later', indicates Majerus.

Similarly, their vocabulary level aged six or seven is strongly correlated to their ability to recall 'serially ordered' information. 'According to our predictions, processing sequential information should also prove to be very important for memorising how to spell words, particularly when these have an unusual structure', he continues. Moreover, studies conducted a few years ago at the University of Brighton showed that in adults, the acquisition of vocabulary in a foreign language is better the greater the ability to process 'serially ordered' information. Finally, other work has shown the importance of these abilities in learning numerical calculations in childhood.

Basic capacity

Images obtained during fMRI show that short-term retention of 'serially ordered' information depends upon a specific cerebral circuit: the anterior part of the right parietal lobe. Using this imaging technique, the ULg team focussed on the cerebral processing of sequences of words and sequences of photographs of faces projected onto a screen to dyslexic adults and control subjects. Four words and four faces were initially presented in a specific order. Combinations of two pairs of words or two faces were then projected onto the screen. The participant had to determine, for each pair, which element had previously been presented first. On the one hand, the dyslexic participants made a greater number of errors, both in terms of faces and the order of words. On the other hand, activation of the anterior part of the right parietal lobe in these same subjects was slightly lower. 'These results indicate that the treatment of 'sequentially ordered' information is a basic ability which operates in learning all new codes which are organised according to a certain sequence', states Majerus.

In his opinion, this experimental data further strengthens the value of using the 'phonic method' of learning to read, which is based on the sequential decoding of letters making up words. Moreover, by highlighting the importance of processing sequence information, they open up a complementary avenue for rehabilitation to treat children with dyslexia.

(1) Professor at the Collège de France, Stanislas Dehaene is head of the Inserm-CEA unit for Cognitive Neuroimaging at Saclay.

(2) Trecy Martinez Perez, Steve Majerus, Aline Mahot et Martine Poncelet (2012). Evidence for a Specific Impairment of Serial Order Short-term Memory in Dyslexic Children, Dyslexia, 18, 94-109
(3) Trecy Martinez Perez, Steve Majerus, and Martine Poncelet. Impaired Short-Term Memory for order in adults with dyslexia, Research in Developmental Disabilities (in press)