Acoustics'08 Paris

[ Lay Language Paper Index | Press Room ]

Can you tell a ‘pea’ from a ‘bee’ if you are a dyslexic child? 

Souhila Messaoud-Galusi, UCL, Department of Speech, Hearing and Phonetic Sciences,
Valerie Hazan, UCL, Department of Speech, Hearing and Phonetic Sciences,
Stuart Rosen, UCL, Department of Speech, Hearing and Phonetic Sciences,

Popular version of paper 2pSCa3
"Is there evidence of an allophonic mode of speech perception in dyslexic children"
Presented at 2:40 p.m. on Tuesday July 1, 2008 in Room 240.

Key summary

Children with dyslexia have problems in learning to read despite having normal intelligence and good educational opportunities. Many researchers have suggested that part of the reason why they do not easily learn the link between speech sounds and written letters is that they have problems in hearing the small differences between similar speech sounds. However, other researchers have argued that only some children may have problems with speech sounds so this cannot be the cause of dyslexia.

Most studies have been limited by the fact that they only tested small numbers of dyslexic children. We tested a large and well-controlled group of 62 children with dyslexia and 51 average-readers. The tests were wide-ranging and investigated how the children heard speech sounds in both quiet and noisy environments. We found that only a small proportion of the dyslexic children consistently found the speech tests to be difficult, while some average-readers also experienced difficulty with the tests. We therefore conclude that difficulties in hearing speech are not the root cause of dyslexia. Also, performing badly in these speech tests is not necessarily due to having some real difficulty in hearing speech sounds. Indeed, the pattern of errors observed in dyslexic children is consistent with them having poorer attention than average-readers, especially during repetitive tasks.

Further details

62 children with a formal diagnosis of dyslexia and 51 children of average reading ability aged between six and fourteen participated in the study. A range of speech tests was presented. First, using highly-controlled synthesised speech, we looked at whether they could hear small differences in the acoustic patterns of the words PEA and BEE in two different ways [LISTEN]. In an identification task, the children heard a single word and had to click on whether they thought it was ‘pea’ or ‘bee’. In the discrimination task, they heard three words and had to decide which was the odd one out. They did these tasks both in quiet and with voices in the background. In these ‘adaptive’ PEA-BEE tests, the difference between the two sounds got smaller and smaller as the test went on. According to previous studies, we expected that dyslexic children, as a group, would find these tasks harder than average-readers.

In another set of tasks, the children had to recognise easy ‘real’ words presented with varying amounts of noise in the background [LISTEN]. We expected that if dyslexic children have poorer speech perception, they should make more errors than average readers when attempting to recognise words in noise.

Our analyses suggest that there were some differences in performance between the two groups of children in the ‘PEA-BEE’ tests, but that the dyslexics and average-readers performed the same on the two tests that involved hearing natural words in noise. Also, when we looked in detail at the performance of many dyslexic children over the course of the PEA-BEE identification tests, we found that they made no errors for the ‘easy’ ‘PEA’ and ‘BEE’ sounds - which were markedly different from each other - at the beginning of the test, but that they made more errors when these ‘easy’ PEA and BEE sounds were slotted in at regular intervals as the test went along. This suggests that their greater difficulty with these tests may be due to increased ‘fatigue’ or inattention, and that dyslexic children may have difficulty in maintaining their attention when the test is repetitive.

Overall, when we examined how individual children performed across the range of tests, we found that under 20% of the dyslexic children were consistently poor at three or more speech perception tasks, while over 30% performed within the range expected for average-readers in all speech perception tasks. By the same criteria, 10% of children in the control group also performed poorly in perception tests despite having no problems with reading. These results do not support the notion that problems in discriminating speech sounds are the root cause of dyslexia.


[ Lay Language Paper Index | Press Room ]